1 #ifndef PEGEN_H
2 #define PEGEN_H
3
4 #define PY_SSIZE_T_CLEAN
5 #include <Python.h>
6 #include <token.h>
7 #include <pycore_ast.h>
8
9 #if 0
10 #define PyPARSE_YIELD_IS_KEYWORD 0x0001
11 #endif
12
13 #define PyPARSE_DONT_IMPLY_DEDENT 0x0002
14
15 #if 0
16 #define PyPARSE_WITH_IS_KEYWORD 0x0003
17 #define PyPARSE_PRINT_IS_FUNCTION 0x0004
18 #define PyPARSE_UNICODE_LITERALS 0x0008
19 #endif
20
21 #define PyPARSE_IGNORE_COOKIE 0x0010
22 #define PyPARSE_BARRY_AS_BDFL 0x0020
23 #define PyPARSE_TYPE_COMMENTS 0x0040
24 #define PyPARSE_ASYNC_HACKS 0x0080
25 #define PyPARSE_ALLOW_INCOMPLETE_INPUT 0x0100
26
27 #define CURRENT_POS (-5)
28
29 typedef struct _memo {
30 int type;
31 void *node;
32 int mark;
33 struct _memo *next;
34 } Memo;
35
36 typedef struct {
37 int type;
38 PyObject *bytes;
39 int level;
40 int lineno, col_offset, end_lineno, end_col_offset;
41 Memo *memo;
42 } Token;
43
44 typedef struct {
45 char *str;
46 int type;
47 } KeywordToken;
48
49
50 typedef struct {
51 struct {
52 int lineno;
53 char *comment; // The " <tag>" in "# type: ignore <tag>"
54 } *items;
55 size_t size;
56 size_t num_items;
57 } growable_comment_array;
58
59 typedef struct {
60 struct tok_state *tok;
61 Token **tokens;
62 int mark;
63 int fill, size;
64 PyArena *arena;
65 KeywordToken **keywords;
66 char **soft_keywords;
67 int n_keyword_lists;
68 int start_rule;
69 int *errcode;
70 int parsing_started;
71 PyObject* normalize;
72 int starting_lineno;
73 int starting_col_offset;
74 int error_indicator;
75 int flags;
76 int feature_version;
77 growable_comment_array type_ignore_comments;
78 Token *known_err_token;
79 int level;
80 int call_invalid_rules;
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 // Internal parser functions
121 #if defined(Py_DEBUG)
122 void _PyPegen_clear_memo_statistics(void);
123 PyObject *_PyPegen_get_memo_statistics(void);
124 #endif
125
126 int _PyPegen_insert_memo(Parser *p, int mark, int type, void *node);
127 int _PyPegen_update_memo(Parser *p, int mark, int type, void *node);
128 int _PyPegen_is_memoized(Parser *p, int type, void *pres);
129
130 int _PyPegen_lookahead_with_name(int, expr_ty (func)(Parser *), Parser *);
131 int _PyPegen_lookahead_with_int(int, Token *(func)(Parser *, int), Parser *, int);
132 int _PyPegen_lookahead_with_string(int , expr_ty (func)(Parser *, const char*), Parser *, const char*);
133 int _PyPegen_lookahead(int, void *(func)(Parser *), Parser *);
134
135 Token *_PyPegen_expect_token(Parser *p, int type);
136 void* _PyPegen_expect_forced_result(Parser *p, void* result, const char* expected);
137 Token *_PyPegen_expect_forced_token(Parser *p, int type, const char* expected);
138 expr_ty _PyPegen_expect_soft_keyword(Parser *p, const char *keyword);
139 expr_ty _PyPegen_soft_keyword_token(Parser *p);
140 Token *_PyPegen_get_last_nonnwhitespace_token(Parser *);
141 int _PyPegen_fill_token(Parser *p);
142 expr_ty _PyPegen_name_token(Parser *p);
143 expr_ty _PyPegen_number_token(Parser *p);
144 void *_PyPegen_string_token(Parser *p);
145 Py_ssize_t _PyPegen_byte_offset_to_character_offset(PyObject *line, Py_ssize_t col_offset);
146 Py_ssize_t _PyPegen_calculate_display_width(PyObject *segment, Py_ssize_t character_offset);
147
148 // Error handling functions and APIs
149 typedef enum {
150 STAR_TARGETS,
151 DEL_TARGETS,
152 FOR_TARGETS
153 } TARGETS_TYPE;
154
155 int _Pypegen_raise_decode_error(Parser *p);
156 void _PyPegen_raise_tokenizer_init_error(PyObject *filename);
157 int _Pypegen_tokenizer_error(Parser *p);
158 void *_PyPegen_raise_error(Parser *p, PyObject *errtype, const char *errmsg, ...);
159 void *_PyPegen_raise_error_known_location(Parser *p, PyObject *errtype,
160 Py_ssize_t lineno, Py_ssize_t col_offset,
161 Py_ssize_t end_lineno, Py_ssize_t end_col_offset,
162 const char *errmsg, va_list va);
163 void _Pypegen_set_syntax_error(Parser* p, Token* last_token);
164 Py_LOCAL_INLINE(void *)
165 RAISE_ERROR_KNOWN_LOCATION(Parser *p, PyObject *errtype,
166 Py_ssize_t lineno, Py_ssize_t col_offset,
167 Py_ssize_t end_lineno, Py_ssize_t end_col_offset,
168 const char *errmsg, ...)
