2004-01-17 10:29:29 -04:00
|
|
|
/*
|
|
|
|
* multibytecodec.c: Common Multibyte Codec Implementation
|
|
|
|
*
|
|
|
|
* Written by Hye-Shik Chang <perky@FreeBSD.org>
|
|
|
|
*/
|
|
|
|
|
2006-03-04 12:08:19 -04:00
|
|
|
#define PY_SSIZE_T_CLEAN
|
2004-01-17 10:29:29 -04:00
|
|
|
#include "Python.h"
|
2006-04-21 07:40:58 -03:00
|
|
|
#include "structmember.h"
|
2004-01-17 10:29:29 -04:00
|
|
|
#include "multibytecodec.h"
|
2014-08-22 12:45:03 -03:00
|
|
|
#include "clinic/multibytecodec.c.h"
|
|
|
|
|
|
|
|
/*[clinic input]
|
|
|
|
module _multibytecodec
|
|
|
|
class _multibytecodec.MultibyteCodec "MultibyteCodecObject *" "&MultibyteCodec_Type"
|
|
|
|
[clinic start generated code]*/
|
2015-04-03 17:53:51 -03:00
|
|
|
/*[clinic end generated code: output=da39a3ee5e6b4b0d input=6ad689546cbb5450]*/
|
2004-01-17 10:29:29 -04:00
|
|
|
|
|
|
|
typedef struct {
|
2013-04-13 21:06:32 -03:00
|
|
|
PyObject *inobj;
|
|
|
|
Py_ssize_t inpos, inlen;
|
2004-01-17 10:29:29 -04:00
|
|
|
unsigned char *outbuf, *outbuf_end;
|
|
|
|
PyObject *excobj, *outobj;
|
|
|
|
} MultibyteEncodeBuffer;
|
|
|
|
|
|
|
|
typedef struct {
|
|
|
|
const unsigned char *inbuf, *inbuf_top, *inbuf_end;
|
2013-04-11 17:09:04 -03:00
|
|
|
PyObject *excobj;
|
|
|
|
_PyUnicodeWriter writer;
|
2004-01-17 10:29:29 -04:00
|
|
|
} MultibyteDecodeBuffer;
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static char *incnewkwarglist[] = {"errors", NULL};
|
2006-02-27 13:20:04 -04:00
|
|
|
static char *streamkwarglist[] = {"stream", "errors", NULL};
|
2004-01-17 10:29:29 -04:00
|
|
|
|
|
|
|
static PyObject *multibytecodec_encode(MultibyteCodec *,
|
2013-04-13 21:06:32 -03:00
|
|
|
MultibyteCodec_State *, PyObject *, Py_ssize_t *,
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject *, int);
|
2004-01-17 10:29:29 -04:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
#define MBENC_RESET MBENC_MAX<<1 /* reset after an encoding session */
|
2004-01-17 10:29:29 -04:00
|
|
|
|
2013-11-12 16:39:02 -04:00
|
|
|
_Py_IDENTIFIER(write);
|
|
|
|
|
2004-01-17 10:29:29 -04:00
|
|
|
static PyObject *
|
2006-03-04 12:08:19 -04:00
|
|
|
make_tuple(PyObject *object, Py_ssize_t len)
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject *v, *w;
|
|
|
|
|
|
|
|
if (object == NULL)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
v = PyTuple_New(2);
|
|
|
|
if (v == NULL) {
|
|
|
|
Py_DECREF(object);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
PyTuple_SET_ITEM(v, 0, object);
|
|
|
|
|
|
|
|
w = PyLong_FromSsize_t(len);
|
|
|
|
if (w == NULL) {
|
|
|
|
Py_DECREF(v);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
PyTuple_SET_ITEM(v, 1, w);
|
|
|
|
|
|
|
|
return v;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
static PyObject *
|
2006-04-21 07:40:58 -03:00
|
|
|
internal_error_callback(const char *errors)
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
if (errors == NULL || strcmp(errors, "strict") == 0)
|
|
|
|
return ERROR_STRICT;
|
|
|
|
else if (strcmp(errors, "ignore") == 0)
|
|
|
|
return ERROR_IGNORE;
|
|
|
|
else if (strcmp(errors, "replace") == 0)
|
|
|
|
return ERROR_REPLACE;
|
|
|
|
else
|
|
|
|
return PyUnicode_FromString(errors);
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
|
|
|
static PyObject *
|
|
|
|
call_error_callback(PyObject *errors, PyObject *exc)
|
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject *args, *cb, *r;
|
|
|
|
const char *str;
|
|
|
|
|
|
|
|
assert(PyUnicode_Check(errors));
|
2016-11-20 03:13:07 -04:00
|
|
|
str = PyUnicode_AsUTF8(errors);
|
2010-05-09 12:52:27 -03:00
|
|
|
if (str == NULL)
|
|
|
|
return NULL;
|
|
|
|
cb = PyCodec_LookupError(str);
|
|
|
|
if (cb == NULL)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
args = PyTuple_New(1);
|
|
|
|
if (args == NULL) {
|
|
|
|
Py_DECREF(cb);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
PyTuple_SET_ITEM(args, 0, exc);
|
|
|
|
Py_INCREF(exc);
|
|
|
|
|
|
|
|
r = PyObject_CallObject(cb, args);
|
|
|
|
Py_DECREF(args);
|
|
|
|
Py_DECREF(cb);
|
|
|
|
return r;
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
|
|
|
static PyObject *
|
|
|
|
codecctx_errors_get(MultibyteStatefulCodecContext *self)
|
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
const char *errors;
|
|
|
|
|
|
|
|
if (self->errors == ERROR_STRICT)
|
|
|
|
errors = "strict";
|
|
|
|
else if (self->errors == ERROR_IGNORE)
|
|
|
|
errors = "ignore";
|
|
|
|
else if (self->errors == ERROR_REPLACE)
|
|
|
|
errors = "replace";
|
|
|
|
else {
|
|
|
|
Py_INCREF(self->errors);
|
|
|
|
return self->errors;
|
|
|
|
}
|
|
|
|
|
|
|
|
return PyUnicode_FromString(errors);
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
codecctx_errors_set(MultibyteStatefulCodecContext *self, PyObject *value,
|
2010-05-09 12:52:27 -03:00
|
|
|
void *closure)
|
2006-04-21 07:40:58 -03:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject *cb;
|
|
|
|
const char *str;
|
2006-04-21 07:40:58 -03:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
if (!PyUnicode_Check(value)) {
|
|
|
|
PyErr_SetString(PyExc_TypeError, "errors must be a string");
|
|
|
|
return -1;
|
|
|
|
}
|
2006-04-21 07:40:58 -03:00
|
|
|
|
2016-11-20 03:13:07 -04:00
|
|
|
str = PyUnicode_AsUTF8(value);
|
2010-05-09 12:52:27 -03:00
|
|
|
if (str == NULL)
|
|
|
|
return -1;
|
2007-08-26 01:19:43 -03:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
cb = internal_error_callback(str);
|
|
|
|
if (cb == NULL)
|
|
|
|
return -1;
|
2006-04-21 07:40:58 -03:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
ERROR_DECREF(self->errors);
|
|
|
|
self->errors = cb;
|
|
|
|
return 0;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
/* This getset handlers list is used by all the stateful codec objects */
|
|
|
|
static PyGetSetDef codecctx_getsets[] = {
|
2010-05-09 12:52:27 -03:00
|
|
|
{"errors", (getter)codecctx_errors_get,
|
|
|
|
(setter)codecctx_errors_set,
|
|
|
|
PyDoc_STR("how to treat errors")},
|
|
|
|
{NULL,}
|
2006-04-21 07:40:58 -03:00
|
|
|
};
|
|
|
|
|
2004-01-17 10:29:29 -04:00
|
|
|
static int
|
2006-03-04 12:08:19 -04:00
|
|
|
expand_encodebuffer(MultibyteEncodeBuffer *buf, Py_ssize_t esize)
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_ssize_t orgpos, orgsize, incsize;
|
2004-01-17 10:29:29 -04:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
orgpos = (Py_ssize_t)((char *)buf->outbuf -
|
|
|
|
PyBytes_AS_STRING(buf->outobj));
|
|
|
|
orgsize = PyBytes_GET_SIZE(buf->outobj);
|
|
|
|
incsize = (esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize);
|
2008-06-17 21:47:36 -03:00
|
|
|
|
2015-02-20 19:19:58 -04:00
|
|
|
if (orgsize > PY_SSIZE_T_MAX - incsize) {
|
|
|
|
PyErr_NoMemory();
|
2010-05-09 12:52:27 -03:00
|
|
|
return -1;
|
2015-02-20 19:19:58 -04:00
|
|
|
}
|
2008-06-17 21:47:36 -03:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
if (_PyBytes_Resize(&buf->outobj, orgsize + incsize) == -1)
|
|
|
|
return -1;
|
2004-01-17 10:29:29 -04:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
buf->outbuf = (unsigned char *)PyBytes_AS_STRING(buf->outobj) +orgpos;
|
|
|
|
buf->outbuf_end = (unsigned char *)PyBytes_AS_STRING(buf->outobj)
|
|
|
|
+ PyBytes_GET_SIZE(buf->outobj);
|
2004-01-17 10:29:29 -04:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
return 0;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
2015-02-20 19:19:58 -04:00
|
|
|
#define REQUIRE_ENCODEBUFFER(buf, s) do { \
|
|
|
|
if ((s) < 0 || (s) > (buf)->outbuf_end - (buf)->outbuf) \
|
2010-05-09 12:52:27 -03:00
|
|
|
if (expand_encodebuffer(buf, s) == -1) \
|
|
|
|
goto errorexit; \
|
2015-02-20 19:19:58 -04:00
|
|
|
} while(0)
|
2004-01-17 10:29:29 -04:00
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
|
|
|
|
/**
|
|
|
|
* MultibyteCodec object
|
|
|
|
*/
|
|
|
|
|
2004-01-17 10:29:29 -04:00
|
|
|
static int
|
|
|
|
multibytecodec_encerror(MultibyteCodec *codec,
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteCodec_State *state,
|
|
|
|
MultibyteEncodeBuffer *buf,
|
|
|
|
PyObject *errors, Py_ssize_t e)
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject *retobj = NULL, *retstr = NULL, *tobj;
|
|
|
|
Py_ssize_t retstrsize, newpos;
|
|
|
|
Py_ssize_t esize, start, end;
|
|
|
|
const char *reason;
|
|
|
|
|
|
|
|
if (e > 0) {
|
|
|
|
reason = "illegal multibyte sequence";
|
|
|
|
esize = e;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
switch (e) {
|
|
|
|
case MBERR_TOOSMALL:
|
|
|
|
REQUIRE_ENCODEBUFFER(buf, -1);
|
|
|
|
return 0; /* retry it */
|
|
|
|
case MBERR_TOOFEW:
|
|
|
|
reason = "incomplete multibyte sequence";
|
2013-04-13 21:06:32 -03:00
|
|
|
esize = (Py_ssize_t)buf->inpos;
|
2010-05-09 12:52:27 -03:00
|
|
|
break;
|
|
|
|
case MBERR_INTERNAL:
|
|
|
|
PyErr_SetString(PyExc_RuntimeError,
|
|
|
|
"internal codec error");
|
|
|
|
return -1;
|
|
|
|
default:
|
|
|
|
PyErr_SetString(PyExc_RuntimeError,
|
|
|
|
"unknown runtime error");
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (errors == ERROR_REPLACE) {
|
2013-04-13 21:06:32 -03:00
|
|
|
PyObject *replchar;
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_ssize_t r;
|
2013-04-13 21:06:32 -03:00
|
|
|
Py_ssize_t inpos;
|
|
|
|
int kind;
|
|
|
|
void *data;
|
|
|
|
|
|
|
|
replchar = PyUnicode_FromOrdinal('?');
|
|
|
|
if (replchar == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
kind = PyUnicode_KIND(replchar);
|
|
|
|
data = PyUnicode_DATA(replchar);
|
2010-05-09 12:52:27 -03:00
|
|
|
|
2013-04-13 21:06:32 -03:00
