+++ /dev/null
-/*\r
- * cjkcodecs.h: common header for cjkcodecs\r
- *\r
- * Written by Hye-Shik Chang <perky@FreeBSD.org>\r
- */\r
-\r
-#ifndef _CJKCODECS_H_\r
-#define _CJKCODECS_H_\r
-\r
-#define PY_SSIZE_T_CLEAN\r
-#include "Python.h"\r
-#include "multibytecodec.h"\r
-\r
-\r
-/* a unicode "undefined" code point */\r
-#define UNIINV 0xFFFE\r
-\r
-/* internal-use DBCS code points which aren't used by any charsets */\r
-#define NOCHAR 0xFFFF\r
-#define MULTIC 0xFFFE\r
-#define DBCINV 0xFFFD\r
-\r
-/* shorter macros to save source size of mapping tables */\r
-#define U UNIINV\r
-#define N NOCHAR\r
-#define M MULTIC\r
-#define D DBCINV\r
-\r
-struct dbcs_index {\r
- const ucs2_t *map;\r
- unsigned char bottom, top;\r
-};\r
-typedef struct dbcs_index decode_map;\r
-\r
-struct widedbcs_index {\r
- const ucs4_t *map;\r
- unsigned char bottom, top;\r
-};\r
-typedef struct widedbcs_index widedecode_map;\r
-\r
-struct unim_index {\r
- const DBCHAR *map;\r
- unsigned char bottom, top;\r
-};\r
-typedef struct unim_index encode_map;\r
-\r
-struct unim_index_bytebased {\r
- const unsigned char *map;\r
- unsigned char bottom, top;\r
-};\r
-\r
-struct dbcs_map {\r
- const char *charset;\r
- const struct unim_index *encmap;\r
- const struct dbcs_index *decmap;\r
-};\r
-\r
-struct pair_encodemap {\r
- ucs4_t uniseq;\r
- DBCHAR code;\r
-};\r
-\r
-static const MultibyteCodec *codec_list;\r
-static const struct dbcs_map *mapping_list;\r
-\r
-#define CODEC_INIT(encoding) \\r
- static int encoding##_codec_init(const void *config)\r
-\r
-#define ENCODER_INIT(encoding) \\r
- static int encoding##_encode_init( \\r
- MultibyteCodec_State *state, const void *config)\r
-#define ENCODER(encoding) \\r
- static Py_ssize_t encoding##_encode( \\r
- MultibyteCodec_State *state, const void *config, \\r
- const Py_UNICODE **inbuf, Py_ssize_t inleft, \\r
- unsigned char **outbuf, Py_ssize_t outleft, int flags)\r
-#define ENCODER_RESET(encoding) \\r
- static Py_ssize_t encoding##_encode_reset( \\r
- MultibyteCodec_State *state, const void *config, \\r
- unsigned char **outbuf, Py_ssize_t outleft)\r
-\r
-#define DECODER_INIT(encoding) \\r
- static int encoding##_decode_init( \\r
- MultibyteCodec_State *state, const void *config)\r
-#define DECODER(encoding) \\r
- static Py_ssize_t encoding##_decode( \\r
- MultibyteCodec_State *state, const void *config, \\r
- const unsigned char **inbuf, Py_ssize_t inleft, \\r
- Py_UNICODE **outbuf, Py_ssize_t outleft)\r
-#define DECODER_RESET(encoding) \\r
- static Py_ssize_t encoding##_decode_reset( \\r
- MultibyteCodec_State *state, const void *config)\r
-\r
-#if Py_UNICODE_SIZE == 4\r
-#define UCS4INVALID(code) \\r
- if ((code) > 0xFFFF) \\r
- return 1;\r
-#else\r
-#define UCS4INVALID(code) \\r
