1 #ifndef PEGEN_H
2 #define PEGEN_H
3
4 #include <Python.h>
5 #include <pycore_ast.h>
6 #include <pycore_token.h>
7
8 #if 0
9 #define PyPARSE_YIELD_IS_KEYWORD 0x0001
10 #endif
11
12 #define PyPARSE_DONT_IMPLY_DEDENT 0x0002
13
14 #if 0
15 #define PyPARSE_WITH_IS_KEYWORD 0x0003
16 #define PyPARSE_PRINT_IS_FUNCTION 0x0004
17 #define PyPARSE_UNICODE_LITERALS 0x0008
18 #endif
19
20 #define PyPARSE_IGNORE_COOKIE 0x0010
21 #define PyPARSE_BARRY_AS_BDFL 0x0020
22 #define PyPARSE_TYPE_COMMENTS 0x0040
23 #define PyPARSE_ALLOW_INCOMPLETE_INPUT 0x0100
24
25 #define CURRENT_POS (-5)
26
27 typedef struct _memo {
28 int type;
29 void *node;
30 int mark;
31 struct _memo *next;
32 } Memo;
33
34 typedef struct {
35 int type;
36 PyObject *bytes;
37 int level;
38 int lineno, col_offset, end_lineno, end_col_offset;
39 Memo *memo;
40 PyObject *metadata;
41 } Token;
42
43 typedef struct {
44 const char *str;
45 int type;
46 } KeywordToken;
47
48
49 typedef struct {
50 struct {
51 int lineno;
52 char *comment; // The " <tag>" in "# type: ignore <tag>"
53 } *items;
54 size_t size;
55 size_t num_items;
56 } growable_comment_array;
57
58 typedef struct {
59 struct tok_state *tok;
60 Token **tokens;
61 int mark;
62 int fill, size;
63 PyArena *arena;
64 KeywordToken **keywords;
65 char **soft_keywords;
66 int n_keyword_lists;
67 int start_rule;
68 int *errcode;
69 int parsing_started;
70 PyObject* normalize;
71 int starting_lineno;
72 int starting_col_offset;
73 int error_indicator;
74 int flags;
75 int feature_version;
76 growable_comment_array type_ignore_comments;
77 Token *known_err_token;
78 int level;
79 int call_invalid_rules;
80 int debug;
81 } Parser;
82
83 typedef struct {
84 cmpop_ty cmpop;
85 expr_ty expr;
86 } CmpopExprPair;
87
88 typedef struct {
89 expr_ty key;
90 expr_ty value;
91 } KeyValuePair;
92
93 typedef struct {
94 expr_ty key;
95 pattern_ty pattern;
96 } KeyPatternPair;
97
98 typedef struct {
99 arg_ty arg;
100 expr_ty value;
101 } NameDefaultPair;
102
103 typedef struct {
104 asdl_arg_seq *plain_names;
105 asdl_seq *names_with_defaults; // asdl_seq* of NameDefaultsPair's
106 } SlashWithDefault;
107
108 typedef struct {
109 arg_ty vararg;
110 asdl_seq *kwonlyargs; // asdl_seq* of NameDefaultsPair's
111 arg_ty kwarg;
112 } StarEtc;
113
114 typedef struct { operator_ty kind; } AugOperator;
115 typedef struct {
116 void *element;
117 int is_keyword;
118 } KeywordOrStarred;
119
120 typedef struct {
121 void *result;
122 PyObject *metadata;
123 } ResultTokenWithMetadata;
124
125 // Internal parser functions
126 #if defined(Py_DEBUG)
127 void _PyPegen_clear_memo_statistics(void);
128 PyObject *_PyPegen_get_memo_statistics(void);