169 {
170 va_list va;
171 va_start(va, errmsg);
172 Py_ssize_t _col_offset = (col_offset == CURRENT_POS ? CURRENT_POS : col_offset + 1);
173 Py_ssize_t _end_col_offset = (end_col_offset == CURRENT_POS ? CURRENT_POS : end_col_offset + 1);
174 _PyPegen_raise_error_known_location(p, errtype, lineno, _col_offset, end_lineno, _end_col_offset, errmsg, va);
175 va_end(va);
176 return NULL;
177 }
178 #define RAISE_SYNTAX_ERROR(msg, ...) _PyPegen_raise_error(p, PyExc_SyntaxError, msg, ##__VA_ARGS__)
179 #define RAISE_INDENTATION_ERROR(msg, ...) _PyPegen_raise_error(p, PyExc_IndentationError, msg, ##__VA_ARGS__)
180 #define RAISE_SYNTAX_ERROR_KNOWN_RANGE(a, b, msg, ...) \
181 RAISE_ERROR_KNOWN_LOCATION(p, PyExc_SyntaxError, (a)->lineno, (a)->col_offset, (b)->end_lineno, (b)->end_col_offset, msg, ##__VA_ARGS__)
182 #define RAISE_SYNTAX_ERROR_KNOWN_LOCATION(a, msg, ...) \
183 RAISE_ERROR_KNOWN_LOCATION(p, PyExc_SyntaxError, (a)->lineno, (a)->col_offset, (a)->end_lineno, (a)->end_col_offset, msg, ##__VA_ARGS__)
184 #define RAISE_SYNTAX_ERROR_STARTING_FROM(a, msg, ...) \
185 RAISE_ERROR_KNOWN_LOCATION(p, PyExc_SyntaxError, (a)->lineno, (a)->col_offset, CURRENT_POS, CURRENT_POS, msg, ##__VA_ARGS__)
186 #define RAISE_SYNTAX_ERROR_INVALID_TARGET(type, e) _RAISE_SYNTAX_ERROR_INVALID_TARGET(p, type, e)
187
188 Py_LOCAL_INLINE(void *)
189 CHECK_CALL(Parser *p, void *result)
190 {
191 if (result == NULL) {
192 assert(PyErr_Occurred());
193 p->error_indicator = 1;
194 }
195 return result;
196 }
197
198 /* This is needed for helper functions that are allowed to
199 return NULL without an error. Example: _PyPegen_seq_extract_starred_exprs */
200 Py_LOCAL_INLINE(void *)
201 CHECK_CALL_NULL_ALLOWED(Parser *p, void *result)
202 {
203 if (result == NULL && PyErr_Occurred()) {
204 p->error_indicator = 1;
205 }
206 return result;
207 }
208
209 #define CHECK(type, result) ((type) CHECK_CALL(p, result))
210 #define CHECK_NULL_ALLOWED(type, result) ((type) CHECK_CALL_NULL_ALLOWED(p, result))
211
212 expr_ty _PyPegen_get_invalid_target(expr_ty e, TARGETS_TYPE targets_type);
213 const char *_PyPegen_get_expr_name(expr_ty);
214 Py_LOCAL_INLINE(void *)
215 _RAISE_SYNTAX_ERROR_INVALID_TARGET(Parser *p, TARGETS_TYPE type, void *e)
216 {
217 expr_ty invalid_target = CHECK_NULL_ALLOWED(expr_ty, _PyPegen_get_invalid_target(e, type));
218 if (invalid_target != NULL) {
219 const char *msg;
220 if (type == STAR_TARGETS || type == FOR_TARGETS) {
221 msg = "cannot assign to %s";
222 }
223 else {
224 msg = "cannot delete %s";
225 }
226 return RAISE_SYNTAX_ERROR_KNOWN_LOCATION(
227 invalid_target,
228 msg,
229 _PyPegen_get_expr_name(invalid_target)
230 );
231 return RAISE_SYNTAX_ERROR_KNOWN_LOCATION(invalid_target, "invalid syntax");