- if (0) ;\r
-#endif\r
-\r
-#define NEXT_IN(i) \\r
- (*inbuf) += (i); \\r
- (inleft) -= (i);\r
-#define NEXT_OUT(o) \\r
- (*outbuf) += (o); \\r
- (outleft) -= (o);\r
-#define NEXT(i, o) \\r
- NEXT_IN(i) NEXT_OUT(o)\r
-\r
-#define REQUIRE_INBUF(n) \\r
- if (inleft < (n)) \\r
- return MBERR_TOOFEW;\r
-#define REQUIRE_OUTBUF(n) \\r
- if (outleft < (n)) \\r
- return MBERR_TOOSMALL;\r
-\r
-#define IN1 ((*inbuf)[0])\r
-#define IN2 ((*inbuf)[1])\r
-#define IN3 ((*inbuf)[2])\r
-#define IN4 ((*inbuf)[3])\r
-\r
-#define OUT1(c) ((*outbuf)[0]) = (c);\r
-#define OUT2(c) ((*outbuf)[1]) = (c);\r
-#define OUT3(c) ((*outbuf)[2]) = (c);\r
-#define OUT4(c) ((*outbuf)[3]) = (c);\r
-\r
-#define WRITE1(c1) \\r
- REQUIRE_OUTBUF(1) \\r
- (*outbuf)[0] = (c1);\r
-#define WRITE2(c1, c2) \\r
- REQUIRE_OUTBUF(2) \\r
- (*outbuf)[0] = (c1); \\r
- (*outbuf)[1] = (c2);\r
-#define WRITE3(c1, c2, c3) \\r
- REQUIRE_OUTBUF(3) \\r
- (*outbuf)[0] = (c1); \\r
- (*outbuf)[1] = (c2); \\r
- (*outbuf)[2] = (c3);\r
-#define WRITE4(c1, c2, c3, c4) \\r
- REQUIRE_OUTBUF(4) \\r
- (*outbuf)[0] = (c1); \\r
- (*outbuf)[1] = (c2); \\r
- (*outbuf)[2] = (c3); \\r
- (*outbuf)[3] = (c4);\r
-\r
-#if Py_UNICODE_SIZE == 2\r
-# define WRITEUCS4(c) \\r
- REQUIRE_OUTBUF(2) \\r
- (*outbuf)[0] = 0xd800 + (((c) - 0x10000) >> 10); \\r
- (*outbuf)[1] = 0xdc00 + (((c) - 0x10000) & 0x3ff); \\r
- NEXT_OUT(2)\r
-#else\r
-# define WRITEUCS4(c) \\r
- REQUIRE_OUTBUF(1) \\r
- **outbuf = (Py_UNICODE)(c); \\r
- NEXT_OUT(1)\r
-#endif\r
-\r
-#define _TRYMAP_ENC(m, assi, val) \\r
- ((m)->map != NULL && (val) >= (m)->bottom && \\r
- (val)<= (m)->top && ((assi) = (m)->map[(val) - \\r
- (m)->bottom]) != NOCHAR)\r
-#define TRYMAP_ENC_COND(charset, assi, uni) \\r
- _TRYMAP_ENC(&charset##_encmap[(uni) >> 8], assi, (uni) & 0xff)\r
-#define TRYMAP_ENC(charset, assi, uni) \\r
- if TRYMAP_ENC_COND(charset, assi, uni)\r
-\r
-#define _TRYMAP_DEC(m, assi, val) \\r
- ((m)->map != NULL && (val) >= (m)->bottom && \\r
- (val)<= (m)->top && ((assi) = (m)->map[(val) - \\r
- (m)->bottom]) != UNIINV)\r
-#define TRYMAP_DEC(charset, assi, c1, c2) \\r
- if _TRYMAP_DEC(&charset##_decmap[c1], assi, c2)\r
-\r
-#define _TRYMAP_ENC_MPLANE(m, assplane, asshi, asslo, val) \\r
- ((m)->map != NULL && (val) >= (m)->bottom && \\r
- (val)<= (m)->top && \\r
- ((assplane) = (m)->map[((val) - (m)->bottom)*3]) != 0 && \\r
- (((asshi) = (m)->map[((val) - (m)->bottom)*3 + 1]), 1) && \\r
- (((asslo) = (m)->map[((val) - (m)->bottom)*3 + 2]), 1))\r
-#define TRYMAP_ENC_MPLANE(charset, assplane, asshi, asslo, uni) \\r
- if _TRYMAP_ENC_MPLANE(&charset##_encmap[(uni) >> 8], \\r
- assplane, asshi, asslo, (uni) & 0xff)\r
-#define TRYMAP_DEC_MPLANE(charset, assi, plane, c1, c2) \\r
