/src/cpython/Parser/action_helpers.c
Line | Count | Source (jump to first uncovered line) |
1 | | #include <Python.h> |
2 | | #include "pycore_pystate.h" // _PyInterpreterState_GET() |
3 | | #include "pycore_runtime.h" // _PyRuntime |
4 | | #include "pycore_unicodeobject.h" // _PyUnicode_InternImmortal() |
5 | | |
6 | | #include "pegen.h" |
7 | | #include "string_parser.h" // _PyPegen_decode_string() |
8 | | |
9 | | |
10 | | void * |
11 | | _PyPegen_dummy_name(Parser *p, ...) |
12 | 76.9k | { |
13 | 76.9k | return &_PyRuntime.parser.dummy_name; |
14 | 76.9k | } |
15 | | |
16 | | /* Creates a single-element asdl_seq* that contains a */ |
17 | | asdl_seq * |
18 | | _PyPegen_singleton_seq(Parser *p, void *a) |
19 | 175k | { |
20 | 175k | assert(a != NULL); |
21 | 175k | asdl_seq *seq = (asdl_seq*)_Py_asdl_generic_seq_new(1, p->arena); |
22 | 175k | if (!seq) { |
23 | 0 | return NULL; |
24 | 0 | } |
25 | 175k | asdl_seq_SET_UNTYPED(seq, 0, a); |
26 | 175k | return seq; |
27 | 175k | } |
28 | | |
29 | | /* Creates a copy of seq and prepends a to it */ |
30 | | asdl_seq * |
31 | | _PyPegen_seq_insert_in_front(Parser *p, void *a, asdl_seq *seq) |
32 | 378k | { |
33 | 378k | assert(a != NULL); |
34 | 378k | if (!seq) { |
35 | 8.66k | return _PyPegen_singleton_seq(p, a); |
36 | 8.66k | } |
37 | | |
38 | 369k | asdl_seq *new_seq = (asdl_seq*)_Py_asdl_generic_seq_new(asdl_seq_LEN(seq) + 1, p->arena); |
39 | 369k | if (!new_seq) { |
40 | 0 | return NULL; |
41 | 0 | } |
42 | | |
43 | 369k | asdl_seq_SET_UNTYPED(new_seq, 0, a); |
44 | 666k | for (Py_ssize_t i = 1, l = asdl_seq_LEN(new_seq); i < l; i++) { |
45 | 296k | asdl_seq_SET_UNTYPED(new_seq, i, asdl_seq_GET_UNTYPED(seq, i - 1)); |
46 | 296k | } |
47 | 369k | return new_seq; |
48 | 369k | } |
49 | | |
50 | | /* Creates a copy of seq and appends a to it */ |
51 | | asdl_seq * |
52 | | _PyPegen_seq_append_to_end(Parser *p, asdl_seq *seq, void *a) |
53 | 0 | { |
54 | 0 | assert(a != NULL); |
55 | 0 | if (!seq) { |
56 | 0 | return _PyPegen_singleton_seq(p, a); |
57 | 0 | } |
58 | | |
59 | 0 | asdl_seq *new_seq = (asdl_seq*)_Py_asdl_generic_seq_new(asdl_seq_LEN(seq) + 1, p->arena); |
60 | 0 | if (!new_seq) { |
61 | 0 | return NULL; |
62 | 0 | } |
63 | | |
64 | 0 | for (Py_ssize_t i = 0, l = asdl_seq_LEN(new_seq); i + 1 < l; i++) { |
65 | 0 | asdl_seq_SET_UNTYPED(new_seq, i, asdl_seq_GET_UNTYPED(seq, i)); |
66 | 0 | } |
67 | 0 | asdl_seq_SET_UNTYPED(new_seq, asdl_seq_LEN(new_seq) - 1, a); |
68 | 0 | return new_seq; |
69 | 0 | } |
70 | | |
71 | | static Py_ssize_t |
72 | | _get_flattened_seq_size(asdl_seq *seqs) |
73 | 29.5k | { |
74 | 29.5k | Py_ssize_t size = 0; |
75 | 144k | for (Py_ssize_t i = 0, l = asdl_seq_LEN(seqs); i < l; i++) { |
76 | 115k | asdl_seq *inner_seq = asdl_seq_GET_UNTYPED(seqs, i); |
77 | 115k | size += asdl_seq_LEN(inner_seq); |
78 | 115k | } |
79 | 29.5k | return size; |
80 | 29.5k | } |
81 | | |
82 | | /* Flattens an asdl_seq* of asdl_seq*s */ |
83 | | asdl_seq * |
84 | | _PyPegen_seq_flatten(Parser *p, asdl_seq *seqs) |
85 | 29.5k | { |
86 | 29.5k | Py_ssize_t flattened_seq_size = _get_flattened_seq_size(seqs); |
87 | 29.5k | assert(flattened_seq_size > 0); |
88 | | |
89 | 29.5k | asdl_seq *flattened_seq = (asdl_seq*)_Py_asdl_generic_seq_new(flattened_seq_size, p->arena); |
90 | 29.5k | if (!flattened_seq) { |
91 | 0 | return NULL; |
92 | 0 | } |
93 | | |
94 | 29.5k | int flattened_seq_idx = 0; |
95 | 144k | for (Py_ssize_t i = 0, l = asdl_seq_LEN(seqs); i < l; i++) { |
96 | 115k | asdl_seq *inner_seq = asdl_seq_GET_UNTYPED(seqs, i); |
97 | 236k | for (Py_ssize_t j = 0, li = asdl_seq_LEN(inner_seq); j < li; j++) { |
98 | 121k | asdl_seq_SET_UNTYPED(flattened_seq, flattened_seq_idx++, asdl_seq_GET_UNTYPED(inner_seq, j)); |
99 | 121k | } |
100 | 115k | } |
101 | 29.5k | assert(flattened_seq_idx == flattened_seq_size); |
102 | | |
103 | 29.5k | return flattened_seq; |
104 | 29.5k | } |
105 | | |
106 | | void * |
107 | | _PyPegen_seq_last_item(asdl_seq *seq) |
108 | 895 | { |
109 | 895 | Py_ssize_t len = asdl_seq_LEN(seq); |
110 | 895 | return asdl_seq_GET_UNTYPED(seq, len - 1); |
111 | 895 | } |
112 | | |
113 | | void * |
114 | | _PyPegen_seq_first_item(asdl_seq *seq) |
115 | 828 | { |
116 | 828 | return asdl_seq_GET_UNTYPED(seq, 0); |
117 | 828 | } |
118 | | |
119 | | /* Creates a new name of the form <first_name>.<second_name> */ |
120 | | expr_ty |
121 | | _PyPegen_join_names_with_dot(Parser *p, expr_ty first_name, expr_ty second_name) |
122 | 2.08k | { |
123 | 2.08k | assert(first_name != NULL && second_name != NULL); |
124 | 2.08k | PyObject *uni = PyUnicode_FromFormat("%U.%U", |
125 | 2.08k | first_name->v.Name.id, second_name->v.Name.id); |
126 | 2.08k | if (!uni) { |
127 | 0 | return NULL; |
128 | 0 | } |
129 | 2.08k | PyInterpreterState *interp = _PyInterpreterState_GET(); |
130 | 2.08k | _PyUnicode_InternImmortal(interp, &uni); |
131 | 2.08k | if (_PyArena_AddPyObject(p->arena, uni) < 0) { |
132 | 0 | Py_DECREF(uni); |
133 | 0 | return NULL; |
134 | 0 | } |
135 | | |
136 | 2.08k | return _PyAST_Name(uni, Load, EXTRA_EXPR(first_name, second_name)); |
137 | 2.08k | } |
138 | | |
139 | | /* Counts the total number of dots in seq's tokens */ |
140 | | int |
141 | | _PyPegen_seq_count_dots(asdl_seq *seq) |
142 | 3.14k | { |
143 | 3.14k | int number_of_dots = 0; |
144 | 7.34k | for (Py_ssize_t i = 0, l = asdl_seq_LEN(seq); i < l; i++) { |
145 | 4.20k | Token *current_expr = asdl_seq_GET_UNTYPED(seq, i); |
146 | 4.20k | switch (current_expr->type) { |
147 | 1.81k | case ELLIPSIS: |
148 | 1.81k | number_of_dots += 3; |
149 | 1.81k | break; |
150 | 2.39k | case DOT: |
151 | 2.39k | number_of_dots += 1; |
152 | 2.39k | break; |
153 | 0 | default: |
154 | 0 | Py_UNREACHABLE(); |
155 | 4.20k | } |
156 | 4.20k | } |
157 | | |
158 | 3.14k | return number_of_dots; |
159 | 3.14k | } |
160 | | |
161 | | /* Creates an alias with '*' as the identifier name */ |
162 | | alias_ty |
163 | | _PyPegen_alias_for_star(Parser *p, int lineno, int col_offset, int end_lineno, |
164 | 608 | int end_col_offset, PyArena *arena) { |
165 | 608 | PyObject *str = PyUnicode_InternFromString("*"); |
166 | 608 | if (!str) { |
167 | 0 | return NULL; |
168 | 0 | } |
169 | 608 | if (_PyArena_AddPyObject(p->arena, str) < 0) { |
170 | 0 | Py_DECREF(str); |
171 | 0 | return NULL; |
172 | 0 | } |
173 | 608 | return _PyAST_alias(str, NULL, lineno, col_offset, end_lineno, end_col_offset, arena); |
174 | 608 | } |
175 | | |
176 | | /* Creates a new asdl_seq* with the identifiers of all the names in seq */ |
177 | | asdl_identifier_seq * |
178 | | _PyPegen_map_names_to_ids(Parser *p, asdl_expr_seq *seq) |
179 | 2.04k | { |
180 | 2.04k | Py_ssize_t len = asdl_seq_LEN(seq); |
181 | 2.04k | assert(len > 0); |
182 | | |
183 | 2.04k | asdl_identifier_seq *new_seq = _Py_asdl_identifier_seq_new(len, p->arena); |
184 | 2.04k | if (!new_seq) { |
185 | 0 | return NULL; |
186 | 0 | } |
187 | 7.18k | for (Py_ssize_t i = 0; i < len; i++) { |
188 | 5.14k | expr_ty e = asdl_seq_GET(seq, i); |
189 | 5.14k | asdl_seq_SET(new_seq, i, e->v.Name.id); |
190 | 5.14k | } |
191 | 2.04k | return new_seq; |
192 | 2.04k | } |
193 | | |
194 | | /* Constructs a CmpopExprPair */ |
195 | | CmpopExprPair * |
196 | | _PyPegen_cmpop_expr_pair(Parser *p, cmpop_ty cmpop, expr_ty expr) |
197 | 22.1k | { |
198 | 22.1k | assert(expr != NULL); |
199 | 22.1k | CmpopExprPair *a = _PyArena_Malloc(p->arena, sizeof(CmpopExprPair)); |
200 | 22.1k | if (!a) { |
201 | 0 | return NULL; |
202 | 0 | } |
203 | 22.1k | a->cmpop = cmpop; |
204 | 22.1k | a->expr = expr; |
205 | 22.1k | return a; |
206 | 22.1k | } |
207 | | |
208 | | asdl_int_seq * |
209 | | _PyPegen_get_cmpops(Parser *p, asdl_seq *seq) |
210 | 10.2k | { |
211 | 10.2k | Py_ssize_t len = asdl_seq_LEN(seq); |
212 | 10.2k | assert(len > 0); |
213 | | |
214 | 10.2k | asdl_int_seq *new_seq = _Py_asdl_int_seq_new(len, p->arena); |
215 | 10.2k | if (!new_seq) { |
216 | 0 | return NULL; |
217 | 0 | } |
218 | 31.6k | for (Py_ssize_t i = 0; i < len; i++) { |
219 | 21.4k | CmpopExprPair *pair = asdl_seq_GET_UNTYPED(seq, i); |
220 | 21.4k | asdl_seq_SET(new_seq, i, pair->cmpop); |
221 | 21.4k | } |
222 | 10.2k | return new_seq; |
223 | 10.2k | } |
224 | | |
225 | | asdl_expr_seq * |
226 | | _PyPegen_get_exprs(Parser *p, asdl_seq *seq) |
227 | 10.2k | { |
228 | 10.2k | Py_ssize_t len = asdl_seq_LEN(seq); |
229 | 10.2k | assert(len > 0); |
230 | | |
231 | 10.2k | asdl_expr_seq *new_seq = _Py_asdl_expr_seq_new(len, p->arena); |
232 | 10.2k | if (!new_seq) { |
233 | 0 | return NULL; |
234 | 0 | } |
235 | 31.6k | for (Py_ssize_t i = 0; i < len; i++) { |
236 | 21.4k | CmpopExprPair *pair = asdl_seq_GET_UNTYPED(seq, i); |
237 | 21.4k | asdl_seq_SET(new_seq, i, pair->expr); |
238 | 21.4k | } |
239 | 10.2k | return new_seq; |
240 | 10.2k | } |
241 | | |
242 | | /* Creates an asdl_seq* where all the elements have been changed to have ctx as context */ |
243 | | static asdl_expr_seq * |
244 | | _set_seq_context(Parser *p, asdl_expr_seq *seq, expr_context_ty ctx) |
245 | 7.97k | { |
246 | 7.97k | Py_ssize_t len = asdl_seq_LEN(seq); |
247 | 7.97k | if (len == 0) { |
248 | 4.00k | return NULL; |
249 | 4.00k | } |
250 | | |
251 | 3.96k | asdl_expr_seq *new_seq = _Py_asdl_expr_seq_new(len, p->arena); |
