cpython/Objects/dictobject.c

2018 lines
50 KiB
C
Raw Normal View History

/* Dictionary object implementation using a hash table */
1997-05-02 00:12:38 -03:00
#include "Python.h"
2001-08-02 01:15:00 -03:00
typedef PyDictEntry dictentry;
typedef PyDictObject dictobject;
/* Define this out if you don't want conversion statistics on exit. */
#undef SHOW_CONVERSION_COUNTS
/* See large comment block below. This must be >= 1. */
#define PERTURB_SHIFT 5
/*
Major subtleties ahead: Most hash schemes depend on having a "good" hash
function, in the sense of simulating randomness. Python doesn't: its most
important hash functions (for strings and ints) are very regular in common
cases:
>>> map(hash, (0, 1, 2, 3))
[0, 1, 2, 3]
>>> map(hash, ("namea", "nameb", "namec", "named"))
[-1658398457, -1658398460, -1658398459, -1658398462]
>>>
This isn't necessarily bad! To the contrary, in a table of size 2**i, taking
the low-order i bits as the initial table index is extremely fast, and there
are no collisions at all for dicts indexed by a contiguous range of ints.
The same is approximately true when keys are "consecutive" strings. So this
gives better-than-random behavior in common cases, and that's very desirable.
OTOH, when collisions occur, the tendency to fill contiguous slices of the
hash table makes a good collision resolution strategy crucial. Taking only
the last i bits of the hash code is also vulnerable: for example, consider
[i << 16 for i in range(20000)] as a set of keys. Since ints are their own
hash codes, and this fits in a dict of size 2**15, the last 15 bits of every
hash code are all 0: they *all* map to the same table index.
But catering to unusual cases should not slow the usual ones, so we just take
the last i bits anyway. It's up to collision resolution to do the rest. If
we *usually* find the key we're looking for on the first try (and, it turns
out, we usually do -- the table load factor is kept under 2/3, so the odds
are solidly in our favor), then it makes best sense to keep the initial index
computation dirt cheap.
The first half of collision resolution is to visit table indices via this
recurrence:
j = ((5*j) + 1) mod 2**i
For any initial j in range(2**i), repeating that 2**i times generates each
int in range(2**i) exactly once (see any text on random-number generation for
proof). By itself, this doesn't help much: like linear probing (setting
j += 1, or j -= 1, on each loop trip), it scans the table entries in a fixed
order. This would be bad, except that's not the only thing we do, and it's
actually *good* in the common cases where hash keys are consecutive. In an
example that's really too small to make this entirely clear, for a table of
size 2**3 the order of indices is:
0 -> 1 -> 6 -> 7 -> 4 -> 5 -> 2 -> 3 -> 0 [and here it's repeating]
If two things come in at index 5, the first place we look after is index 2,
not 6, so if another comes in at index 6 the collision at 5 didn't hurt it.
Linear probing is deadly in this case because there the fixed probe order
is the *same* as the order consecutive keys are likely to arrive. But it's
extremely unlikely hash codes will follow a 5*j+1 recurrence by accident,
and certain that consecutive hash codes do not.
The other half of the strategy is to get the other bits of the hash code
into play. This is done by initializing a (unsigned) vrbl "perturb" to the
full hash code, and changing the recurrence to:
j = (5*j) + 1 + perturb;
perturb >>= PERTURB_SHIFT;
use j % 2**i as the next table index;
Now the probe sequence depends (eventually) on every bit in the hash code,
and the pseudo-scrambling property of recurring on 5*j+1 is more valuable,
because it quickly magnifies small differences in the bits that didn't affect
the initial index. Note that because perturb is unsigned, if the recurrence
is executed often enough perturb eventually becomes and remains 0. At that
point (very rarely reached) the recurrence is on (just) 5*j+1 again, and
that's certain to find an empty slot eventually (since it generates every int
in range(2**i), and we make sure there's always at least one empty slot).
Selecting a good value for PERTURB_SHIFT is a balancing act. You want it
small so that the high bits of the hash code continue to affect the probe
sequence across iterations; but you want it large so that in really bad cases
the high-order hash bits have an effect on early iterations. 5 was "the
best" in minimizing total collisions across experiments Tim Peters ran (on
both normal and pathological cases), but 4 and 6 weren't significantly worse.
Historical: Reimer Behrends contributed the idea of using a polynomial-based
approach, using repeated multiplication by x in GF(2**n) where an irreducible
polynomial for each table size was chosen such that x was a primitive root.
Christian Tismer later extended that to use division by x instead, as an
efficient way to get the high bits of the hash code into play. This scheme
also gave excellent collision statistics, but was more expensive: two
if-tests were required inside the loop; computing "the next" index took about
the same number of operations but without as much potential parallelism
(e.g., computing 5*j can go on at the same time as computing 1+perturb in the
above, and then shifting perturb can be done while the table index is being
masked); and the dictobject struct required a member to hold the table's
polynomial. In Tim's experiments the current scheme ran faster, produced
equally good collision statistics, needed less code & used less memory.
*/
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
/* Object used as dummy key to fill deleted entries */
static PyObject *dummy; /* Initialized by first call to newdictobject() */
/* forward declarations */
static dictentry *
lookdict_string(dictobject *mp, PyObject *key, long hash);
#ifdef SHOW_CONVERSION_COUNTS
static long created = 0L;
static long converted = 0L;
static void
show_counts(void)
{
fprintf(stderr, "created %ld string dicts\n", created);
fprintf(stderr, "converted %ld to normal dicts\n", converted);
fprintf(stderr, "%.2f%% conversion rate\n", (100.0*converted)/created);
}
#endif
2001-08-02 01:15:00 -03:00
/* Initialization macros.
There are two ways to create a dict: PyDict_New() is the main C API
function, and the tp_new slot maps to dict_new(). In the latter case we
can save a little time over what PyDict_New does because it's guaranteed
that the PyDictObject struct is already zeroed out.
Everyone except dict_new() should use EMPTY_TO_MINSIZE (unless they have
an excellent reason not to).
*/
#define INIT_NONZERO_DICT_SLOTS(mp) do { \
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
(mp)->ma_table = (mp)->ma_smalltable; \
2001-08-02 01:15:00 -03:00
(mp)->ma_mask = PyDict_MINSIZE - 1; \
} while(0)
#define EMPTY_TO_MINSIZE(mp) do { \
memset((mp)->ma_smalltable, 0, sizeof((mp)->ma_smalltable)); \
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
(mp)->ma_used = (mp)->ma_fill = 0; \
2001-08-02 01:15:00 -03:00
INIT_NONZERO_DICT_SLOTS(mp); \
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
} while(0)
1997-05-02 00:12:38 -03:00
PyObject *
PyDict_New(void)
{
register dictobject *mp;
if (dummy == NULL) { /* Auto-initialize dummy */
1997-05-02 00:12:38 -03:00
dummy = PyString_FromString("<dummy key>");
if (dummy == NULL)
return NULL;
#ifdef SHOW_CONVERSION_COUNTS
Py_AtExit(show_counts);
#endif
}
2001-08-29 20:54:21 -03:00
mp = PyObject_GC_New(dictobject, &PyDict_Type);
if (mp == NULL)
return NULL;
2001-08-02 01:15:00 -03:00
EMPTY_TO_MINSIZE(mp);
mp->ma_lookup = lookdict_string;
#ifdef SHOW_CONVERSION_COUNTS
++created;
#endif
2001-08-29 20:54:21 -03:00
_PyObject_GC_TRACK(mp);
1997-05-02 00:12:38 -03:00
return (PyObject *)mp;
}
/*
The basic lookup function used by all operations.
This is based on Algorithm D from Knuth Vol. 3, Sec. 6.4.
Open addressing is preferred over chaining since the link overhead for
chaining would be substantial (100% with typical malloc overhead).
The initial probe index is computed as hash mod the table size. Subsequent
probe indices are computed as explained earlier.
All arithmetic on hash should ignore overflow.
(The details in this version are due to Tim Peters, building on many past
contributions by Reimer Behrends, Jyrki Alakuijala, Vladimir Marangozov and
Christian Tismer).
This function must never return NULL; failures are indicated by returning
a dictentry* for which the me_value field is NULL. Exceptions are never
reported by this function, and outstanding exceptions are maintained.
*/
static dictentry *
lookdict(dictobject *mp, PyObject *key, register long hash)
{
register int i;
register unsigned int perturb;
register dictentry *freeslot;
register unsigned int mask = mp->ma_mask;
dictentry *ep0 = mp->ma_table;
register dictentry *ep;
register int restore_error;
register int checked_error;
register int cmp;
PyObject *err_type, *err_value, *err_tb;
PyObject *startkey;
Get rid of the superstitious "~" in dict hashing's "i = (~hash) & mask". The comment following used to say: /* We use ~hash instead of hash, as degenerate hash functions, such as for ints <sigh>, can have lots of leading zeros. It's not really a performance risk, but better safe than sorry. 12-Dec-00 tim: so ~hash produces lots of leading ones instead -- what's the gain? */ That is, there was never a good reason for doing it. And to the contrary, as explained on Python-Dev last December, it tended to make the *sum* (i + incr) & mask (which is the first table index examined in case of collison) the same "too often" across distinct hashes. Changing to the simpler "i = hash & mask" reduced the number of string-dict collisions (== # number of times we go around the lookup for-loop) from about 6 million to 5 million during a full run of the test suite (these are approximate because the test suite does some random stuff from run to run). The number of collisions in non-string dicts also decreased, but not as dramatically. Note that this may, for a given dict, change the order (wrt previous releases) of entries exposed by .keys(), .values() and .items(). A number of std tests suffered bogus failures as a result. For dicts keyed by small ints, or (less so) by characters, the order is much more likely to be in increasing order of key now; e.g., >>> d = {} >>> for i in range(10): ... d[i] = i ... >>> d {0: 0, 1: 1, 2: 2, 3: 3, 4: 4, 5: 5, 6: 6, 7: 7, 8: 8, 9: 9} >>> Unfortunately. people may latch on to that in small examples and draw a bogus conclusion. test_support.py Moved test_extcall's sortdict() into test_support, made it stronger, and imported sortdict into other std tests that needed it. test_unicode.py Excluced cp875 from the "roundtrip over range(128)" test, because cp875 doesn't have a well-defined inverse for unicode("?", "cp875"). See Python-Dev for excruciating details. Cookie.py Chaged various output functions to sort dicts before building strings from them. test_extcall Fiddled the expected-result file. This remains sensitive to native dict ordering, because, e.g., if there are multiple errors in a keyword-arg dict (and test_extcall sets up many cases like that), the specific error Python complains about first depends on native dict ordering.