|
|
|
inpos = 0;
|
2010-05-09 12:52:27 -03:00
|
|
|
for (;;) {
|
2013-04-13 21:06:32 -03:00
|
|
|
Py_ssize_t outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf);
|
2010-05-09 12:52:27 -03:00
|
|
|
|
2013-04-13 21:06:32 -03:00
|
|
|
r = codec->encode(state, codec->config,
|
|
|
|
kind, data, &inpos, 1,
|
2010-05-09 12:52:27 -03:00
|
|
|
&buf->outbuf, outleft, 0);
|
|
|
|
if (r == MBERR_TOOSMALL) {
|
|
|
|
REQUIRE_ENCODEBUFFER(buf, -1);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2013-04-13 21:06:32 -03:00
|
|
|
Py_DECREF(replchar);
|
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
if (r != 0) {
|
|
|
|
REQUIRE_ENCODEBUFFER(buf, 1);
|
|
|
|
*buf->outbuf++ = '?';
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) {
|
2013-04-13 21:06:32 -03:00
|
|
|
buf->inpos += esize;
|
2010-05-09 12:52:27 -03:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2013-04-13 21:06:32 -03:00
|
|
|
start = (Py_ssize_t)buf->inpos;
|
2010-05-09 12:52:27 -03:00
|
|
|
end = start + esize;
|
|
|
|
|
|
|
|
/* use cached exception object if available */
|
|
|
|
if (buf->excobj == NULL) {
|
2013-04-13 21:06:32 -03:00
|
|
|
buf->excobj = PyObject_CallFunction(PyExc_UnicodeEncodeError,
|
|
|
|
"sOnns",
|
|
|
|
codec->encoding, buf->inobj,
|
|
|
|
start, end, reason);
|
2010-05-09 12:52:27 -03:00
|
|
|
if (buf->excobj == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
if (PyUnicodeEncodeError_SetStart(buf->excobj, start) != 0 ||
|
|
|
|
PyUnicodeEncodeError_SetEnd(buf->excobj, end) != 0 ||
|
|
|
|
PyUnicodeEncodeError_SetReason(buf->excobj, reason) != 0)
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
if (errors == ERROR_STRICT) {
|
|
|
|
PyCodec_StrictErrors(buf->excobj);
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
|
|
|
retobj = call_error_callback(errors, buf->excobj);
|
|
|
|
if (retobj == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 ||
|
2012-12-02 12:20:28 -04:00
|
|
|
(!PyUnicode_Check((tobj = PyTuple_GET_ITEM(retobj, 0))) && !PyBytes_Check(tobj)) ||
|
2010-05-09 12:52:27 -03:00
|
|
|
!PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) {
|
|
|
|
PyErr_SetString(PyExc_TypeError,
|
|
|
|
"encoding error handler must return "
|
2012-12-02 11:53:41 -04:00
|
|
|
"(str, int) tuple");
|
2010-05-09 12:52:27 -03:00
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
2012-12-02 12:20:28 -04:00
|
|
|
if (PyUnicode_Check(tobj)) {
|
2013-04-13 21:06:32 -03:00
|
|
|
Py_ssize_t inpos;
|
2010-05-09 12:52:27 -03:00
|
|
|
|
2013-04-13 21:06:32 -03:00
|
|
|
retstr = multibytecodec_encode(codec, state, tobj,
|
|
|
|
&inpos, ERROR_STRICT,
|
2010-05-09 12:52:27 -03:00
|
|
|
MBENC_FLUSH);
|
|
|
|
if (retstr == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
}
|
2012-12-02 12:20:28 -04:00
|
|
|
else {
|
|
|
|
Py_INCREF(tobj);
|
|
|
|
retstr = tobj;
|
|
|
|
}
|
2010-05-09 12:52:27 -03:00
|
|
|
|
|
|
|
assert(PyBytes_Check(retstr));
|
|
|
|
retstrsize = PyBytes_GET_SIZE(retstr);
|
2015-02-20 19:19:58 -04:00
|
|
|
if (retstrsize > 0) {
|
|
|
|
REQUIRE_ENCODEBUFFER(buf, retstrsize);
|
|
|
|
memcpy(buf->outbuf, PyBytes_AS_STRING(retstr), retstrsize);
|
|
|
|
buf->outbuf += retstrsize;
|
|
|
|
}
|
2010-05-09 12:52:27 -03:00
|
|
|
|
|
|
|
newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1));
|
|
|
|
if (newpos < 0 && !PyErr_Occurred())
|
2013-04-13 21:06:32 -03:00
|
|
|
newpos += (Py_ssize_t)buf->inlen;
|
|
|
|
if (newpos < 0 || newpos > buf->inlen) {
|
2010-05-09 12:52:27 -03:00
|
|
|
PyErr_Clear();
|
|
|
|
PyErr_Format(PyExc_IndexError,
|
|
|
|
"position %zd from error handler out of bounds",
|
|
|
|
newpos);
|
|
|
|
goto errorexit;
|
|
|
|
}
|
2013-04-13 21:06:32 -03:00
|
|
|
buf->inpos = newpos;
|
2010-05-09 12:52:27 -03:00
|
|
|
|
|
|
|
Py_DECREF(retobj);
|
|
|
|
Py_DECREF(retstr);
|
|
|
|
return 0;
|
2004-01-17 10:29:29 -04:00
|
|
|
|
|
|
|
errorexit:
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_XDECREF(retobj);
|
|
|
|
Py_XDECREF(retstr);
|
|
|
|
return -1;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
multibytecodec_decerror(MultibyteCodec *codec,
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteCodec_State *state,
|
|
|
|
MultibyteDecodeBuffer *buf,
|
|
|
|
PyObject *errors, Py_ssize_t e)
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject *retobj = NULL, *retuni = NULL;
|
2013-04-11 17:09:04 -03:00
|
|
|
Py_ssize_t newpos;
|
2010-05-09 12:52:27 -03:00
|
|
|
const char *reason;
|
|
|
|
Py_ssize_t esize, start, end;
|
|
|
|
|
|
|
|
if (e > 0) {
|
|
|
|
reason = "illegal multibyte sequence";
|
|
|
|
esize = e;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
switch (e) {
|
|
|
|
case MBERR_TOOSMALL:
|
|
|
|
return 0; /* retry it */
|
|
|
|
case MBERR_TOOFEW:
|
|
|
|
reason = "incomplete multibyte sequence";
|
|
|
|
esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
|
|
|
|
break;
|
|
|
|
case MBERR_INTERNAL:
|
|
|
|
PyErr_SetString(PyExc_RuntimeError,
|
|
|
|
"internal codec error");
|
|
|
|
return -1;
|
2013-07-16 16:41:43 -03:00
|
|
|
case MBERR_EXCEPTION:
|
|
|
|
return -1;
|
2010-05-09 12:52:27 -03:00
|
|
|
default:
|
|
|
|
PyErr_SetString(PyExc_RuntimeError,
|
|
|
|
"unknown runtime error");
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (errors == ERROR_REPLACE) {
|
2013-04-11 17:09:04 -03:00
|
|
|
if (_PyUnicodeWriter_WriteChar(&buf->writer,
|
|
|
|
Py_UNICODE_REPLACEMENT_CHARACTER) < 0)
|
|
|
|
goto errorexit;
|
2010-05-09 12:52:27 -03:00
|
|
|
}
|
|
|
|
if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) {
|
|
|
|
buf->inbuf += esize;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top);
|
|
|
|
end = start + esize;
|
|
|
|
|
|
|
|
/* use cached exception object if available */
|
|
|
|
if (buf->excobj == NULL) {
|
|
|
|
buf->excobj = PyUnicodeDecodeError_Create(codec->encoding,
|
|
|
|
(const char *)buf->inbuf_top,
|
|
|
|
(Py_ssize_t)(buf->inbuf_end - buf->inbuf_top),
|
|
|
|
start, end, reason);
|
|
|
|
if (buf->excobj == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
if (PyUnicodeDecodeError_SetStart(buf->excobj, start) ||
|
|
|
|
PyUnicodeDecodeError_SetEnd(buf->excobj, end) ||
|
|
|
|
PyUnicodeDecodeError_SetReason(buf->excobj, reason))
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
if (errors == ERROR_STRICT) {
|
|
|
|
PyCodec_StrictErrors(buf->excobj);
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
|
|
|
retobj = call_error_callback(errors, buf->excobj);
|
|
|
|
if (retobj == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 ||
|
|
|
|
!PyUnicode_Check((retuni = PyTuple_GET_ITEM(retobj, 0))) ||
|
|
|
|
!PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) {
|
|
|
|
PyErr_SetString(PyExc_TypeError,
|
|
|
|
"decoding error handler must return "
|
2012-12-02 11:53:41 -04:00
|
|
|
"(str, int) tuple");
|
2010-05-09 12:52:27 -03:00
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
2013-04-11 17:09:04 -03:00
|
|
|
if (_PyUnicodeWriter_WriteStr(&buf->writer, retuni) < 0)
|
2011-11-20 22:01:27 -04:00
|
|
|
goto errorexit;
|
2010-05-09 12:52:27 -03:00
|
|
|
|
|
|
|
newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1));
|
|
|
|
if (newpos < 0 && !PyErr_Occurred())
|
|
|
|
newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top);
|
|
|
|
if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) {
|
|
|
|
PyErr_Clear();
|
|
|
|
PyErr_Format(PyExc_IndexError,
|
|
|
|
"position %zd from error handler out of bounds",
|
|
|
|
newpos);
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
buf->inbuf = buf->inbuf_top + newpos;
|
|
|
|
Py_DECREF(retobj);
|
|
|
|
return 0;
|
2004-01-17 10:29:29 -04:00
|
|
|
|
|
|
|
errorexit:
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_XDECREF(retobj);
|
|
|
|
return -1;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
static PyObject *
|
|
|
|
multibytecodec_encode(MultibyteCodec *codec,
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteCodec_State *state,
|
2013-04-13 21:06:32 -03:00
|
|
|
PyObject *text, Py_ssize_t *inpos_t,
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject *errors, int flags)
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteEncodeBuffer buf;
|
|
|
|
Py_ssize_t finalsize, r = 0;
|
2013-04-13 21:06:32 -03:00
|
|
|
Py_ssize_t datalen;
|
|
|
|
int kind;
|
|
|
|
void *data;
|
|
|
|
|
|
|
|
if (PyUnicode_READY(text) < 0)
|
|
|
|
return NULL;
|
|
|
|
datalen = PyUnicode_GET_LENGTH(text);
|
2010-05-09 12:52:27 -03:00
|
|
|
|
2011-05-24 17:17:55 -03:00
|
|
|
if (datalen == 0 && !(flags & MBENC_RESET))
|
2010-05-09 12:52:27 -03:00
|
|
|
return PyBytes_FromStringAndSize(NULL, 0);
|
|
|
|
|
|
|
|
buf.excobj = NULL;
|
2011-02-22 16:15:44 -04:00
|
|
|
buf.outobj = NULL;
|
2013-04-13 21:06:32 -03:00
|
|
|
buf.inobj = text; /* borrowed reference */
|
|
|
|
buf.inpos = 0;
|
|
|
|
buf.inlen = datalen;
|
|
|
|
kind = PyUnicode_KIND(buf.inobj);
|
|
|
|
data = PyUnicode_DATA(buf.inobj);
|
2010-05-09 12:52:27 -03:00
|
|
|
|
|
|
|
if (datalen > (PY_SSIZE_T_MAX - 16) / 2) {
|
|
|
|
PyErr_NoMemory();
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
|
|
|
buf.outobj = PyBytes_FromStringAndSize(NULL, datalen * 2 + 16);
|
|
|
|
if (buf.outobj == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
buf.outbuf = (unsigned char *)PyBytes_AS_STRING(buf.outobj);
|
|
|
|
buf.outbuf_end = buf.outbuf + PyBytes_GET_SIZE(buf.outobj);
|
|
|
|
|
2013-04-13 21:06:32 -03:00
|
|
|
while (buf.inpos < buf.inlen) {
|
2010-05-09 12:52:27 -03:00
|
|
|
/* we don't reuse inleft and outleft here.
