bpo-1635741: _PyUnicode_Name_CAPI moves to internal C API (GH-22713)

The private _PyUnicode_Name_CAPI structure of the PyCapsule API
unicodedata.ucnhash_CAPI moves to the internal C API. Moreover, the
structure gets a new state member which must be passed to the
getcode() and getname() functions.

* Move Include/ucnhash.h to Include/internal/pycore_ucnhash.h
* unicodedata module is now built with Py_BUILD_CORE_MODULE.
* unicodedata: move hashAPI variable into unicodedata_module_state.

Doc/whatsnew/3.10.rst

@@ -407,6 +407,12 @@ Porting to Python 3.10
   Unicode object without initial data.
   (Contributed by Inada Naoki in :issue:`36346`.)
 
+* The private ``_PyUnicode_Name_CAPI`` structure of the PyCapsule API
+  ``unicodedata.ucnhash_CAPI`` moves to the internal C API. Moreover,
+  the structure gets a new ``state`` member which must be passed to the
+  ``getcode()`` and ``getname()`` functions.
+  (Contributed by Victor Stinner in :issue:`1635741`.)
+
 Deprecated
 ----------
 

Include/internal/pycore_ucnhash.h

@@ -1,11 +1,14 @@
 /* Unicode name database interface */
-#ifndef Py_LIMITED_API
+#ifndef Py_INTERNAL_UCNHASH_H
-#ifndef Py_UCNHASH_H
+#define Py_INTERNAL_UCNHASH_H
-#define Py_UCNHASH_H
 #ifdef __cplusplus
 extern "C" {
 #endif
 
+#ifndef Py_BUILD_CORE
+#  error "this header requires Py_BUILD_CORE define"
+#endif
+
 /* revised ucnhash CAPI interface (exported through a "wrapper") */
 
 #define PyUnicodeData_CAPSULE_NAME "unicodedata.ucnhash_CAPI"
@@ -15,16 +18,22 @@ typedef struct {
     /* Size of this struct */
     int size;
 
+    // state which must be passed as the first parameter to getname()
+    // and getcode()
+    void *state;
+
     /* Get name for a given character code.  Returns non-zero if
        success, zero if not.  Does not set Python exceptions.
        If self is NULL, data come from the default version of the database.
        If it is not NULL, it should be a unicodedata.ucd_X_Y_Z object */
-    int (*getname)(PyObject *self, Py_UCS4 code, char* buffer, int buflen,
+    int (*getname)(void *state, PyObject *self, Py_UCS4 code,
+                   char* buffer, int buflen,
                    int with_alias_and_seq);
 
     /* Get character code for a given name.  Same error handling
        as for getname. */
-    int (*getcode)(PyObject *self, const char* name, int namelen, Py_UCS4* code,
+    int (*getcode)(void *state, PyObject *self,
+                   const char* name, int namelen, Py_UCS4* code,
                    int with_named_seq);
 
 } _PyUnicode_Name_CAPI;
@@ -32,5 +41,4 @@ typedef struct {
 #ifdef __cplusplus
 }
 #endif
-#endif /* !Py_UCNHASH_H */
+#endif /* !Py_INTERNAL_UCNHASH_H */
-#endif /* !Py_LIMITED_API */

Makefile.pre.in

@@ -1065,7 +1065,6 @@ PYTHON_HEADERS= \
 		$(srcdir)/Include/traceback.h \
 		$(srcdir)/Include/tracemalloc.h \
 		$(srcdir)/Include/tupleobject.h \
-		$(srcdir)/Include/ucnhash.h \
 		$(srcdir)/Include/unicodeobject.h \
 		$(srcdir)/Include/warnings.h \
 		$(srcdir)/Include/weakrefobject.h \
@@ -1129,6 +1128,7 @@ PYTHON_HEADERS= \
 		$(srcdir)/Include/internal/pycore_sysmodule.h \
 		$(srcdir)/Include/internal/pycore_traceback.h \
 		$(srcdir)/Include/internal/pycore_tuple.h \
+		$(srcdir)/Include/internal/pycore_ucnhash.h \
 		$(srcdir)/Include/internal/pycore_unionobject.h \
 		$(srcdir)/Include/internal/pycore_warnings.h \
 		$(DTRACE_HEADERS)