129 #endif
130
131 int _PyPegen_insert_memo(Parser *p, int mark, int type, void *node);
132 int _PyPegen_update_memo(Parser *p, int mark, int type, void *node);
133 int _PyPegen_is_memoized(Parser *p, int type, void *pres);
134
135 int _PyPegen_lookahead_with_name(int, expr_ty (func)(Parser *), Parser *);
136 int _PyPegen_lookahead_with_int(int, Token *(func)(Parser *, int), Parser *, int);
137 int _PyPegen_lookahead_with_string(int , expr_ty (func)(Parser *, const char*), Parser *, const char*);
138 int _PyPegen_lookahead(int, void *(func)(Parser *), Parser *);
139
140 Token *_PyPegen_expect_token(Parser *p, int type);
141 void* _PyPegen_expect_forced_result(Parser *p, void* result, const char* expected);
142 Token *_PyPegen_expect_forced_token(Parser *p, int type, const char* expected);
143 expr_ty _PyPegen_expect_soft_keyword(Parser *p, const char *keyword);
144 expr_ty _PyPegen_soft_keyword_token(Parser *p);
145 expr_ty _PyPegen_fstring_middle_token(Parser* p);
146 Token *_PyPegen_get_last_nonnwhitespace_token(Parser *);
147 int _PyPegen_fill_token(Parser *p);
148 expr_ty _PyPegen_name_token(Parser *p);
149 expr_ty _PyPegen_number_token(Parser *p);
150 void *_PyPegen_string_token(Parser *p);
151 Py_ssize_t _PyPegen_byte_offset_to_character_offset_line(PyObject *line, Py_ssize_t col_offset, Py_ssize_t end_col_offset);
152 Py_ssize_t _PyPegen_byte_offset_to_character_offset(PyObject *line, Py_ssize_t col_offset);
153 Py_ssize_t _PyPegen_byte_offset_to_character_offset_raw(const char*, Py_ssize_t col_offset);
154
155 // Error handling functions and APIs
156 typedef enum {
157 STAR_TARGETS,
158 DEL_TARGETS,
159 FOR_TARGETS
160 } TARGETS_TYPE;
161
162 int _Pypegen_raise_decode_error(Parser *p);
163 void _PyPegen_raise_tokenizer_init_error(PyObject *filename);
164 int _Pypegen_tokenizer_error(Parser *p);
165 void *_PyPegen_raise_error(Parser *p, PyObject *errtype, int use_mark, const char *errmsg, ...);
166 void *_PyPegen_raise_error_known_location(Parser *p, PyObject *errtype,
167 Py_ssize_t lineno, Py_ssize_t col_offset,
168 Py_ssize_t end_lineno, Py_ssize_t end_col_offset,
169 const char *errmsg, va_list va);
170 void _Pypegen_set_syntax_error(Parser* p, Token* last_token);
171 void _Pypegen_stack_overflow(Parser *p);
172
173 Py_LOCAL_INLINE(void *)
RAISE_ERROR_KNOWN_LOCATION(Parser * p,PyObject * errtype,Py_ssize_t lineno,Py_ssize_t col_offset,Py_ssize_t end_lineno,Py_ssize_t end_col_offset,const char * errmsg,...)174 RAISE_ERROR_KNOWN_LOCATION(Parser *p, PyObject *errtype,
175 Py_ssize_t lineno, Py_ssize_t col_offset,
176 Py_ssize_t end_lineno, Py_ssize_t end_col_offset,
177 const char *errmsg, ...)