252 | 3.96k | if (!new_seq) { |
253 | 0 | return NULL; |
254 | 0 | } |
255 | 11.8k | for (Py_ssize_t i = 0; i < len; i++) { |
256 | 7.92k | expr_ty e = asdl_seq_GET(seq, i); |
257 | 7.92k | asdl_seq_SET(new_seq, i, _PyPegen_set_expr_context(p, e, ctx)); |
258 | 7.92k | } |
259 | 3.96k | return new_seq; |
260 | 3.96k | } |
261 | | |
262 | | static expr_ty |
263 | | _set_name_context(Parser *p, expr_ty e, expr_context_ty ctx) |
264 | 210k | { |
265 | 210k | return _PyAST_Name(e->v.Name.id, ctx, EXTRA_EXPR(e, e)); |
266 | 210k | } |
267 | | |
268 | | static expr_ty |
269 | | _set_tuple_context(Parser *p, expr_ty e, expr_context_ty ctx) |
270 | 6.66k | { |
271 | 6.66k | return _PyAST_Tuple( |
272 | 6.66k | _set_seq_context(p, e->v.Tuple.elts, ctx), |
273 | 6.66k | ctx, |
274 | 6.66k | EXTRA_EXPR(e, e)); |
275 | 6.66k | } |
276 | | |
277 | | static expr_ty |
278 | | _set_list_context(Parser *p, expr_ty e, expr_context_ty ctx) |
279 | 1.31k | { |
280 | 1.31k | return _PyAST_List( |
281 | 1.31k | _set_seq_context(p, e->v.List.elts, ctx), |
282 | 1.31k | ctx, |
283 | 1.31k | EXTRA_EXPR(e, e)); |
284 | 1.31k | } |
285 | | |
286 | | static expr_ty |
287 | | _set_subscript_context(Parser *p, expr_ty e, expr_context_ty ctx) |
288 | 419 | { |
289 | 419 | return _PyAST_Subscript(e->v.Subscript.value, e->v.Subscript.slice, |
290 | 419 | ctx, EXTRA_EXPR(e, e)); |
291 | 419 | } |
292 | | |
293 | | static expr_ty |
294 | | _set_attribute_context(Parser *p, expr_ty e, expr_context_ty ctx) |
295 | 80 | { |
296 | 80 | return _PyAST_Attribute(e->v.Attribute.value, e->v.Attribute.attr, |
297 | 80 | ctx, EXTRA_EXPR(e, e)); |
298 | 80 | } |
299 | | |
300 | | static expr_ty |
301 | | _set_starred_context(Parser *p, expr_ty e, expr_context_ty ctx) |
302 | 674 | { |
303 | 674 | return _PyAST_Starred(_PyPegen_set_expr_context(p, e->v.Starred.value, ctx), |
304 | 674 | ctx, EXTRA_EXPR(e, e)); |
305 | 674 | } |
306 | | |
307 | | /* Creates an `expr_ty` equivalent to `expr` but with `ctx` as context */ |
308 | | expr_ty |
309 | | _PyPegen_set_expr_context(Parser *p, expr_ty expr, expr_context_ty ctx) |
310 | 219k | { |
311 | 219k | assert(expr != NULL); |
312 | | |
313 | 219k | expr_ty new = NULL; |
314 | 219k | switch (expr->kind) { |
315 | 210k | case Name_kind: |
316 | 210k | new = _set_name_context(p, expr, ctx); |
317 | 210k | break; |
318 | 6.66k | case Tuple_kind: |
319 | 6.66k | new = _set_tuple_context(p, expr, ctx); |
320 | 6.66k | break; |
321 | 1.31k | case List_kind: |
322 | 1.31k | new = _set_list_context(p, expr, ctx); |
323 | 1.31k | break; |
324 | 419 | case Subscript_kind: |
325 | 419 | new = _set_subscript_context(p, expr, ctx); |
326 | 419 | break; |
327 | 80 | case Attribute_kind: |
328 | 80 | new = _set_attribute_context(p, expr, ctx); |
329 | 80 | break; |
330 | 674 | case Starred_kind: |
331 | 674 | new = _set_starred_context(p, expr, ctx); |
332 | 674 | break; |
333 | 0 | default: |
334 | 0 | new = expr; |
335 | 219k | } |
336 | 219k | return new; |
337 | 219k | } |
338 | | |
339 | | /* Constructs a KeyValuePair that is used when parsing a dict's key value pairs */ |
340 | | KeyValuePair * |
341 | | _PyPegen_key_value_pair(Parser *p, expr_ty key, expr_ty value) |
342 | 47.2k | { |
343 | 47.2k | KeyValuePair *a = _PyArena_Malloc(p->arena, sizeof(KeyValuePair)); |
344 | 47.2k | if (!a) { |
345 | 0 | return NULL; |
346 | 0 | } |
347 | 47.2k | a->key = key; |
348 | 47.2k | a->value = value; |
349 | 47.2k | return a; |
350 | 47.2k | } |
351 | | |
352 | | /* Extracts all keys from an asdl_seq* of KeyValuePair*'s */ |
353 | | asdl_expr_seq * |
354 | | _PyPegen_get_keys(Parser *p, asdl_seq *seq) |
355 | 4.40k | { |
356 | 4.40k | Py_ssize_t len = asdl_seq_LEN(seq); |
357 | 4.40k | asdl_expr_seq *new_seq = _Py_asdl_expr_seq_new(len, p->arena); |
358 | 4.40k | if (!new_seq) { |
359 | 0 | return NULL; |
360 | 0 | } |
361 | 40.2k | for (Py_ssize_t i = 0; i < len; i++) { |
362 | 35.8k | KeyValuePair *pair = asdl_seq_GET_UNTYPED(seq, i); |
363 | 35.8k | asdl_seq_SET(new_seq, i, pair->key); |
364 | 35.8k | } |
365 | 4.40k | return new_seq; |
366 | 4.40k | } |
367 | | |
368 | | /* Extracts all values from an asdl_seq* of KeyValuePair*'s */ |
369 | | asdl_expr_seq * |
370 | | _PyPegen_get_values(Parser *p, asdl_seq *seq) |
371 | 4.40k | { |
372 | 4.40k | Py_ssize_t len = asdl_seq_LEN(seq); |
373 | 4.40k | asdl_expr_seq *new_seq = _Py_asdl_expr_seq_new(len, p->arena); |
374 | 4.40k | if (!new_seq) { |
375 | 0 | return NULL; |
376 | 0 | } |
377 | 40.2k | for (Py_ssize_t i = 0; i < len; i++) { |
378 | 35.8k | KeyValuePair *pair = asdl_seq_GET_UNTYPED(seq, i); |
379 | 35.8k | asdl_seq_SET(new_seq, i, pair->value); |
380 | 35.8k | } |
381 | 4.40k | return new_seq; |
382 | 4.40k | } |
383 | | |
384 | | /* Constructs a KeyPatternPair that is used when parsing mapping & class patterns */ |
385 | | KeyPatternPair * |
386 | | _PyPegen_key_pattern_pair(Parser *p, expr_ty key, pattern_ty pattern) |
387 | 11.7k | { |
388 | 11.7k | KeyPatternPair *a = _PyArena_Malloc(p->arena, sizeof(KeyPatternPair)); |
389 | 11.7k | if (!a) { |
390 | 0 | return NULL; |
391 | 0 | } |
392 | 11.7k | a->key = key; |
393 | 11.7k | a->pattern = pattern; |
394 | 11.7k | return a; |
395 | 11.7k | } |
396 | | |
397 | | /* Extracts all keys from an asdl_seq* of KeyPatternPair*'s */ |
398 | | asdl_expr_seq * |
399 | | _PyPegen_get_pattern_keys(Parser *p, asdl_seq *seq) |
400 | 1.90k | { |
401 | 1.90k | Py_ssize_t len = asdl_seq_LEN(seq); |
402 | 1.90k | asdl_expr_seq *new_seq = _Py_asdl_expr_seq_new(len, p->arena); |
403 | 1.90k | if (!new_seq) { |
404 | 0 | return NULL; |
405 | 0 | } |
406 | 4.77k | for (Py_ssize_t i = 0; i < len; i++) { |
407 | 2.87k | KeyPatternPair *pair = asdl_seq_GET_UNTYPED(seq, i); |
408 | 2.87k | asdl_seq_SET(new_seq, i, pair->key); |
409 | 2.87k | } |
410 | 1.90k | return new_seq; |
411 | 1.90k | } |
412 | | |
413 | | /* Extracts all patterns from an asdl_seq* of KeyPatternPair*'s */ |
414 | | asdl_pattern_seq * |
415 | | _PyPegen_get_patterns(Parser *p, asdl_seq *seq) |
416 | 1.90k | { |
417 | 1.90k | Py_ssize_t len = asdl_seq_LEN(seq); |
418 | 1.90k | asdl_pattern_seq *new_seq = _Py_asdl_pattern_seq_new(len, p->arena); |
419 | 1.90k | if (!new_seq) { |
420 | 0 | return NULL; |
421 | 0 | } |
422 | 4.77k | for (Py_ssize_t i = 0; i < len; i++) { |
423 | 2.87k | KeyPatternPair *pair = asdl_seq_GET_UNTYPED(seq, i); |
424 | 2.87k | asdl_seq_SET(new_seq, i, pair->pattern); |
425 | 2.87k | } |
426 | 1.90k | return new_seq; |
427 | 1.90k | } |
428 | | |
429 | | /* Constructs a NameDefaultPair */ |
430 | | NameDefaultPair * |
431 | | _PyPegen_name_default_pair(Parser *p, arg_ty arg, expr_ty value, Token *tc) |
432 | 82.8k | { |
433 | 82.8k | NameDefaultPair *a = _PyArena_Malloc(p->arena, sizeof(NameDefaultPair)); |
434 | 82.8k | if (!a) { |
435 | 0 | return NULL; |
436 | 0 | } |
437 | 82.8k | a->arg = _PyPegen_add_type_comment_to_arg(p, arg, tc); |
438 | 82.8k | a->value = value; |
439 | 82.8k | return a; |
440 | 82.8k | } |
441 | | |
442 | | /* Constructs a SlashWithDefault */ |
443 | | SlashWithDefault * |
444 | | _PyPegen_slash_with_default(Parser *p, asdl_arg_seq *plain_names, asdl_seq *names_with_defaults) |
445 | 5.82k | { |
446 | 5.82k | SlashWithDefault *a = _PyArena_Malloc(p->arena, sizeof(SlashWithDefault)); |
447 | 5.82k | if (!a) { |
448 | 0 | return NULL; |
449 | 0 | } |
450 | 5.82k | a->plain_names = plain_names; |
451 | 5.82k | a->names_with_defaults = names_with_defaults; |
452 | 5.82k | return a; |
453 | 5.82k | } |
454 | | |
455 | | /* Constructs a StarEtc */ |
456 | | StarEtc * |
457 | | _PyPegen_star_etc(Parser *p, arg_ty vararg, asdl_seq *kwonlyargs, arg_ty kwarg) |
458 | 6.12k | { |
459 | 6.12k | StarEtc *a = _PyArena_Malloc(p->arena, sizeof(StarEtc)); |
460 | 6.12k | if (!a) { |
461 | 0 | return NULL; |
462 | 0 | } |
463 | 6.12k | a->vararg = vararg; |
464 | 6.12k | a->kwonlyargs = kwonlyargs; |
465 | 6.12k | a->kwarg = kwarg; |
466 | 6.12k | return a; |
467 | 6.12k | } |
468 | | |
469 | | asdl_seq * |
470 | | _PyPegen_join_sequences(Parser *p, asdl_seq *a, asdl_seq *b) |
471 | 13.7k | { |
472 | 13.7k | Py_ssize_t first_len = asdl_seq_LEN(a); |
473 | 13.7k | Py_ssize_t second_len = asdl_seq_LEN(b); |
474 | 13.7k | asdl_seq *new_seq = (asdl_seq*)_Py_asdl_generic_seq_new(first_len + second_len, p->arena); |
475 | 13.7k | if (!new_seq) { |
476 | 0 | return NULL; |
477 | 0 | } |
478 | | |
479 | 13.7k | int k = 0; |
480 | 37.0k | for (Py_ssize_t i = 0; i < first_len; i++) { |
481 | 23.2k | asdl_seq_SET_UNTYPED(new_seq, k++, asdl_seq_GET_UNTYPED(a, i)); |
482 | 23.2k | } |
483 | 23.2k | for (Py_ssize_t i = 0; i < second_len; i++) { |
484 | 9.43k | asdl_seq_SET_UNTYPED(new_seq, k++, asdl_seq_GET_UNTYPED(b, i)); |
485 | 9.43k | } |
486 | | |
487 | 13.7k | return new_seq; |
488 | 13.7k | } |
489 | | |
490 | | static asdl_arg_seq* |
491 | | _get_names(Parser *p, asdl_seq *names_with_defaults) |
492 | 20.9k | { |
493 | 20.9k | Py_ssize_t len = asdl_seq_LEN(names_with_defaults); |
494 | 20.9k | asdl_arg_seq *seq = _Py_asdl_arg_seq_new(len, p->arena); |
495 | 20.9k | if (!seq) { |
496 | 0 | return NULL; |
497 | 0 | } |
498 | 35.4k | for (Py_ssize_t i = 0; i < len; i++) { |