2001-05-12 21:19:31 -03:00
i = hash & mask;
ep = &ep0[i];
if (ep->me_key == NULL || ep->me_key == key)
return ep;
restore_error = checked_error = 0;
if (ep->me_key == dummy)
freeslot = ep;
else {
if (ep->me_hash == hash) {
/* error can't have been checked yet */
checked_error = 1;
if (PyErr_Occurred()) {
restore_error = 1;
PyErr_Fetch(&err_type, &err_value, &err_tb);
}
startkey = ep->me_key;
cmp = PyObject_RichCompareBool(startkey, key, Py_EQ);
if (cmp < 0)
PyErr_Clear();
if (ep0 == mp->ma_table && ep->me_key == startkey) {
if (cmp > 0)
goto Done;
}
else {
/* The compare did major nasty stuff to the
* dict: start over.
* XXX A clever adversary could prevent this
* XXX from terminating.
*/
ep = lookdict(mp, key, hash);
goto Done;
}
}
freeslot = NULL;
1997-01-18 03:55:05 -04:00
}
/* In the loop, me_key == dummy is by far (factor of 100s) the
least likely outcome, so test for that last. */
for (perturb = hash; ; perturb >>= PERTURB_SHIFT) {
i = (i << 2) + i + perturb + 1;
ep = &ep0[i & mask];
if (ep->me_key == NULL) {
if (freeslot != NULL)
ep = freeslot;
break;
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
}
if (ep->me_key == key)
break;
if (ep->me_hash == hash && ep->me_key != dummy) {
if (!checked_error) {
checked_error = 1;
if (PyErr_Occurred()) {
restore_error = 1;
PyErr_Fetch(&err_type, &err_value,
&err_tb);
}
}
startkey = ep->me_key;
cmp = PyObject_RichCompareBool(startkey, key, Py_EQ);
if (cmp < 0)
PyErr_Clear();
if (ep0 == mp->ma_table && ep->me_key == startkey) {
if (cmp > 0)
break;
}
else {
/* The compare did major nasty stuff to the
* dict: start over.
* XXX A clever adversary could prevent this
* XXX from terminating.
*/
ep = lookdict(mp, key, hash);
break;
}
}
else if (ep->me_key == dummy && freeslot == NULL)
freeslot = ep;
}
Done:
if (restore_error)
PyErr_Restore(err_type, err_value, err_tb);
return ep;
}
/*
* Hacked up version of lookdict which can assume keys are always strings;
* this assumption allows testing for errors during PyObject_Compare() to
* be dropped; string-string comparisons never raise exceptions. This also
* means we don't need to go through PyObject_Compare(); we can always use
* _PyString_Eq directly.
*
* This is valuable because the general-case error handling in lookdict() is
* expensive, and dicts with pure-string keys are very common.
*/
static dictentry *
lookdict_string(dictobject *mp, PyObject *key, register long hash)
{
register int i;
register unsigned int perturb;
register dictentry *freeslot;
register unsigned int mask = mp->ma_mask;
dictentry *ep0 = mp->ma_table;
register dictentry *ep;
/* Make sure this function doesn't have to handle non-string keys,
including subclasses of str; e.g., one reason to subclass
strings is to override __eq__, and for speed we don't cater to
that here. */
if (!PyString_CheckExact(key)) {
#ifdef SHOW_CONVERSION_COUNTS
++converted;
#endif
mp->ma_lookup = lookdict;
return lookdict(mp, key, hash);
}
Get rid of the superstitious "~" in dict hashing's "i = (~hash) & mask". The comment following used to say: /* We use ~hash instead of hash, as degenerate hash functions, such as for ints <sigh>, can have lots of leading zeros. It's not really a performance risk, but better safe than sorry. 12-Dec-00 tim: so ~hash produces lots of leading ones instead -- what's the gain? */ That is, there was never a good reason for doing it. And to the contrary, as explained on Python-Dev last December, it tended to make the *sum* (i + incr) & mask (which is the first table index examined in case of collison) the same "too often" across distinct hashes. Changing to the simpler "i = hash & mask" reduced the number of string-dict collisions (== # number of times we go around the lookup for-loop) from about 6 million to 5 million during a full run of the test suite (these are approximate because the test suite does some random stuff from run to run). The number of collisions in non-string dicts also decreased, but not as dramatically. Note that this may, for a given dict, change the order (wrt previous releases) of entries exposed by .keys(), .values() and .items(). A number of std tests suffered bogus failures as a result. For dicts keyed by small ints, or (less so) by characters, the order is much more likely to be in increasing order of key now; e.g., >>> d = {} >>> for i in range(10): ... d[i] = i ... >>> d {0: 0, 1: 1, 2: 2, 3: 3, 4: 4, 5: 5, 6: 6, 7: 7, 8: 8, 9: 9} >>> Unfortunately. people may latch on to that in small examples and draw a bogus conclusion. test_support.py Moved test_extcall's sortdict() into test_support, made it stronger, and imported sortdict into other std tests that needed it. test_unicode.py Excluced cp875 from the "roundtrip over range(128)" test, because cp875 doesn't have a well-defined inverse for unicode("?", "cp875"). See Python-Dev for excruciating details. Cookie.py Chaged various output functions to sort dicts before building strings from them. test_extcall Fiddled the expected-result file. This remains sensitive to native dict ordering, because, e.g., if there are multiple errors in a keyword-arg dict (and test_extcall sets up many cases like that), the specific error Python complains about first depends on native dict ordering.
2001-05-12 21:19:31 -03:00
i = hash & mask;
ep = &ep0[i];
if (ep->me_key == NULL || ep->me_key == key)
return ep;
if (ep->me_key == dummy)
freeslot = ep;
else {
if (ep->me_hash == hash
&& _PyString_Eq(ep->me_key, key)) {
return ep;
}
freeslot = NULL;
}
/* In the loop, me_key == dummy is by far (factor of 100s) the
least likely outcome, so test for that last. */
for (perturb = hash; ; perturb >>= PERTURB_SHIFT) {
i = (i << 2) + i + perturb + 1;
ep = &ep0[i & mask];
if (ep->me_key == NULL)
return freeslot == NULL ? ep : freeslot;
if (ep->me_key == key
|| (ep->me_hash == hash
&& ep->me_key != dummy
&& _PyString_Eq(ep->me_key, key)))
return ep;
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
if (ep->me_key == dummy && freeslot == NULL)
freeslot = ep;
}
}
/*
Internal routine to insert a new item into the table.
Used both by the internal resize routine and by the public insert routine.
Eats a reference to key and one to value.
*/
static void
insertdict(register dictobject *mp, PyObject *key, long hash, PyObject *value)
{
1997-05-02 00:12:38 -03:00
PyObject *old_value;
register dictentry *ep;
2001-08-02 01:15:00 -03:00
typedef PyDictEntry *(*lookupfunc)(PyDictObject *, PyObject *, long);
assert(mp->ma_lookup != NULL);
ep = mp->ma_lookup(mp, key, hash);
if (ep->me_value != NULL) {
old_value = ep->me_value;
ep->me_value = value;
1997-05-02 00:12:38 -03:00
Py_DECREF(old_value); /* which **CAN** re-enter */
Py_DECREF(key);
}
else {
if (ep->me_key == NULL)
mp->ma_fill++;
else
1997-05-02 00:12:38 -03:00
Py_DECREF(ep->me_key);
ep->me_key = key;
ep->me_hash = hash;
ep->me_value = value;
mp->ma_used++;
}
}
/*
Restructure the table by allocating a new table and reinserting all
items again. When entries have been deleted, the new table may
actually be smaller than the old one.
*/
static int
dictresize(dictobject *mp, int minused)
{
int newsize;
dictentry *oldtable, *newtable, *ep;
int i;
int is_oldtable_malloced;
2001-08-02 01:15:00 -03:00
dictentry small_copy[PyDict_MINSIZE];
assert(minused >= 0);
/* Find the smallest table size > minused. */
2001-08-02 01:15:00 -03:00
for (newsize = PyDict_MINSIZE;
newsize <= minused && newsize > 0;
newsize <<= 1)
;
if (newsize <= 0) {
1997-05-02 00:12:38 -03:00
PyErr_NoMemory();
return -1;
}
/* Get space for a new table. */
oldtable = mp->ma_table;
assert(oldtable != NULL);
is_oldtable_malloced = oldtable != mp->ma_smalltable;
2001-08-02 01:15:00 -03:00
if (newsize == PyDict_MINSIZE) {
/* A large table is shrinking, or we can't get any smaller. */
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
newtable = mp->ma_smalltable;
if (newtable == oldtable) {
if (mp->ma_fill == mp->ma_used) {
/* No dummies, so no point doing anything. */
return 0;
}
/* We're not going to resize it, but rebuild the
table anyway to purge old dummy entries.
Subtle: This is *necessary* if fill==size,
as lookdict needs at least one virgin slot to
terminate failing searches. If fill < size, it's
merely desirable, as dummies slow searches. */
assert(mp->ma_fill > mp->ma_used);
memcpy(small_copy, oldtable, sizeof(small_copy));
oldtable = small_copy;
}
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
}
else {
newtable = PyMem_NEW(dictentry, newsize);
if (newtable == NULL) {
PyErr_NoMemory();
return -1;
}
}
/* Make the dict empty, using the new table. */
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
assert(newtable != oldtable);
mp->ma_table = newtable;
mp->ma_mask = newsize - 1;
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
memset(newtable, 0, sizeof(dictentry) * newsize);
mp->ma_used = 0;
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
i = mp->ma_fill;
mp->ma_fill = 0;
/* Copy the data over; this is refcount-neutral for active entries;
dummy entries aren't copied over, of course */
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
for (ep = oldtable; i > 0; ep++) {
if (ep->me_value != NULL) { /* active entry */
--i;
insertdict(mp, ep->me_key, ep->me_hash, ep->me_value);
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
}
else if (ep->me_key != NULL) { /* dummy entry */
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
--i;
assert(ep->me_key == dummy);
Py_DECREF(ep->me_key);
}
/* else key == value == NULL: nothing to do */
}
if (is_oldtable_malloced)
PyMem_DEL(oldtable);
return 0;
}
1997-05-02 00:12:38 -03:00
PyObject *
PyDict_GetItem(PyObject *op, PyObject *key)
{
long hash;
dictobject *mp = (dictobject *)op;
1997-05-02 00:12:38 -03:00
if (!PyDict_Check(op)) {
return NULL;
}
#ifdef CACHE_HASH
if (!PyString_CheckExact(key) ||
1997-05-02 00:12:38 -03:00
(hash = ((PyStringObject *) key)->ob_shash) == -1)
#endif
{
1997-05-02 00:12:38 -03:00
hash = PyObject_Hash(key);
if (hash == -1) {
PyErr_Clear();
return NULL;
}
}
return (mp->ma_lookup)(mp, key, hash)->me_value;
}
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
/* CAUTION: PyDict_SetItem() must guarantee that it won't resize the
* dictionary if it is merely replacing the value for an existing key.