|
|
|
|
* error callbacks can relocate the cursor anywhere on buffer*/
|
2013-04-13 21:06:32 -03:00
|
|
|
Py_ssize_t outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
|
|
|
|
|
|
|
|
r = codec->encode(state, codec->config,
|
|
|
|
kind, data,
|
|
|
|
&buf.inpos, buf.inlen,
|
2010-05-09 12:52:27 -03:00
|
|
|
&buf.outbuf, outleft, flags);
|
|
|
|
if ((r == 0) || (r == MBERR_TOOFEW && !(flags & MBENC_FLUSH)))
|
|
|
|
break;
|
|
|
|
else if (multibytecodec_encerror(codec, state, &buf, errors,r))
|
|
|
|
goto errorexit;
|
|
|
|
else if (r == MBERR_TOOFEW)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2011-05-24 17:17:55 -03:00
|
|
|
if (codec->encreset != NULL && (flags & MBENC_RESET))
|
2010-05-09 12:52:27 -03:00
|
|
|
for (;;) {
|
|
|
|
Py_ssize_t outleft;
|
|
|
|
|
|
|
|
outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
|
|
|
|
r = codec->encreset(state, codec->config, &buf.outbuf,
|
|
|
|
outleft);
|
|
|
|
if (r == 0)
|
|
|
|
break;
|
|
|
|
else if (multibytecodec_encerror(codec, state,
|
|
|
|
&buf, errors, r))
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
|
|
|
finalsize = (Py_ssize_t)((char *)buf.outbuf -
|
|
|
|
PyBytes_AS_STRING(buf.outobj));
|
|
|
|
|
|
|
|
if (finalsize != PyBytes_GET_SIZE(buf.outobj))
|
|
|
|
if (_PyBytes_Resize(&buf.outobj, finalsize) == -1)
|
|
|
|
goto errorexit;
|
|
|
|
|
2013-04-13 21:06:32 -03:00
|
|
|
if (inpos_t)
|
|
|
|
*inpos_t = buf.inpos;
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_XDECREF(buf.excobj);
|
|
|
|
return buf.outobj;
|
2004-01-17 10:29:29 -04:00
|
|
|
|
|
|
|
errorexit:
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_XDECREF(buf.excobj);
|
|
|
|
Py_XDECREF(buf.outobj);
|
|
|
|
return NULL;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
_multibytecodec.MultibyteCodec.encode
|
|
|
|
|
|
|
|
input: object
|
2015-05-04 10:59:46 -03:00
|
|
|
errors: str(accept={str, NoneType}) = NULL
|
2014-08-22 12:45:03 -03:00
|
|
|
|
|
|
|
Return an encoded string version of `input'.
|
|
|
|
|
|
|
|
'errors' may be given to set a different error handling scheme. Default is
|
|
|
|
'strict' meaning that encoding errors raise a UnicodeEncodeError. Other possible
|
|
|
|
values are 'ignore', 'replace' and 'xmlcharrefreplace' as well as any other name
|
|
|
|
registered with codecs.register_error that can handle UnicodeEncodeErrors.
|
|
|
|
[clinic start generated code]*/
|
|
|
|
|
2004-01-17 10:29:29 -04:00
|
|
|
static PyObject *
|
2015-04-14 19:07:59 -03:00
|
|
|
_multibytecodec_MultibyteCodec_encode_impl(MultibyteCodecObject *self,
|
|
|
|
PyObject *input,
|
|
|
|
const char *errors)
|
2015-05-04 10:59:46 -03:00
|
|
|
/*[clinic end generated code: output=7b26652045ba56a9 input=05f6ced3c8dd0582]*/
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteCodec_State state;
|
2014-08-22 12:45:03 -03:00
|
|
|
PyObject *errorcb, *r, *ucvt;
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_ssize_t datalen;
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
if (PyUnicode_Check(input))
|
2010-05-09 12:52:27 -03:00
|
|
|
ucvt = NULL;
|
|
|
|
else {
|
2014-08-22 12:45:03 -03:00
|
|
|
input = ucvt = PyObject_Str(input);
|
|
|
|
if (input == NULL)
|
2010-05-09 12:52:27 -03:00
|
|
|
return NULL;
|
2014-08-22 12:45:03 -03:00
|
|
|
else if (!PyUnicode_Check(input)) {
|
2010-05-09 12:52:27 -03:00
|
|
|
PyErr_SetString(PyExc_TypeError,
|
|
|
|
"couldn't convert the object to unicode.");
|
|
|
|
Py_DECREF(ucvt);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
if (PyUnicode_READY(input) < 0) {
|
2011-11-20 21:50:14 -04:00
|
|
|
Py_XDECREF(ucvt);
|
|
|
|
return NULL;
|
|
|
|
}
|
2014-08-22 12:45:03 -03:00
|
|
|
datalen = PyUnicode_GET_LENGTH(input);
|
2010-05-09 12:52:27 -03:00
|
|
|
|
|
|
|
errorcb = internal_error_callback(errors);
|
|
|
|
if (errorcb == NULL) {
|
|
|
|
Py_XDECREF(ucvt);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (self->codec->encinit != NULL &&
|
|
|
|
self->codec->encinit(&state, self->codec->config) != 0)
|
|
|
|
goto errorexit;
|
|
|
|
r = multibytecodec_encode(self->codec, &state,
|
2014-08-22 12:45:03 -03:00
|
|
|
input, NULL, errorcb,
|
2010-05-09 12:52:27 -03:00
|
|
|
MBENC_FLUSH | MBENC_RESET);
|
|
|
|
if (r == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
ERROR_DECREF(errorcb);
|
|
|
|
Py_XDECREF(ucvt);
|
|
|
|
return make_tuple(r, datalen);
|
2004-01-17 10:29:29 -04:00
|
|
|
|
|
|
|
errorexit:
|
2010-05-09 12:52:27 -03:00
|
|
|
ERROR_DECREF(errorcb);
|
|
|
|
Py_XDECREF(ucvt);
|
|
|
|
return NULL;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
_multibytecodec.MultibyteCodec.decode
|
|
|
|
|
|
|
|
input: Py_buffer
|
2015-05-04 10:59:46 -03:00
|
|
|
errors: str(accept={str, NoneType}) = NULL
|
2014-08-22 12:45:03 -03:00
|
|
|
|
|
|
|
Decodes 'input'.
|
|
|
|
|
|
|
|
'errors' may be given to set a different error handling scheme. Default is
|
|
|
|
'strict' meaning that encoding errors raise a UnicodeDecodeError. Other possible
|
|
|
|
values are 'ignore' and 'replace' as well as any other name registered with
|
|
|
|
codecs.register_error that is able to handle UnicodeDecodeErrors."
|
|
|
|
[clinic start generated code]*/
|
|
|
|
|
2004-01-17 10:29:29 -04:00
|
|
|
static PyObject *
|
2015-04-14 19:07:59 -03:00
|
|
|
_multibytecodec_MultibyteCodec_decode_impl(MultibyteCodecObject *self,
|
|
|
|
Py_buffer *input,
|
|
|
|
const char *errors)
|
2015-05-04 10:59:46 -03:00
|
|
|
/*[clinic end generated code: output=ff419f65bad6cc77 input=a7d45f87f75e5e02]*/
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteCodec_State state;
|
|
|
|
MultibyteDecodeBuffer buf;
|
2013-04-11 17:09:04 -03:00
|
|
|
PyObject *errorcb, *res;
|
2014-08-22 12:45:03 -03:00
|
|
|
const char *data;
|
2013-04-11 17:09:04 -03:00
|
|
|
Py_ssize_t datalen;
|
2010-05-09 12:52:27 -03:00
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
data = input->buf;
|
|
|
|
datalen = input->len;
|
2010-05-09 12:52:27 -03:00
|
|
|
|
|
|
|
errorcb = internal_error_callback(errors);
|
|
|
|
if (errorcb == NULL) {
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (datalen == 0) {
|
|
|
|
ERROR_DECREF(errorcb);
|
2011-11-30 22:18:59 -04:00
|
|
|
return make_tuple(PyUnicode_New(0, 0), 0);
|
2010-05-09 12:52:27 -03:00
|
|
|
}
|
|
|
|
|
2013-04-17 18:02:17 -03:00
|
|
|
_PyUnicodeWriter_Init(&buf.writer);
|
|
|
|
buf.writer.min_length = datalen;
|
2010-05-09 12:52:27 -03:00
|
|
|
buf.excobj = NULL;
|
|
|
|
buf.inbuf = buf.inbuf_top = (unsigned char *)data;
|
|
|
|
buf.inbuf_end = buf.inbuf_top + datalen;
|
|
|
|
|
|
|
|
if (self->codec->decinit != NULL &&
|
|
|
|
self->codec->decinit(&state, self->codec->config) != 0)
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
while (buf.inbuf < buf.inbuf_end) {
|
2013-04-11 17:09:04 -03:00
|
|
|
Py_ssize_t inleft, r;
|
2010-05-09 12:52:27 -03:00
|
|
|
|
|
|
|
inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
|
|
|
|
|
|
|
|
r = self->codec->decode(&state, self->codec->config,
|
2013-04-11 17:09:04 -03:00
|
|
|
&buf.inbuf, inleft, &buf.writer);
|
2010-05-09 12:52:27 -03:00
|
|
|
if (r == 0)
|
|
|
|
break;
|
|
|
|
else if (multibytecodec_decerror(self->codec, &state,
|
|
|
|
&buf, errorcb, r))
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
2013-04-11 17:09:04 -03:00
|
|
|
res = _PyUnicodeWriter_Finish(&buf.writer);
|
|
|
|
if (res == NULL)
|
|
|
|
goto errorexit;
|
2010-05-09 12:52:27 -03:00
|
|
|
|
|
|
|
Py_XDECREF(buf.excobj);
|
|
|
|
ERROR_DECREF(errorcb);
|
2013-04-11 17:09:04 -03:00
|
|
|
return make_tuple(res, datalen);
|
2004-01-17 10:29:29 -04:00
|
|
|
|
|
|
|
errorexit:
|
2010-05-09 12:52:27 -03:00
|
|
|
ERROR_DECREF(errorcb);
|
|
|
|
Py_XDECREF(buf.excobj);
|
2013-04-11 17:09:04 -03:00
|
|
|
_PyUnicodeWriter_Dealloc(&buf.writer);
|
2004-01-17 10:29:29 -04:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
return NULL;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
static struct PyMethodDef multibytecodec_methods[] = {
|
2014-08-22 12:45:03 -03:00
|
|
|
_MULTIBYTECODEC_MULTIBYTECODEC_ENCODE_METHODDEF
|
|
|
|
_MULTIBYTECODEC_MULTIBYTECODEC_DECODE_METHODDEF
|
|
|
|
{NULL, NULL},
|
2004-01-17 10:29:29 -04:00
|
|
|
};
|
|
|
|
|
|
|
|
static void
|
|
|
|
multibytecodec_dealloc(MultibyteCodecObject *self)
|
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject_Del(self);
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
static PyTypeObject MultibyteCodec_Type = {
|
2010-05-09 12:52:27 -03:00
|
|
|
PyVarObject_HEAD_INIT(NULL, 0)
|
|
|
|
"MultibyteCodec", /* tp_name */
|
|
|
|
sizeof(MultibyteCodecObject), /* tp_basicsize */
|
|
|
|
0, /* tp_itemsize */
|
|
|
|
/* methods */
|
|
|
|
(destructor)multibytecodec_dealloc, /* tp_dealloc */
|
|
|
|
0, /* tp_print */
|
|
|
|
0, /* tp_getattr */
|
|
|
|
0, /* tp_setattr */
|
|
|
|
0, /* tp_reserved */
|
|
|
|
0, /* tp_repr */
|
|
|
|
0, /* tp_as_number */
|
|
|
|
0, /* tp_as_sequence */
|
|
|
|
0, /* tp_as_mapping */
|
|
|
|
0, /* tp_hash */
|
|
|
|
0, /* tp_call */
|
|
|
|
0, /* tp_str */
|
|
|
|
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 */
|
|
|
|
0, /* tp_iter */
|
|
|
|
0, /* tp_iterext */
|
|
|
|
multibytecodec_methods, /* tp_methods */
|
2004-01-17 10:29:29 -04:00
|
|
|
};
|
|
|
|
|