499 | 14.5k | NameDefaultPair *pair = asdl_seq_GET_UNTYPED(names_with_defaults, i); |
500 | 14.5k | asdl_seq_SET(seq, i, pair->arg); |
501 | 14.5k | } |
502 | 20.9k | return seq; |
503 | 20.9k | } |
504 | | |
505 | | static asdl_expr_seq * |
506 | | _get_defaults(Parser *p, asdl_seq *names_with_defaults) |
507 | 20.9k | { |
508 | 20.9k | Py_ssize_t len = asdl_seq_LEN(names_with_defaults); |
509 | 20.9k | asdl_expr_seq *seq = _Py_asdl_expr_seq_new(len, p->arena); |
510 | 20.9k | if (!seq) { |
511 | 0 | return NULL; |
512 | 0 | } |
513 | 35.4k | for (Py_ssize_t i = 0; i < len; i++) { |
514 | 14.5k | NameDefaultPair *pair = asdl_seq_GET_UNTYPED(names_with_defaults, i); |
515 | 14.5k | asdl_seq_SET(seq, i, pair->value); |
516 | 14.5k | } |
517 | 20.9k | return seq; |
518 | 20.9k | } |
519 | | |
520 | | static int |
521 | | _make_posonlyargs(Parser *p, |
522 | | asdl_arg_seq *slash_without_default, |
523 | | SlashWithDefault *slash_with_default, |
524 | 19.2k | asdl_arg_seq **posonlyargs) { |
525 | 19.2k | if (slash_without_default != NULL) { |
526 | 1.37k | *posonlyargs = slash_without_default; |
527 | 1.37k | } |
528 | 17.8k | else if (slash_with_default != NULL) { |
529 | 2.07k | asdl_arg_seq *slash_with_default_names = |
530 | 2.07k | _get_names(p, slash_with_default->names_with_defaults); |
531 | 2.07k | if (!slash_with_default_names) { |
532 | 0 | return -1; |
533 | 0 | } |
534 | 2.07k | *posonlyargs = (asdl_arg_seq*)_PyPegen_join_sequences( |
535 | 2.07k | p, |
536 | 2.07k | (asdl_seq*)slash_with_default->plain_names, |
537 | 2.07k | (asdl_seq*)slash_with_default_names); |
538 | 2.07k | } |
539 | 15.7k | else { |
540 | 15.7k | *posonlyargs = _Py_asdl_arg_seq_new(0, p->arena); |
541 | 15.7k | } |
542 | 19.2k | return *posonlyargs == NULL ? -1 : 0; |
543 | 19.2k | } |
544 | | |
545 | | static int |
546 | | _make_posargs(Parser *p, |
547 | | asdl_arg_seq *plain_names, |
548 | | asdl_seq *names_with_default, |
549 | 19.2k | asdl_arg_seq **posargs) { |
550 | | |
551 | 19.2k | if (names_with_default != NULL) { |
552 | 13.8k | if (plain_names != NULL) { |
553 | 9.27k | asdl_arg_seq *names_with_default_names = _get_names(p, names_with_default); |
554 | 9.27k | if (!names_with_default_names) { |
555 | 0 | return -1; |
556 | 0 | } |
557 | 9.27k | *posargs = (asdl_arg_seq*)_PyPegen_join_sequences( |
558 | 9.27k | p,(asdl_seq*)plain_names, (asdl_seq*)names_with_default_names); |
559 | 9.27k | } |
560 | 4.55k | else { |
561 | 4.55k | *posargs = _get_names(p, names_with_default); |
562 | 4.55k | } |
563 | 13.8k | } |
564 | 5.37k | else { |
565 | 5.37k | if (plain_names != NULL) { |
566 | | // With the current grammar, we never get here. |
567 | | // If that has changed, remove the assert, and test thoroughly. |
568 | 0 | assert(0); |
569 | 0 | *posargs = plain_names; |
570 | 0 | } |
571 | 5.37k | else { |
572 | 5.37k | *posargs = _Py_asdl_arg_seq_new(0, p->arena); |
573 | 5.37k | } |
574 | 5.37k | } |
575 | 19.2k | return *posargs == NULL ? -1 : 0; |
576 | 19.2k | } |
577 | | |
578 | | static int |
579 | | _make_posdefaults(Parser *p, |
580 | | SlashWithDefault *slash_with_default, |
581 | | asdl_seq *names_with_default, |
582 | 19.2k | asdl_expr_seq **posdefaults) { |
583 | 19.2k | if (slash_with_default != NULL && names_with_default != NULL) { |
584 | 2.07k | asdl_expr_seq *slash_with_default_values = |
585 | 2.07k | _get_defaults(p, slash_with_default->names_with_defaults); |
586 | 2.07k | if (!slash_with_default_values) { |
587 | 0 | return -1; |
588 | 0 | } |
589 | 2.07k | asdl_expr_seq *names_with_default_values = _get_defaults(p, names_with_default); |
590 | 2.07k | if (!names_with_default_values) { |
591 | 0 | return -1; |
592 | 0 | } |
593 | 2.07k | *posdefaults = (asdl_expr_seq*)_PyPegen_join_sequences( |
594 | 2.07k | p, |
595 | 2.07k | (asdl_seq*)slash_with_default_values, |
596 | 2.07k | (asdl_seq*)names_with_default_values); |
597 | 2.07k | } |
598 | 17.1k | else if (slash_with_default == NULL && names_with_default != NULL) { |
599 | 11.7k | *posdefaults = _get_defaults(p, names_with_default); |
600 | 11.7k | } |
601 | 5.37k | else if (slash_with_default != NULL && names_with_default == NULL) { |
602 | 0 | *posdefaults = _get_defaults(p, slash_with_default->names_with_defaults); |
603 | 0 | } |
604 | 5.37k | else { |
605 | 5.37k | *posdefaults = _Py_asdl_expr_seq_new(0, p->arena); |
606 | 5.37k | } |
607 | 19.2k | return *posdefaults == NULL ? -1 : 0; |
608 | 19.2k | } |
609 | | |
610 | | static int |
611 | | _make_kwargs(Parser *p, StarEtc *star_etc, |
612 | | asdl_arg_seq **kwonlyargs, |
613 | 19.2k | asdl_expr_seq **kwdefaults) { |
614 | 19.2k | if (star_etc != NULL && star_etc->kwonlyargs != NULL) { |
615 | 5.02k | *kwonlyargs = _get_names(p, star_etc->kwonlyargs); |
616 | 5.02k | } |
617 | 14.1k | else { |
618 | 14.1k | *kwonlyargs = _Py_asdl_arg_seq_new(0, p->arena); |
619 | 14.1k | } |
620 | | |
621 | 19.2k | if (*kwonlyargs == NULL) { |
622 | 0 | return -1; |
623 | 0 | } |
624 | | |
625 | 19.2k | if (star_etc != NULL && star_etc->kwonlyargs != NULL) { |
626 | 5.02k | *kwdefaults = _get_defaults(p, star_etc->kwonlyargs); |
627 | 5.02k | } |
628 | 14.1k | else { |
629 | 14.1k | *kwdefaults = _Py_asdl_expr_seq_new(0, p->arena); |
630 | 14.1k | } |
631 | | |
632 | 19.2k | if (*kwdefaults == NULL) { |
633 | 0 | return -1; |
634 | 0 | } |
635 | | |
636 | 19.2k | return 0; |
637 | 19.2k | } |
638 | | |
639 | | /* Constructs an arguments_ty object out of all the parsed constructs in the parameters rule */ |
640 | | arguments_ty |
641 | | _PyPegen_make_arguments(Parser *p, asdl_arg_seq *slash_without_default, |
642 | | SlashWithDefault *slash_with_default, asdl_arg_seq *plain_names, |
643 | | asdl_seq *names_with_default, StarEtc *star_etc) |
644 | 19.2k | { |
645 | 19.2k | asdl_arg_seq *posonlyargs; |
646 | 19.2k | if (_make_posonlyargs(p, slash_without_default, slash_with_default, &posonlyargs) == -1) { |
647 | 0 | return NULL; |
648 | 0 | } |
649 | | |
650 | 19.2k | asdl_arg_seq *posargs; |
651 | 19.2k | if (_make_posargs(p, plain_names, names_with_default, &posargs) == -1) { |
652 | 0 | return NULL; |
653 | 0 | } |
654 | | |
655 | 19.2k | asdl_expr_seq *posdefaults; |
656 | 19.2k | if (_make_posdefaults(p,slash_with_default, names_with_default, &posdefaults) == -1) { |
657 | 0 | return NULL; |
658 | 0 | } |
659 | | |
660 | 19.2k | arg_ty vararg = NULL; |
661 | 19.2k | if (star_etc != NULL && star_etc->vararg != NULL) { |
662 | 3.13k | vararg = star_etc->vararg; |
663 | 3.13k | } |
664 | | |
665 | 19.2k | asdl_arg_seq *kwonlyargs; |
666 | 19.2k | asdl_expr_seq *kwdefaults; |
667 | 19.2k | if (_make_kwargs(p, star_etc, &kwonlyargs, &kwdefaults) == -1) { |
668 | 0 | return NULL; |
669 | 0 | } |
670 | | |
671 | 19.2k | arg_ty kwarg = NULL; |
672 | 19.2k | if (star_etc != NULL && star_etc->kwarg != NULL) { |
673 | 1.19k | kwarg = star_etc->kwarg; |
674 | 1.19k | } |
675 | | |
676 | 19.2k | return _PyAST_arguments(posonlyargs, posargs, vararg, kwonlyargs, |
677 | 19.2k | kwdefaults, kwarg, posdefaults, p->arena); |
678 | 19.2k | } |
679 | | |
680 | | |
681 | | /* Constructs an empty arguments_ty object, that gets used when a function accepts no |
682 | | * arguments. */ |
683 | | arguments_ty |
684 | | _PyPegen_empty_arguments(Parser *p) |
685 | 2.04k | { |
686 | 2.04k | asdl_arg_seq *posonlyargs = _Py_asdl_arg_seq_new(0, p->arena); |
687 | 2.04k | if (!posonlyargs) { |
688 | 0 | return NULL; |
689 | 0 | } |
690 | 2.04k | asdl_arg_seq *posargs = _Py_asdl_arg_seq_new(0, p->arena); |
691 | 2.04k | if (!posargs) { |
692 | 0 | return NULL; |
693 | 0 | } |
694 | 2.04k | asdl_expr_seq *posdefaults = _Py_asdl_expr_seq_new(0, p->arena); |
695 | 2.04k | if (!posdefaults) { |
696 | 0 | return NULL; |
697 | 0 | } |
698 | 2.04k | asdl_arg_seq *kwonlyargs = _Py_asdl_arg_seq_new(0, p->arena); |
699 | 2.04k | if (!kwonlyargs) { |
700 | 0 | return NULL; |
701 | 0 | } |
702 | 2.04k | asdl_expr_seq *kwdefaults = _Py_asdl_expr_seq_new(0, p->arena); |
703 | 2.04k | if (!kwdefaults) { |
704 | 0 | return NULL; |
705 | 0 | } |
706 | | |
707 | 2.04k | return _PyAST_arguments(posonlyargs, posargs, NULL, kwonlyargs, |
708 | 2.04k | kwdefaults, NULL, posdefaults, p->arena); |
709 | 2.04k | } |
710 | | |
711 | | /* Encapsulates the value of an operator_ty into an AugOperator struct */ |
712 | | AugOperator * |
713 | | _PyPegen_augoperator(Parser *p, operator_ty kind) |
714 | 2.96k | { |
715 | 2.96k | AugOperator *a = _PyArena_Malloc(p->arena, sizeof(AugOperator)); |
716 | 2.96k | if (!a) { |
717 | 0 | return NULL; |
718 | 0 | } |
719 | 2.96k | a->kind = kind; |
720 | 2.96k | return a; |
721 | 2.96k | } |
722 | | |
723 | | /* Construct a FunctionDef equivalent to function_def, but with decorators */ |
724 | | stmt_ty |
725 | | _PyPegen_function_def_decorators(Parser *p, asdl_expr_seq *decorators, stmt_ty function_def) |
726 | 1.00k | { |
727 | 1.00k | assert(function_def != NULL); |
728 | 1.00k | if (function_def->kind == AsyncFunctionDef_kind) { |
729 | 209 | return _PyAST_AsyncFunctionDef( |
730 | 209 | function_def->v.AsyncFunctionDef.name, |
731 | 209 | function_def->v.AsyncFunctionDef.args, |
732 | 209 | function_def->v.AsyncFunctionDef.body, decorators, |
733 | 209 | function_def->v.AsyncFunctionDef.returns, |