Misc/NEWS.d/next/C API/2020-10-16-10-47-17.bpo-1635741.e3BcPM.rst

@@ -0,0 +1,4 @@
+The private ``_PyUnicode_Name_CAPI`` structure of the PyCapsule API
+``unicodedata.ucnhash_CAPI`` moves to the internal C API. Moreover, the
+structure gets a new ``state`` member which must be passed to the
+``getcode()`` and ``getname()`` functions. Patch by Victor Stinner.

Modules/Setup

@@ -185,7 +185,7 @@ _symtable symtablemodule.c
 #_json -I$(srcdir)/Include/internal -DPy_BUILD_CORE_BUILTIN _json.c	# _json speedups
 #_statistics _statisticsmodule.c # statistics accelerator
 
-#unicodedata unicodedata.c    # static Unicode character database
+#unicodedata unicodedata.c -DPy_BUILD_CORE_BUILTIN   # static Unicode character database
 
 
 # Modules with some UNIX dependencies -- on by default:

Modules/unicodedata.c

@@ -16,7 +16,7 @@
 #define PY_SSIZE_T_CLEAN
 
 #include "Python.h"
-#include "ucnhash.h"
+#include "pycore_ucnhash.h"       // _PyUnicode_Name_CAPI
 #include "structmember.h"         // PyMemberDef
 
 #include <stdbool.h>
@@ -97,6 +97,8 @@ typedef struct {
     // Borrowed reference to &UCD_Type. It is used to prepare the code
     // to convert the UCD_Type static type to a heap type.
     PyTypeObject *ucd_type;
+
+    _PyUnicode_Name_CAPI capi;
 } unicodedata_module_state;
 
 // bpo-1635741: Temporary global state until the unicodedata module
@@ -1180,10 +1182,11 @@ _getucname(unicodedata_module_state *state, PyObject *self,
 }
 
 static int
-capi_getucname(PyObject *self, Py_UCS4 code, char* buffer, int buflen,
+capi_getucname(void *state_raw, PyObject *self, Py_UCS4 code,
+               char* buffer, int buflen,
                int with_alias_and_seq)
 {
-    unicodedata_module_state *state = &global_module_state;
+    unicodedata_module_state *state = (unicodedata_module_state *)state_raw;
     return _getucname(state, self, code, buffer, buflen, with_alias_and_seq);
 
 }
@@ -1323,21 +1326,15 @@ _getcode(unicodedata_module_state *state, PyObject* self,
 }
 
 static int
-capi_getcode(PyObject* self, const char* name, int namelen, Py_UCS4* code,
+capi_getcode(void *state_raw, PyObject* self,
+             const char* name, int namelen, Py_UCS4* code,
              int with_named_seq)
 {
-    unicodedata_module_state *state = &global_module_state;
+    unicodedata_module_state *state = (unicodedata_module_state *)state_raw;
     return _getcode(state, self, name, namelen, code, with_named_seq);
 
 }
 
-static const _PyUnicode_Name_CAPI hashAPI =
-{
-    sizeof(_PyUnicode_Name_CAPI),
-    capi_getucname,
-    capi_getcode
-};
-
 /* -------------------------------------------------------------------- */
 /* Python bindings */
 
@@ -1510,6 +1507,11 @@ PyInit_unicodedata(void)
     PyObject *m, *v;
     unicodedata_module_state *state = &global_module_state;
 
+    state->capi.size = sizeof(_PyUnicode_Name_CAPI);
+    state->capi.state = state;
+    state->capi.getname = capi_getucname;
+    state->capi.getcode = capi_getcode;
+
     Py_SET_TYPE(&UCD_Type, &PyType_Type);
     state->ucd_type = &UCD_Type;
 
@@ -1528,7 +1530,7 @@ PyInit_unicodedata(void)
         PyModule_AddObject(m, "ucd_3_2_0", v);
 