2004-07-18 00:06:29 -03:00
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
/**
|
|
|
|
* Utility functions for stateful codec mechanism
|
|
|
|
*/
|
2004-07-18 00:06:29 -03:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
#define STATEFUL_DCTX(o) ((MultibyteStatefulDecoderContext *)(o))
|
|
|
|
#define STATEFUL_ECTX(o) ((MultibyteStatefulEncoderContext *)(o))
|
2004-07-18 00:06:29 -03:00
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static PyObject *
|
|
|
|
encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx,
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject *unistr, int final)
|
2006-04-21 07:40:58 -03:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject *ucvt, *r = NULL;
|
2013-04-13 21:06:32 -03:00
|
|
|
PyObject *inbuf = NULL;
|
|
|
|
Py_ssize_t inpos, datalen;
|
|
|
|
PyObject *origpending = NULL;
|
2010-05-09 12:52:27 -03:00
|
|
|
|
|
|
|
if (PyUnicode_Check(unistr))
|
|
|
|
ucvt = NULL;
|
|
|
|
else {
|
|
|
|
unistr = ucvt = PyObject_Str(unistr);
|
|
|
|
if (unistr == NULL)
|
|
|
|
return NULL;
|
|
|
|
else if (!PyUnicode_Check(unistr)) {
|
|
|
|
PyErr_SetString(PyExc_TypeError,
|
2012-12-02 11:53:41 -04:00
|
|
|
"couldn't convert the object to str.");
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_DECREF(ucvt);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2013-04-13 21:06:32 -03:00
|
|
|
if (ctx->pending) {
|
|
|
|
PyObject *inbuf_tmp;
|
|
|
|
|
|
|
|
Py_INCREF(ctx->pending);
|
|
|
|
origpending = ctx->pending;
|
|
|
|
|
|
|
|
Py_INCREF(ctx->pending);
|
|
|
|
inbuf_tmp = ctx->pending;
|
|
|
|
PyUnicode_Append(&inbuf_tmp, unistr);
|
2010-05-09 12:52:27 -03:00
|
|
|
if (inbuf_tmp == NULL)
|
|
|
|
goto errorexit;
|
2013-04-13 21:06:32 -03:00
|
|
|
Py_CLEAR(ctx->pending);
|
2010-05-09 12:52:27 -03:00
|
|
|
inbuf = inbuf_tmp;
|
|
|
|
}
|
2013-04-13 21:06:32 -03:00
|
|
|
else {
|
|
|
|
origpending = NULL;
|
2010-05-09 12:52:27 -03:00
|
|
|
|
2013-04-13 21:06:32 -03:00
|
|
|
Py_INCREF(unistr);
|
|
|
|
inbuf = unistr;
|
|
|
|
}
|
|
|
|
if (PyUnicode_READY(inbuf) < 0)
|
|
|
|
goto errorexit;
|
|
|
|
inpos = 0;
|
|
|
|
datalen = PyUnicode_GET_LENGTH(inbuf);
|
2010-05-09 12:52:27 -03:00
|
|
|
|
|
|
|
r = multibytecodec_encode(ctx->codec, &ctx->state,
|
2013-04-13 21:06:32 -03:00
|
|
|
inbuf, &inpos,
|
|
|
|
ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0);
|
2010-05-09 12:52:27 -03:00
|
|
|
if (r == NULL) {
|
|
|
|
/* recover the original pending buffer */
|
2016-04-06 03:45:48 -03:00
|
|
|
Py_XSETREF(ctx->pending, origpending);
|
2013-04-13 21:06:32 -03:00
|
|
|
origpending = NULL;
|
2010-05-09 12:52:27 -03:00
|
|
|
goto errorexit;
|
|
|
|
}
|
2013-04-14 13:11:41 -03:00
|
|
|
Py_XDECREF(origpending);
|
2010-05-09 12:52:27 -03:00
|
|
|
|
2013-04-13 21:06:32 -03:00
|
|
|
if (inpos < datalen) {
|
|
|
|
if (datalen - inpos > MAXENCPENDING) {
|
2010-05-09 12:52:27 -03:00
|
|
|
/* normal codecs can't reach here */
|
|
|
|
PyErr_SetString(PyExc_UnicodeError,
|
|
|
|
"pending buffer overflow");
|
|
|
|
goto errorexit;
|
|
|
|
}
|
2013-04-13 21:06:32 -03:00
|
|
|
ctx->pending = PyUnicode_Substring(inbuf, inpos, datalen);
|
|
|
|
if (ctx->pending == NULL) {
|
|
|
|
/* normal codecs can't reach here */
|
|
|
|
goto errorexit;
|
|
|
|
}
|
2010-05-09 12:52:27 -03:00
|
|
|
}
|
|
|
|
|
2013-04-14 13:11:41 -03:00
|
|
|
Py_DECREF(inbuf);
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_XDECREF(ucvt);
|
|
|
|
return r;
|
2004-01-17 10:29:29 -04:00
|
|
|
|
|
|
|
errorexit:
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_XDECREF(r);
|
|
|
|
Py_XDECREF(ucvt);
|
2013-04-13 21:06:32 -03:00
|
|
|
Py_XDECREF(origpending);
|
2013-04-14 13:11:41 -03:00
|
|
|
Py_XDECREF(inbuf);
|
2010-05-09 12:52:27 -03:00
|
|
|
return NULL;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static int
|
|
|
|
decoder_append_pending(MultibyteStatefulDecoderContext *ctx,
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteDecodeBuffer *buf)
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_ssize_t npendings;
|
|
|
|
|
|
|
|
npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
|
|
|
|
if (npendings + ctx->pendingsize > MAXDECPENDING ||
|
|
|
|
npendings > PY_SSIZE_T_MAX - ctx->pendingsize) {
|
|
|
|
PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow");
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings);
|
|
|
|
ctx->pendingsize += npendings;
|
|
|
|
return 0;
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
2004-07-18 00:06:29 -03:00
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static int
|
|
|
|
decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data,
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_ssize_t size)
|
2006-04-21 07:40:58 -03:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
buf->inbuf = buf->inbuf_top = (const unsigned char *)data;
|
|
|
|
buf->inbuf_end = buf->inbuf_top + size;
|
2013-04-17 18:02:17 -03:00
|
|
|
buf->writer.min_length += size;
|
2010-05-09 12:52:27 -03:00
|
|
|
return 0;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static int
|
|
|
|
decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx,
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteDecodeBuffer *buf)
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
while (buf->inbuf < buf->inbuf_end) {
|
2013-04-11 17:09:04 -03:00
|
|
|
Py_ssize_t inleft;
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_ssize_t r;
|
|
|
|
|
|
|
|
inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
|
|
|
|
|
|
|
|
r = ctx->codec->decode(&ctx->state, ctx->codec->config,
|
2013-04-11 17:09:04 -03:00
|
|
|
&buf->inbuf, inleft, &buf->writer);
|
2010-05-09 12:52:27 -03:00
|
|
|
if (r == 0 || r == MBERR_TOOFEW)
|
|
|
|
break;
|
|
|
|
else if (multibytecodec_decerror(ctx->codec, &ctx->state,
|
|
|
|
buf, ctx->errors, r))
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
return 0;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
2004-07-18 00:06:29 -03:00
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
class _multibytecodec.MultibyteIncrementalEncoder "MultibyteIncrementalEncoderObject *" "&MultibyteIncrementalEncoder_Type"
|
|
|
|
[clinic start generated code]*/
|
|
|
|
/*[clinic end generated code: output=da39a3ee5e6b4b0d input=3be82909cd08924d]*/
|
2004-07-18 00:06:29 -03:00
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
_multibytecodec.MultibyteIncrementalEncoder.encode
|
2004-07-18 00:06:29 -03:00
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
input: object
|
2017-03-12 05:10:47 -03:00
|
|
|
final: bool(accept={int}) = False
|
2014-08-22 12:45:03 -03:00
|
|
|
[clinic start generated code]*/
|
2004-07-18 00:06:29 -03:00
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
static PyObject *
|
2015-04-14 19:07:59 -03:00
|
|
|
_multibytecodec_MultibyteIncrementalEncoder_encode_impl(MultibyteIncrementalEncoderObject *self,
|
|
|
|
PyObject *input,
|
|
|
|
int final)
|
2017-03-12 05:10:47 -03:00
|
|
|
/*[clinic end generated code: output=123361b6c505e2c1 input=093a1ddbb2fc6721]*/
|
2014-08-22 12:45:03 -03:00
|
|
|
{
|
|
|
|
return encoder_encode_stateful(STATEFUL_ECTX(self), input, final);
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
_multibytecodec.MultibyteIncrementalEncoder.reset
|
|
|
|
[clinic start generated code]*/
|
|
|
|
|
2004-01-17 10:29:29 -04:00
|
|
|
static PyObject *
|
2014-08-22 12:45:03 -03:00
|
|
|
_multibytecodec_MultibyteIncrementalEncoder_reset_impl(MultibyteIncrementalEncoderObject *self)
|
|
|
|
/*[clinic end generated code: output=b4125d8f537a253f input=930f06760707b6ea]*/
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2011-05-30 17:56:00 -03:00
|
|
|
/* Longest output: 4 bytes (b'\x0F\x1F(B') with ISO 2022 */
|
|
|
|
unsigned char buffer[4], *outbuf;
|
|
|
|
Py_ssize_t r;
|
|
|
|
if (self->codec->encreset != NULL) {
|
|
|
|
outbuf = buffer;
|
|
|
|
r = self->codec->encreset(&self->state, self->codec->config,
|
|
|
|
&outbuf, sizeof(buffer));
|
|
|
|
if (r != 0)
|
|
|
|
return NULL;
|
|
|
|
}
|
2013-04-13 21:06:32 -03:00
|
|
|
Py_CLEAR(self->pending);
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_RETURN_NONE;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static struct PyMethodDef mbiencoder_methods[] = {
|
2014-08-22 12:45:03 -03:00
|
|
|
_MULTIBYTECODEC_MULTIBYTEINCREMENTALENCODER_ENCODE_METHODDEF
|
|
|
|
_MULTIBYTECODEC_MULTIBYTEINCREMENTALENCODER_RESET_METHODDEF
|
|
|
|
{NULL, NULL},
|
2006-04-21 07:40:58 -03:00
|
|
|
};
|
|
|
|
|
|
|
|
static PyObject *
|
|
|
|
mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
|
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteIncrementalEncoderObject *self;
|
|
|
|
PyObject *codec = NULL;
|
|
|
|
char *errors = NULL;
|
|
|
|
|
|
|
|
if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder",
|
|
|
|
incnewkwarglist, &errors))
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0);
|
|
|
|
if (self == NULL)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
codec = PyObject_GetAttrString((PyObject *)type, "codec");
|
|
|
|
if (codec == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
if (!MultibyteCodec_Check(codec)) {
|
|
|
|
PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
|
|
|