178 {
179 va_list va;
180 va_start(va, errmsg);
181 Py_ssize_t _col_offset = (col_offset == CURRENT_POS ? CURRENT_POS : col_offset + 1);
182 Py_ssize_t _end_col_offset = (end_col_offset == CURRENT_POS ? CURRENT_POS : end_col_offset + 1);
183 _PyPegen_raise_error_known_location(p, errtype, lineno, _col_offset, end_lineno, _end_col_offset, errmsg, va);
184 va_end(va);
185 return NULL;
186 }
187 #define RAISE_SYNTAX_ERROR(msg, ...) _PyPegen_raise_error(p, PyExc_SyntaxError, 0, msg, ##__VA_ARGS__)
188 #define RAISE_INDENTATION_ERROR(msg, ...) _PyPegen_raise_error(p, PyExc_IndentationError, 0, msg, ##__VA_ARGS__)
189 #define RAISE_SYNTAX_ERROR_ON_NEXT_TOKEN(msg, ...) _PyPegen_raise_error(p, PyExc_SyntaxError, 1, msg, ##__VA_ARGS__)
190 #define RAISE_SYNTAX_ERROR_KNOWN_RANGE(a, b, msg, ...) \
191 RAISE_ERROR_KNOWN_LOCATION(p, PyExc_SyntaxError, (a)->lineno, (a)->col_offset, (b)->end_lineno, (b)->end_col_offset, msg, ##__VA_ARGS__)
192 #define RAISE_SYNTAX_ERROR_KNOWN_LOCATION(a, msg, ...) \
193 RAISE_ERROR_KNOWN_LOCATION(p, PyExc_SyntaxError, (a)->lineno, (a)->col_offset, (a)->end_lineno, (a)->end_col_offset, msg, ##__VA_ARGS__)
194 #define RAISE_SYNTAX_ERROR_STARTING_FROM(a, msg, ...) \
195 RAISE_ERROR_KNOWN_LOCATION(p, PyExc_SyntaxError, (a)->lineno, (a)->col_offset, CURRENT_POS, CURRENT_POS, msg, ##__VA_ARGS__)
196 #define RAISE_SYNTAX_ERROR_INVALID_TARGET(type, e) _RAISE_SYNTAX_ERROR_INVALID_TARGET(p, type, e)
197
198 Py_LOCAL_INLINE(void *)
CHECK_CALL(Parser * p,void * result)199 CHECK_CALL(Parser *p, void *result)
200 {
201 if (result == NULL) {
202 assert(PyErr_Occurred());
203 p->error_indicator = 1;
204 }
205 return result;
206 }
207
208 /* This is needed for helper functions that are allowed to
209 return NULL without an error. Example: _PyPegen_seq_extract_starred_exprs */
210 Py_LOCAL_INLINE(void *)
CHECK_CALL_NULL_ALLOWED(Parser * p,void * result)211 CHECK_CALL_NULL_ALLOWED(Parser *p, void *result)
212 {
213 if (result == NULL && PyErr_Occurred()) {
214 p->error_indicator = 1;
215 }
216 return result;
217 }
218
219 #define CHECK(type, result) ((type) CHECK_CALL(p, result))
220 #define CHECK_NULL_ALLOWED(type, result) ((type) CHECK_CALL_NULL_ALLOWED(p, result))
221
222 expr_ty _PyPegen_get_invalid_target(expr_ty e, TARGETS_TYPE targets_type);
223 const char *_PyPegen_get_expr_name(expr_ty);
224 Py_LOCAL_INLINE(void *)
_RAISE_SYNTAX_ERROR_INVALID_TARGET(Parser * p,TARGETS_TYPE type,void * e)225 _RAISE_SYNTAX_ERROR_INVALID_TARGET(Parser *p, TARGETS_TYPE type, void *e)
226 {
227 expr_ty invalid_target = CHECK_NULL_ALLOWED(expr_ty, _PyPegen_get_invalid_target(e, type));
228 if (invalid_target != NULL) {
229 const char *msg;
230 if (type == STAR_TARGETS || type == FOR_TARGETS) {
231 msg = "cannot assign to %s";
232 }
233 else {
234 msg = "cannot delete %s";
235 }
236 return RAISE_SYNTAX_ERROR_KNOWN_LOCATION(
237 invalid_target,
238 msg,
239 _PyPegen_get_expr_name(invalid_target)
240 );
241 return RAISE_SYNTAX_ERROR_KNOWN_LOCATION(invalid_target, "invalid syntax");
242 }
243 return NULL;
244 }
245
246 // Action utility functions
247
248 void *_PyPegen_dummy_name(Parser *p, ...);
