| @@ -43,6 +43,18 @@ SOFTWARE. | |||||
| #define malloc PyObject_Malloc // XXX: yuck | #define malloc PyObject_Malloc // XXX: yuck | ||||
| #define free PyObject_Free | #define free PyObject_Free | ||||
| /* Unicode support macros */ | |||||
| #if defined(IS_PY3K) && PYTHON_MINOR_VERSION >= 3 | |||||
| #define PEP_393 | |||||
| #endif | |||||
| #ifdef PEP_393 | |||||
| #define Unicode Py_UCS4 | |||||
| #else | |||||
| #define Unicode Py_UNICODE | |||||
| #endif | |||||
| /* Error handling macros */ | /* Error handling macros */ | ||||
| #define BAD_ROUTE self->route_state | #define BAD_ROUTE self->route_state | ||||
| @@ -63,18 +75,15 @@ extern PyObject* definitions; | |||||
| /* Structs */ | /* Structs */ | ||||
| struct Textbuffer { | |||||
| typedef struct { | |||||
| Py_ssize_t size; | Py_ssize_t size; | ||||
| Py_UNICODE* data; | Py_UNICODE* data; | ||||
| struct Textbuffer* prev; | |||||
| struct Textbuffer* next; | |||||
| }; | |||||
| typedef struct Textbuffer Textbuffer; | |||||
| } Textbuffer; | |||||
| struct Stack { | struct Stack { | ||||
| PyObject* stack; | PyObject* stack; | ||||
| uint64_t context; | uint64_t context; | ||||
| struct Textbuffer* textbuffer; | |||||
| Textbuffer* textbuffer; | |||||
| struct Stack* next; | struct Stack* next; | ||||
| }; | }; | ||||
| typedef struct Stack Stack; | typedef struct Stack Stack; | ||||
| @@ -30,11 +30,6 @@ SOFTWARE. | |||||
| #define HEXDIGITS "0123456789abcdefABCDEF" | #define HEXDIGITS "0123456789abcdefABCDEF" | ||||
| #define ALPHANUM "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ" | #define ALPHANUM "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ" | ||||
| static const char MARKERS[] = { | |||||
| '{', '}', '[', ']', '<', '>', '|', '=', '&', '\'', '#', '*', ';', ':', '/', | |||||
| '-', '!', '\n', '\0'}; | |||||
| #define NUM_MARKERS 19 | |||||
| #define MAX_BRACES 255 | #define MAX_BRACES 255 | ||||
| #define MAX_ENTITY_SIZE 8 | #define MAX_ENTITY_SIZE 8 | ||||
| @@ -45,12 +40,6 @@ static const char MARKERS[] = { | |||||
| #define IS_SCHEME(scheme, slashes, reverse) \ | #define IS_SCHEME(scheme, slashes, reverse) \ | ||||
| (call_def_func("is_scheme", scheme, slashes ? Py_True : Py_False, reverse ? Py_True : Py_False)) | (call_def_func("is_scheme", scheme, slashes ? Py_True : Py_False, reverse ? Py_True : Py_False)) | ||||
| #ifdef IS_PY3K | |||||
| #define NEW_INT_FUNC PyLong_FromSsize_t | |||||
| #else | |||||
| #define NEW_INT_FUNC PyInt_FromSsize_t | |||||
| #endif | |||||
| typedef struct { | typedef struct { | ||||
| PyObject* title; | PyObject* title; | ||||
| int level; | int level; | ||||
| @@ -798,7 +787,11 @@ static int Tokenizer_parse_heading(Tokenizer* self) | |||||
| self->global ^= GL_HEADING; | self->global ^= GL_HEADING; | ||||
| return 0; | return 0; | ||||
| } | } | ||||
| level = NEW_INT_FUNC(heading->level); | |||||
| #ifdef IS_PY3K | |||||
| level = PyLong_FromSsize_t(heading->level); | |||||
| #else | |||||
| level = PyInt_FromSsize_t(heading->level); | |||||
| #endif | |||||
| if (!level) { | if (!level) { | ||||
| Py_DECREF(heading->title); | Py_DECREF(heading->title); | ||||
| free(heading); | free(heading); | ||||
| @@ -24,6 +24,12 @@ SOFTWARE. | |||||
| #include "common.h" | #include "common.h" | ||||
| static const char MARKERS[] = { | |||||
| '{', '}', '[', ']', '<', '>', '|', '=', '&', '\'', '#', '*', ';', ':', '/', | |||||
| '-', '!', '\n', '\0'}; | |||||
| #define NUM_MARKERS 19 | |||||
| /* Functions */ | /* Functions */ | ||||
| PyObject* Tokenizer_parse(Tokenizer*, uint64_t, int); | PyObject* Tokenizer_parse(Tokenizer*, uint64_t, int); | ||||
| @@ -27,6 +27,6 @@ output: [Text(text="𐌲𐌿𐍄𐌰𐍂𐌰𐌶𐌳𐌰")] | |||||
| --- | --- | ||||
| name: large | name: large | ||||
| label: a lot of text, requiring multiple textbuffer blocks in the C tokenizer | |||||
| label: a lot of text, requiring proper storage in the C tokenizer | |||||
| input: "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" | input: "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" | ||||
| output: [Text(text="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")] | output: [Text(text="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")] | ||||