self->codec = ((MultibyteCodecObject *)codec)->codec;
|
2013-04-13 21:06:32 -03:00
|
|
|
self->pending = NULL;
|
2010-05-09 12:52:27 -03:00
|
|
|
self->errors = internal_error_callback(errors);
|
|
|
|
if (self->errors == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
if (self->codec->encinit != NULL &&
|
|
|
|
self->codec->encinit(&self->state, self->codec->config) != 0)
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
Py_DECREF(codec);
|
|
|
|
return (PyObject *)self;
|
2006-04-21 07:40:58 -03:00
|
|
|
|
|
|
|
errorexit:
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_XDECREF(self);
|
|
|
|
Py_XDECREF(codec);
|
|
|
|
return NULL;
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
2006-04-21 13:21:44 -03:00
|
|
|
static int
|
|
|
|
mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds)
|
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
return 0;
|
2006-04-21 13:21:44 -03:00
|
|
|
}
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static int
|
|
|
|
mbiencoder_traverse(MultibyteIncrementalEncoderObject *self,
|
2010-05-09 12:52:27 -03:00
|
|
|
visitproc visit, void *arg)
|
2006-04-21 07:40:58 -03:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
if (ERROR_ISCUSTOM(self->errors))
|
|
|
|
Py_VISIT(self->errors);
|
|
|
|
return 0;
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self)
|
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject_GC_UnTrack(self);
|
|
|
|
ERROR_DECREF(self->errors);
|
|
|
|
Py_TYPE(self)->tp_free(self);
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
|
|
|
static PyTypeObject MultibyteIncrementalEncoder_Type = {
|
2010-05-09 12:52:27 -03:00
|
|
|
PyVarObject_HEAD_INIT(NULL, 0)
|
|
|
|
"MultibyteIncrementalEncoder", /* tp_name */
|
|
|
|
sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */
|
|
|
|
0, /* tp_itemsize */
|
|
|
|
/* methods */
|
|
|
|
(destructor)mbiencoder_dealloc, /* tp_dealloc */
|
|
|
|
0, /* tp_print */
|
|
|
|
0, /* tp_getattr */
|
|
|
|
0, /* tp_setattr */
|
|
|
|
0, /* tp_reserved */
|
|
|
|
0, /* tp_repr */
|
|
|
|
0, /* tp_as_number */
|
|
|
|
0, /* tp_as_sequence */
|
|
|
|
0, /* tp_as_mapping */
|
|
|
|
0, /* tp_hash */
|
|
|
|
0, /* tp_call */
|
|
|
|
0, /* tp_str */
|
|
|
|
PyObject_GenericGetAttr, /* tp_getattro */
|
|
|
|
0, /* tp_setattro */
|
|
|
|
0, /* tp_as_buffer */
|
|
|
|
Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
|
|
|
|
| Py_TPFLAGS_BASETYPE, /* tp_flags */
|
|
|
|
0, /* tp_doc */
|
|
|
|
(traverseproc)mbiencoder_traverse, /* tp_traverse */
|
|
|
|
0, /* tp_clear */
|
|
|
|
0, /* tp_richcompare */
|
|
|
|
0, /* tp_weaklistoffset */
|
|
|
|
0, /* tp_iter */
|
|
|
|
0, /* tp_iterext */
|
|
|
|
mbiencoder_methods, /* tp_methods */
|
|
|
|
0, /* tp_members */
|
|
|
|
codecctx_getsets, /* tp_getset */
|
|
|
|
0, /* tp_base */
|
|
|
|
0, /* tp_dict */
|
|
|
|
0, /* tp_descr_get */
|
|
|
|
0, /* tp_descr_set */
|
|
|
|
0, /* tp_dictoffset */
|
|
|
|
mbiencoder_init, /* tp_init */
|
|
|
|
0, /* tp_alloc */
|
|
|
|
mbiencoder_new, /* tp_new */
|
2006-04-21 07:40:58 -03:00
|
|
|
};
|
|
|
|
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
class _multibytecodec.MultibyteIncrementalDecoder "MultibyteIncrementalDecoderObject *" "&MultibyteIncrementalDecoder_Type"
|
|
|
|
[clinic start generated code]*/
|
|
|
|
/*[clinic end generated code: output=da39a3ee5e6b4b0d input=f6003faaf2cea692]*/
|
|
|
|
|
|
|
|
/*[clinic input]
|
|
|
|
_multibytecodec.MultibyteIncrementalDecoder.decode
|
|
|
|
|
|
|
|
input: Py_buffer
|
2017-03-12 05:10:47 -03:00
|
|
|
final: bool(accept={int}) = False
|
2014-08-22 12:45:03 -03:00
|
|
|
[clinic start generated code]*/
|
2006-04-21 07:40:58 -03:00
|
|
|
|
|
|
|
static PyObject *
|
2015-04-14 19:07:59 -03:00
|
|
|
_multibytecodec_MultibyteIncrementalDecoder_decode_impl(MultibyteIncrementalDecoderObject *self,
|
|
|
|
Py_buffer *input,
|
|
|
|
int final)
|
2017-03-12 05:10:47 -03:00
|
|
|
/*[clinic end generated code: output=b9b9090e8a9ce2ba input=c9132b24d503eb1d]*/
|
2006-04-21 07:40:58 -03:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteDecodeBuffer buf;
|
|
|
|
char *data, *wdata = NULL;
|
2013-04-11 17:09:04 -03:00
|
|
|
Py_ssize_t wsize, size, origpending;
|
|
|
|
PyObject *res;
|
2010-05-09 12:52:27 -03:00
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
data = input->buf;
|
|
|
|
size = input->len;
|
2010-05-09 12:52:27 -03:00
|
|
|
|
2013-04-17 18:02:17 -03:00
|
|
|
_PyUnicodeWriter_Init(&buf.writer);
|
2013-04-11 17:09:04 -03:00
|
|
|
buf.excobj = NULL;
|
2010-05-09 12:52:27 -03:00
|
|
|
origpending = self->pendingsize;
|
|
|
|
|
|
|
|
if (self->pendingsize == 0) {
|
|
|
|
wsize = size;
|
|
|
|
wdata = data;
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
if (size > PY_SSIZE_T_MAX - self->pendingsize) {
|
|
|
|
PyErr_NoMemory();
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
wsize = size + self->pendingsize;
|
|
|
|
wdata = PyMem_Malloc(wsize);
|
2013-07-15 12:47:39 -03:00
|
|
|
if (wdata == NULL) {
|
|
|
|
PyErr_NoMemory();
|
2010-05-09 12:52:27 -03:00
|
|
|
goto errorexit;
|
2013-07-15 12:47:39 -03:00
|
|
|
}
|
2010-05-09 12:52:27 -03:00
|
|
|
memcpy(wdata, self->pending, self->pendingsize);
|
|
|
|
memcpy(wdata + self->pendingsize, data, size);
|
|
|
|
self->pendingsize = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (decoder_prepare_buffer(&buf, wdata, wsize) != 0)
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf))
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
if (final && buf.inbuf < buf.inbuf_end) {
|
|
|
|
if (multibytecodec_decerror(self->codec, &self->state,
|
|
|
|
&buf, self->errors, MBERR_TOOFEW)) {
|
|
|
|
/* recover the original pending buffer */
|
|
|
|
memcpy(self->pending, wdata, origpending);
|
|
|
|
self->pendingsize = origpending;
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */
|
|
|
|
if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0)
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
2013-04-11 17:09:04 -03:00
|
|
|
res = _PyUnicodeWriter_Finish(&buf.writer);
|
|
|
|
if (res == NULL)
|
|
|
|
goto errorexit;
|
2010-05-09 12:52:27 -03:00
|
|
|
|
|
|
|
if (wdata != data)
|
|
|
|
PyMem_Del(wdata);
|
|
|
|
Py_XDECREF(buf.excobj);
|
2013-04-11 17:09:04 -03:00
|
|
|
return res;
|
2006-04-21 07:40:58 -03:00
|
|
|
|
|
|
|
errorexit:
|
2010-05-09 12:52:27 -03:00
|
|
|
if (wdata != NULL && wdata != data)
|
|
|
|
PyMem_Del(wdata);
|
|
|
|
Py_XDECREF(buf.excobj);
|
2013-04-11 17:09:04 -03:00
|
|
|
_PyUnicodeWriter_Dealloc(&buf.writer);
|
2010-05-09 12:52:27 -03:00
|
|
|
return NULL;
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
_multibytecodec.MultibyteIncrementalDecoder.reset
|
|
|
|
[clinic start generated code]*/
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static PyObject *
|
2014-08-22 12:45:03 -03:00
|
|
|
_multibytecodec_MultibyteIncrementalDecoder_reset_impl(MultibyteIncrementalDecoderObject *self)
|
|
|
|
/*[clinic end generated code: output=da423b1782c23ed1 input=3b63b3be85b2fb45]*/
|
2006-04-21 07:40:58 -03:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
if (self->codec->decreset != NULL &&
|
|
|
|
self->codec->decreset(&self->state, self->codec->config) != 0)
|
|
|
|
return NULL;
|
|
|
|
self->pendingsize = 0;
|
2006-04-21 07:40:58 -03:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_RETURN_NONE;
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
|
|
|
static struct PyMethodDef mbidecoder_methods[] = {
|
2014-08-22 12:45:03 -03:00
|
|
|
_MULTIBYTECODEC_MULTIBYTEINCREMENTALDECODER_DECODE_METHODDEF
|
|
|
|
_MULTIBYTECODEC_MULTIBYTEINCREMENTALDECODER_RESET_METHODDEF
|
|
|
|
{NULL, NULL},
|
2006-04-21 07:40:58 -03:00
|
|
|
};
|
|
|
|
|
|
|
|
static PyObject *
|
|
|
|
mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
|
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteIncrementalDecoderObject *self;
|
|
|
|
PyObject *codec = NULL;
|
|
|
|
char *errors = NULL;
|
|
|
|
|
|
|
|
if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder",
|
|
|
|
incnewkwarglist, &errors))
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0);
|
|
|
|
if (self == NULL)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
codec = PyObject_GetAttrString((PyObject *)type, "codec");
|
|
|
|
if (codec == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
if (!MultibyteCodec_Check(codec)) {
|
|
|
|
PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
|
|
|
self->codec = ((MultibyteCodecObject *)codec)->codec;
|
|
|
|
self->pendingsize = 0;
|
|
|
|
self->errors = internal_error_callback(errors);
|
|
|
|
if (self->errors == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
if (self->codec->decinit != NULL &&
|
|
|
|
self->codec->decinit(&self->state, self->codec->config) != 0)
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
Py_DECREF(codec);
|
|
|
|
return (PyObject *)self;
|
2006-04-21 07:40:58 -03:00
|
|
|
|
|
|
|
errorexit:
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_XDECREF(self);
|
|
|
|
Py_XDECREF(codec);
|
|
|
|
return NULL;
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
2006-04-21 13:21:44 -03:00
|
|
|
static int
|
|
|
|
mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds)
|
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