232 }
233 return NULL;
234 }
235
236 // Action utility functions
237
238 void *_PyPegen_dummy_name(Parser *p, ...);
239 void * _PyPegen_seq_last_item(asdl_seq *seq);
240 #define PyPegen_last_item(seq, type) ((type)_PyPegen_seq_last_item((asdl_seq*)seq))
241 void * _PyPegen_seq_first_item(asdl_seq *seq);
242 #define PyPegen_first_item(seq, type) ((type)_PyPegen_seq_first_item((asdl_seq*)seq))
243 #define UNUSED(expr) do { (void)(expr); } while (0)
244 #define EXTRA_EXPR(head, tail) head->lineno, (head)->col_offset, (tail)->end_lineno, (tail)->end_col_offset, p->arena
245 #define EXTRA _start_lineno, _start_col_offset, _end_lineno, _end_col_offset, p->arena
246 PyObject *_PyPegen_new_type_comment(Parser *, const char *);
247
248 Py_LOCAL_INLINE(PyObject *)
249 NEW_TYPE_COMMENT(Parser *p, Token *tc)
250 {
251 if (tc == NULL) {
252 return NULL;
253 }
254 const char *bytes = PyBytes_AsString(tc->bytes);
255 if (bytes == NULL) {
256 goto error;
257 }
258 PyObject *tco = _PyPegen_new_type_comment(p, bytes);
259 if (tco == NULL) {
260 goto error;
261 }
262 return tco;
263 error:
264 p->error_indicator = 1; // Inline CHECK_CALL
265 return NULL;
266 }
267
268 Py_LOCAL_INLINE(void *)
269 INVALID_VERSION_CHECK(Parser *p, int version, char *msg, void *node)
270 {
271 if (node == NULL) {
272 p->error_indicator = 1; // Inline CHECK_CALL
273 return NULL;
274 }
275 if (p->feature_version < version) {
276 p->error_indicator = 1;
277 return RAISE_SYNTAX_ERROR("%s only supported in Python 3.%i and greater",
278 msg, version);
279 }
280 return node;
281 }
282
283 #define CHECK_VERSION(type, version, msg, node) ((type) INVALID_VERSION_CHECK(p, version, msg, node))
284
285 arg_ty _PyPegen_add_type_comment_to_arg(Parser *, arg_ty, Token *);
286 PyObject *_PyPegen_new_identifier(Parser *, const char *);
287 asdl_seq *_PyPegen_singleton_seq(Parser *, void *);
288 asdl_seq *_PyPegen_seq_insert_in_front(Parser *, void *, asdl_seq *);
289 asdl_seq *_PyPegen_seq_append_to_end(Parser *, asdl_seq *, void *);
290 asdl_seq *_PyPegen_seq_flatten(Parser *, asdl_seq *);
291 expr_ty _PyPegen_join_names_with_dot(Parser *, expr_ty, expr_ty);
292 int _PyPegen_seq_count_dots(asdl_seq *);
293 alias_ty _PyPegen_alias_for_star(Parser *, int, int, int, int, PyArena *);
294 asdl_identifier_seq *_PyPegen_map_names_to_ids(Parser *, asdl_expr_seq *);
295 CmpopExprPair *_PyPegen_cmpop_expr_pair(Parser *, cmpop_ty, expr_ty);
296 asdl_int_seq *_PyPegen_get_cmpops(Parser *p, asdl_seq *);
297 asdl_expr_seq *_PyPegen_get_exprs(Parser *, asdl_seq *);
298 expr_ty _PyPegen_set_expr_context(Parser *, expr_ty, expr_context_ty);
299 KeyValuePair *_PyPegen_key_value_pair(Parser *, expr_ty, expr_ty);