* This is means that it's safe to loop over a dictionary with
* PyDict_Next() and occasionally replace a value -- but you can't
* insert new keys or remove them.
*/
int
PyDict_SetItem(register PyObject *op, PyObject *key, PyObject *value)
{
register dictobject *mp;
register long hash;
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
register int n_used;
1997-05-02 00:12:38 -03:00
if (!PyDict_Check(op)) {
PyErr_BadInternalCall();
return -1;
}
mp = (dictobject *)op;
#ifdef CACHE_HASH
if (PyString_CheckExact(key)) {
1997-01-18 03:55:05 -04:00
#ifdef INTERN_STRINGS
1997-05-02 00:12:38 -03:00
if (((PyStringObject *)key)->ob_sinterned != NULL) {
key = ((PyStringObject *)key)->ob_sinterned;
hash = ((PyStringObject *)key)->ob_shash;
1997-01-18 03:55:05 -04:00
}
else
#endif
1997-01-18 03:55:05 -04:00
{
1997-05-02 00:12:38 -03:00
hash = ((PyStringObject *)key)->ob_shash;
1997-01-18 03:55:05 -04:00
if (hash == -1)
1997-05-02 00:12:38 -03:00
hash = PyObject_Hash(key);
1997-01-18 03:55:05 -04:00
}
}
else
#endif
{
1997-05-02 00:12:38 -03:00
hash = PyObject_Hash(key);
1997-01-18 03:55:05 -04:00
if (hash == -1)
return -1;
}
assert(mp->ma_fill <= mp->ma_mask); /* at least one empty slot */
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
n_used = mp->ma_used;
Py_INCREF(value);
Py_INCREF(key);
insertdict(mp, key, hash, value);
/* If we added a key, we can safely resize. Otherwise skip this!
* If fill >= 2/3 size, adjust size. Normally, this doubles the
* size, but it's also possible for the dict to shrink (if ma_fill is
* much larger than ma_used, meaning a lot of dict keys have been
* deleted).
*/
if (mp->ma_used > n_used && mp->ma_fill*3 >= (mp->ma_mask+1)*2) {
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
if (dictresize(mp, mp->ma_used*2) != 0)
return -1;
}
return 0;
}
int
PyDict_DelItem(PyObject *op, PyObject *key)
{
register dictobject *mp;
register long hash;
register dictentry *ep;
1997-05-02 00:12:38 -03:00
PyObject *old_value, *old_key;
1997-05-02 00:12:38 -03:00
if (!PyDict_Check(op)) {
PyErr_BadInternalCall();
return -1;
}
#ifdef CACHE_HASH
if (!PyString_CheckExact(key) ||
1997-05-02 00:12:38 -03:00
(hash = ((PyStringObject *) key)->ob_shash) == -1)
#endif
{
1997-05-02 00:12:38 -03:00
hash = PyObject_Hash(key);
if (hash == -1)
return -1;
}
mp = (dictobject *)op;
ep = (mp->ma_lookup)(mp, key, hash);
if (ep->me_value == NULL) {
1997-05-02 00:12:38 -03:00
PyErr_SetObject(PyExc_KeyError, key);
return -1;
}
old_key = ep->me_key;
1997-05-02 00:12:38 -03:00
Py_INCREF(dummy);
ep->me_key = dummy;
old_value = ep->me_value;
ep->me_value = NULL;
mp->ma_used--;
Py_DECREF(old_value);
Py_DECREF(old_key);
return 0;
}
void
PyDict_Clear(PyObject *op)
{
dictobject *mp;
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
dictentry *ep, *table;
int table_is_malloced;
int fill;
2001-08-02 01:15:00 -03:00
dictentry small_copy[PyDict_MINSIZE];
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
#ifdef Py_DEBUG
int i, n;
#endif
1997-05-02 00:12:38 -03:00
if (!PyDict_Check(op))
return;
mp = (dictobject *)op;
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
#ifdef Py_DEBUG
n = mp->ma_mask + 1;
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
i = 0;
#endif
table = mp->ma_table;
assert(table != NULL);
table_is_malloced = table != mp->ma_smalltable;
/* This is delicate. During the process of clearing the dict,
* decrefs can cause the dict to mutate. To avoid fatal confusion
* (voice of experience), we have to make the dict empty before
* clearing the slots, and never refer to anything via mp->xxx while
* clearing.
*/
fill = mp->ma_fill;
if (table_is_malloced)
2001-08-02 01:15:00 -03:00
EMPTY_TO_MINSIZE(mp);
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
else if (fill > 0) {
/* It's a small table with something that needs to be cleared.
* Afraid the only safe way is to copy the dict entries into
* another small table first.
*/
memcpy(small_copy, table, sizeof(small_copy));
table = small_copy;
2001-08-02 01:15:00 -03:00
EMPTY_TO_MINSIZE(mp);
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
}
/* else it's a small table that's already empty */
/* Now we can finally clear things. If C had refcounts, we could
* assert that the refcount on table is 1 now, i.e. that this function
* has unique access to it, so decref side-effects can't alter it.
*/
for (ep = table; fill > 0; ++ep) {
#ifdef Py_DEBUG
assert(i < n);
++i;
#endif
if (ep->me_key) {
--fill;
Py_DECREF(ep->me_key);
Py_XDECREF(ep->me_value);
}
#ifdef Py_DEBUG
else
assert(ep->me_value == NULL);
#endif
}
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
if (table_is_malloced)
PyMem_DEL(table);
}
/* CAUTION: In general, it isn't safe to use PyDict_Next in a loop that
* mutates the dict. One exception: it is safe if the loop merely changes
* the values associated with the keys (but doesn't insert new keys or
* delete keys), via PyDict_SetItem().
*/
int
PyDict_Next(PyObject *op, int *ppos, PyObject **pkey, PyObject **pvalue)
{
int i;
register dictobject *mp;
1997-05-02 00:12:38 -03:00
if (!PyDict_Check(op))
return 0;
mp = (dictobject *)op;
i = *ppos;
if (i < 0)
return 0;
while (i <= mp->ma_mask && mp->ma_table[i].me_value == NULL)
i++;
*ppos = i+1;
if (i > mp->ma_mask)
return 0;
if (pkey)
*pkey = mp->ma_table[i].me_key;
if (pvalue)
*pvalue = mp->ma_table[i].me_value;
return 1;
}
/* Methods */
static void
dict_dealloc(register dictobject *mp)
{
register dictentry *ep;
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
int fill = mp->ma_fill;
Py_TRASHCAN_SAFE_BEGIN(mp)
2001-08-29 20:54:21 -03:00
_PyObject_GC_UNTRACK(mp);
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
for (ep = mp->ma_table; fill > 0; ep++) {
if (ep->me_key) {
--fill;
1997-05-02 00:12:38 -03:00
Py_DECREF(ep->me_key);
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
Py_XDECREF(ep->me_value);
}
}
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
if (mp->ma_table != mp->ma_smalltable)
PyMem_DEL(mp->ma_table);
mp->ob_type->tp_free((PyObject *)mp);
Py_TRASHCAN_SAFE_END(mp)
}
static int
dict_print(register dictobject *mp, register FILE *fp, register int flags)
{
register int i;
register int any;
i = Py_ReprEnter((PyObject*)mp);
if (i != 0) {
if (i < 0)
return i;
fprintf(fp, "{...}");
return 0;
}
fprintf(fp, "{");
any = 0;
for (i = 0; i <= mp->ma_mask; i++) {
dictentry *ep = mp->ma_table + i;
PyObject *pvalue = ep->me_value;
if (pvalue != NULL) {
/* Prevent PyObject_Repr from deleting value during
key format */
Py_INCREF(pvalue);
if (any++ > 0)
fprintf(fp, ", ");
if (PyObject_Print((PyObject *)ep->me_key, fp, 0)!=0) {
Py_DECREF(pvalue);
Py_ReprLeave((PyObject*)mp);
return -1;
}
fprintf(fp, ": ");
if (PyObject_Print(pvalue, fp, 0) != 0) {
Py_DECREF(pvalue);
Py_ReprLeave((PyObject*)mp);
return -1;
}
Py_DECREF(pvalue);
}
}
fprintf(fp, "}");
Py_ReprLeave((PyObject*)mp);
return 0;
}
1997-05-02 00:12:38 -03:00
static PyObject *
dict_repr(dictobject *mp)
{
int i;
PyObject *s, *temp, *colon = NULL;
PyObject *pieces = NULL, *result = NULL;
PyObject *key, *value;
i = Py_ReprEnter((PyObject *)mp);
if (i != 0) {
return i > 0 ? PyString_FromString("{...}") : NULL;
}
if (mp->ma_used == 0) {
result = PyString_FromString("{}");
goto Done;
}
pieces = PyList_New(0);
if (pieces == NULL)
goto Done;
1997-05-02 00:12:38 -03:00
colon = PyString_FromString(": ");
if (colon == NULL)
goto Done;
/* Do repr() on each key+value pair, and insert ": " between them.