return 0;
|
2006-04-21 13:21:44 -03:00
|
|
|
}
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static int
|
|
|
|
mbidecoder_traverse(MultibyteIncrementalDecoderObject *self,
|
2010-05-09 12:52:27 -03:00
|
|
|
visitproc visit, void *arg)
|
2006-04-21 07:40:58 -03:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
if (ERROR_ISCUSTOM(self->errors))
|
|
|
|
Py_VISIT(self->errors);
|
|
|
|
return 0;
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self)
|
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject_GC_UnTrack(self);
|
|
|
|
ERROR_DECREF(self->errors);
|
|
|
|
Py_TYPE(self)->tp_free(self);
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
|
|
|
static PyTypeObject MultibyteIncrementalDecoder_Type = {
|
2010-05-09 12:52:27 -03:00
|
|
|
PyVarObject_HEAD_INIT(NULL, 0)
|
|
|
|
"MultibyteIncrementalDecoder", /* tp_name */
|
|
|
|
sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */
|
|
|
|
0, /* tp_itemsize */
|
|
|
|
/* methods */
|
|
|
|
(destructor)mbidecoder_dealloc, /* tp_dealloc */
|
|
|
|
0, /* tp_print */
|
|
|
|
0, /* tp_getattr */
|
|
|
|
0, /* tp_setattr */
|
|
|
|
0, /* tp_reserved */
|
|
|
|
0, /* tp_repr */
|
|
|
|
0, /* tp_as_number */
|
|
|
|
0, /* tp_as_sequence */
|
|
|
|
0, /* tp_as_mapping */
|
|
|
|
0, /* tp_hash */
|
|
|
|
0, /* tp_call */
|
|
|
|
0, /* tp_str */
|
|
|
|
PyObject_GenericGetAttr, /* tp_getattro */
|
|
|
|
0, /* tp_setattro */
|
|
|
|
0, /* tp_as_buffer */
|
|
|
|
Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
|
|
|
|
| Py_TPFLAGS_BASETYPE, /* tp_flags */
|
|
|
|
0, /* tp_doc */
|
|
|
|
(traverseproc)mbidecoder_traverse, /* tp_traverse */
|
|
|
|
0, /* tp_clear */
|
|
|
|
0, /* tp_richcompare */
|
|
|
|
0, /* tp_weaklistoffset */
|
|
|
|
0, /* tp_iter */
|
|
|
|
0, /* tp_iterext */
|
|
|
|
mbidecoder_methods, /* tp_methods */
|
|
|
|
0, /* tp_members */
|
|
|
|
codecctx_getsets, /* tp_getset */
|
|
|
|
0, /* tp_base */
|
|
|
|
0, /* tp_dict */
|
|
|
|
0, /* tp_descr_get */
|
|
|
|
0, /* tp_descr_set */
|
|
|
|
0, /* tp_dictoffset */
|
|
|
|
mbidecoder_init, /* tp_init */
|
|
|
|
0, /* tp_alloc */
|
|
|
|
mbidecoder_new, /* tp_new */
|
2006-04-21 07:40:58 -03:00
|
|
|
};
|
|
|
|
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
class _multibytecodec.MultibyteStreamReader "MultibyteStreamReaderObject *" "MultibyteStreamReader_Type"
|
|
|
|
[clinic start generated code]*/
|
|
|
|
/*[clinic end generated code: output=da39a3ee5e6b4b0d input=d323634b74976f09]*/
|
2006-04-21 07:40:58 -03:00
|
|
|
|
|
|
|
static PyObject *
|
|
|
|
mbstreamreader_iread(MultibyteStreamReaderObject *self,
|
2010-05-09 12:52:27 -03:00
|
|
|
const char *method, Py_ssize_t sizehint)
|
2006-04-21 07:40:58 -03:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteDecodeBuffer buf;
|
2013-04-11 17:09:04 -03:00
|
|
|
PyObject *cres, *res;
|
|
|
|
Py_ssize_t rsize;
|
2010-05-09 12:52:27 -03:00
|
|
|
|
|
|
|
if (sizehint == 0)
|
2011-11-30 22:18:59 -04:00
|
|
|
return PyUnicode_New(0, 0);
|
2010-05-09 12:52:27 -03:00
|
|
|
|
2013-04-17 18:02:17 -03:00
|
|
|
_PyUnicodeWriter_Init(&buf.writer);
|
2013-04-11 17:09:04 -03:00
|
|
|
buf.excobj = NULL;
|
2010-05-09 12:52:27 -03:00
|
|
|
cres = NULL;
|
|
|
|
|
|
|
|
for (;;) {
|
|
|
|
int endoffile;
|
|
|
|
|
|
|
|
if (sizehint < 0)
|
|
|
|
cres = PyObject_CallMethod(self->stream,
|
2014-09-28 05:27:24 -03:00
|
|
|
method, NULL);
|
2010-05-09 12:52:27 -03:00
|
|
|
else
|
|
|
|
cres = PyObject_CallMethod(self->stream,
|
2014-09-28 05:27:24 -03:00
|
|
|
method, "i", sizehint);
|
2010-05-09 12:52:27 -03:00
|
|
|
if (cres == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
if (!PyBytes_Check(cres)) {
|
|
|
|
PyErr_Format(PyExc_TypeError,
|
|
|
|
"stream function returned a "
|
|
|
|
"non-bytes object (%.100s)",
|
|
|
|
cres->ob_type->tp_name);
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
|
|
|
endoffile = (PyBytes_GET_SIZE(cres) == 0);
|
|
|
|
|
|
|
|
if (self->pendingsize > 0) {
|
|
|
|
PyObject *ctr;
|
|
|
|
char *ctrdata;
|
|
|
|
|
|
|
|
if (PyBytes_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) {
|
|
|
|
PyErr_NoMemory();
|
|
|
|
goto errorexit;
|
2013-07-08 17:28:27 -03:00
|
|
|
}
|
|
|
|
rsize = PyBytes_GET_SIZE(cres) + self->pendingsize;
|
|
|
|
ctr = PyBytes_FromStringAndSize(NULL, rsize);
|
|
|
|
if (ctr == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
ctrdata = PyBytes_AS_STRING(ctr);
|
|
|
|
memcpy(ctrdata, self->pending, self->pendingsize);
|
|
|
|
memcpy(ctrdata + self->pendingsize,
|
|
|
|
PyBytes_AS_STRING(cres),
|
|
|
|
PyBytes_GET_SIZE(cres));
|
|
|
|
Py_DECREF(cres);
|
|
|
|
cres = ctr;
|
|
|
|
self->pendingsize = 0;
|
2010-05-09 12:52:27 -03:00
|
|
|
}
|
|
|
|
|
|
|
|
rsize = PyBytes_GET_SIZE(cres);
|
|
|
|
if (decoder_prepare_buffer(&buf, PyBytes_AS_STRING(cres),
|
|
|
|
rsize) != 0)
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
if (rsize > 0 && decoder_feed_buffer(
|
|
|
|
(MultibyteStatefulDecoderContext *)self, &buf))
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
if (endoffile || sizehint < 0) {
|
|
|
|
if (buf.inbuf < buf.inbuf_end &&
|
|
|
|
multibytecodec_decerror(self->codec, &self->state,
|
|
|
|
&buf, self->errors, MBERR_TOOFEW))
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */
|
|
|
|
if (decoder_append_pending(STATEFUL_DCTX(self),
|
|
|
|
&buf) != 0)
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
|
|
|
Py_DECREF(cres);
|
|
|
|
cres = NULL;
|
|
|
|
|
2013-04-11 17:09:04 -03:00
|
|
|
if (sizehint < 0 || buf.writer.pos != 0 || rsize == 0)
|
2010-05-09 12:52:27 -03:00
|
|
|
break;
|
|
|
|
|
|
|
|
sizehint = 1; /* read 1 more byte and retry */
|
|
|
|
}
|
|
|
|
|
2013-04-11 17:09:04 -03:00
|
|
|
res = _PyUnicodeWriter_Finish(&buf.writer);
|
|
|
|
if (res == NULL)
|
|
|
|
goto errorexit;
|
2010-05-09 12:52:27 -03:00
|
|
|
|
|
|
|
Py_XDECREF(cres);
|
|
|
|
Py_XDECREF(buf.excobj);
|
2013-04-11 17:09:04 -03:00
|
|
|
return res;
|
2006-04-21 07:40:58 -03:00
|
|
|
|
|
|
|
errorexit:
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_XDECREF(cres);
|
|
|
|
Py_XDECREF(buf.excobj);
|
2013-04-11 17:09:04 -03:00
|
|
|
_PyUnicodeWriter_Dealloc(&buf.writer);
|
2010-05-09 12:52:27 -03:00
|
|
|
return NULL;
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
_multibytecodec.MultibyteStreamReader.read
|
|
|
|
|
|
|
|
sizeobj: object = None
|
|
|
|
/
|
|
|
|
[clinic start generated code]*/
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static PyObject *
|
2015-04-14 19:07:59 -03:00
|
|
|
_multibytecodec_MultibyteStreamReader_read_impl(MultibyteStreamReaderObject *self,
|
|
|
|
PyObject *sizeobj)
|
|
|
|
/*[clinic end generated code: output=35621eb75355d5b8 input=015b0d3ff2fca485]*/
|
2006-04-21 07:40:58 -03:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_ssize_t size;
|
2006-04-21 07:40:58 -03:00
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
if (sizeobj == Py_None)
|
2010-05-09 12:52:27 -03:00
|
|
|
size = -1;
|
|
|
|
else if (PyLong_Check(sizeobj))
|
|
|
|
size = PyLong_AsSsize_t(sizeobj);
|
|
|
|
else {
|
|
|
|
PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
|
|
|
|
return NULL;
|
|
|
|
}
|
2006-04-21 07:40:58 -03:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
if (size == -1 && PyErr_Occurred())
|
|
|
|
return NULL;
|
2007-01-13 23:31:43 -04:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
return mbstreamreader_iread(self, "read", size);
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
_multibytecodec.MultibyteStreamReader.readline
|
|
|
|
|
|
|
|
sizeobj: object = None
|
|
|
|
/
|
|
|
|
[clinic start generated code]*/
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static PyObject *
|
2015-04-14 19:07:59 -03:00
|
|
|
_multibytecodec_MultibyteStreamReader_readline_impl(MultibyteStreamReaderObject *self,
|
|
|
|
PyObject *sizeobj)
|
|
|
|
/*[clinic end generated code: output=4fbfaae1ed457a11 input=41ccc64f9bb0cec3]*/
|
2006-04-21 07:40:58 -03:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_ssize_t size;
|
2006-04-21 07:40:58 -03:00
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
if (sizeobj == Py_None)
|
2010-05-09 12:52:27 -03:00
|
|
|
size = -1;
|
|
|
|
else if (PyLong_Check(sizeobj))
|
|
|
|
size = PyLong_AsSsize_t(sizeobj);
|
|
|
|
else {
|
|
|
|
PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
|
|
|
|
return NULL;
|
|
|
|
}
|
2006-04-21 07:40:58 -03:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
if (size == -1 && PyErr_Occurred())
|
|
|
|
return NULL;
|
2007-01-13 23:31:43 -04:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
return mbstreamreader_iread(self, "readline", size);
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
_multibytecodec.MultibyteStreamReader.readlines
|
|
|
|
|
|
|
|
sizehintobj: object = None
|
|
|
|
/
|
|
|
|
[clinic start generated code]*/
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static PyObject *
|
2015-04-14 19:07:59 -03:00
|
|
|
_multibytecodec_MultibyteStreamReader_readlines_impl(MultibyteStreamReaderObject *self,
|
|
|
|
PyObject *sizehintobj)
|
|
|
|
/*[clinic end generated code: output=e7c4310768ed2ad4 input=54932f5d4d88e880]*/
|
2006-04-21 07:40:58 -03:00
|
|
|
{
|
2014-08-22 12:45:03 -03:00
|
|
|
PyObject *r, *sr;