734 | 209 | function_def->v.AsyncFunctionDef.type_comment, |
735 | 209 | function_def->v.AsyncFunctionDef.type_params, |
736 | 209 | function_def->lineno, function_def->col_offset, |
737 | 209 | function_def->end_lineno, function_def->end_col_offset, p->arena); |
738 | 209 | } |
739 | | |
740 | 791 | return _PyAST_FunctionDef( |
741 | 791 | function_def->v.FunctionDef.name, |
742 | 791 | function_def->v.FunctionDef.args, |
743 | 791 | function_def->v.FunctionDef.body, decorators, |
744 | 791 | function_def->v.FunctionDef.returns, |
745 | 791 | function_def->v.FunctionDef.type_comment, |
746 | 791 | function_def->v.FunctionDef.type_params, |
747 | 791 | function_def->lineno, function_def->col_offset, |
748 | 791 | function_def->end_lineno, function_def->end_col_offset, p->arena); |
749 | 1.00k | } |
750 | | |
751 | | /* Construct a ClassDef equivalent to class_def, but with decorators */ |
752 | | stmt_ty |
753 | | _PyPegen_class_def_decorators(Parser *p, asdl_expr_seq *decorators, stmt_ty class_def) |
754 | 389 | { |
755 | 389 | assert(class_def != NULL); |
756 | 389 | return _PyAST_ClassDef( |
757 | 389 | class_def->v.ClassDef.name, |
758 | 389 | class_def->v.ClassDef.bases, class_def->v.ClassDef.keywords, |
759 | 389 | class_def->v.ClassDef.body, decorators, |
760 | 389 | class_def->v.ClassDef.type_params, |
761 | 389 | class_def->lineno, class_def->col_offset, class_def->end_lineno, |
762 | 389 | class_def->end_col_offset, p->arena); |
763 | 389 | } |
764 | | |
765 | | /* Construct a KeywordOrStarred */ |
766 | | KeywordOrStarred * |
767 | | _PyPegen_keyword_or_starred(Parser *p, void *element, int is_keyword) |
768 | 37.0k | { |
769 | 37.0k | KeywordOrStarred *a = _PyArena_Malloc(p->arena, sizeof(KeywordOrStarred)); |
770 | 37.0k | if (!a) { |
771 | 0 | return NULL; |
772 | 0 | } |
773 | 37.0k | a->element = element; |
774 | 37.0k | a->is_keyword = is_keyword; |
775 | 37.0k | return a; |
776 | 37.0k | } |
777 | | |
778 | | /* Get the number of starred expressions in an asdl_seq* of KeywordOrStarred*s */ |
779 | | static int |
780 | | _seq_number_of_starred_exprs(asdl_seq *seq) |
781 | 18.5k | { |
782 | 18.5k | int n = 0; |
783 | 58.0k | for (Py_ssize_t i = 0, l = asdl_seq_LEN(seq); i < l; i++) { |
784 | 39.5k | KeywordOrStarred *k = asdl_seq_GET_UNTYPED(seq, i); |
785 | 39.5k | if (!k->is_keyword) { |
786 | 2.42k | n++; |
787 | 2.42k | } |
788 | 39.5k | } |
789 | 18.5k | return n; |
790 | 18.5k | } |
791 | | |
792 | | /* Extract the starred expressions of an asdl_seq* of KeywordOrStarred*s */ |
793 | | asdl_expr_seq * |
794 | | _PyPegen_seq_extract_starred_exprs(Parser *p, asdl_seq *kwargs) |
795 | 9.25k | { |
796 | 9.25k | int new_len = _seq_number_of_starred_exprs(kwargs); |
797 | 9.25k | if (new_len == 0) { |
798 | 8.73k | return NULL; |
799 | 8.73k | } |
800 | 513 | asdl_expr_seq *new_seq = _Py_asdl_expr_seq_new(new_len, p->arena); |
801 | 513 | if (!new_seq) { |
802 | 0 | return NULL; |
803 | 0 | } |
804 | | |
805 | 513 | int idx = 0; |
806 | 2.72k | for (Py_ssize_t i = 0, len = asdl_seq_LEN(kwargs); i < len; i++) { |
807 | 2.20k | KeywordOrStarred *k = asdl_seq_GET_UNTYPED(kwargs, i); |
808 | 2.20k | if (!k->is_keyword) { |
809 | 1.21k | asdl_seq_SET(new_seq, idx++, k->element); |
810 | 1.21k | } |
811 | 2.20k | } |
812 | 513 | return new_seq; |
813 | 513 | } |
814 | | |
815 | | /* Return a new asdl_seq* with only the keywords in kwargs */ |
816 | | asdl_keyword_seq* |
817 | | _PyPegen_seq_delete_starred_exprs(Parser *p, asdl_seq *kwargs) |
818 | 9.25k | { |
819 | 9.25k | Py_ssize_t len = asdl_seq_LEN(kwargs); |
820 | 9.25k | Py_ssize_t new_len = len - _seq_number_of_starred_exprs(kwargs); |
821 | 9.25k | if (new_len == 0) { |
822 | 0 | return NULL; |
823 | 0 | } |
824 | 9.25k | asdl_keyword_seq *new_seq = _Py_asdl_keyword_seq_new(new_len, p->arena); |
825 | 9.25k | if (!new_seq) { |
826 | 0 | return NULL; |
827 | 0 | } |
828 | | |
829 | 9.25k | int idx = 0; |
830 | 29.0k | for (Py_ssize_t i = 0; i < len; i++) { |
831 | 19.7k | KeywordOrStarred *k = asdl_seq_GET_UNTYPED(kwargs, i); |
832 | 19.7k | if (k->is_keyword) { |
833 | 18.5k | asdl_seq_SET(new_seq, idx++, k->element); |
834 | 18.5k | } |
835 | 19.7k | } |
836 | 9.25k | return new_seq; |
837 | 9.25k | } |
838 | | |
839 | | expr_ty |
840 | | _PyPegen_ensure_imaginary(Parser *p, expr_ty exp) |
841 | 1.00k | { |
842 | 1.00k | if (exp->kind != Constant_kind || !PyComplex_CheckExact(exp->v.Constant.value)) { |
843 | 8 | RAISE_SYNTAX_ERROR_KNOWN_LOCATION(exp, "imaginary number required in complex literal"); |
844 | 8 | return NULL; |
845 | 8 | } |
846 | 996 | return exp; |
847 | 1.00k | } |
848 | | |
849 | | expr_ty |
850 | | _PyPegen_ensure_real(Parser *p, expr_ty exp) |
851 | 1.72k | { |
852 | 1.72k | if (exp->kind != Constant_kind || PyComplex_CheckExact(exp->v.Constant.value)) { |
853 | 3 | RAISE_SYNTAX_ERROR_KNOWN_LOCATION(exp, "real number required in complex literal"); |
854 | 3 | return NULL; |
855 | 3 | } |
856 | 1.71k | return exp; |
857 | 1.72k | } |
858 | | |
859 | | mod_ty |
860 | 7.63k | _PyPegen_make_module(Parser *p, asdl_stmt_seq *a) { |
861 | 7.63k | asdl_type_ignore_seq *type_ignores = NULL; |
862 | 7.63k | Py_ssize_t num = p->type_ignore_comments.num_items; |
863 | 7.63k | if (num > 0) { |
864 | | // Turn the raw (comment, lineno) pairs into TypeIgnore objects in the arena |
865 | 0 | type_ignores = _Py_asdl_type_ignore_seq_new(num, p->arena); |
866 | 0 | if (type_ignores == NULL) { |
867 | 0 | return NULL; |
868 | 0 | } |
869 | 0 | for (Py_ssize_t i = 0; i < num; i++) { |
870 | 0 | PyObject *tag = _PyPegen_new_type_comment(p, p->type_ignore_comments.items[i].comment); |
871 | 0 | if (tag == NULL) { |
872 | 0 | return NULL; |
873 | 0 | } |
874 | 0 | type_ignore_ty ti = _PyAST_TypeIgnore(p->type_ignore_comments.items[i].lineno, |
875 | 0 | tag, p->arena); |
876 | 0 | if (ti == NULL) { |
877 | 0 | return NULL; |
878 | 0 | } |
879 | 0 | asdl_seq_SET(type_ignores, i, ti); |
880 | 0 | } |
881 | 0 | } |
882 | 7.63k | return _PyAST_Module(a, type_ignores, p->arena); |
883 | 7.63k | } |
884 | | |
885 | | PyObject * |
886 | | _PyPegen_new_type_comment(Parser *p, const char *s) |
887 | 0 | { |
888 | 0 | PyObject *res = PyUnicode_DecodeUTF8(s, strlen(s), NULL); |
889 | 0 | if (res == NULL) { |
890 | 0 | return NULL; |
891 | 0 | } |
892 | 0 | if (_PyArena_AddPyObject(p->arena, res) < 0) { |
893 | 0 | Py_DECREF(res); |
894 | 0 | return NULL; |
895 | 0 | } |
896 | 0 | return res; |
897 | 0 | } |
898 | | |
899 | | arg_ty |
900 | | _PyPegen_add_type_comment_to_arg(Parser *p, arg_ty a, Token *tc) |
901 | 161k | { |
902 | 161k | if (tc == NULL) { |
903 | 161k | return a; |
904 | 161k | } |
905 | 0 | const char *bytes = PyBytes_AsString(tc->bytes); |
906 | 0 | if (bytes == NULL) { |
907 | 0 | return NULL; |
908 | 0 | } |
909 | 0 | PyObject *tco = _PyPegen_new_type_comment(p, bytes); |
910 | 0 | if (tco == NULL) { |
911 | 0 | return NULL; |
912 | 0 | } |
913 | 0 | return _PyAST_arg(a->arg, a->annotation, tco, |
914 | 0 | a->lineno, a->col_offset, a->end_lineno, a->end_col_offset, |
915 | 0 | p->arena); |
916 | 0 | } |
917 | | |
918 | | /* Checks if the NOTEQUAL token is valid given the current parser flags |
919 | | 0 indicates success and nonzero indicates failure (an exception may be set) */ |
920 | | int |
921 | 1.28k | _PyPegen_check_barry_as_flufl(Parser *p, Token* t) { |
922 | 1.28k | assert(t->bytes != NULL); |
923 | 1.28k | assert(t->type == NOTEQUAL); |
924 | | |
925 | 1.28k | const char* tok_str = PyBytes_AS_STRING(t->bytes); |
926 | 1.28k | if (p->flags & PyPARSE_BARRY_AS_BDFL && strcmp(tok_str, "<>") != 0) { |
927 | 1 | RAISE_SYNTAX_ERROR("with Barry as BDFL, use '<>' instead of '!='"); |
928 | 1 | return -1; |
929 | 1 | } |
930 | 1.28k | if (!(p->flags & PyPARSE_BARRY_AS_BDFL)) { |
931 | 1.28k | return strcmp(tok_str, "!="); |
932 | 1.28k | } |
933 | 0 | return 0; |
934 | 1.28k | } |
935 | | |
936 | | int |
937 | 8.58k | _PyPegen_check_legacy_stmt(Parser *p, expr_ty name) { |
938 | 8.58k | if (name->kind != Name_kind) { |
939 | 2.56k | return 0; |
940 | 2.56k | } |
941 | 6.02k | const char* candidates[2] = {"print", "exec"}; |
942 | 17.4k | for (int i=0; i<2; i++) { |
943 | 11.8k | if (PyUnicode_CompareWithASCIIString(name->v.Name.id, candidates[i]) == 0) { |
944 | 407 | return 1; |
945 | 407 | } |
946 | 11.8k | } |
947 | 5.61k | return 0; |
948 | 6.02k | } |
949 | | |
950 | | static ResultTokenWithMetadata * |
951 | | result_token_with_metadata(Parser *p, void *result, PyObject *metadata) |
952 | 7.73k | { |
953 | 7.73k | ResultTokenWithMetadata *res = _PyArena_Malloc(p->arena, sizeof(ResultTokenWithMetadata)); |
954 | 7.73k | if (res == NULL) { |
955 | 0 | return NULL; |
956 | 0 | } |
957 | 7.73k | res->metadata = metadata; |
958 | 7.73k | res->result = result; |
959 | 7.73k | return res; |
960 | 7.73k | } |
961 | | |
962 | | ResultTokenWithMetadata * |
963 | | _PyPegen_check_fstring_conversion(Parser *p, Token* conv_token, expr_ty conv) |
964 | 1.59k | { |
965 | 1.59k | if (conv_token->lineno != conv->lineno || conv_token->end_col_offset != conv->col_offset) { |
966 | 3 | return RAISE_SYNTAX_ERROR_KNOWN_RANGE( |
967 | 3 | conv_token, conv, |
968 | 3 | "%c-string: conversion type must come right after the exclamation mark", |