Note that repr may mutate the dict. */
i = 0;
while (PyDict_Next((PyObject *)mp, &i, &key, &value)) {
int status;
/* Prevent repr from deleting value during key format. */
Py_INCREF(value);
s = PyObject_Repr(key);
PyString_Concat(&s, colon);
PyString_ConcatAndDel(&s, PyObject_Repr(value));
Py_DECREF(value);
if (s == NULL)
goto Done;
status = PyList_Append(pieces, s);
Py_DECREF(s); /* append created a new ref */
if (status < 0)
goto Done;
}
/* Add "{}" decorations to the first and last items. */
assert(PyList_GET_SIZE(pieces) > 0);
s = PyString_FromString("{");
if (s == NULL)
goto Done;
temp = PyList_GET_ITEM(pieces, 0);
PyString_ConcatAndDel(&s, temp);
PyList_SET_ITEM(pieces, 0, s);
if (s == NULL)
goto Done;
s = PyString_FromString("}");
if (s == NULL)
goto Done;
temp = PyList_GET_ITEM(pieces, PyList_GET_SIZE(pieces) - 1);
PyString_ConcatAndDel(&temp, s);
PyList_SET_ITEM(pieces, PyList_GET_SIZE(pieces) - 1, temp);
if (temp == NULL)
goto Done;
/* Paste them all together with ", " between. */
s = PyString_FromString(", ");
if (s == NULL)
goto Done;
result = _PyString_Join(s, pieces);
Py_DECREF(s);
Done:
Py_XDECREF(pieces);
1997-05-02 00:12:38 -03:00
Py_XDECREF(colon);
Py_ReprLeave((PyObject *)mp);
return result;
}
static int
dict_length(dictobject *mp)
{
return mp->ma_used;
}
1997-05-02 00:12:38 -03:00
static PyObject *
dict_subscript(dictobject *mp, register PyObject *key)
{
1997-05-02 00:12:38 -03:00
PyObject *v;
long hash;
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
assert(mp->ma_table != NULL);
#ifdef CACHE_HASH
if (!PyString_CheckExact(key) ||
1997-05-02 00:12:38 -03:00
(hash = ((PyStringObject *) key)->ob_shash) == -1)
#endif
{
1997-05-02 00:12:38 -03:00
hash = PyObject_Hash(key);
if (hash == -1)
return NULL;
}
v = (mp->ma_lookup)(mp, key, hash) -> me_value;
if (v == NULL)
1997-05-02 00:12:38 -03:00
PyErr_SetObject(PyExc_KeyError, key);
else
1997-05-02 00:12:38 -03:00
Py_INCREF(v);
return v;
}
static int
dict_ass_sub(dictobject *mp, PyObject *v, PyObject *w)
{
if (w == NULL)
1997-05-02 00:12:38 -03:00
return PyDict_DelItem((PyObject *)mp, v);
else
1997-05-02 00:12:38 -03:00
return PyDict_SetItem((PyObject *)mp, v, w);
}
static PyMappingMethods dict_as_mapping = {
(inquiry)dict_length, /*mp_length*/
(binaryfunc)dict_subscript, /*mp_subscript*/
(objobjargproc)dict_ass_sub, /*mp_ass_subscript*/
};
1997-05-02 00:12:38 -03:00
static PyObject *
dict_keys(register dictobject *mp)
{
1997-05-02 00:12:38 -03:00
register PyObject *v;
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
register int i, j, n;
again:
n = mp->ma_used;
v = PyList_New(n);
if (v == NULL)
return NULL;
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
if (n != mp->ma_used) {
/* Durnit. The allocations caused the dict to resize.
* Just start over, this shouldn't normally happen.
*/
Py_DECREF(v);
goto again;
}
for (i = 0, j = 0; i <= mp->ma_mask; i++) {
if (mp->ma_table[i].me_value != NULL) {
1997-05-02 00:12:38 -03:00
PyObject *key = mp->ma_table[i].me_key;
Py_INCREF(key);
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
PyList_SET_ITEM(v, j, key);
j++;
}
}
return v;
}
1997-05-02 00:12:38 -03:00
static PyObject *
dict_values(register dictobject *mp)
{
1997-05-02 00:12:38 -03:00
register PyObject *v;
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
register int i, j, n;
again:
n = mp->ma_used;
v = PyList_New(n);
if (v == NULL)
return NULL;
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
if (n != mp->ma_used) {
/* Durnit. The allocations caused the dict to resize.
* Just start over, this shouldn't normally happen.
*/
Py_DECREF(v);
goto again;
}
for (i = 0, j = 0; i <= mp->ma_mask; i++) {
if (mp->ma_table[i].me_value != NULL) {
1997-05-02 00:12:38 -03:00
PyObject *value = mp->ma_table[i].me_value;
Py_INCREF(value);
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
PyList_SET_ITEM(v, j, value);
j++;
}
}
return v;
}
1997-05-02 00:12:38 -03:00
static PyObject *
dict_items(register dictobject *mp)
{
1997-05-02 00:12:38 -03:00
register PyObject *v;
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
register int i, j, n;
PyObject *item, *key, *value;
/* Preallocate the list of tuples, to avoid allocations during
* the loop over the items, which could trigger GC, which
* could resize the dict. :-(
*/
again:
n = mp->ma_used;
v = PyList_New(n);
if (v == NULL)
return NULL;
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
for (i = 0; i < n; i++) {
item = PyTuple_New(2);
if (item == NULL) {
Py_DECREF(v);
return NULL;
}
PyList_SET_ITEM(v, i, item);
}
if (n != mp->ma_used) {
/* Durnit. The allocations caused the dict to resize.
* Just start over, this shouldn't normally happen.
*/
Py_DECREF(v);
goto again;
}
/* Nothing we do below makes any function calls. */
for (i = 0, j = 0; i <= mp->ma_mask; i++) {
if (mp->ma_table[i].me_value != NULL) {
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
key = mp->ma_table[i].me_key;
value = mp->ma_table[i].me_value;
item = PyList_GET_ITEM(v, j);
1997-05-02 00:12:38 -03:00
Py_INCREF(key);
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
PyTuple_SET_ITEM(item, 0, key);
1997-05-02 00:12:38 -03:00
Py_INCREF(value);
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
PyTuple_SET_ITEM(item, 1, value);
j++;
}
}
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
assert(j == n);
return v;
}
1997-05-28 16:15:28 -03:00
static PyObject *
dict_update(PyObject *mp, PyObject *other)
2001-08-02 01:15:00 -03:00
{
if (PyDict_Update(mp, other) < 0)
return NULL;
Py_INCREF(Py_None);
return Py_None;
}
/* Update unconditionally replaces existing items.
Merge has a 3rd argument 'override'; if set, it acts like Update,
Generalize dictionary() to accept a sequence of 2-sequences. At the outer level, the iterator protocol is used for memory-efficiency (the outer sequence may be very large if fully materialized); at the inner level, PySequence_Fast() is used for time-efficiency (these should always be sequences of length 2). dictobject.c, new functions PyDict_{Merge,Update}FromSeq2. These are wholly analogous to PyDict_{Merge,Update}, but process a sequence-of-2- sequences argument instead of a mapping object. For now, I left these functions file static, so no corresponding doc changes. It's tempting to change dict.update() to allow a sequence-of-2-seqs argument too. Also changed the name of dictionary's keyword argument from "mapping" to "x". Got a better name? "mapping_or_sequence_of_pairs" isn't attractive, although more so than "mosop" <wink>. abstract.h, abstract.tex: Added new PySequence_Fast_GET_SIZE function, much faster than going thru the all-purpose PySequence_Size. libfuncs.tex: - Document dictionary(). - Fiddle tuple() and list() to admit that their argument is optional. - The long-winded repetitions of "a sequence, a container that supports iteration, or an iterator object" is getting to be a PITA. Many months ago I suggested factoring this out into "iterable object", where the definition of that could include being explicit about generators too (as is, I'm not sure a reader outside of PythonLabs could guess that "an iterator object" includes a generator call). - Please check my curly braces -- I'm going blind <0.9 wink>. abstract.c, PySequence_Tuple(): When PyObject_GetIter() fails, leave its error msg alone now (the msg it produces has improved since PySequence_Tuple was generalized to accept iterable objects, and PySequence_Tuple was also stomping on the msg in cases it shouldn't have even before PyObject_GetIter grew a better msg).
2001-10-26 02:06:50 -03:00
otherwise it leaves existing items unchanged.
PyDict_{Update,Merge} update/merge from a mapping object.
PyDict_MergeFromSeq2 updates/merges from any iterable object
Generalize dictionary() to accept a sequence of 2-sequences. At the outer level, the iterator protocol is used for memory-efficiency (the outer sequence may be very large if fully materialized); at the inner level, PySequence_Fast() is used for time-efficiency (these should always be sequences of length 2). dictobject.c, new functions PyDict_{Merge,Update}FromSeq2. These are wholly analogous to PyDict_{Merge,Update}, but process a sequence-of-2- sequences argument instead of a mapping object. For now, I left these functions file static, so no corresponding doc changes. It's tempting to change dict.update() to allow a sequence-of-2-seqs argument too. Also changed the name of dictionary's keyword argument from "mapping" to "x". Got a better name? "mapping_or_sequence_of_pairs" isn't attractive, although more so than "mosop" <wink>. abstract.h, abstract.tex: Added new PySequence_Fast_GET_SIZE function, much faster than going thru the all-purpose PySequence_Size. libfuncs.tex: - Document dictionary(). - Fiddle tuple() and list() to admit that their argument is optional. - The long-winded repetitions of "a sequence, a container that supports iteration, or an iterator object" is getting to be a PITA. Many months ago I suggested factoring this out into "iterable object", where the definition of that could include being explicit about generators too (as is, I'm not sure a reader outside of PythonLabs could guess that "an iterator object" includes a generator call). - Please check my curly braces -- I'm going blind <0.9 wink>. abstract.c, PySequence_Tuple(): When PyObject_GetIter() fails, leave its error msg alone now (the msg it produces has improved since PySequence_Tuple was generalized to accept iterable objects, and PySequence_Tuple was also stomping on the msg in cases it shouldn't have even before PyObject_GetIter grew a better msg).
2001-10-26 02:06:50 -03:00
producing iterable objects of length 2.
*/
int
Generalize dictionary() to accept a sequence of 2-sequences. At the outer level, the iterator protocol is used for memory-efficiency (the outer sequence may be very large if fully materialized); at the inner level, PySequence_Fast() is used for time-efficiency (these should always be sequences of length 2). dictobject.c, new functions PyDict_{Merge,Update}FromSeq2. These are wholly analogous to PyDict_{Merge,Update}, but process a sequence-of-2- sequences argument instead of a mapping object. For now, I left these functions file static, so no corresponding doc changes. It's tempting to change dict.update() to allow a sequence-of-2-seqs argument too. Also changed the name of dictionary's keyword argument from "mapping" to "x". Got a better name? "mapping_or_sequence_of_pairs" isn't attractive, although more so than "mosop" <wink>. abstract.h, abstract.tex: Added new PySequence_Fast_GET_SIZE function, much faster than going thru the all-purpose PySequence_Size. libfuncs.tex: - Document dictionary(). - Fiddle tuple() and list() to admit that their argument is optional. - The long-winded repetitions of "a sequence, a container that supports iteration, or an iterator object" is getting to be a PITA. Many months ago I suggested factoring this out into "iterable object", where the definition of that could include being explicit about generators too (as is, I'm not sure a reader outside of PythonLabs could guess that "an iterator object" includes a generator call). - Please check my curly braces -- I'm going blind <0.9 wink>. abstract.c, PySequence_Tuple(): When PyObject_GetIter() fails, leave its error msg alone now (the msg it produces has improved since PySequence_Tuple was generalized to accept iterable objects, and PySequence_Tuple was also stomping on the msg in cases it shouldn't have even before PyObject_GetIter grew a better msg).