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_ssize_t sizehint;
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
if (sizehintobj == Py_None)
|
2010-05-09 12:52:27 -03:00
|
|
|
sizehint = -1;
|
|
|
|
else if (PyLong_Check(sizehintobj))
|
|
|
|
sizehint = PyLong_AsSsize_t(sizehintobj);
|
|
|
|
else {
|
|
|
|
PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (sizehint == -1 && PyErr_Occurred())
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
r = mbstreamreader_iread(self, "read", sizehint);
|
|
|
|
if (r == NULL)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
sr = PyUnicode_Splitlines(r, 1);
|
|
|
|
Py_DECREF(r);
|
|
|
|
return sr;
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
_multibytecodec.MultibyteStreamReader.reset
|
|
|
|
[clinic start generated code]*/
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static PyObject *
|
2014-08-22 12:45:03 -03:00
|
|
|
_multibytecodec_MultibyteStreamReader_reset_impl(MultibyteStreamReaderObject *self)
|
|
|
|
/*[clinic end generated code: output=138490370a680abc input=5d4140db84b5e1e2]*/
|
2006-04-21 07:40:58 -03:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
if (self->codec->decreset != NULL &&
|
|
|
|
self->codec->decreset(&self->state, self->codec->config) != 0)
|
|
|
|
return NULL;
|
|
|
|
self->pendingsize = 0;
|
2006-04-21 07:40:58 -03:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_RETURN_NONE;
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
|
|
|
static struct PyMethodDef mbstreamreader_methods[] = {
|
2014-08-22 12:45:03 -03:00
|
|
|
_MULTIBYTECODEC_MULTIBYTESTREAMREADER_READ_METHODDEF
|
|
|
|
_MULTIBYTECODEC_MULTIBYTESTREAMREADER_READLINE_METHODDEF
|
|
|
|
_MULTIBYTECODEC_MULTIBYTESTREAMREADER_READLINES_METHODDEF
|
|
|
|
_MULTIBYTECODEC_MULTIBYTESTREAMREADER_RESET_METHODDEF
|
2010-05-09 12:52:27 -03:00
|
|
|
{NULL, NULL},
|
2004-01-17 10:29:29 -04:00
|
|
|
};
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static PyMemberDef mbstreamreader_members[] = {
|
2010-05-09 12:52:27 -03:00
|
|
|
{"stream", T_OBJECT,
|
|
|
|
offsetof(MultibyteStreamReaderObject, stream),
|
|
|
|
READONLY, NULL},
|
|
|
|
{NULL,}
|
2006-04-21 07:40:58 -03:00
|
|
|
};
|
|
|
|
|
|
|
|
static PyObject *
|
|
|
|
mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteStreamReaderObject *self;
|
|
|
|
PyObject *stream, *codec = NULL;
|
|
|
|
char *errors = NULL;
|
|
|
|
|
|
|
|
if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader",
|
|
|
|
streamkwarglist, &stream, &errors))
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0);
|
|
|
|
if (self == NULL)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
codec = PyObject_GetAttrString((PyObject *)type, "codec");
|
|
|
|
if (codec == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
if (!MultibyteCodec_Check(codec)) {
|
|
|
|
PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
|
|
|
self->codec = ((MultibyteCodecObject *)codec)->codec;
|
|
|
|
self->stream = stream;
|
|
|
|
Py_INCREF(stream);
|
|
|
|
self->pendingsize = 0;
|
|
|
|
self->errors = internal_error_callback(errors);
|
|
|
|
if (self->errors == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
if (self->codec->decinit != NULL &&
|
|
|
|
self->codec->decinit(&self->state, self->codec->config) != 0)
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
Py_DECREF(codec);
|
|
|
|
return (PyObject *)self;
|
2006-04-21 07:40:58 -03:00
|
|
|
|
|
|
|
errorexit:
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_XDECREF(self);
|
|
|
|
Py_XDECREF(codec);
|
|
|
|
return NULL;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
2006-04-21 13:21:44 -03:00
|
|
|
static int
|
|
|
|
mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds)
|
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
return 0;
|
2006-04-21 13:21:44 -03:00
|
|
|
}
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static int
|
|
|
|
mbstreamreader_traverse(MultibyteStreamReaderObject *self,
|
2010-05-09 12:52:27 -03:00
|
|
|
visitproc visit, void *arg)
|
2006-04-21 07:40:58 -03:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
if (ERROR_ISCUSTOM(self->errors))
|
|
|
|
Py_VISIT(self->errors);
|
|
|
|
Py_VISIT(self->stream);
|
|
|
|
return 0;
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
2004-01-17 10:29:29 -04:00
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static void
|
|
|
|
mbstreamreader_dealloc(MultibyteStreamReaderObject *self)
|
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject_GC_UnTrack(self);
|
|
|
|
ERROR_DECREF(self->errors);
|
|
|
|
Py_XDECREF(self->stream);
|
|
|
|
Py_TYPE(self)->tp_free(self);
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
2004-01-17 10:29:29 -04:00
|
|
|
|
|
|
|
static PyTypeObject MultibyteStreamReader_Type = {
|
2010-05-09 12:52:27 -03:00
|
|
|
PyVarObject_HEAD_INIT(NULL, 0)
|
|
|
|
"MultibyteStreamReader", /* tp_name */
|
|
|
|
sizeof(MultibyteStreamReaderObject), /* tp_basicsize */
|
|
|
|
0, /* tp_itemsize */
|
|
|
|
/* methods */
|
|
|
|
(destructor)mbstreamreader_dealloc, /* tp_dealloc */
|
|
|
|
0, /* tp_print */
|
|
|
|
0, /* tp_getattr */
|
|
|
|
0, /* tp_setattr */
|
|
|
|
0, /* tp_reserved */
|
|
|
|
0, /* tp_repr */
|
|
|
|
0, /* tp_as_number */
|
|
|
|
0, /* tp_as_sequence */
|
|
|
|
0, /* tp_as_mapping */
|
|
|
|
0, /* tp_hash */
|
|
|
|
0, /* tp_call */
|
|
|
|
0, /* tp_str */
|
|
|
|
PyObject_GenericGetAttr, /* tp_getattro */
|
|
|
|
0, /* tp_setattro */
|
|
|
|
0, /* tp_as_buffer */
|
|
|
|
Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
|
|
|
|
| Py_TPFLAGS_BASETYPE, /* tp_flags */
|
|
|
|
0, /* tp_doc */
|
|
|
|
(traverseproc)mbstreamreader_traverse, /* tp_traverse */
|
|
|
|
0, /* tp_clear */
|
|
|
|
0, /* tp_richcompare */
|
|
|
|
0, /* tp_weaklistoffset */
|
|
|
|
0, /* tp_iter */
|
|
|
|
0, /* tp_iterext */
|
|
|
|
mbstreamreader_methods, /* tp_methods */
|
|
|
|
mbstreamreader_members, /* tp_members */
|
|
|
|
codecctx_getsets, /* tp_getset */
|
|
|
|
0, /* tp_base */
|
|
|
|
0, /* tp_dict */
|
|
|
|
0, /* tp_descr_get */
|
|
|
|
0, /* tp_descr_set */
|
|
|
|
0, /* tp_dictoffset */
|
|
|
|
mbstreamreader_init, /* tp_init */
|
|
|
|
0, /* tp_alloc */
|
|
|
|
mbstreamreader_new, /* tp_new */
|
2004-01-17 10:29:29 -04:00
|
|
|
};
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
class _multibytecodec.MultibyteStreamWriter "MultibyteStreamWriterObject *" "&MultibyteStreamWriter_Type"
|
|
|
|
[clinic start generated code]*/
|
|
|
|
/*[clinic end generated code: output=da39a3ee5e6b4b0d input=cde22780a215d6ac]*/
|
2006-04-21 07:40:58 -03:00
|
|
|
|
2004-01-17 10:29:29 -04:00
|
|
|
static int
|
|
|
|
mbstreamwriter_iwrite(MultibyteStreamWriterObject *self,
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject *unistr)
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject *str, *wr;
|
2004-07-18 00:06:29 -03:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0);
|
|
|
|
if (str == NULL)
|
|
|
|
return -1;
|
2004-07-18 00:06:29 -03:00
|
|
|
|
2016-12-09 11:09:30 -04:00
|
|
|
wr = _PyObject_CallMethodIdObjArgs(self->stream, &PyId_write, str, NULL);
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_DECREF(str);
|
|
|
|
if (wr == NULL)
|
|
|
|
return -1;
|
2004-07-18 00:06:29 -03:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_DECREF(wr);
|
|
|
|
return 0;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
_multibytecodec.MultibyteStreamWriter.write
|
|
|
|
|
|
|
|
strobj: object
|
|
|
|
/
|
|
|
|
[clinic start generated code]*/
|
|
|
|
|
2004-01-17 10:29:29 -04:00
|
|
|
static PyObject *
|
2015-04-14 19:07:59 -03:00
|
|
|
_multibytecodec_MultibyteStreamWriter_write(MultibyteStreamWriterObject *self,
|
|
|
|
PyObject *strobj)
|
|
|
|
/*[clinic end generated code: output=e13ae841c895251e input=551dc4c018c10a2b]*/
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
if (mbstreamwriter_iwrite(self, strobj))
|
|
|
|
return NULL;
|
|
|
|
else
|
|
|
|
Py_RETURN_NONE;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
_multibytecodec.MultibyteStreamWriter.writelines
|
|
|
|
|
|
|
|
lines: object
|
|
|
|
/
|
|
|
|
[clinic start generated code]*/
|
|
|
|
|
2004-01-17 10:29:29 -04:00
|
|
|
static PyObject *
|
2015-04-14 19:07:59 -03:00
|
|
|
_multibytecodec_MultibyteStreamWriter_writelines(MultibyteStreamWriterObject *self,
|
|
|
|
PyObject *lines)
|
|
|
|
/*[clinic end generated code: output=e5c4285ac8e7d522 input=57797fe7008d4e96]*/
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject *strobj;
|
|
|
|
int i, r;
|
|
|
|
|
|
|
|
if (!PySequence_Check(lines)) {
|
|
|
|
PyErr_SetString(PyExc_TypeError,
|
|
|
|
"arg must be a sequence object");
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
for (i = 0; i < PySequence_Length(lines); i++) {
|
|
|
|
/* length can be changed even within this loop */
|
|
|
|
strobj = PySequence_GetItem(lines, i);
|
|
|
|
if (strobj == NULL)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
r = mbstreamwriter_iwrite(self, strobj);
|
|
|
|
Py_DECREF(strobj);
|
|
|
|
if (r == -1)
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
Py_RETURN_NONE;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
_multibytecodec.MultibyteStreamWriter.reset
|
|
|
|