969 | 3 | TOK_GET_STRING_PREFIX(p->tok) |
970 | 3 | ); |
971 | 3 | } |
972 | | |
973 | 1.59k | Py_UCS4 first = PyUnicode_READ_CHAR(conv->v.Name.id, 0); |
974 | 1.59k | if (PyUnicode_GET_LENGTH(conv->v.Name.id) > 1 || |
975 | 1.59k | !(first == 's' || first == 'r' || first == 'a')) { |
976 | 12 | RAISE_SYNTAX_ERROR_KNOWN_LOCATION(conv, |
977 | 12 | "%c-string: invalid conversion character %R: expected 's', 'r', or 'a'", |
978 | 12 | TOK_GET_STRING_PREFIX(p->tok), |
979 | 12 | conv->v.Name.id); |
980 | 12 | return NULL; |
981 | 12 | } |
982 | | |
983 | 1.57k | return result_token_with_metadata(p, conv, conv_token->metadata); |
984 | 1.59k | } |
985 | | |
986 | | ResultTokenWithMetadata * |
987 | | _PyPegen_setup_full_format_spec(Parser *p, Token *colon, asdl_expr_seq *spec, int lineno, int col_offset, |
988 | | int end_lineno, int end_col_offset, PyArena *arena) |
989 | 6.15k | { |
990 | 6.15k | if (!spec) { |
991 | 0 | return NULL; |
992 | 0 | } |
993 | | |
994 | | // This is needed to keep compatibility with 3.11, where an empty format |
995 | | // spec is parsed as an *empty* JoinedStr node, instead of having an empty |
996 | | // constant in it. |
997 | 6.15k | Py_ssize_t n_items = asdl_seq_LEN(spec); |
998 | 6.15k | Py_ssize_t non_empty_count = 0; |
999 | 17.8k | for (Py_ssize_t i = 0; i < n_items; i++) { |
1000 | 11.6k | expr_ty item = asdl_seq_GET(spec, i); |
1001 | 11.6k | non_empty_count += !(item->kind == Constant_kind && |
1002 | 11.6k | PyUnicode_CheckExact(item->v.Constant.value) && |
1003 | 11.6k | PyUnicode_GET_LENGTH(item->v.Constant.value) == 0); |
1004 | 11.6k | } |
1005 | 6.15k | if (non_empty_count != n_items) { |
1006 | 2.12k | asdl_expr_seq *resized_spec = |
1007 | 2.12k | _Py_asdl_expr_seq_new(non_empty_count, p->arena); |
1008 | 2.12k | if (resized_spec == NULL) { |
1009 | 0 | return NULL; |
1010 | 0 | } |
1011 | 2.12k | Py_ssize_t j = 0; |
1012 | 6.08k | for (Py_ssize_t i = 0; i < n_items; i++) { |
1013 | 3.96k | expr_ty item = asdl_seq_GET(spec, i); |
1014 | 3.96k | if (item->kind == Constant_kind && |
1015 | 3.96k | PyUnicode_CheckExact(item->v.Constant.value) && |
1016 | 3.96k | PyUnicode_GET_LENGTH(item->v.Constant.value) == 0) { |
1017 | 2.12k | continue; |
1018 | 2.12k | } |
1019 | 1.83k | asdl_seq_SET(resized_spec, j++, item); |
1020 | 1.83k | } |
1021 | 2.12k | assert(j == non_empty_count); |
1022 | 2.12k | spec = resized_spec; |
1023 | 2.12k | } |
1024 | 6.15k | expr_ty res; |
1025 | 6.15k | Py_ssize_t n = asdl_seq_LEN(spec); |
1026 | 6.15k | if (n == 0 || (n == 1 && asdl_seq_GET(spec, 0)->kind == Constant_kind)) { |
1027 | 5.21k | res = _PyAST_JoinedStr(spec, lineno, col_offset, end_lineno, |
1028 | 5.21k | end_col_offset, p->arena); |
1029 | 5.21k | } else { |
1030 | 942 | res = _PyPegen_concatenate_strings(p, spec, |
1031 | 942 | lineno, col_offset, end_lineno, |
1032 | 942 | end_col_offset, arena); |
1033 | 942 | } |
1034 | 6.15k | if (!res) { |
1035 | 0 | return NULL; |
1036 | 0 | } |
1037 | 6.15k | return result_token_with_metadata(p, res, colon->metadata); |
1038 | 6.15k | } |
1039 | | |
1040 | | const char * |
1041 | | _PyPegen_get_expr_name(expr_ty e) |
1042 | 177 | { |
1043 | 177 | assert(e != NULL); |
1044 | 177 | switch (e->kind) { |
1045 | 1 | case Attribute_kind: |
1046 | 1 | return "attribute"; |
1047 | 1 | case Subscript_kind: |
1048 | 1 | return "subscript"; |
1049 | 1 | case Starred_kind: |
1050 | 1 | return "starred"; |
1051 | 4 | case Name_kind: |
1052 | 4 | return "name"; |
1053 | 1 | case List_kind: |
1054 | 1 | return "list"; |
1055 | 5 | case Tuple_kind: |
1056 | 5 | return "tuple"; |
1057 | 2 | case Lambda_kind: |
1058 | 2 | return "lambda"; |
1059 | 11 | case Call_kind: |
1060 | 11 | return "function call"; |
1061 | 3 | case BoolOp_kind: |
1062 | 25 | case BinOp_kind: |
1063 | 31 | case UnaryOp_kind: |
1064 | 31 | return "expression"; |
1065 | 1 | case GeneratorExp_kind: |
1066 | 1 | return "generator expression"; |
1067 | 1 | case Yield_kind: |
1068 | 2 | case YieldFrom_kind: |
1069 | 2 | return "yield expression"; |
1070 | 1 | case Await_kind: |
1071 | 1 | return "await expression"; |
1072 | 1 | case ListComp_kind: |
1073 | 1 | return "list comprehension"; |
1074 | 1 | case SetComp_kind: |
1075 | 1 | return "set comprehension"; |
1076 | 1 | case DictComp_kind: |
1077 | 1 | return "dict comprehension"; |
1078 | 1 | case Dict_kind: |
1079 | 1 | return "dict literal"; |
1080 | 1 | case Set_kind: |
1081 | 1 | return "set display"; |
1082 | 6 | case JoinedStr_kind: |
1083 | 6 | case FormattedValue_kind: |
1084 | 6 | return "f-string expression"; |
1085 | 7 | case TemplateStr_kind: |
1086 | 7 | case Interpolation_kind: |
1087 | 7 | return "t-string expression"; |
1088 | 81 | case Constant_kind: { |
1089 | 81 | PyObject *value = e->v.Constant.value; |
1090 | 81 | if (value == Py_None) { |
1091 | 1 | return "None"; |
1092 | 1 | } |
1093 | 80 | if (value == Py_False) { |
1094 | 1 | return "False"; |
1095 | 1 | } |
1096 | 79 | if (value == Py_True) { |
1097 | 1 | return "True"; |
1098 | 1 | } |
1099 | 78 | if (value == Py_Ellipsis) { |
1100 | 1 | return "ellipsis"; |
1101 | 1 | } |
1102 | 77 | return "literal"; |
1103 | 78 | } |
1104 | 15 | case Compare_kind: |
1105 | 15 | return "comparison"; |
1106 | 1 | case IfExp_kind: |
1107 | 1 | return "conditional expression"; |
1108 | 1 | case NamedExpr_kind: |
1109 | 1 | return "named expression"; |
1110 | 0 | default: |
1111 | 0 | PyErr_Format(PyExc_SystemError, |
1112 | 0 | "unexpected expression in assignment %d (line %d)", |
1113 | 0 | e->kind, e->lineno); |
1114 | 0 | return NULL; |
1115 | 177 | } |
1116 | 177 | } |
1117 | | |
1118 | | expr_ty |
1119 | 18 | _PyPegen_get_last_comprehension_item(comprehension_ty comprehension) { |
1120 | 18 | if (comprehension->ifs == NULL || asdl_seq_LEN(comprehension->ifs) == 0) { |
1121 | 14 | return comprehension->iter; |
1122 | 14 | } |
1123 | 4 | return PyPegen_last_item(comprehension->ifs, expr_ty); |
1124 | 18 | } |
1125 | | |
1126 | | expr_ty _PyPegen_collect_call_seqs(Parser *p, asdl_expr_seq *a, asdl_seq *b, |
1127 | | int lineno, int col_offset, int end_lineno, |
1128 | 34.6k | int end_col_offset, PyArena *arena) { |
1129 | 34.6k | Py_ssize_t args_len = asdl_seq_LEN(a); |
1130 | 34.6k | Py_ssize_t total_len = args_len; |
1131 | | |
1132 | 34.6k | if (b == NULL) { |
1133 | 32.1k | return _PyAST_Call(_PyPegen_dummy_name(p), a, NULL, lineno, col_offset, |
1134 | 32.1k | end_lineno, end_col_offset, arena); |
1135 | | |
1136 | 32.1k | } |
1137 | | |
1138 | 2.48k | asdl_expr_seq *starreds = _PyPegen_seq_extract_starred_exprs(p, b); |
1139 | 2.48k | asdl_keyword_seq *keywords = _PyPegen_seq_delete_starred_exprs(p, b); |
1140 | | |
1141 | 2.48k | if (starreds) { |
1142 | 284 | total_len += asdl_seq_LEN(starreds); |
1143 | 284 | } |
1144 | | |
1145 | 2.48k | asdl_expr_seq *args = _Py_asdl_expr_seq_new(total_len, arena); |
1146 | 2.48k | if (args == NULL) { |
1147 | 0 | return NULL; |
1148 | 0 | } |
1149 | | |
1150 | 2.48k | Py_ssize_t i = 0; |
1151 | 5.98k | for (i = 0; i < args_len; i++) { |
1152 | 3.50k | asdl_seq_SET(args, i, asdl_seq_GET(a, i)); |
1153 | 3.50k | } |
1154 | 3.18k | for (; i < total_len; i++) { |
1155 | 692 | asdl_seq_SET(args, i, asdl_seq_GET(starreds, i - args_len)); |
1156 | 692 | } |
1157 | | |
1158 | 2.48k | return _PyAST_Call(_PyPegen_dummy_name(p), args, keywords, lineno, |
1159 | 2.48k | col_offset, end_lineno, end_col_offset, arena); |
1160 | 2.48k | } |
1161 | | |
1162 | | // AST Error reporting helpers |
1163 | | |
1164 | | expr_ty |
1165 | | _PyPegen_get_invalid_target(expr_ty e, TARGETS_TYPE targets_type) |
1166 | 10.1k | { |
1167 | 10.1k | if (e == NULL) { |
1168 | 0 | return NULL; |
1169 | 0 | } |
1170 | | |
1171 | 10.1k | #define VISIT_CONTAINER(CONTAINER, TYPE) do { \ |
1172 | 2.59k | Py_ssize_t len = asdl_seq_LEN((CONTAINER)->v.TYPE.elts);\ |
1173 | 8.40k | for (Py_ssize_t i = 0; i < len; i++) {\ |
1174 | 5.92k | expr_ty other = asdl_seq_GET((CONTAINER)->v.TYPE.elts, i);\ |
1175 | 5.92k | expr_ty child = _PyPegen_get_invalid_target(other, targets_type);\ |
1176 | 5.92k | if (child != NULL) {\ |
1177 | 111 | return child;\ |
1178 | 111 | }\ |
1179 | 5.92k | }\ |
1180 | 2.59k | } while (0) |
1181 | | |
1182 | | // We only need to visit List and Tuple nodes recursively as those |
1183 | | // are the only ones that can contain valid names in targets when |
1184 | | // they are parsed as expressions. Any other kind of expression |
1185 | | // that is a container (like Sets or Dicts) is directly invalid and |
1186 | | // we don't need to visit it recursively. |
1187 | | |
1188 | 10.1k | switch (e->kind) { |
1189 | 857 | case List_kind: |
1190 | 857 | VISIT_CONTAINER(e, List); |
1191 | 838 | return NULL; |
1192 | 1.73k | case Tuple_kind: |
1193 | 1.73k | VISIT_CONTAINER(e, Tuple); |
1194 | 1.64k | return NULL; |
1195 | 1.90k | case Starred_kind: |
1196 | 1.90k | if (targets_type == DEL_TARGETS) { |
1197 | 1 | return e; |
1198 | 1 | } |
1199 | 1.90k | return _PyPegen_get_invalid_target(e->v.Starred.value, targets_type); |
1200 | 377 | case Compare_kind: |
1201 | | // This is needed, because the `a in b` in `for a in b` gets parsed |
1202 | | // as a comparison, and so we need to search the left side of the comparison |