     /* Export C API */
-    v = PyCapsule_New((void *)&hashAPI, PyUnicodeData_CAPSULE_NAME, NULL);
+    v = PyCapsule_New((void *)&state->capi, PyUnicodeData_CAPSULE_NAME, NULL);
     if (v != NULL)
         PyModule_AddObject(m, "ucnhash_CAPI", v);
     return m;

Objects/unicodeobject.c

@@ -48,7 +48,7 @@ OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 #include "pycore_pathconfig.h"    // _Py_DumpPathConfig()
 #include "pycore_pylifecycle.h"   // _Py_SetFileSystemEncoding()
 #include "pycore_pystate.h"       // _PyInterpreterState_GET()
-#include "ucnhash.h"               // _PyUnicode_Name_CAPI
+#include "pycore_ucnhash.h"       // _PyUnicode_Name_CAPI
 #include "stringlib/eq.h"         // unicode_eq()
 
 #ifdef MS_WINDOWS
@@ -6344,7 +6344,7 @@ PyUnicode_AsUTF16String(PyObject *unicode)
 
 /* --- Unicode Escape Codec ----------------------------------------------- */
 
-static _PyUnicode_Name_CAPI *ucnhash_CAPI = NULL;
+static _PyUnicode_Name_CAPI *ucnhash_capi = NULL;
 
 PyObject *
 _PyUnicode_DecodeUnicodeEscape(const char *s,
@@ -6497,11 +6497,11 @@ _PyUnicode_DecodeUnicodeEscape(const char *s,
 
             /* \N{name} */
         case 'N':
-            if (ucnhash_CAPI == NULL) {
+            if (ucnhash_capi == NULL) {
                 /* load the unicode data module */
-                ucnhash_CAPI = (_PyUnicode_Name_CAPI *)PyCapsule_Import(
+                ucnhash_capi = (_PyUnicode_Name_CAPI *)PyCapsule_Import(
                                                 PyUnicodeData_CAPSULE_NAME, 1);
-                if (ucnhash_CAPI == NULL) {
+                if (ucnhash_capi == NULL) {
                     PyErr_SetString(
                         PyExc_UnicodeError,
                         "\\N escapes not supported (can't load unicodedata module)"
@@ -6523,7 +6523,8 @@ _PyUnicode_DecodeUnicodeEscape(const char *s,
                     s++;
                     ch = 0xffffffff; /* in case 'getcode' messes up */
                     if (namelen <= INT_MAX &&
-                        ucnhash_CAPI->getcode(NULL, start, (int)namelen,
+                        ucnhash_capi->getcode(ucnhash_capi->state, NULL,
+                                              start, (int)namelen,
                                               &ch, 0)) {
                         assert(ch <= MAX_UNICODE);
                         WRITE_CHAR(ch);

PCbuild/pythoncore.vcxproj

@@ -196,6 +196,7 @@
     <ClInclude Include="..\Include\internal\pycore_sysmodule.h" />
     <ClInclude Include="..\Include\internal\pycore_traceback.h" />
     <ClInclude Include="..\Include\internal\pycore_tuple.h" />
+    <ClInclude Include="..\Include\internal\pycore_ucnhash.h" />
     <ClInclude Include="..\Include\internal\pycore_unionobject.h" />
     <ClInclude Include="..\Include\internal\pycore_warnings.h" />
     <ClInclude Include="..\Include\interpreteridobject.h" />
@@ -252,7 +253,6 @@
     <ClInclude Include="..\Include\traceback.h" />
     <ClInclude Include="..\Include\tracemalloc.h" />
     <ClInclude Include="..\Include\tupleobject.h" />
-    <ClInclude Include="..\Include\ucnhash.h" />
     <ClInclude Include="..\Include\unicodeobject.h" />
     <ClInclude Include="..\Include\weakrefobject.h" />
     <ClInclude Include="..\Modules\_math.h" />