249 void * _PyPegen_seq_last_item(asdl_seq *seq);
250 #define PyPegen_last_item(seq, type) ((type)_PyPegen_seq_last_item((asdl_seq*)seq))
251 void * _PyPegen_seq_first_item(asdl_seq *seq);
252 #define PyPegen_first_item(seq, type) ((type)_PyPegen_seq_first_item((asdl_seq*)seq))
253 #define UNUSED(expr) do { (void)(expr); } while (0)
254 #define EXTRA_EXPR(head, tail) head->lineno, (head)->col_offset, (tail)->end_lineno, (tail)->end_col_offset, p->arena
255 #define EXTRA _start_lineno, _start_col_offset, _end_lineno, _end_col_offset, p->arena
256 PyObject *_PyPegen_new_type_comment(Parser *, const char *);
257
258 Py_LOCAL_INLINE(PyObject *)
NEW_TYPE_COMMENT(Parser * p,Token * tc)259 NEW_TYPE_COMMENT(Parser *p, Token *tc)
260 {
261 if (tc == NULL) {
262 return NULL;
263 }
264 const char *bytes = PyBytes_AsString(tc->bytes);
265 if (bytes == NULL) {
266 goto error;
267 }
268 PyObject *tco = _PyPegen_new_type_comment(p, bytes);
269 if (tco == NULL) {
270 goto error;
271 }
272 return tco;
273 error:
274 p->error_indicator = 1; // Inline CHECK_CALL
275 return NULL;
276 }
277
278 Py_LOCAL_INLINE(void *)
INVALID_VERSION_CHECK(Parser * p,int version,char * msg,void * node)279 INVALID_VERSION_CHECK(Parser *p, int version, char *msg, void *node)
280 {
281 if (node == NULL) {
282 p->error_indicator = 1; // Inline CHECK_CALL
283 return NULL;
284 }
285 if (p->feature_version < version) {
286 p->error_indicator = 1;
287 return RAISE_SYNTAX_ERROR("%s only supported in Python 3.%i and greater",
288 msg, version);
289 }
290 return node;
291 }
292
293 #define CHECK_VERSION(type, version, msg, node) ((type) INVALID_VERSION_CHECK(p, version, msg, node))
294
295 arg_ty _PyPegen_add_type_comment_to_arg(Parser *, arg_ty, Token *);
296 PyObject *_PyPegen_new_identifier(Parser *, const char *);
297 asdl_seq *_PyPegen_singleton_seq(Parser *, void *);
298 asdl_seq *_PyPegen_seq_insert_in_front(Parser *, void *, asdl_seq *);
299 asdl_seq *_PyPegen_seq_append_to_end(Parser *, asdl_seq *, void *);
300 asdl_seq *_PyPegen_seq_flatten(Parser *, asdl_seq *);
301 expr_ty _PyPegen_join_names_with_dot(Parser *, expr_ty, expr_ty);
302 int _PyPegen_seq_count_dots(asdl_seq *);
303 alias_ty _PyPegen_alias_for_star(Parser *, int, int, int, int, PyArena *);
304 asdl_identifier_seq *_PyPegen_map_names_to_ids(Parser *, asdl_expr_seq *);
305 CmpopExprPair *_PyPegen_cmpop_expr_pair(Parser *, cmpop_ty, expr_ty);
306 asdl_int_seq *_PyPegen_get_cmpops(Parser *p, asdl_seq *);
307 asdl_expr_seq *_PyPegen_get_exprs(Parser *, asdl_seq *);
308 expr_ty _PyPegen_set_expr_context(Parser *, expr_ty, expr_context_ty);
309 KeyValuePair *_PyPegen_key_value_pair(Parser *, expr_ty, expr_ty);
310 asdl_expr_seq *_PyPegen_get_keys(Parser *, asdl_seq *);
311 asdl_expr_seq *_PyPegen_get_values(Parser *, asdl_seq *);
312 KeyPatternPair *_PyPegen_key_pattern_pair(Parser *, expr_ty, pattern_ty);
313 asdl_expr_seq *_PyPegen_get_pattern_keys(Parser *, asdl_seq *);
314 asdl_pattern_seq *_PyPegen_get_patterns(Parser *, asdl_seq *);
315 NameDefaultPair *_PyPegen_name_default_pair(Parser *, arg_ty, expr_ty, Token *);