1203 | | // for invalid targets. |
1204 | 377 | if (targets_type == FOR_TARGETS) { |
1205 | 365 | cmpop_ty cmpop = (cmpop_ty) asdl_seq_GET(e->v.Compare.ops, 0); |
1206 | 365 | if (cmpop == In) { |
1207 | 75 | return _PyPegen_get_invalid_target(e->v.Compare.left, targets_type); |
1208 | 75 | } |
1209 | 290 | return NULL; |
1210 | 365 | } |
1211 | 12 | return e; |
1212 | 3.86k | case Name_kind: |
1213 | 4.63k | case Subscript_kind: |
1214 | 5.11k | case Attribute_kind: |
1215 | 5.11k | return NULL; |
1216 | 121 | default: |
1217 | 121 | return e; |
1218 | 10.1k | } |
1219 | 10.1k | } |
1220 | | |
1221 | 35 | void *_PyPegen_arguments_parsing_error(Parser *p, expr_ty e) { |
1222 | 35 | int kwarg_unpacking = 0; |
1223 | 629 | for (Py_ssize_t i = 0, l = asdl_seq_LEN(e->v.Call.keywords); i < l; i++) { |
1224 | 594 | keyword_ty keyword = asdl_seq_GET(e->v.Call.keywords, i); |
1225 | 594 | if (!keyword->arg) { |
1226 | 275 | kwarg_unpacking = 1; |
1227 | 275 | } |
1228 | 594 | } |
1229 | | |
1230 | 35 | const char *msg = NULL; |
1231 | 35 | if (kwarg_unpacking) { |
1232 | 20 | msg = "positional argument follows keyword argument unpacking"; |
1233 | 20 | } else { |
1234 | 15 | msg = "positional argument follows keyword argument"; |
1235 | 15 | } |
1236 | | |
1237 | 35 | return RAISE_SYNTAX_ERROR(msg); |
1238 | 35 | } |
1239 | | |
1240 | | void * |
1241 | | _PyPegen_nonparen_genexp_in_call(Parser *p, expr_ty args, asdl_comprehension_seq *comprehensions) |
1242 | 294 | { |
1243 | | /* The rule that calls this function is 'args for_if_clauses'. |
1244 | | For the input f(L, x for x in y), L and x are in args and |
1245 | | the for is parsed as a for_if_clause. We have to check if |
1246 | | len <= 1, so that input like dict((a, b) for a, b in x) |
1247 | | gets successfully parsed and then we pass the last |
1248 | | argument (x in the above example) as the location of the |
1249 | | error */ |
1250 | 294 | Py_ssize_t len = asdl_seq_LEN(args->v.Call.args); |
1251 | 294 | if (len <= 1) { |
1252 | 291 | return NULL; |
1253 | 291 | } |
1254 | | |
1255 | 3 | comprehension_ty last_comprehension = PyPegen_last_item(comprehensions, comprehension_ty); |
1256 | | |
1257 | 3 | return RAISE_SYNTAX_ERROR_KNOWN_RANGE( |
1258 | 294 | (expr_ty) asdl_seq_GET(args->v.Call.args, len - 1), |
1259 | 294 | _PyPegen_get_last_comprehension_item(last_comprehension), |
1260 | 294 | "Generator expression must be parenthesized" |
1261 | 294 | ); |
1262 | 294 | } |
1263 | | |
1264 | | // Fstring stuff |
1265 | | |
1266 | | static expr_ty |
1267 | 25.3k | _PyPegen_decode_fstring_part(Parser* p, int is_raw, expr_ty constant, Token* token) { |
1268 | 25.3k | assert(PyUnicode_CheckExact(constant->v.Constant.value)); |
1269 | | |
1270 | 25.3k | const char* bstr = PyUnicode_AsUTF8(constant->v.Constant.value); |
1271 | 25.3k | if (bstr == NULL) { |
1272 | 0 | return NULL; |
1273 | 0 | } |
1274 | | |
1275 | 25.3k | size_t len; |
1276 | 25.3k | if (strcmp(bstr, "{{") == 0 || strcmp(bstr, "}}") == 0) { |
1277 | 0 | len = 1; |
1278 | 25.3k | } else { |
1279 | 25.3k | len = strlen(bstr); |
1280 | 25.3k | } |
1281 | | |
1282 | 25.3k | is_raw = is_raw || strchr(bstr, '\\') == NULL; |
1283 | 25.3k | PyObject *str = _PyPegen_decode_string(p, is_raw, bstr, len, token); |
1284 | 25.3k | if (str == NULL) { |
1285 | 5 | _Pypegen_raise_decode_error(p); |
1286 | 5 | return NULL; |
1287 | 5 | } |
1288 | 25.3k | if (_PyArena_AddPyObject(p->arena, str) < 0) { |
1289 | 0 | Py_DECREF(str); |
1290 | 0 | return NULL; |
1291 | 0 | } |
1292 | 25.3k | return _PyAST_Constant(str, NULL, constant->lineno, constant->col_offset, |
1293 | 25.3k | constant->end_lineno, constant->end_col_offset, |
1294 | 25.3k | p->arena); |
1295 | 25.3k | } |
1296 | | |
1297 | | static asdl_expr_seq * |
1298 | | _get_resized_exprs(Parser *p, Token *a, asdl_expr_seq *raw_expressions, Token *b, enum string_kind_t string_kind) |
1299 | 18.1k | { |
1300 | 18.1k | Py_ssize_t n_items = asdl_seq_LEN(raw_expressions); |
1301 | 18.1k | Py_ssize_t total_items = n_items; |
1302 | 61.9k | for (Py_ssize_t i = 0; i < n_items; i++) { |
1303 | 43.7k | expr_ty item = asdl_seq_GET(raw_expressions, i); |
1304 | 43.7k | if (item->kind == JoinedStr_kind) { |
1305 | 5.25k | total_items += asdl_seq_LEN(item->v.JoinedStr.values) - 1; |
1306 | 5.25k | } |
1307 | 43.7k | } |
1308 | | |
1309 | 18.1k | const char* quote_str = PyBytes_AsString(a->bytes); |
1310 | 18.1k | if (quote_str == NULL) { |
1311 | 0 | return NULL; |
1312 | 0 | } |
1313 | 18.1k | int is_raw = strpbrk(quote_str, "rR") != NULL; |
1314 | | |
1315 | 18.1k | asdl_expr_seq *seq = _Py_asdl_expr_seq_new(total_items, p->arena); |
1316 | 18.1k | if (seq == NULL) { |
1317 | 0 | return NULL; |
1318 | 0 | } |
1319 | | |
1320 | 18.1k | Py_ssize_t index = 0; |
1321 | 61.9k | for (Py_ssize_t i = 0; i < n_items; i++) { |
1322 | 43.7k | expr_ty item = asdl_seq_GET(raw_expressions, i); |
1323 | | |
1324 | | // This should correspond to a JoinedStr node of two elements |
1325 | | // created _PyPegen_formatted_value. This situation can only be the result of |
1326 | | // a (f|t)-string debug expression where the first element is a constant with the text and the second |
1327 | | // a formatted value with the expression. |
1328 | 43.7k | if (item->kind == JoinedStr_kind) { |
1329 | 5.25k | asdl_expr_seq *values = item->v.JoinedStr.values; |
1330 | 5.25k | if (asdl_seq_LEN(values) != 2) { |
1331 | 0 | PyErr_Format(PyExc_SystemError, |
1332 | 0 | string_kind == TSTRING |
1333 | 0 | ? "unexpected TemplateStr node without debug data in t-string at line %d" |
1334 | 0 | : "unexpected JoinedStr node without debug data in f-string at line %d", |
1335 | 0 | item->lineno); |
1336 | 0 | return NULL; |
1337 | 0 | } |
1338 | | |
1339 | 5.25k | expr_ty first = asdl_seq_GET(values, 0); |
1340 | 5.25k | assert(first->kind == Constant_kind); |
1341 | 5.25k | asdl_seq_SET(seq, index++, first); |
1342 | | |
1343 | 5.25k | expr_ty second = asdl_seq_GET(values, 1); |
1344 | 5.25k | assert((string_kind == TSTRING && second->kind == Interpolation_kind) || second->kind == FormattedValue_kind); |
1345 | 5.25k | asdl_seq_SET(seq, index++, second); |
1346 | | |
1347 | 5.25k | continue; |
1348 | 5.25k | } |
1349 | | |
1350 | 38.5k | if (item->kind == Constant_kind) { |
1351 | 25.3k | item = _PyPegen_decode_fstring_part(p, is_raw, item, b); |
1352 | 25.3k | if (item == NULL) { |
1353 | 5 | return NULL; |
1354 | 5 | } |
1355 | | |
1356 | | /* Tokenizer emits string parts even when the underlying string |
1357 | | might become an empty value (e.g. FSTRING_MIDDLE with the value \\n) |
1358 | | so we need to check for them and simplify it here. */ |
1359 | 25.3k | if (PyUnicode_CheckExact(item->v.Constant.value) |
1360 | 25.3k | && PyUnicode_GET_LENGTH(item->v.Constant.value) == 0) { |
1361 | 424 | continue; |
1362 | 424 | } |
1363 | 25.3k | } |
1364 | 38.0k | asdl_seq_SET(seq, index++, item); |
1365 | 38.0k | } |
1366 | | |
1367 | 18.1k | asdl_expr_seq *resized_exprs; |
1368 | 18.1k | if (index != total_items) { |
1369 | 424 | resized_exprs = _Py_asdl_expr_seq_new(index, p->arena); |
1370 | 424 | if (resized_exprs == NULL) { |
1371 | 0 | return NULL; |
1372 | 0 | } |
1373 | 869 | for (Py_ssize_t i = 0; i < index; i++) { |
1374 | 445 | asdl_seq_SET(resized_exprs, i, asdl_seq_GET(seq, i)); |
1375 | 445 | } |
1376 | 424 | } |
1377 | 17.7k | else { |
1378 | 17.7k | resized_exprs = seq; |
1379 | 17.7k | } |
1380 | 18.1k | return resized_exprs; |
1381 | 18.1k | } |
1382 | | |
1383 | | expr_ty |
1384 | 3.86k | _PyPegen_template_str(Parser *p, Token *a, asdl_expr_seq *raw_expressions, Token *b) { |
1385 | | |
1386 | 3.86k | asdl_expr_seq *resized_exprs = _get_resized_exprs(p, a, raw_expressions, b, TSTRING); |
1387 | 3.86k | return _PyAST_TemplateStr(resized_exprs, a->lineno, a->col_offset, |
1388 | 3.86k | b->end_lineno, b->end_col_offset, |
1389 | 3.86k | p->arena); |
1390 | 3.86k | } |
1391 | | |
1392 | | expr_ty |
1393 | 14.2k | _PyPegen_joined_str(Parser *p, Token* a, asdl_expr_seq* raw_expressions, Token*b) { |
1394 | | |
1395 | 14.2k | asdl_expr_seq *resized_exprs = _get_resized_exprs(p, a, raw_expressions, b, FSTRING); |
1396 | 14.2k | return _PyAST_JoinedStr(resized_exprs, a->lineno, a->col_offset, |
1397 | 14.2k | b->end_lineno, b->end_col_offset, |
1398 | 14.2k | p->arena); |
1399 | 14.2k | } |
1400 | | |
1401 | 8.21k | expr_ty _PyPegen_decoded_constant_from_token(Parser* p, Token* tok) { |
1402 | 8.21k | Py_ssize_t bsize; |
1403 | 8.21k | char* bstr; |
1404 | 8.21k | if (PyBytes_AsStringAndSize(tok->bytes, &bstr, &bsize) == -1) { |
1405 | 0 | return NULL; |
1406 | 0 | } |
1407 | | |
1408 | | // Check if we're inside a raw f-string for format spec decoding |
1409 | 8.21k | int is_raw = 0; |
1410 | 8.21k | if (INSIDE_FSTRING(p->tok)) { |
1411 | 5.28k | tokenizer_mode *mode = TOK_GET_MODE(p->tok); |
1412 | 5.28k | is_raw = mode->raw; |
1413 | 5.28k | } |
1414 | | |
1415 | 8.21k | PyObject* str = _PyPegen_decode_string(p, is_raw, bstr, bsize, tok); |
1416 | 8.21k | if (str == NULL) { |
1417 | 2 | return NULL; |
1418 | 2 | } |
1419 | 8.21k | if (_PyArena_AddPyObject(p->arena, str) < 0) { |
1420 | 0 | Py_DECREF(str); |
1421 | 0 | return NULL; |