- if _TRYMAP_DEC(&charset##_decmap[plane][c1], assi, c2)\r
-\r
-#if Py_UNICODE_SIZE == 2\r
-#define DECODE_SURROGATE(c) \\r
- if (c >> 10 == 0xd800 >> 10) { /* high surrogate */ \\r
- REQUIRE_INBUF(2) \\r
- if (IN2 >> 10 == 0xdc00 >> 10) { /* low surrogate */ \\r
- c = 0x10000 + ((ucs4_t)(c - 0xd800) << 10) + \\r
- ((ucs4_t)(IN2) - 0xdc00); \\r
- } \\r
- }\r
-#define GET_INSIZE(c) ((c) > 0xffff ? 2 : 1)\r
-#else\r
-#define DECODE_SURROGATE(c) {;}\r
-#define GET_INSIZE(c) 1\r
-#endif\r
-\r
-#define BEGIN_MAPPINGS_LIST static const struct dbcs_map _mapping_list[] = {\r
-#define MAPPING_ENCONLY(enc) {#enc, (void*)enc##_encmap, NULL},\r
-#define MAPPING_DECONLY(enc) {#enc, NULL, (void*)enc##_decmap},\r
-#define MAPPING_ENCDEC(enc) {#enc, (void*)enc##_encmap, (void*)enc##_decmap},\r
-#define END_MAPPINGS_LIST \\r
- {"", NULL, NULL} }; \\r
- static const struct dbcs_map *mapping_list = \\r
- (const struct dbcs_map *)_mapping_list;\r
-\r
-#define BEGIN_CODECS_LIST static const MultibyteCodec _codec_list[] = {\r
-#define _STATEFUL_METHODS(enc) \\r
- enc##_encode, \\r
- enc##_encode_init, \\r
- enc##_encode_reset, \\r
- enc##_decode, \\r
- enc##_decode_init, \\r
- enc##_decode_reset,\r
-#define _STATELESS_METHODS(enc) \\r
- enc##_encode, NULL, NULL, \\r
- enc##_decode, NULL, NULL,\r
-#define CODEC_STATEFUL(enc) { \\r
- #enc, NULL, NULL, \\r
- _STATEFUL_METHODS(enc) \\r
-},\r
-#define CODEC_STATELESS(enc) { \\r
- #enc, NULL, NULL, \\r
- _STATELESS_METHODS(enc) \\r
-},\r
-#define CODEC_STATELESS_WINIT(enc) { \\r
- #enc, NULL, \\r
- enc##_codec_init, \\r
- _STATELESS_METHODS(enc) \\r
-},\r
-#define END_CODECS_LIST \\r
- {"", NULL,} }; \\r
- static const MultibyteCodec *codec_list = \\r
- (const MultibyteCodec *)_codec_list;\r
-\r
-static PyObject *\r
-getmultibytecodec(void)\r
-{\r
- static PyObject *cofunc = NULL;\r
-\r
- if (cofunc == NULL) {\r
- PyObject *mod = PyImport_ImportModuleNoBlock("_multibytecodec");\r
- if (mod == NULL)\r
- return NULL;\r
- cofunc = PyObject_GetAttrString(mod, "__create_codec");\r
- Py_DECREF(mod);\r
- }\r
- return cofunc;\r
-}\r
-\r
-static PyObject *\r
-getcodec(PyObject *self, PyObject *encoding)\r
-{\r
- PyObject *codecobj, *r, *cofunc;\r
- const MultibyteCodec *codec;\r
- const char *enc;\r
-\r
- if (!PyString_Check(encoding)) {\r
- PyErr_SetString(PyExc_TypeError,\r
- "encoding name must be a string.");\r
- return NULL;\r
- }\r
-\r
- cofunc = getmultibytecodec();\r
- if (cofunc == NULL)\r
- return NULL;\r
-\r
- enc = PyString_AS_STRING(encoding);\r
- for (codec = codec_list; codec->encoding[0]; codec++)\r
- if (strcmp(codec->encoding, enc) == 0)\r
- break;\r
-\r
- if (codec->encoding[0] == '\0') {\r
- PyErr_SetString(PyExc_LookupError,\r
- "no such codec is supported.");\r