2001-10-26 02:06:50 -03:00
PyDict_MergeFromSeq2(PyObject *d, PyObject *seq2, int override)
{
PyObject *it; /* iter(seq2) */
int i; /* index into seq2 of current element */
PyObject *item; /* seq2[i] */
PyObject *fast; /* item as a 2-tuple or 2-list */
assert(d != NULL);
assert(PyDict_Check(d));
assert(seq2 != NULL);
it = PyObject_GetIter(seq2);
if (it == NULL)
return -1;
for (i = 0; ; ++i) {
PyObject *key, *value;
int n;
fast = NULL;
item = PyIter_Next(it);
if (item == NULL) {
if (PyErr_Occurred())
goto Fail;
break;
}
/* Convert item to sequence, and verify length 2. */
fast = PySequence_Fast(item, "");
if (fast == NULL) {
if (PyErr_ExceptionMatches(PyExc_TypeError))
PyErr_Format(PyExc_TypeError,
"cannot convert dictionary update "
"sequence element #%d to a sequence",
i);
goto Fail;
}
n = PySequence_Fast_GET_SIZE(fast);
if (n != 2) {
PyErr_Format(PyExc_ValueError,
"dictionary update sequence element #%d "
"has length %d; 2 is required",
i, n);
goto Fail;
}
/* Update/merge with this (key, value) pair. */
key = PySequence_Fast_GET_ITEM(fast, 0);
value = PySequence_Fast_GET_ITEM(fast, 1);
if (override || PyDict_GetItem(d, key) == NULL) {
int status = PyDict_SetItem(d, key, value);
if (status < 0)
goto Fail;
}
Py_DECREF(fast);
Py_DECREF(item);
}
i = 0;
goto Return;
Fail:
Py_XDECREF(item);
Py_XDECREF(fast);
i = -1;
Return:
Py_DECREF(it);
return i;
}
2001-08-02 01:15:00 -03:00
int
PyDict_Update(PyObject *a, PyObject *b)
{
return PyDict_Merge(a, b, 1);
}
int
PyDict_Merge(PyObject *a, PyObject *b, int override)
1997-05-28 16:15:28 -03:00
{
2001-08-02 01:15:00 -03:00
register PyDictObject *mp, *other;
1997-05-28 16:15:28 -03:00
register int i;
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
dictentry *entry;
2001-08-02 01:15:00 -03:00
/* We accept for the argument either a concrete dictionary object,
* or an abstract "mapping" object. For the former, we can do
* things quite efficiently. For the latter, we only require that
* PyMapping_Keys() and PyObject_GetItem() be supported.
*/
2001-08-02 01:15:00 -03:00
if (a == NULL || !PyDict_Check(a) || b == NULL) {
PyErr_BadInternalCall();
return -1;
}
mp = (dictobject*)a;
if (PyDict_Check(b)) {
other = (dictobject*)b;
if (other == mp || other->ma_used == 0)
/* a.update(a) or a.update({}); nothing to do */
2001-08-02 01:15:00 -03:00
return 0;
/* Do one big resize at the start, rather than
* incrementally resizing as we insert new items. Expect
* that there will be no (or few) overlapping keys.
*/
if ((mp->ma_fill + other->ma_used)*3 >= (mp->ma_mask+1)*2) {
if (dictresize(mp, (mp->ma_used + other->ma_used)*3/2) != 0)
2001-08-02 01:15:00 -03:00
return -1;
}
for (i = 0; i <= other->ma_mask; i++) {
entry = &other->ma_table[i];
if (entry->me_value != NULL &&
(override ||
PyDict_GetItem(a, entry->me_key) == NULL)) {
Py_INCREF(entry->me_key);
Py_INCREF(entry->me_value);
insertdict(mp, entry->me_key, entry->me_hash,
entry->me_value);
}
}
1997-05-28 16:15:28 -03:00
}
else {
/* Do it the generic, slower way */
2001-08-02 01:15:00 -03:00
PyObject *keys = PyMapping_Keys(b);
PyObject *iter;
PyObject *key, *value;
int status;
if (keys == NULL)
/* Docstring says this is equivalent to E.keys() so
* if E doesn't have a .keys() method we want
* AttributeError to percolate up. Might as well
* do the same for any other error.
*/
2001-08-02 01:15:00 -03:00
return -1;
iter = PyObject_GetIter(keys);
Py_DECREF(keys);
if (iter == NULL)
2001-08-02 01:15:00 -03:00
return -1;
for (key = PyIter_Next(iter); key; key = PyIter_Next(iter)) {
if (!override && PyDict_GetItem(a, key) != NULL) {
Py_DECREF(key);
continue;
}
2001-08-02 01:15:00 -03:00
value = PyObject_GetItem(b, key);
if (value == NULL) {
Py_DECREF(iter);
Py_DECREF(key);
2001-08-02 01:15:00 -03:00
return -1;
}
status = PyDict_SetItem(a, key, value);
Py_DECREF(key);
Py_DECREF(value);
if (status < 0) {
Py_DECREF(iter);
2001-08-02 01:15:00 -03:00
return -1;
}
1997-05-28 16:15:28 -03:00
}
Py_DECREF(iter);
if (PyErr_Occurred())
/* Iterator completed, via error */
2001-08-02 01:15:00 -03:00
return -1;
1997-05-28 16:15:28 -03:00
}
2001-08-02 01:15:00 -03:00
return 0;
1997-05-28 16:15:28 -03:00
}
static PyObject *
dict_copy(register dictobject *mp)
1997-05-28 16:15:28 -03:00
{
return PyDict_Copy((PyObject*)mp);
}
PyObject *
PyDict_Copy(PyObject *o)
{
register dictobject *mp;
1997-05-28 16:15:28 -03:00
register int i;
dictobject *copy;
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
dictentry *entry;
if (o == NULL || !PyDict_Check(o)) {
PyErr_BadInternalCall();
1997-05-28 16:15:28 -03:00
return NULL;
}
mp = (dictobject *)o;
1997-05-28 16:15:28 -03:00
copy = (dictobject *)PyDict_New();
if (copy == NULL)
return NULL;
if (mp->ma_used > 0) {
if (dictresize(copy, mp->ma_used*3/2) != 0)
return NULL;
for (i = 0; i <= mp->ma_mask; i++) {
1997-05-28 16:15:28 -03:00
entry = &mp->ma_table[i];
if (entry->me_value != NULL) {
Py_INCREF(entry->me_key);
Py_INCREF(entry->me_value);
insertdict(copy, entry->me_key, entry->me_hash,
entry->me_value);
}
}
}
return (PyObject *)copy;
}
int
PyDict_Size(PyObject *mp)
{
1997-05-02 00:12:38 -03:00
if (mp == NULL || !PyDict_Check(mp)) {
PyErr_BadInternalCall();
return 0;
}
return ((dictobject *)mp)->ma_used;
}
1997-05-02 00:12:38 -03:00
PyObject *
PyDict_Keys(PyObject *mp)
{
1997-05-02 00:12:38 -03:00
if (mp == NULL || !PyDict_Check(mp)) {
PyErr_BadInternalCall();
return NULL;
}
return dict_keys((dictobject *)mp);
}
1997-05-02 00:12:38 -03:00
PyObject *
PyDict_Values(PyObject *mp)
{
1997-05-02 00:12:38 -03:00
if (mp == NULL || !PyDict_Check(mp)) {
PyErr_BadInternalCall();
return NULL;
}
return dict_values((dictobject *)mp);
}
1997-05-02 00:12:38 -03:00
PyObject *
PyDict_Items(PyObject *mp)
{
1997-05-02 00:12:38 -03:00
if (mp == NULL || !PyDict_Check(mp)) {
PyErr_BadInternalCall();
return NULL;
}
return dict_items((dictobject *)mp);
}
/* Subroutine which returns the smallest key in a for which b's value
is different or absent. The value is returned too, through the
pval argument. Both are NULL if no key in a is found for which b's status
differs. The refcounts on (and only on) non-NULL *pval and function return
values must be decremented by the caller (characterize() increments them
to ensure that mutating comparison and PyDict_GetItem calls can't delete
them before the caller is done looking at them). */
1997-05-02 00:12:38 -03:00
static PyObject *
characterize(dictobject *a, dictobject *b, PyObject **pval)
{
PyObject *akey = NULL; /* smallest key in a s.t. a[akey] != b[akey] */
PyObject *aval = NULL; /* a[akey] */
int i, cmp;
for (i = 0; i <= a->ma_mask; i++) {
PyObject *thiskey, *thisaval, *thisbval;
if (a->ma_table[i].me_value == NULL)
continue;
thiskey = a->ma_table[i].me_key;
Py_INCREF(thiskey); /* keep alive across compares */
if (akey != NULL) {
cmp = PyObject_RichCompareBool(akey, thiskey, Py_LT);
if (cmp < 0) {
Py_DECREF(thiskey);
goto Fail;
}
if (cmp > 0 ||
i > a->ma_mask ||
a->ma_table[i].me_value == NULL)
1997-05-02 00:12:38 -03:00
{
/* Not the *smallest* a key; or maybe it is
* but the compare shrunk the dict so we can't
* find its associated value anymore; or
* maybe it is but the compare deleted the
* a[thiskey] entry.