1422 | 0 | } |
1423 | 8.21k | return _PyAST_Constant(str, NULL, tok->lineno, tok->col_offset, |
1424 | 8.21k | tok->end_lineno, tok->end_col_offset, |
1425 | 8.21k | p->arena); |
1426 | 8.21k | } |
1427 | | |
1428 | 32.0k | expr_ty _PyPegen_constant_from_token(Parser* p, Token* tok) { |
1429 | 32.0k | char* bstr = PyBytes_AsString(tok->bytes); |
1430 | 32.0k | if (bstr == NULL) { |
1431 | 0 | return NULL; |
1432 | 0 | } |
1433 | 32.0k | PyObject* str = PyUnicode_FromString(bstr); |
1434 | 32.0k | if (str == NULL) { |
1435 | 12 | return NULL; |
1436 | 12 | } |
1437 | 32.0k | if (_PyArena_AddPyObject(p->arena, str) < 0) { |
1438 | 0 | Py_DECREF(str); |
1439 | 0 | return NULL; |
1440 | 0 | } |
1441 | 32.0k | return _PyAST_Constant(str, NULL, tok->lineno, tok->col_offset, |
1442 | 32.0k | tok->end_lineno, tok->end_col_offset, |
1443 | 32.0k | p->arena); |
1444 | 32.0k | } |
1445 | | |
1446 | 74.7k | expr_ty _PyPegen_constant_from_string(Parser* p, Token* tok) { |
1447 | 74.7k | char* the_str = PyBytes_AsString(tok->bytes); |
1448 | 74.7k | if (the_str == NULL) { |
1449 | 0 | return NULL; |
1450 | 0 | } |
1451 | 74.7k | PyObject *s = _PyPegen_parse_string(p, tok); |
1452 | 74.7k | if (s == NULL) { |
1453 | 370 | _Pypegen_raise_decode_error(p); |
1454 | 370 | return NULL; |
1455 | 370 | } |
1456 | 74.4k | if (_PyArena_AddPyObject(p->arena, s) < 0) { |
1457 | 0 | Py_DECREF(s); |
1458 | 0 | return NULL; |
1459 | 0 | } |
1460 | 74.4k | PyObject *kind = NULL; |
1461 | 74.4k | if (the_str && the_str[0] == 'u') { |
1462 | 517 | kind = _PyPegen_new_identifier(p, "u"); |
1463 | 517 | if (kind == NULL) { |
1464 | 0 | return NULL; |
1465 | 0 | } |
1466 | 517 | } |
1467 | 74.4k | return _PyAST_Constant(s, kind, tok->lineno, tok->col_offset, tok->end_lineno, tok->end_col_offset, p->arena); |
1468 | 74.4k | } |
1469 | | |
1470 | | static int |
1471 | | _get_interpolation_conversion(Parser *p, Token *debug, ResultTokenWithMetadata *conversion, |
1472 | | ResultTokenWithMetadata *format) |
1473 | 25.6k | { |
1474 | 25.6k | if (conversion != NULL) { |
1475 | 1.55k | expr_ty conversion_expr = (expr_ty) conversion->result; |
1476 | 1.55k | assert(conversion_expr->kind == Name_kind); |
1477 | 1.55k | Py_UCS4 first = PyUnicode_READ_CHAR(conversion_expr->v.Name.id, 0); |
1478 | 1.55k | return Py_SAFE_DOWNCAST(first, Py_UCS4, int); |
1479 | 1.55k | } |
1480 | 24.1k | else if (debug && !format) { |
1481 | | /* If no conversion is specified, use !r for debug expressions */ |
1482 | 5.35k | return (int)'r'; |
1483 | 5.35k | } |
1484 | 18.7k | return -1; |
1485 | 25.6k | } |
1486 | | |
1487 | | static PyObject * |
1488 | | _strip_interpolation_expr(PyObject *exprstr) |
1489 | 5.12k | { |
1490 | 5.12k | Py_ssize_t len = PyUnicode_GET_LENGTH(exprstr); |
1491 | | |
1492 | 10.4k | for (Py_ssize_t i = len - 1; i >= 0; i--) { |
1493 | 10.4k | Py_UCS4 c = PyUnicode_READ_CHAR(exprstr, i); |
1494 | 10.4k | if (_PyUnicode_IsWhitespace(c) || c == '=') { |
1495 | 5.35k | len--; |
1496 | 5.35k | } |
1497 | 5.12k | else { |
1498 | 5.12k | break; |
1499 | 5.12k | } |
1500 | 10.4k | } |
1501 | | |
1502 | 5.12k | return PyUnicode_Substring(exprstr, 0, len); |
1503 | 5.12k | } |
1504 | | |
1505 | | expr_ty _PyPegen_interpolation(Parser *p, expr_ty expression, Token *debug, ResultTokenWithMetadata *conversion, |
1506 | | ResultTokenWithMetadata *format, Token *closing_brace, int lineno, int col_offset, |
1507 | 5.12k | int end_lineno, int end_col_offset, PyArena *arena) { |
1508 | | |
1509 | 5.12k | int conversion_val = _get_interpolation_conversion(p, debug, conversion, format); |
1510 | | |
1511 | | /* Find the non whitespace token after the "=" */ |
1512 | 5.12k | int debug_end_line, debug_end_offset; |
1513 | 5.12k | PyObject *debug_metadata; |
1514 | 5.12k | constant exprstr; |
1515 | | |
1516 | 5.12k | if (conversion) { |
1517 | 543 | debug_end_line = ((expr_ty) conversion->result)->lineno; |
1518 | 543 | debug_end_offset = ((expr_ty) conversion->result)->col_offset; |
1519 | 543 | debug_metadata = exprstr = conversion->metadata; |
1520 | 543 | } |
1521 | 4.58k | else if (format) { |
1522 | 672 | debug_end_line = ((expr_ty) format->result)->lineno; |
1523 | 672 | debug_end_offset = ((expr_ty) format->result)->col_offset + 1; |
1524 | 672 | debug_metadata = exprstr = format->metadata; |
1525 | 672 | } |
1526 | 3.90k | else { |
1527 | 3.90k | debug_end_line = end_lineno; |
1528 | 3.90k | debug_end_offset = end_col_offset; |
1529 | 3.90k | debug_metadata = exprstr = closing_brace->metadata; |
1530 | 3.90k | } |
1531 | | |
1532 | 5.12k | assert(exprstr != NULL); |
1533 | 5.12k | PyObject *final_exprstr = _strip_interpolation_expr(exprstr); |
1534 | 5.12k | if (!final_exprstr || _PyArena_AddPyObject(arena, final_exprstr) < 0) { |
1535 | 0 | Py_XDECREF(final_exprstr); |
1536 | 0 | return NULL; |
1537 | 0 | } |
1538 | | |
1539 | 5.12k | expr_ty interpolation = _PyAST_Interpolation( |
1540 | 5.12k | expression, final_exprstr, conversion_val, format ? (expr_ty) format->result : NULL, |
1541 | 5.12k | lineno, col_offset, end_lineno, |
1542 | 5.12k | end_col_offset, arena |
1543 | 5.12k | ); |
1544 | | |
1545 | 5.12k | if (!debug) { |
1546 | 4.77k | return interpolation; |
1547 | 4.77k | } |
1548 | | |
1549 | 353 | expr_ty debug_text = _PyAST_Constant(debug_metadata, NULL, lineno, col_offset + 1, debug_end_line, |
1550 | 353 | debug_end_offset - 1, p->arena); |
1551 | 353 | if (!debug_text) { |
1552 | 0 | return NULL; |
1553 | 0 | } |
1554 | | |
1555 | 353 | asdl_expr_seq *values = _Py_asdl_expr_seq_new(2, arena); |
1556 | 353 | asdl_seq_SET(values, 0, debug_text); |
1557 | 353 | asdl_seq_SET(values, 1, interpolation); |
1558 | 353 | return _PyAST_JoinedStr(values, lineno, col_offset, debug_end_line, debug_end_offset, p->arena); |
1559 | 353 | } |
1560 | | |
1561 | | expr_ty _PyPegen_formatted_value(Parser *p, expr_ty expression, Token *debug, ResultTokenWithMetadata *conversion, |
1562 | | ResultTokenWithMetadata *format, Token *closing_brace, int lineno, int col_offset, |
1563 | 20.5k | int end_lineno, int end_col_offset, PyArena *arena) { |
1564 | 20.5k | int conversion_val = _get_interpolation_conversion(p, debug, conversion, format); |
1565 | | |
1566 | 20.5k | expr_ty formatted_value = _PyAST_FormattedValue( |
1567 | 20.5k | expression, conversion_val, format ? (expr_ty) format->result : NULL, |
1568 | 20.5k | lineno, col_offset, end_lineno, |
1569 | 20.5k | end_col_offset, arena |
1570 | 20.5k | ); |
1571 | | |
1572 | 20.5k | if (!debug) { |
1573 | 14.7k | return formatted_value; |
1574 | 14.7k | } |
1575 | | |
1576 | | /* Find the non whitespace token after the "=" */ |
1577 | 5.80k | int debug_end_line, debug_end_offset; |
1578 | 5.80k | PyObject *debug_metadata; |
1579 | | |
1580 | 5.80k | if (conversion) { |
1581 | 565 | debug_end_line = ((expr_ty) conversion->result)->lineno; |
1582 | 565 | debug_end_offset = ((expr_ty) conversion->result)->col_offset; |
1583 | 565 | debug_metadata = conversion->metadata; |
1584 | 565 | } |
1585 | 5.23k | else if (format) { |
1586 | 113 | debug_end_line = ((expr_ty) format->result)->lineno; |
1587 | 113 | debug_end_offset = ((expr_ty) format->result)->col_offset + 1; |
1588 | 113 | debug_metadata = format->metadata; |
1589 | 113 | } |
1590 | 5.12k | else { |
1591 | 5.12k | debug_end_line = end_lineno; |
1592 | 5.12k | debug_end_offset = end_col_offset; |
1593 | 5.12k | debug_metadata = closing_brace->metadata; |
1594 | 5.12k | } |
1595 | 5.80k | expr_ty debug_text = _PyAST_Constant(debug_metadata, NULL, lineno, col_offset + 1, debug_end_line, |
1596 | 5.80k | debug_end_offset - 1, p->arena); |
1597 | 5.80k | if (!debug_text) { |
1598 | 2 | return NULL; |
1599 | 2 | } |
1600 | | |
1601 | 5.79k | asdl_expr_seq *values = _Py_asdl_expr_seq_new(2, arena); |
1602 | 5.79k | asdl_seq_SET(values, 0, debug_text); |
1603 | 5.79k | asdl_seq_SET(values, 1, formatted_value); |
1604 | 5.79k | return _PyAST_JoinedStr(values, lineno, col_offset, debug_end_line, debug_end_offset, p->arena); |
1605 | 5.80k | } |
1606 | | |
1607 | | static expr_ty |
1608 | | _build_concatenated_bytes(Parser *p, asdl_expr_seq *strings, int lineno, |
1609 | | int col_offset, int end_lineno, int end_col_offset, |
1610 | | PyArena *arena) |
1611 | 844 | { |
1612 | 844 | Py_ssize_t len = asdl_seq_LEN(strings); |
1613 | 844 | assert(len > 0); |
1614 | | |
1615 | 844 | PyObject* res = Py_GetConstant(Py_CONSTANT_EMPTY_BYTES); |
1616 | | |
1617 | | /* Bytes literals never get a kind, but just for consistency |
1618 | | since they are represented as Constant nodes, we'll mirror |
1619 | | the same behavior as unicode strings for determining the |
1620 | | kind. */ |
1621 | 844 | PyObject* kind = asdl_seq_GET(strings, 0)->v.Constant.kind; |
1622 | 4.10k | for (Py_ssize_t i = 0; i < len; i++) { |
1623 | 3.25k | expr_ty elem = asdl_seq_GET(strings, i); |
1624 | 3.25k | PyBytes_Concat(&res, elem->v.Constant.value); |
1625 | 3.25k | } |
1626 | 844 | if (!res || _PyArena_AddPyObject(arena, res) < 0) { |
1627 | 0 | Py_XDECREF(res); |
1628 | 0 | return NULL; |
1629 | 0 | } |
1630 | 844 | return _PyAST_Constant(res, kind, lineno, col_offset, end_lineno, end_col_offset, p->arena); |
1631 | 844 | } |
1632 | | |
1633 | | static expr_ty |
