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