[clinic start generated code]*/
|
|
|
|
|
2004-01-17 10:29:29 -04:00
|
|
|
static PyObject *
|
2014-08-22 12:45:03 -03:00
|
|
|
_multibytecodec_MultibyteStreamWriter_reset_impl(MultibyteStreamWriterObject *self)
|
|
|
|
/*[clinic end generated code: output=8f54a4d9b03db5ff input=b56dbcbaf35cc10c]*/
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject *pwrt;
|
|
|
|
|
2015-07-16 17:17:31 -03:00
|
|
|
if (!self->pending)
|
|
|
|
Py_RETURN_NONE;
|
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
pwrt = multibytecodec_encode(self->codec, &self->state,
|
2013-04-13 21:06:32 -03:00
|
|
|
self->pending, NULL, self->errors,
|
2010-05-09 12:52:27 -03:00
|
|
|
MBENC_FLUSH | MBENC_RESET);
|
|
|
|
/* some pending buffer can be truncated when UnicodeEncodeError is
|
|
|
|
* raised on 'strict' mode. but, 'reset' method is designed to
|
|
|
|
* reset the pending buffer or states so failed string sequence
|
|
|
|
* ought to be missed */
|
2013-04-13 21:06:32 -03:00
|
|
|
Py_CLEAR(self->pending);
|
2010-05-09 12:52:27 -03:00
|
|
|
if (pwrt == NULL)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
assert(PyBytes_Check(pwrt));
|
|
|
|
if (PyBytes_Size(pwrt) > 0) {
|
|
|
|
PyObject *wr;
|
2011-10-09 05:38:36 -03:00
|
|
|
|
2016-12-09 11:09:30 -04:00
|
|
|
wr = _PyObject_CallMethodIdObjArgs(self->stream, &PyId_write, pwrt);
|
2010-05-09 12:52:27 -03:00
|
|
|
if (wr == NULL) {
|
|
|
|
Py_DECREF(pwrt);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
Py_DECREF(pwrt);
|
|
|
|
|
|
|
|
Py_RETURN_NONE;
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
|
|
|
static PyObject *
|
|
|
|
mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
|
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteStreamWriterObject *self;
|
|
|
|
PyObject *stream, *codec = NULL;
|
|
|
|
char *errors = NULL;
|
|
|
|
|
|
|
|
if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter",
|
|
|
|
streamkwarglist, &stream, &errors))
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0);
|
|
|
|
if (self == NULL)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
codec = PyObject_GetAttrString((PyObject *)type, "codec");
|
|
|
|
if (codec == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
if (!MultibyteCodec_Check(codec)) {
|
|
|
|
PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
|
|
|
|
goto errorexit;
|
|
|
|
}
|
|
|
|
|
|
|
|
self->codec = ((MultibyteCodecObject *)codec)->codec;
|
|
|
|
self->stream = stream;
|
|
|
|
Py_INCREF(stream);
|
2013-04-13 21:06:32 -03:00
|
|
|
self->pending = NULL;
|
2010-05-09 12:52:27 -03:00
|
|
|
self->errors = internal_error_callback(errors);
|
|
|
|
if (self->errors == NULL)
|
|
|
|
goto errorexit;
|
|
|
|
if (self->codec->encinit != NULL &&
|
|
|
|
self->codec->encinit(&self->state, self->codec->config) != 0)
|
|
|
|
goto errorexit;
|
|
|
|
|
|
|
|
Py_DECREF(codec);
|
|
|
|
return (PyObject *)self;
|
2006-04-21 07:40:58 -03:00
|
|
|
|
|
|
|
errorexit:
|
2010-05-09 12:52:27 -03:00
|
|
|
Py_XDECREF(self);
|
|
|
|
Py_XDECREF(codec);
|
|
|
|
return NULL;
|
2006-04-21 07:40:58 -03:00
|
|
|
}
|
|
|
|
|
2006-04-21 13:21:44 -03:00
|
|
|
static int
|
|
|
|
mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds)
|
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
return 0;
|
2006-04-21 13:21:44 -03:00
|
|
|
}
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static int
|
|
|
|
mbstreamwriter_traverse(MultibyteStreamWriterObject *self,
|
2010-05-09 12:52:27 -03:00
|
|
|
visitproc visit, void *arg)
|
2006-04-21 07:40:58 -03:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
if (ERROR_ISCUSTOM(self->errors))
|
|
|
|
Py_VISIT(self->errors);
|
|
|
|
Py_VISIT(self->stream);
|
|
|
|
return 0;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
mbstreamwriter_dealloc(MultibyteStreamWriterObject *self)
|
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
PyObject_GC_UnTrack(self);
|
|
|
|
ERROR_DECREF(self->errors);
|
|
|
|
Py_XDECREF(self->stream);
|
|
|
|
Py_TYPE(self)->tp_free(self);
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
static struct PyMethodDef mbstreamwriter_methods[] = {
|
2014-08-22 12:45:03 -03:00
|
|
|
_MULTIBYTECODEC_MULTIBYTESTREAMWRITER_WRITE_METHODDEF
|
|
|
|
_MULTIBYTECODEC_MULTIBYTESTREAMWRITER_WRITELINES_METHODDEF
|
|
|
|
_MULTIBYTECODEC_MULTIBYTESTREAMWRITER_RESET_METHODDEF
|
|
|
|
{NULL, NULL},
|
2004-01-17 10:29:29 -04:00
|
|
|
};
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
static PyMemberDef mbstreamwriter_members[] = {
|
2010-05-09 12:52:27 -03:00
|
|
|
{"stream", T_OBJECT,
|
|
|
|
offsetof(MultibyteStreamWriterObject, stream),
|
|
|
|
READONLY, NULL},
|
|
|
|
{NULL,}
|
2006-04-21 07:40:58 -03:00
|
|
|
};
|
2004-01-17 10:29:29 -04:00
|
|
|
|
|
|
|
static PyTypeObject MultibyteStreamWriter_Type = {
|
2010-05-09 12:52:27 -03:00
|
|
|
PyVarObject_HEAD_INIT(NULL, 0)
|
|
|
|
"MultibyteStreamWriter", /* tp_name */
|
|
|
|
sizeof(MultibyteStreamWriterObject), /* tp_basicsize */
|
|
|
|
0, /* tp_itemsize */
|
|
|
|
/* methods */
|
|
|
|
(destructor)mbstreamwriter_dealloc, /* tp_dealloc */
|
|
|
|
0, /* tp_print */
|
|
|
|
0, /* tp_getattr */
|
|
|
|
0, /* tp_setattr */
|
|
|
|
0, /* tp_reserved */
|
|
|
|
0, /* tp_repr */
|
|
|
|
0, /* tp_as_number */
|
|
|
|
0, /* tp_as_sequence */
|
|
|
|
0, /* tp_as_mapping */
|
|
|
|
0, /* tp_hash */
|
|
|
|
0, /* tp_call */
|
|
|
|
0, /* tp_str */
|
|
|
|
PyObject_GenericGetAttr, /* tp_getattro */
|
|
|
|
0, /* tp_setattro */
|
|
|
|
0, /* tp_as_buffer */
|
|
|
|
Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
|
|
|
|
| Py_TPFLAGS_BASETYPE, /* tp_flags */
|
|
|
|
0, /* tp_doc */
|
|
|
|
(traverseproc)mbstreamwriter_traverse, /* tp_traverse */
|
|
|
|
0, /* tp_clear */
|
|
|
|
0, /* tp_richcompare */
|
|
|
|
0, /* tp_weaklistoffset */
|
|
|
|
0, /* tp_iter */
|
|
|
|
0, /* tp_iterext */
|
|
|
|
mbstreamwriter_methods, /* tp_methods */
|
|
|
|
mbstreamwriter_members, /* tp_members */
|
|
|
|
codecctx_getsets, /* tp_getset */
|
|
|
|
0, /* tp_base */
|
|
|
|
0, /* tp_dict */
|
|
|
|
0, /* tp_descr_get */
|
|
|
|
0, /* tp_descr_set */
|
|
|
|
0, /* tp_dictoffset */
|
|
|
|
mbstreamwriter_init, /* tp_init */
|
|
|
|
0, /* tp_alloc */
|
|
|
|
mbstreamwriter_new, /* tp_new */
|
2004-01-17 10:29:29 -04:00
|
|
|
};
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
|
2014-08-22 12:45:03 -03:00
|
|
|
/*[clinic input]
|
|
|
|
_multibytecodec.__create_codec
|
|
|
|
|
|
|
|
arg: object
|
|
|
|
/
|
|
|
|
[clinic start generated code]*/
|
2006-04-21 07:40:58 -03:00
|
|
|
|
2004-01-17 10:29:29 -04:00
|
|
|
static PyObject *
|
2016-07-07 11:35:15 -03:00
|
|
|
_multibytecodec___create_codec(PyObject *module, PyObject *arg)
|
|
|
|
/*[clinic end generated code: output=cfa3dce8260e809d input=6840b2a6b183fcfa]*/
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
MultibyteCodecObject *self;
|
|
|
|
MultibyteCodec *codec;
|
2004-01-17 10:29:29 -04:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) {
|
|
|
|
PyErr_SetString(PyExc_ValueError, "argument type invalid");
|
|
|
|
return NULL;
|
|
|
|
}
|
2004-01-17 10:29:29 -04:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME);
|
|
|
|
if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0)
|
|
|
|
return NULL;
|
2004-01-17 10:29:29 -04:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type);
|
|
|
|
if (self == NULL)
|
|
|
|
return NULL;
|
|
|
|
self->codec = codec;
|
2004-01-17 10:29:29 -04:00
|
|
|
|
2010-05-09 12:52:27 -03:00
|
|
|
return (PyObject *)self;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
static struct PyMethodDef __methods[] = {
|
2014-08-22 12:45:03 -03:00
|
|
|
_MULTIBYTECODEC___CREATE_CODEC_METHODDEF
|
2010-05-09 12:52:27 -03:00
|
|
|
{NULL, NULL},
|
2004-01-17 10:29:29 -04:00
|
|
|
};
|
|
|
|
|
2008-06-11 02:26:20 -03:00
|
|
|
|
|
|
|
static struct PyModuleDef _multibytecodecmodule = {
|
2010-05-09 12:52:27 -03:00
|
|
|
PyModuleDef_HEAD_INIT,
|
|
|
|
"_multibytecodec",
|
|
|
|
NULL,
|
|
|
|
-1,
|
|
|
|
__methods,
|
|
|
|
NULL,
|
|
|
|
NULL,
|
|
|
|
NULL,
|
|
|
|
NULL
|
2008-06-11 02:26:20 -03:00
|
|
|
};
|
|
|
|
|
2006-04-21 07:40:58 -03:00
|
|
|
PyMODINIT_FUNC
|
2008-06-11 02:26:20 -03:00
|
|
|
PyInit__multibytecodec(void)
|
2004-01-17 10:29:29 -04:00
|
|
|
{
|
2010-05-09 12:52:27 -03:00
|
|
|
int i;
|
|
|
|
PyObject *m;
|
|
|
|
PyTypeObject *typelist[] = {
|
|
|
|
&MultibyteIncrementalEncoder_Type,
|
|
|
|
&MultibyteIncrementalDecoder_Type,
|
|
|
|
&MultibyteStreamReader_Type,
|
|
|
|
&MultibyteStreamWriter_Type,
|
|
|
|
NULL
|
|
|
|
};
|
|
|
|
|
|
|
|
if (PyType_Ready(&MultibyteCodec_Type) < 0)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
m = PyModule_Create(&_multibytecodecmodule);
|
|
|
|
if (m == NULL)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
for (i = 0; typelist[i] != NULL; i++) {
|
|
|
|
if (PyType_Ready(typelist[i]) < 0)
|
|
|
|
return NULL;
|
|
|
|
Py_INCREF(typelist[i]);
|
|
|
|
PyModule_AddObject(m, typelist[i]->tp_name,
|
|
|
|
(PyObject *)typelist[i]);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (PyErr_Occurred()) {
|
|
|
|
Py_FatalError("can't initialize the _multibytecodec module");
|
|
|
|
Py_DECREF(m);
|
|
|
|
m = NULL;
|
|
|
|
}
|
|
|
|
return m;
|
2004-01-17 10:29:29 -04:00
|
|
|
}
|