316 SlashWithDefault *_PyPegen_slash_with_default(Parser *, asdl_arg_seq *, asdl_seq *);
317 StarEtc *_PyPegen_star_etc(Parser *, arg_ty, asdl_seq *, arg_ty);
318 arguments_ty _PyPegen_make_arguments(Parser *, asdl_arg_seq *, SlashWithDefault *,
319 asdl_arg_seq *, asdl_seq *, StarEtc *);
320 arguments_ty _PyPegen_empty_arguments(Parser *);
321 expr_ty _PyPegen_formatted_value(Parser *, expr_ty, Token *, ResultTokenWithMetadata *, ResultTokenWithMetadata *, Token *,
322 int, int, int, int, PyArena *);
323 AugOperator *_PyPegen_augoperator(Parser*, operator_ty type);
324 stmt_ty _PyPegen_function_def_decorators(Parser *, asdl_expr_seq *, stmt_ty);
325 stmt_ty _PyPegen_class_def_decorators(Parser *, asdl_expr_seq *, stmt_ty);
326 KeywordOrStarred *_PyPegen_keyword_or_starred(Parser *, void *, int);
327 asdl_expr_seq *_PyPegen_seq_extract_starred_exprs(Parser *, asdl_seq *);
328 asdl_keyword_seq *_PyPegen_seq_delete_starred_exprs(Parser *, asdl_seq *);
329 expr_ty _PyPegen_collect_call_seqs(Parser *, asdl_expr_seq *, asdl_seq *,
330 int lineno, int col_offset, int end_lineno,
331 int end_col_offset, PyArena *arena);
332 expr_ty _PyPegen_constant_from_token(Parser* p, Token* tok);
333 expr_ty _PyPegen_decoded_constant_from_token(Parser* p, Token* tok);
334 expr_ty _PyPegen_constant_from_string(Parser* p, Token* tok);
335 expr_ty _PyPegen_concatenate_strings(Parser *p, asdl_expr_seq *, int, int, int, int, PyArena *);
336 expr_ty _PyPegen_FetchRawForm(Parser *p, int, int, int, int);
337 expr_ty _PyPegen_ensure_imaginary(Parser *p, expr_ty);
338 expr_ty _PyPegen_ensure_real(Parser *p, expr_ty);
339 asdl_seq *_PyPegen_join_sequences(Parser *, asdl_seq *, asdl_seq *);
340 int _PyPegen_check_barry_as_flufl(Parser *, Token *);
341 int _PyPegen_check_legacy_stmt(Parser *p, expr_ty t);
342 ResultTokenWithMetadata *_PyPegen_check_fstring_conversion(Parser *p, Token *, expr_ty t);
343 ResultTokenWithMetadata *_PyPegen_setup_full_format_spec(Parser *, Token *, asdl_expr_seq *, int, int,
344 int, int, PyArena *);
345 mod_ty _PyPegen_make_module(Parser *, asdl_stmt_seq *);
346 void *_PyPegen_arguments_parsing_error(Parser *, expr_ty);
347 expr_ty _PyPegen_get_last_comprehension_item(comprehension_ty comprehension);
348 void *_PyPegen_nonparen_genexp_in_call(Parser *p, expr_ty args, asdl_comprehension_seq *comprehensions);
349
350 // Parser API
351
352 Parser *_PyPegen_Parser_New(struct tok_state *, int, int, int, int *, PyArena *);
353 void _PyPegen_Parser_Free(Parser *);
354 mod_ty _PyPegen_run_parser_from_file_pointer(FILE *, int, PyObject *, const char *,
355 const char *, const char *, PyCompilerFlags *, int *, PyObject **,
356 PyArena *);
357 void *_PyPegen_run_parser(Parser *);
358 mod_ty _PyPegen_run_parser_from_string(const char *, int, PyObject *, PyCompilerFlags *, PyArena *);
359 asdl_stmt_seq *_PyPegen_interactive_exit(Parser *);
360
361 // TODO: move to the correct place in this file
362 expr_ty _PyPegen_joined_str(Parser *p, Token* a, asdl_expr_seq* expr, Token*b);
363
364 // Generated function in parse.c - function definition in python.gram
365 void *_PyPegen_parse(Parser *);
366
367 #endif
368