- return NULL;\r
- }\r
-\r
- codecobj = PyCapsule_New((void *)codec, PyMultibyteCodec_CAPSULE_NAME, NULL);\r
- if (codecobj == NULL)\r
- return NULL;\r
-\r
- r = PyObject_CallFunctionObjArgs(cofunc, codecobj, NULL);\r
- Py_DECREF(codecobj);\r
-\r
- return r;\r
-}\r
-\r
-static struct PyMethodDef __methods[] = {\r
- {"getcodec", (PyCFunction)getcodec, METH_O, ""},\r
- {NULL, NULL},\r
-};\r
-\r
-static int\r
-register_maps(PyObject *module)\r
-{\r
- const struct dbcs_map *h;\r
-\r
- for (h = mapping_list; h->charset[0] != '\0'; h++) {\r
- char mhname[256] = "__map_";\r
- int r;\r
- strcpy(mhname + sizeof("__map_") - 1, h->charset);\r
- r = PyModule_AddObject(module, mhname,\r
- PyCapsule_New((void *)h, PyMultibyteCodec_CAPSULE_NAME, NULL));\r
- if (r == -1)\r
- return -1;\r
- }\r
- return 0;\r
-}\r
-\r
-#ifdef USING_BINARY_PAIR_SEARCH\r
-static DBCHAR\r
-find_pairencmap(ucs2_t body, ucs2_t modifier,\r
- const struct pair_encodemap *haystack, int haystacksize)\r
-{\r
- int pos, min, max;\r
- ucs4_t value = body << 16 | modifier;\r
-\r
- min = 0;\r
- max = haystacksize;\r
-\r
- for (pos = haystacksize >> 1; min != max; pos = (min + max) >> 1)\r
- if (value < haystack[pos].uniseq) {\r
- if (max == pos) break;\r
- else max = pos;\r
- }\r
- else if (value > haystack[pos].uniseq) {\r
- if (min == pos) break;\r
- else min = pos;\r
- }\r
- else\r
- break;\r
-\r
- if (value == haystack[pos].uniseq)\r
- return haystack[pos].code;\r
- else\r
- return DBCINV;\r
-}\r
-#endif\r
-\r
-#ifdef USING_IMPORTED_MAPS\r
-#define IMPORT_MAP(locale, charset, encmap, decmap) \\r
- importmap("_codecs_" #locale, "__map_" #charset, \\r
- (const void**)encmap, (const void**)decmap)\r
-\r
-static int\r
-importmap(const char *modname, const char *symbol,\r
- const void **encmap, const void **decmap)\r
-{\r
- PyObject *o, *mod;\r
-\r
- mod = PyImport_ImportModule((char *)modname);\r
- if (mod == NULL)\r
- return -1;\r
-\r
- o = PyObject_GetAttrString(mod, (char*)symbol);\r
- if (o == NULL)\r
- goto errorexit;\r
- else if (!PyCapsule_IsValid(o, PyMultibyteCodec_CAPSULE_NAME)) {\r
- PyErr_SetString(PyExc_ValueError,\r
- "map data must be a Capsule.");\r
- goto errorexit;\r
- }\r
- else {\r
- struct dbcs_map *map;\r
- map = PyCapsule_GetPointer(o, PyMultibyteCodec_CAPSULE_NAME);\r
- if (encmap != NULL)\r
- *encmap = map->encmap;\r
- if (decmap != NULL)\r
- *decmap = map->decmap;\r
- Py_DECREF(o);\r
- }\r
-\r
- Py_DECREF(mod);\r
- return 0;\r
-\r
-errorexit:\r
- Py_DECREF(mod);\r
- return -1;\r
-}\r
-#endif\r
-\r
-#define I_AM_A_MODULE_FOR(loc) \\r
- void \\r
- init_codecs_##loc(void) \\r
- { \\r
- PyObject *m = Py_InitModule("_codecs_" #loc, __methods);\\r
- if (m != NULL) \\r
- (void)register_maps(m); \\r
- }\r
-\r
-#endif\r