*/
Py_DECREF(thiskey);
continue;
}
}
/* Compare a[thiskey] to b[thiskey]; cmp <- true iff equal. */
thisaval = a->ma_table[i].me_value;
assert(thisaval);
Py_INCREF(thisaval); /* keep alive */
thisbval = PyDict_GetItem((PyObject *)b, thiskey);
if (thisbval == NULL)
cmp = 0;
else {
/* both dicts have thiskey: same values? */
cmp = PyObject_RichCompareBool(
thisaval, thisbval, Py_EQ);
if (cmp < 0) {
Py_DECREF(thiskey);
Py_DECREF(thisaval);
goto Fail;
}
}
if (cmp == 0) {
/* New winner. */
Py_XDECREF(akey);
Py_XDECREF(aval);
akey = thiskey;
aval = thisaval;
}
else {
Py_DECREF(thiskey);
Py_DECREF(thisaval);
}
}
*pval = aval;
return akey;
Fail:
Py_XDECREF(akey);
Py_XDECREF(aval);
*pval = NULL;
return NULL;
}
static int
dict_compare(dictobject *a, dictobject *b)
{
1997-05-02 00:12:38 -03:00
PyObject *adiff, *bdiff, *aval, *bval;
int res;
/* Compare lengths first */
if (a->ma_used < b->ma_used)
return -1; /* a is shorter */
else if (a->ma_used > b->ma_used)
return 1; /* b is shorter */
/* Same length -- check all keys */
bdiff = bval = NULL;
adiff = characterize(a, b, &aval);
if (adiff == NULL) {
assert(!aval);
2001-05-10 15:58:31 -03:00
/* Either an error, or a is a subset with the same length so
* must be equal.
*/
res = PyErr_Occurred() ? -1 : 0;
goto Finished;
}
bdiff = characterize(b, a, &bval);
if (bdiff == NULL && PyErr_Occurred()) {
assert(!bval);
res = -1;
goto Finished;
}
res = 0;
if (bdiff) {
/* bdiff == NULL "should be" impossible now, but perhaps
* the last comparison done by the characterize() on a had
* the side effect of making the dicts equal!
*/
res = PyObject_Compare(adiff, bdiff);
}
if (res == 0 && bval != NULL)
1997-05-02 00:12:38 -03:00
res = PyObject_Compare(aval, bval);
Finished:
Py_XDECREF(adiff);
Py_XDECREF(bdiff);
Py_XDECREF(aval);
Py_XDECREF(bval);
return res;
}
/* Return 1 if dicts equal, 0 if not, -1 if error.
* Gets out as soon as any difference is detected.
* Uses only Py_EQ comparison.
*/
static int
dict_equal(dictobject *a, dictobject *b)
{
int i;
if (a->ma_used != b->ma_used)
/* can't be equal if # of entries differ */
return 0;
/* Same # of entries -- check all of 'em. Exit early on any diff. */
for (i = 0; i <= a->ma_mask; i++) {
PyObject *aval = a->ma_table[i].me_value;
if (aval != NULL) {
int cmp;
PyObject *bval;
PyObject *key = a->ma_table[i].me_key;
/* temporarily bump aval's refcount to ensure it stays
alive until we're done with it */
Py_INCREF(aval);
bval = PyDict_GetItem((PyObject *)b, key);
if (bval == NULL) {
Py_DECREF(aval);
return 0;
}
cmp = PyObject_RichCompareBool(aval, bval, Py_EQ);
Py_DECREF(aval);
if (cmp <= 0) /* error or not equal */
return cmp;
}
}
return 1;
}
static PyObject *
dict_richcompare(PyObject *v, PyObject *w, int op)
{
int cmp;
PyObject *res;
if (!PyDict_Check(v) || !PyDict_Check(w)) {
res = Py_NotImplemented;
}
else if (op == Py_EQ || op == Py_NE) {
cmp = dict_equal((dictobject *)v, (dictobject *)w);
if (cmp < 0)
return NULL;
res = (cmp == (op == Py_EQ)) ? Py_True : Py_False;
}
else
res = Py_NotImplemented;
Py_INCREF(res);
return res;
}
1997-05-02 00:12:38 -03:00
static PyObject *
dict_has_key(register dictobject *mp, PyObject *key)
{
long hash;
register long ok;
#ifdef CACHE_HASH
if (!PyString_CheckExact(key) ||
1997-05-02 00:12:38 -03:00
(hash = ((PyStringObject *) key)->ob_shash) == -1)
#endif
{
1997-05-02 00:12:38 -03:00
hash = PyObject_Hash(key);
if (hash == -1)
return NULL;
}
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
ok = (mp->ma_lookup)(mp, key, hash)->me_value != NULL;
1997-05-02 00:12:38 -03:00
return PyInt_FromLong(ok);
}
static PyObject *
dict_get(register dictobject *mp, PyObject *args)
{
PyObject *key;
PyObject *failobj = Py_None;
PyObject *val = NULL;
long hash;
if (!PyArg_ParseTuple(args, "O|O:get", &key, &failobj))
return NULL;
#ifdef CACHE_HASH
if (!PyString_CheckExact(key) ||
(hash = ((PyStringObject *) key)->ob_shash) == -1)
#endif
{
hash = PyObject_Hash(key);
if (hash == -1)
return NULL;
}
val = (mp->ma_lookup)(mp, key, hash)->me_value;
if (val == NULL)
val = failobj;
Py_INCREF(val);
return val;
}
static PyObject *
dict_setdefault(register dictobject *mp, PyObject *args)
{
PyObject *key;
PyObject *failobj = Py_None;
PyObject *val = NULL;
long hash;
if (!PyArg_ParseTuple(args, "O|O:setdefault", &key, &failobj))
return NULL;
#ifdef CACHE_HASH
if (!PyString_CheckExact(key) ||
(hash = ((PyStringObject *) key)->ob_shash) == -1)
#endif
{
hash = PyObject_Hash(key);
if (hash == -1)
return NULL;
}
val = (mp->ma_lookup)(mp, key, hash)->me_value;
if (val == NULL) {
val = failobj;
if (PyDict_SetItem((PyObject*)mp, key, failobj))
val = NULL;
}
Py_XINCREF(val);
return val;
}
1997-05-02 00:12:38 -03:00
static PyObject *
dict_clear(register dictobject *mp)
{
1997-05-02 00:12:38 -03:00
PyDict_Clear((PyObject *)mp);
Py_INCREF(Py_None);
return Py_None;
}
2000-12-12 18:02:18 -04:00
static PyObject *
dict_popitem(dictobject *mp)
2000-12-12 18:02:18 -04:00
{
int i = 0;
dictentry *ep;
PyObject *res;
/* Allocate the result tuple before checking the size. Believe it
* or not, this allocation could trigger a garbage collection which
* could empty the dict, so if we checked the size first and that
* happened, the result would be an infinite loop (searching for an
* entry that no longer exists). Note that the usual popitem()
* idiom is "while d: k, v = d.popitem()". so needing to throw the
* tuple away if the dict *is* empty isn't a significant
* inefficiency -- possible, but unlikely in practice.
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
*/
res = PyTuple_New(2);
if (res == NULL)
return NULL;
if (mp->ma_used == 0) {
Py_DECREF(res);
PyErr_SetString(PyExc_KeyError,
"popitem(): dictionary is empty");
return NULL;
}
2000-12-12 18:02:18 -04:00
/* Set ep to "the first" dict entry with a value. We abuse the hash
* field of slot 0 to hold a search finger:
* If slot 0 has a value, use slot 0.
* Else slot 0 is being used to hold a search finger,
* and we use its hash value as the first index to look.
*/
ep = &mp->ma_table[0];
if (ep->me_value == NULL) {
i = (int)ep->me_hash;
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
/* The hash field may be a real hash value, or it may be a
* legit search finger, or it may be a once-legit search
* finger that's out of bounds now because it wrapped around
* or the table shrunk -- simply make sure it's in bounds now.
2000-12-12 18:02:18 -04:00
*/
if (i > mp->ma_mask || i < 1)
2000-12-12 18:02:18 -04:00
i = 1; /* skip slot 0 */
while ((ep = &mp->ma_table[i])->me_value == NULL) {
i++;
if (i > mp->ma_mask)
2000-12-12 18:02:18 -04:00
i = 1;
}
}
Tentative fix for a problem that Tim discovered at the last moment, and reported to python-dev: because we were calling dict_resize() in PyDict_Next(), and because GC's dict_traverse() uses PyDict_Next(), and because PyTuple_New() can cause GC, and because dict_items() calls PyTuple_New(), it was possible for dict_items() to have the dict resized right under its nose. The solution is convoluted, and touches several places: keys(), values(), items(), popitem(), PyDict_Next(), and PyDict_SetItem(). There are two parts to it. First, we no longer call dict_resize() in PyDict_Next(), which seems to solve the immediate problem. But then PyDict_SetItem() must have a different policy about when *it* calls dict_resize(), because we want to guarantee (e.g. for an algorithm that Jeremy uses in the compiler) that you can loop over a dict using PyDict_Next() and make changes to the dict as long as those changes are only value replacements for existing keys using PyDict_SetItem(). This is done by resizing *after* the insertion instead of before, and by remembering the size before we insert the item, and if the size is still the same, we don't bother to even check if we might need to resize. An additional detail is that if the dict starts out empty, we must still resize it before the insertion. That was the first part. :-) The second part is to make keys(), values(), items(), and popitem() safe against side effects on the dict caused by allocations, under the assumption that if the GC can cause arbitrary Python code to run, it can cause other threads to run, and it's not inconceivable that our dict could be resized -- it would be insane to write code that relies on this, but not all code is sane. Now, I have this nagging feeling that the loops in lookdict probably are blissfully assuming that doing a simple key comparison does not change the dict's size. This is not necessarily true (the keys could be class instances after all). But that's a battle for another day.