300 asdl_expr_seq *_PyPegen_get_keys(Parser *, asdl_seq *);
301 asdl_expr_seq *_PyPegen_get_values(Parser *, asdl_seq *);
302 KeyPatternPair *_PyPegen_key_pattern_pair(Parser *, expr_ty, pattern_ty);
303 asdl_expr_seq *_PyPegen_get_pattern_keys(Parser *, asdl_seq *);
304 asdl_pattern_seq *_PyPegen_get_patterns(Parser *, asdl_seq *);
305 NameDefaultPair *_PyPegen_name_default_pair(Parser *, arg_ty, expr_ty, Token *);
306 SlashWithDefault *_PyPegen_slash_with_default(Parser *, asdl_arg_seq *, asdl_seq *);
307 StarEtc *_PyPegen_star_etc(Parser *, arg_ty, asdl_seq *, arg_ty);
308 arguments_ty _PyPegen_make_arguments(Parser *, asdl_arg_seq *, SlashWithDefault *,
309 asdl_arg_seq *, asdl_seq *, StarEtc *);
310 arguments_ty _PyPegen_empty_arguments(Parser *);
311 AugOperator *_PyPegen_augoperator(Parser*, operator_ty type);
312 stmt_ty _PyPegen_function_def_decorators(Parser *, asdl_expr_seq *, stmt_ty);
313 stmt_ty _PyPegen_class_def_decorators(Parser *, asdl_expr_seq *, stmt_ty);
314 KeywordOrStarred *_PyPegen_keyword_or_starred(Parser *, void *, int);
315 asdl_expr_seq *_PyPegen_seq_extract_starred_exprs(Parser *, asdl_seq *);
316 asdl_keyword_seq *_PyPegen_seq_delete_starred_exprs(Parser *, asdl_seq *);
317 expr_ty _PyPegen_collect_call_seqs(Parser *, asdl_expr_seq *, asdl_seq *,
318 int lineno, int col_offset, int end_lineno,
319 int end_col_offset, PyArena *arena);
320 expr_ty _PyPegen_concatenate_strings(Parser *p, asdl_seq *);
321 expr_ty _PyPegen_ensure_imaginary(Parser *p, expr_ty);
322 expr_ty _PyPegen_ensure_real(Parser *p, expr_ty);
323 asdl_seq *_PyPegen_join_sequences(Parser *, asdl_seq *, asdl_seq *);
324 int _PyPegen_check_barry_as_flufl(Parser *, Token *);
325 int _PyPegen_check_legacy_stmt(Parser *p, expr_ty t);
326 mod_ty _PyPegen_make_module(Parser *, asdl_stmt_seq *);
327 void *_PyPegen_arguments_parsing_error(Parser *, expr_ty);
328 expr_ty _PyPegen_get_last_comprehension_item(comprehension_ty comprehension);
329 void *_PyPegen_nonparen_genexp_in_call(Parser *p, expr_ty args, asdl_comprehension_seq *comprehensions);
330
331 // Parser API
332
333 Parser *_PyPegen_Parser_New(struct tok_state *, int, int, int, int *, PyArena *);
334 void _PyPegen_Parser_Free(Parser *);
335 mod_ty _PyPegen_run_parser_from_file_pointer(FILE *, int, PyObject *, const char *,
336 const char *, const char *, PyCompilerFlags *, int *, PyArena *);
337 void *_PyPegen_run_parser(Parser *);
338 mod_ty _PyPegen_run_parser_from_string(const char *, int, PyObject *, PyCompilerFlags *, PyArena *);
339 asdl_stmt_seq *_PyPegen_interactive_exit(Parser *);
340
341 // Generated function in parse.c - function definition in python.gram
342 void *_PyPegen_parse(Parser *);
343
344 #endif