PCbuild/pythoncore.vcxproj.filters

@@ -273,9 +273,6 @@
     <ClInclude Include="..\Include\tupleobject.h">
       <Filter>Include</Filter>
     </ClInclude>
-    <ClInclude Include="..\Include\ucnhash.h">
-      <Filter>Include</Filter>
-    </ClInclude>
     <ClInclude Include="..\Include\unicodeobject.h">
       <Filter>Include</Filter>
     </ClInclude>
@@ -573,6 +570,9 @@
     <ClInclude Include="..\Include\internal\pycore_tuple.h">
       <Filter>Include\internal</Filter>
     </ClInclude>
+    <ClInclude Include="..\Include\internal\pycore_ucnhash.h">
+      <Filter>Include\internal</Filter>
+    </ClInclude>
     <ClInclude Include="..\Include\internal\pycore_unionobject.h">
       <Filter>Include\internal</Filter>
     </ClInclude>

Python/codecs.c

@@ -11,7 +11,7 @@ Copyright (c) Corporation for National Research Initiatives.
 #include "Python.h"
 #include "pycore_interp.h"        // PyInterpreterState.codec_search_path
 #include "pycore_pystate.h"       // _PyInterpreterState_GET()
-#include "ucnhash.h"
+#include "pycore_ucnhash.h"       // _PyUnicode_Name_CAPI
 #include <ctype.h>
 
 const char *Py_hexdigits = "0123456789abcdef";
@@ -954,7 +954,7 @@ PyObject *PyCodec_BackslashReplaceErrors(PyObject *exc)
     return Py_BuildValue("(Nn)", res, end);
 }
 
-static _PyUnicode_Name_CAPI *ucnhash_CAPI = NULL;
+static _PyUnicode_Name_CAPI *ucnhash_capi = NULL;
 
 PyObject *PyCodec_NameReplaceErrors(PyObject *exc)
 {
@@ -976,17 +976,19 @@ PyObject *PyCodec_NameReplaceErrors(PyObject *exc)
             return NULL;
         if (!(object = PyUnicodeEncodeError_GetObject(exc)))
             return NULL;
-        if (!ucnhash_CAPI) {
+        if (!ucnhash_capi) {
             /* load the unicode data module */
-            ucnhash_CAPI = (_PyUnicode_Name_CAPI *)PyCapsule_Import(
+            ucnhash_capi = (_PyUnicode_Name_CAPI *)PyCapsule_Import(
                                             PyUnicodeData_CAPSULE_NAME, 1);
-            if (!ucnhash_CAPI)
+            if (!ucnhash_capi) {
                 return NULL;
             }
+        }
         for (i = start, ressize = 0; i < end; ++i) {
             /* object is guaranteed to be "ready" */
             c = PyUnicode_READ_CHAR(object, i);
-            if (ucnhash_CAPI->getname(NULL, c, buffer, sizeof(buffer), 1)) {
+            if (ucnhash_capi->getname(ucnhash_capi->state, NULL,
+                                      c, buffer, sizeof(buffer), 1)) {
                 replsize = 1+1+1+(int)strlen(buffer)+1;
             }
             else if (c >= 0x10000) {
@@ -1009,7 +1011,8 @@ PyObject *PyCodec_NameReplaceErrors(PyObject *exc)
             i < end; ++i) {
             c = PyUnicode_READ_CHAR(object, i);
             *outp++ = '\\';
-            if (ucnhash_CAPI->getname(NULL, c, buffer, sizeof(buffer), 1)) {
+            if (ucnhash_capi->getname(ucnhash_capi->state, NULL,
+                                      c, buffer, sizeof(buffer), 1)) {
                 *outp++ = 'N';
                 *outp++ = '{';
                 strcpy((char *)outp, buffer);

setup.py

@@ -878,7 +878,8 @@ class PyBuildExt(build_ext):
         self.add(Extension('_lsprof', ['_lsprof.c', 'rotatingtree.c']))
         # static Unicode character database
         self.add(Extension('unicodedata', ['unicodedata.c'],
-                           depends=['unicodedata_db.h', 'unicodename_db.h']))
+                           depends=['unicodedata_db.h', 'unicodename_db.h'],
+                           extra_compile_args=['-DPy_BUILD_CORE_MODULE']))
         # _opcode module
         self.add(Extension('_opcode', ['_opcode.c']))
         # asyncio speedups