2001-04-15 19:16:26 -03:00
PyTuple_SET_ITEM(res, 0, ep->me_key);
PyTuple_SET_ITEM(res, 1, ep->me_value);
Py_INCREF(dummy);
ep->me_key = dummy;
ep->me_value = NULL;
mp->ma_used--;
assert(mp->ma_table[0].me_value == NULL);
mp->ma_table[0].me_hash = i + 1; /* next place to start */
2000-12-12 18:02:18 -04:00
return res;
}
static int
dict_traverse(PyObject *op, visitproc visit, void *arg)
{
int i = 0, err;
PyObject *pk;
PyObject *pv;
while (PyDict_Next(op, &i, &pk, &pv)) {
err = visit(pk, arg);
if (err)
return err;
err = visit(pv, arg);
if (err)
return err;
}
return 0;
}
static int
dict_tp_clear(PyObject *op)
{
PyDict_Clear(op);
return 0;
}
staticforward PyObject *dictiter_new(dictobject *, binaryfunc);
static PyObject *
select_key(PyObject *key, PyObject *value)
{
Py_INCREF(key);
return key;
}
static PyObject *
select_value(PyObject *key, PyObject *value)
{
Py_INCREF(value);
return value;
}
static PyObject *
select_item(PyObject *key, PyObject *value)
{
PyObject *res = PyTuple_New(2);
if (res != NULL) {
Py_INCREF(key);
Py_INCREF(value);
PyTuple_SET_ITEM(res, 0, key);
PyTuple_SET_ITEM(res, 1, value);
}
return res;
}
static PyObject *
dict_iterkeys(dictobject *dict)
{
return dictiter_new(dict, select_key);
}
static PyObject *
dict_itervalues(dictobject *dict)
{
return dictiter_new(dict, select_value);
}
static PyObject *
dict_iteritems(dictobject *dict)
{
return dictiter_new(dict, select_item);
}
static char has_key__doc__[] =
"D.has_key(k) -> 1 if D has a key k, else 0";
static char get__doc__[] =
"D.get(k[,d]) -> D[k] if D.has_key(k), else d. d defaults to None.";
static char setdefault_doc__[] =
"D.setdefault(k[,d]) -> D.get(k,d), also set D[k]=d if not D.has_key(k)";
static char popitem__doc__[] =
"D.popitem() -> (k, v), remove and return some (key, value) pair as a\n\
2-tuple; but raise KeyError if D is empty";
static char keys__doc__[] =
"D.keys() -> list of D's keys";
static char items__doc__[] =
"D.items() -> list of D's (key, value) pairs, as 2-tuples";
static char values__doc__[] =
"D.values() -> list of D's values";
static char update__doc__[] =
"D.update(E) -> None. Update D from E: for k in E.keys(): D[k] = E[k]";
static char clear__doc__[] =
"D.clear() -> None. Remove all items from D.";
static char copy__doc__[] =
"D.copy() -> a shallow copy of D";
static char iterkeys__doc__[] =
"D.iterkeys() -> an iterator over the keys of D";
static char itervalues__doc__[] =
"D.itervalues() -> an iterator over the values of D";
static char iteritems__doc__[] =
"D.iteritems() -> an iterator over the (key, value) items of D";
1997-05-02 00:12:38 -03:00
static PyMethodDef mapp_methods[] = {
{"has_key", (PyCFunction)dict_has_key, METH_O,
has_key__doc__},
{"get", (PyCFunction)dict_get, METH_VARARGS,
get__doc__},
{"setdefault", (PyCFunction)dict_setdefault, METH_VARARGS,
setdefault_doc__},
{"popitem", (PyCFunction)dict_popitem, METH_NOARGS,
popitem__doc__},
{"keys", (PyCFunction)dict_keys, METH_NOARGS,
keys__doc__},
{"items", (PyCFunction)dict_items, METH_NOARGS,
items__doc__},
{"values", (PyCFunction)dict_values, METH_NOARGS,
values__doc__},
{"update", (PyCFunction)dict_update, METH_O,
update__doc__},
{"clear", (PyCFunction)dict_clear, METH_NOARGS,
clear__doc__},
{"copy", (PyCFunction)dict_copy, METH_NOARGS,
copy__doc__},
{"iterkeys", (PyCFunction)dict_iterkeys, METH_NOARGS,
iterkeys__doc__},
{"itervalues", (PyCFunction)dict_itervalues, METH_NOARGS,
itervalues__doc__},
{"iteritems", (PyCFunction)dict_iteritems, METH_NOARGS,
iteritems__doc__},
{NULL, NULL} /* sentinel */
};
static int
dict_contains(dictobject *mp, PyObject *key)
{
long hash;
#ifdef CACHE_HASH
if (!PyString_CheckExact(key) ||
(hash = ((PyStringObject *) key)->ob_shash) == -1)
#endif
{
hash = PyObject_Hash(key);
if (hash == -1)
return -1;
}
SF patch #425242: Patch which "inlines" small dictionaries. The idea is Marc-Andre Lemburg's, the implementation is Tim's. Add a new ma_smalltable member to dictobjects, an embedded vector of MINSIZE (8) dictentry structs. Short course is that this lets us avoid additional malloc(s) for dicts with no more than 5 entries. The changes are widespread but mostly small. Long course: WRT speed, all scalar operations (getitem, setitem, delitem) on non-empty dicts benefit from no longer needing NULL-pointer checks (ma_table is never NULL anymore). Bulk operations (copy, update, resize, clearing slots during dealloc) benefit in some cases from now looping on the ma_fill count rather than on ma_size, but that was an unexpected benefit: the original reason to loop on ma_fill was to let bulk operations on empty dicts end quickly (since the NULL-pointer checks went away, empty dicts aren't special-cased any more). Special considerations: For dicts that remain empty, this change is a lose on two counts: the dict object contains 8 new dictentry slots now that weren't needed before, and dict object creation also spends time memset'ing these doomed-to-be-unsused slots to NULLs. For dicts with one or two entries that never get larger than 2, it's a mix: a malloc()/free() pair is no longer needed, and the 2-entry case gets to use 8 slots (instead of 4) thus decreasing the chance of collision. Against that, dict object creation spends time memset'ing 4 slots that aren't strictly needed in this case. For dicts with 3 through 5 entries that never get larger than 5, it's a pure win: the dict is created with all the space they need, and they never need to resize. Before they suffered two malloc()/free() calls, plus 1 dict resize, to get enough space. In addition, the 8-slot table they ended with consumed more memory overall, because of the hidden overhead due to the additional malloc. For dicts with 6 or more entries, the ma_smalltable member is wasted space, but then these are large(r) dicts so 8 slots more or less doesn't make much difference. They still benefit all the time from removing ubiquitous dynamic null-pointer checks, and get a small benefit (but relatively smaller the larger the dict) from not having to do two mallocs, two frees, and a resize on the way *to* getting their sixth entry. All in all it appears a small but definite general win, with larger benefits in specific cases. It's especially nice that it allowed to get rid of several branches, gotos and labels, and overall made the code smaller.
2001-05-22 17:40:22 -03:00
return (mp->ma_lookup)(mp, key, hash)->me_value != NULL;
}
/* Hack to implement "key in dict" */
static PySequenceMethods dict_as_sequence = {
0, /* sq_length */
0, /* sq_concat */
0, /* sq_repeat */
0, /* sq_item */
0, /* sq_slice */
0, /* sq_ass_item */
0, /* sq_ass_slice */
(objobjproc)dict_contains, /* sq_contains */
0, /* sq_inplace_concat */
0, /* sq_inplace_repeat */
};
2001-08-02 01:15:00 -03:00
static PyObject *
dict_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
{
PyObject *self;
assert(type != NULL && type->tp_alloc != NULL);
self = type->tp_alloc(type, 0);
if (self != NULL) {
PyDictObject *d = (PyDictObject *)self;
/* It's guaranteed that tp->alloc zeroed out the struct. */
assert(d->ma_table == NULL && d->ma_fill == 0 && d->ma_used == 0);
INIT_NONZERO_DICT_SLOTS(d);
d->ma_lookup = lookdict_string;
#ifdef SHOW_CONVERSION_COUNTS
++created;
#endif
}
return self;
}
static int
dict_init(PyObject *self, PyObject *args, PyObject *kwds)
{
PyObject *arg = NULL;
static char *kwlist[] = {"items", 0};
Generalize dictionary() to accept a sequence of 2-sequences. At the outer level, the iterator protocol is used for memory-efficiency (the outer sequence may be very large if fully materialized); at the inner level, PySequence_Fast() is used for time-efficiency (these should always be sequences of length 2). dictobject.c, new functions PyDict_{Merge,Update}FromSeq2. These are wholly analogous to PyDict_{Merge,Update}, but process a sequence-of-2- sequences argument instead of a mapping object. For now, I left these functions file static, so no corresponding doc changes. It's tempting to change dict.update() to allow a sequence-of-2-seqs argument too. Also changed the name of dictionary's keyword argument from "mapping" to "x". Got a better name? "mapping_or_sequence_of_pairs" isn't attractive, although more so than "mosop" <wink>. abstract.h, abstract.tex: Added new PySequence_Fast_GET_SIZE function, much faster than going thru the all-purpose PySequence_Size. libfuncs.tex: - Document dictionary(). - Fiddle tuple() and list() to admit that their argument is optional. - The long-winded repetitions of "a sequence, a container that supports iteration, or an iterator object" is getting to be a PITA. Many months ago I suggested factoring this out into "iterable object", where the definition of that could include being explicit about generators too (as is, I'm not sure a reader outside of PythonLabs could guess that "an iterator object" includes a generator call). - Please check my curly braces -- I'm going blind <0.9 wink>. abstract.c, PySequence_Tuple(): When PyObject_GetIter() fails, leave its error msg alone now (the msg it produces has improved since PySequence_Tuple was generalized to accept iterable objects, and PySequence_Tuple was also stomping on the msg in cases it shouldn't have even before PyObject_GetIter grew a better msg).
2001-10-26 02:06:50 -03:00
int result = 0;
if (!PyArg_ParseTupleAndKeywords(args, kwds, "|O:dict",
kwlist, &arg))
Generalize dictionary() to accept a sequence of 2-sequences. At the outer level, the iterator protocol is used for memory-efficiency (the outer sequence may be very large if fully materialized); at the inner level, PySequence_Fast() is used for time-efficiency (these should always be sequences of length 2). dictobject.c, new functions PyDict_{Merge,Update}FromSeq2. These are wholly analogous to PyDict_{Merge,Update}, but process a sequence-of-2- sequences argument instead of a mapping object. For now, I left these functions file static, so no corresponding doc changes. It's tempting to change dict.update() to allow a sequence-of-2-seqs argument too. Also changed the name of dictionary's keyword argument from "mapping" to "x". Got a better name? "mapping_or_sequence_of_pairs" isn't attractive, although more so than "mosop" <wink>. abstract.h, abstract.tex: Added new PySequence_Fast_GET_SIZE function, much faster than going thru the all-purpose PySequence_Size. libfuncs.tex: - Document dictionary(). - Fiddle tuple() and list() to admit that their argument is optional. - The long-winded repetitions of "a sequence, a container that supports iteration, or an iterator object" is getting to be a PITA. Many months ago I suggested factoring this out into "iterable object", where the definition of that could include being explicit about generators too (as is, I'm not sure a reader outside of PythonLabs could guess that "an iterator object" includes a generator call). - Please check my curly braces -- I'm going blind <0.9 wink>. abstract.c, PySequence_Tuple(): When PyObject_GetIter() fails, leave its error msg alone now (the msg it produces has improved since PySequence_Tuple was generalized to accept iterable objects, and PySequence_Tuple was also stomping on the msg in cases it shouldn't have even before PyObject_GetIter grew a better msg).
2001-10-26 02:06:50 -03:00
result = -1;
else if (arg != NULL) {
if (PyObject_HasAttrString(arg, "keys"))
result = PyDict_Merge(self, arg, 1);
else
result = PyDict_MergeFromSeq2(self, arg, 1);
}
Generalize dictionary() to accept a sequence of 2-sequences. At the outer level, the iterator protocol is used for memory-efficiency (the outer sequence may be very large if fully materialized); at the inner level, PySequence_Fast() is used for time-efficiency (these should always be sequences of length 2). dictobject.c, new functions PyDict_{Merge,Update}FromSeq2. These are wholly analogous to PyDict_{Merge,Update}, but process a sequence-of-2- sequences argument instead of a mapping object. For now, I left these functions file static, so no corresponding doc changes. It's tempting to change dict.update() to allow a sequence-of-2-seqs argument too. Also changed the name of dictionary's keyword argument from "mapping" to "x". Got a better name? "mapping_or_sequence_of_pairs" isn't attractive, although more so than "mosop" <wink>. abstract.h, abstract.tex: Added new PySequence_Fast_GET_SIZE function, much faster than going thru the all-purpose PySequence_Size. libfuncs.tex: - Document dictionary(). - Fiddle tuple() and list() to admit that their argument is optional. - The long-winded repetitions of "a sequence, a container that supports iteration, or an iterator object" is getting to be a PITA. Many months ago I suggested factoring this out into "iterable object", where the definition of that could include being explicit about generators too (as is, I'm not sure a reader outside of PythonLabs could guess that "an iterator object" includes a generator call). - Please check my curly braces -- I'm going blind <0.9 wink>. abstract.c, PySequence_Tuple(): When PyObject_GetIter() fails, leave its error msg alone now (the msg it produces has improved since PySequence_Tuple was generalized to accept iterable objects, and PySequence_Tuple was also stomping on the msg in cases it shouldn't have even before PyObject_GetIter grew a better msg).
2001-10-26 02:06:50 -03:00
return result;
}
static long
dict_nohash(PyObject *self)
{
PyErr_SetString(PyExc_TypeError, "dict objects are unhashable");
return -1;
}
static PyObject *
dict_iter(dictobject *dict)
{
return dictiter_new(dict, select_key);
}
static char dictionary_doc[] =
"dict() -> new empty dictionary.\n"
"dict(mapping) -> new dictionary initialized from a mapping object's\n"
Generalize dictionary() to accept a sequence of 2-sequences. At the outer level, the iterator protocol is used for memory-efficiency (the outer sequence may be very large if fully materialized); at the inner level, PySequence_Fast() is used for time-efficiency (these should always be sequences of length 2). dictobject.c, new functions PyDict_{Merge,Update}FromSeq2. These are wholly analogous to PyDict_{Merge,Update}, but process a sequence-of-2- sequences argument instead of a mapping object. For now, I left these functions file static, so no corresponding doc changes. It's tempting to change dict.update() to allow a sequence-of-2-seqs argument too. Also changed the name of dictionary's keyword argument from "mapping" to "x". Got a better name? "mapping_or_sequence_of_pairs" isn't attractive, although more so than "mosop" <wink>. abstract.h, abstract.tex: Added new PySequence_Fast_GET_SIZE function, much faster than going thru the all-purpose PySequence_Size. libfuncs.tex: - Document dictionary(). - Fiddle tuple() and list() to admit that their argument is optional. - The long-winded repetitions of "a sequence, a container that supports iteration, or an iterator object" is getting to be a PITA. Many months ago I suggested factoring this out into "iterable object", where the definition of that could include being explicit about generators too (as is, I'm not sure a reader outside of PythonLabs could guess that "an iterator object" includes a generator call). - Please check my curly braces -- I'm going blind <0.9 wink>. abstract.c, PySequence_Tuple(): When PyObject_GetIter() fails, leave its error msg alone now (the msg it produces has improved since PySequence_Tuple was generalized to accept iterable objects, and PySequence_Tuple was also stomping on the msg in cases it shouldn't have even before PyObject_GetIter grew a better msg).
2001-10-26 02:06:50 -03:00
" (key, value) pairs.\n"
"dict(seq) -> new dictionary initialized as if via:\n"
" d = {}\n"
" for k, v in seq:\n"
" d[k] = v";
1997-05-02 00:12:38 -03:00
PyTypeObject PyDict_Type = {
PyObject_HEAD_INIT(&PyType_Type)
0,
"dict",
2001-08-29 20:54:21 -03:00
sizeof(dictobject),
0,
(destructor)dict_dealloc, /* tp_dealloc */
(printfunc)dict_print, /* tp_print */
2001-08-02 01:15:00 -03:00
0, /* tp_getattr */
0, /* tp_setattr */
(cmpfunc)dict_compare, /* tp_compare */
(reprfunc)dict_repr, /* tp_repr */
0, /* tp_as_number */
&dict_as_sequence, /* tp_as_sequence */
&dict_as_mapping, /* tp_as_mapping */
dict_nohash, /* tp_hash */
0, /* tp_call */
0, /* tp_str */
2001-08-02 01:15:00 -03:00
PyObject_GenericGetAttr, /* tp_getattro */
0, /* tp_setattro */
0, /* tp_as_buffer */
2001-08-29 20:54:21 -03:00
Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
2001-08-02 01:15:00 -03:00
Py_TPFLAGS_BASETYPE, /* tp_flags */
dictionary_doc, /* tp_doc */
(traverseproc)dict_traverse, /* tp_traverse */
(inquiry)dict_tp_clear, /* tp_clear */
dict_richcompare, /* tp_richcompare */
0, /* tp_weaklistoffset */
(getiterfunc)dict_iter, /* tp_iter */
0, /* tp_iternext */
2001-08-02 01:15:00 -03:00
mapp_methods, /* tp_methods */
0, /* tp_members */
0, /* tp_getset */
0, /* tp_base */
0, /* tp_dict */
0, /* tp_descr_get */
0, /* tp_descr_set */
0, /* tp_dictoffset */
(initproc)dict_init, /* tp_init */
2001-08-02 01:15:00 -03:00
PyType_GenericAlloc, /* tp_alloc */
dict_new, /* tp_new */
_PyObject_GC_Del, /* tp_free */
};
/* For backward compatibility with old dictionary interface */
1997-05-02 00:12:38 -03:00
PyObject *
PyDict_GetItemString(PyObject *v, char *key)
{
PyObject *kv, *rv;
kv = PyString_FromString(key);
if (kv == NULL)
return NULL;
rv = PyDict_GetItem(v, kv);
Py_DECREF(kv);
return rv;
}
int
PyDict_SetItemString(PyObject *v, char *key, PyObject *item)
{
PyObject *kv;
int err;
kv = PyString_FromString(key);
if (kv == NULL)
return -1;
PyString_InternInPlace(&kv); /* XXX Should we really? */
err = PyDict_SetItem(v, kv, item);
Py_DECREF(kv);
return err;
}
int
PyDict_DelItemString(PyObject *v, char *key)
{
PyObject *kv;
int err;
kv = PyString_FromString(key);
if (kv == NULL)
return -1;
err = PyDict_DelItem(v, kv);
Py_DECREF(kv);
return err;
}
/* Dictionary iterator type */
extern PyTypeObject PyDictIter_Type; /* Forward */
typedef struct {
PyObject_HEAD
dictobject *di_dict;
int di_used;
int di_pos;
binaryfunc di_select;
} dictiterobject;
static PyObject *
dictiter_new(dictobject *dict, binaryfunc select)
{
dictiterobject *di;
di = PyObject_NEW(dictiterobject, &PyDictIter_Type);
if (di == NULL)
return NULL;
Py_INCREF(dict);
di->di_dict = dict;
di->di_used = dict->ma_used;
di->di_pos = 0;
di->di_select = select;
return (PyObject *)di;
}
static void
dictiter_dealloc(dictiterobject *di)
{
Py_DECREF(di->di_dict);
PyObject_DEL(di);
}
static PyObject *
dictiter_next(dictiterobject *di, PyObject *args)
{
PyObject *key, *value;
if (di->di_used != di->di_dict->ma_used) {
PyErr_SetString(PyExc_RuntimeError,
"dictionary changed size during iteration");
return NULL;
}
if (PyDict_Next((PyObject *)(di->di_dict), &di->di_pos, &key, &value)) {
return (*di->di_select)(key, value);
}
PyErr_SetObject(PyExc_StopIteration, Py_None);
return NULL;
}
static PyObject *
dictiter_getiter(PyObject *it)
{
Py_INCREF(it);
return it;
}
static PyMethodDef dictiter_methods[] = {
{"next", (PyCFunction)dictiter_next, METH_VARARGS,
"it.next() -- get the next value, or raise StopIteration"},
{NULL, NULL} /* sentinel */
};
static PyObject *dictiter_iternext(dictiterobject *di)
{
PyObject *key, *value;
if (di->di_used != di->di_dict->ma_used) {
PyErr_SetString(PyExc_RuntimeError,
"dictionary changed size during iteration");
return NULL;
}
if (PyDict_Next((PyObject *)(di->di_dict), &di->di_pos, &key, &value)) {
return (*di->di_select)(key, value);
}
return NULL;
}
PyTypeObject PyDictIter_Type = {
PyObject_HEAD_INIT(&PyType_Type)
0, /* ob_size */
"dictionary-iterator", /* tp_name */
sizeof(dictiterobject), /* tp_basicsize */
0, /* tp_itemsize */
/* methods */
(destructor)dictiter_dealloc, /* tp_dealloc */
0, /* tp_print */
2001-08-02 01:15:00 -03:00
0, /* tp_getattr */
0, /* tp_setattr */
0, /* tp_compare */
0, /* tp_repr */
0, /* tp_as_number */
0, /* tp_as_sequence */
0, /* tp_as_mapping */
0, /* tp_hash */
0, /* tp_call */
0, /* tp_str */
2001-08-02 01:15:00 -03:00
PyObject_GenericGetAttr, /* tp_getattro */
0, /* tp_setattro */
0, /* tp_as_buffer */
Py_TPFLAGS_DEFAULT, /* tp_flags */
0, /* tp_doc */
0, /* tp_traverse */
0, /* tp_clear */
0, /* tp_richcompare */
0, /* tp_weaklistoffset */
(getiterfunc)dictiter_getiter, /* tp_iter */
(iternextfunc)dictiter_iternext, /* tp_iternext */
2001-08-02 01:15:00 -03:00
dictiter_methods, /* tp_methods */
0, /* tp_members */
0, /* tp_getset */
0, /* tp_base */
0, /* tp_dict */
0, /* tp_descr_get */
0, /* tp_descr_set */
};