1634 | | _build_concatenated_unicode(Parser *p, asdl_expr_seq *strings, int lineno, |
1635 | | int col_offset, int end_lineno, int end_col_offset, |
1636 | | PyArena *arena) |
1637 | 1.63k | { |
1638 | 1.63k | Py_ssize_t len = asdl_seq_LEN(strings); |
1639 | 1.63k | assert(len > 1); |
1640 | | |
1641 | 1.63k | expr_ty first = asdl_seq_GET(strings, 0); |
1642 | | |
1643 | | /* When a string is getting concatenated, the kind of the string |
1644 | | is determined by the first string in the concatenation |
1645 | | sequence. |
1646 | | |
1647 | | u"abc" "def" -> u"abcdef" |
1648 | | "abc" u"abc" -> "abcabc" */ |
1649 | 1.63k | PyObject *kind = first->v.Constant.kind; |
1650 | | |
1651 | 1.63k | PyUnicodeWriter *writer = PyUnicodeWriter_Create(0); |
1652 | 1.63k | if (writer == NULL) { |
1653 | 0 | return NULL; |
1654 | 0 | } |
1655 | | |
1656 | 22.8k | for (Py_ssize_t i = 0; i < len; i++) { |
1657 | 21.1k | expr_ty current_elem = asdl_seq_GET(strings, i); |
1658 | 21.1k | assert(current_elem->kind == Constant_kind); |
1659 | | |
1660 | 21.1k | if (PyUnicodeWriter_WriteStr(writer, |
1661 | 21.1k | current_elem->v.Constant.value)) { |
1662 | 0 | PyUnicodeWriter_Discard(writer); |
1663 | 0 | return NULL; |
1664 | 0 | } |
1665 | 21.1k | } |
1666 | | |
1667 | 1.63k | PyObject *final = PyUnicodeWriter_Finish(writer); |
1668 | 1.63k | if (final == NULL) { |
1669 | 0 | return NULL; |
1670 | 0 | } |
1671 | 1.63k | if (_PyArena_AddPyObject(p->arena, final) < 0) { |
1672 | 0 | Py_DECREF(final); |
1673 | 0 | return NULL; |
1674 | 0 | } |
1675 | 1.63k | return _PyAST_Constant(final, kind, lineno, col_offset, |
1676 | 1.63k | end_lineno, end_col_offset, arena); |
1677 | 1.63k | } |
1678 | | |
1679 | | static asdl_expr_seq * |
1680 | | _build_concatenated_str(Parser *p, asdl_expr_seq *strings, |
1681 | | int lineno, int col_offset, int end_lineno, |
1682 | | int end_col_offset, PyArena *arena) |
1683 | 13.0k | { |
1684 | 13.0k | Py_ssize_t len = asdl_seq_LEN(strings); |
1685 | 13.0k | assert(len > 0); |
1686 | | |
1687 | 13.0k | Py_ssize_t n_flattened_elements = 0; |
1688 | 41.1k | for (Py_ssize_t i = 0; i < len; i++) { |
1689 | 28.0k | expr_ty elem = asdl_seq_GET(strings, i); |
1690 | 28.0k | switch(elem->kind) { |
1691 | 12.3k | case JoinedStr_kind: |
1692 | 12.3k | n_flattened_elements += asdl_seq_LEN(elem->v.JoinedStr.values); |
1693 | 12.3k | break; |
1694 | 3.68k | case TemplateStr_kind: |
1695 | 3.68k | n_flattened_elements += asdl_seq_LEN(elem->v.TemplateStr.values); |
1696 | 3.68k | break; |
1697 | 12.0k | default: |
1698 | 12.0k | n_flattened_elements++; |
1699 | 12.0k | break; |
1700 | 28.0k | } |
1701 | 28.0k | } |
1702 | | |
1703 | | |
1704 | 13.0k | asdl_expr_seq* flattened = _Py_asdl_expr_seq_new(n_flattened_elements, p->arena); |
1705 | 13.0k | if (flattened == NULL) { |
1706 | 0 | return NULL; |
1707 | 0 | } |
1708 | | |
1709 | | /* build flattened list */ |
1710 | 13.0k | Py_ssize_t current_pos = 0; |
1711 | 41.1k | for (Py_ssize_t i = 0; i < len; i++) { |
1712 | 28.0k | expr_ty elem = asdl_seq_GET(strings, i); |
1713 | 28.0k | switch(elem->kind) { |
1714 | 12.3k | case JoinedStr_kind: |
1715 | 50.2k | for (Py_ssize_t j = 0; j < asdl_seq_LEN(elem->v.JoinedStr.values); j++) { |
1716 | 37.8k | expr_ty subvalue = asdl_seq_GET(elem->v.JoinedStr.values, j); |
1717 | 37.8k | if (subvalue == NULL) { |
1718 | 0 | return NULL; |
1719 | 0 | } |
1720 | 37.8k | asdl_seq_SET(flattened, current_pos++, subvalue); |
1721 | 37.8k | } |
1722 | 12.3k | break; |
1723 | 12.3k | case TemplateStr_kind: |
1724 | 11.1k | for (Py_ssize_t j = 0; j < asdl_seq_LEN(elem->v.TemplateStr.values); j++) { |
1725 | 7.51k | expr_ty subvalue = asdl_seq_GET(elem->v.TemplateStr.values, j); |
1726 | 7.51k | if (subvalue == NULL) { |
1727 | 0 | return NULL; |
1728 | 0 | } |
1729 | 7.51k | asdl_seq_SET(flattened, current_pos++, subvalue); |
1730 | 7.51k | } |
1731 | 3.68k | break; |
1732 | 12.0k | default: |
1733 | 12.0k | asdl_seq_SET(flattened, current_pos++, elem); |
1734 | 12.0k | break; |
1735 | 28.0k | } |
1736 | 28.0k | } |
1737 | | |
1738 | | /* calculate folded element count */ |
1739 | 13.0k | Py_ssize_t n_elements = 0; |
1740 | 13.0k | int prev_is_constant = 0; |
1741 | 70.4k | for (Py_ssize_t i = 0; i < n_flattened_elements; i++) { |
1742 | 57.3k | expr_ty elem = asdl_seq_GET(flattened, i); |
1743 | | |
1744 | | /* The concatenation of a FormattedValue and an empty Constant should |
1745 | | lead to the FormattedValue itself. Thus, we will not take any empty |
1746 | | constants into account, just as in `_PyPegen_joined_str` */ |
1747 | 57.3k | if (elem->kind == Constant_kind && |
1748 | 57.3k | PyUnicode_CheckExact(elem->v.Constant.value) && |
1749 | 57.3k | PyUnicode_GET_LENGTH(elem->v.Constant.value) == 0) |
1750 | 1.53k | continue; |
1751 | | |
1752 | 55.8k | if (!prev_is_constant || elem->kind != Constant_kind) { |
1753 | 46.0k | n_elements++; |
1754 | 46.0k | } |
1755 | 55.8k | prev_is_constant = elem->kind == Constant_kind; |
1756 | 55.8k | } |
1757 | | |
1758 | 13.0k | asdl_expr_seq* values = _Py_asdl_expr_seq_new(n_elements, p->arena); |
1759 | 13.0k | if (values == NULL) { |
1760 | 0 | return NULL; |
1761 | 0 | } |
1762 | | |
1763 | | /* build folded list */ |
1764 | 13.0k | current_pos = 0; |
1765 | 59.3k | for (Py_ssize_t i = 0; i < n_flattened_elements; i++) { |
1766 | 46.2k | expr_ty elem = asdl_seq_GET(flattened, i); |
1767 | | |
1768 | | /* if the current elem and the following are constants, |
1769 | | fold them and all consequent constants */ |
1770 | 46.2k | if (elem->kind == Constant_kind) { |
1771 | 24.5k | if (i + 1 < n_flattened_elements && |
1772 | 24.5k | asdl_seq_GET(flattened, i + 1)->kind == Constant_kind) { |
1773 | 5.00k | expr_ty first_elem = elem; |
1774 | | |
1775 | | /* When a string is getting concatenated, the kind of the string |
1776 | | is determined by the first string in the concatenation |
1777 | | sequence. |
1778 | | |
1779 | | u"abc" "def" -> u"abcdef" |
1780 | | "abc" u"abc" -> "abcabc" */ |
1781 | 5.00k | PyObject *kind = elem->v.Constant.kind; |
1782 | | |
1783 | 5.00k | PyUnicodeWriter *writer = PyUnicodeWriter_Create(0); |
1784 | 5.00k | if (writer == NULL) { |
1785 | 0 | return NULL; |
1786 | 0 | } |
1787 | 5.00k | expr_ty last_elem = elem; |
1788 | 5.00k | Py_ssize_t j; |
1789 | 21.1k | for (j = i; j < n_flattened_elements; j++) { |
1790 | 18.8k | expr_ty current_elem = asdl_seq_GET(flattened, j); |
1791 | 18.8k | if (current_elem->kind == Constant_kind) { |
1792 | 16.1k | if (PyUnicodeWriter_WriteStr(writer, |
1793 | 16.1k | current_elem->v.Constant.value)) { |
1794 | 0 | PyUnicodeWriter_Discard(writer); |
1795 | 0 | return NULL; |
1796 | 0 | } |
1797 | 16.1k | last_elem = current_elem; |
1798 | 16.1k | } else { |
1799 | 2.76k | break; |
1800 | 2.76k | } |
1801 | 18.8k | } |
1802 | 5.00k | i = j - 1; |
1803 | | |
1804 | 5.00k | PyObject *concat_str = PyUnicodeWriter_Finish(writer); |
1805 | 5.00k | if (concat_str == NULL) { |
1806 | 0 | return NULL; |
1807 | 0 | } |
1808 | 5.00k | if (_PyArena_AddPyObject(p->arena, concat_str) < 0) { |
1809 | 0 | Py_DECREF(concat_str); |
1810 | 0 | return NULL; |
1811 | 0 | } |
1812 | 5.00k | elem = _PyAST_Constant(concat_str, kind, first_elem->lineno, |
1813 | 5.00k | first_elem->col_offset, |
1814 | 5.00k | last_elem->end_lineno, |
1815 | 5.00k | last_elem->end_col_offset, p->arena); |
1816 | 5.00k | if (elem == NULL) { |
1817 | 0 | return NULL; |
1818 | 0 | } |
1819 | 5.00k | } |
1820 | | |
1821 | | /* Drop all empty contanst strings */ |
1822 | 24.5k | if (PyUnicode_CheckExact(elem->v.Constant.value) && |
1823 | 24.5k | PyUnicode_GET_LENGTH(elem->v.Constant.value) == 0) { |
1824 | 264 | continue; |
1825 | 264 | } |
1826 | 24.5k | } |
1827 | | |
1828 | 46.0k | asdl_seq_SET(values, current_pos++, elem); |
1829 | 46.0k | } |
1830 | | |
1831 | 13.0k | assert(current_pos == n_elements); |
1832 | 13.0k | return values; |
1833 | 13.0k | } |
1834 | | |
1835 | | static expr_ty |
1836 | | _build_concatenated_joined_str(Parser *p, asdl_expr_seq *strings, |
1837 | | int lineno, int col_offset, int end_lineno, |
1838 | | int end_col_offset, PyArena *arena) |
1839 | 10.0k | { |
1840 | 10.0k | asdl_expr_seq *values = _build_concatenated_str(p, strings, lineno, |
1841 | 10.0k | col_offset, end_lineno, end_col_offset, arena); |
1842 | 10.0k | return _PyAST_JoinedStr(values, lineno, col_offset, end_lineno, end_col_offset, p->arena); |
1843 | 10.0k | } |
1844 | | |
1845 | | expr_ty |
1846 | | _PyPegen_concatenate_tstrings(Parser *p, asdl_expr_seq *strings, |
1847 | | int lineno, int col_offset, int end_lineno, |
1848 | | int end_col_offset, PyArena *arena) |
1849 | 3.00k | { |
1850 | 3.00k | asdl_expr_seq *values = _build_concatenated_str(p, strings, lineno, |
1851 | 3.00k | col_offset, end_lineno, end_col_offset, arena); |
1852 | 3.00k | return _PyAST_TemplateStr(values, lineno, col_offset, end_lineno, |
1853 | 3.00k | end_col_offset, arena); |
1854 | 3.00k | } |
1855 | | |
1856 | | expr_ty |
1857 | | _PyPegen_concatenate_strings(Parser *p, asdl_expr_seq *strings, |
1858 | | int lineno, int col_offset, int end_lineno, |
1859 | | int end_col_offset, PyArena *arena) |
1860 | 49.4k | { |
1861 | 49.4k | Py_ssize_t len = asdl_seq_LEN(strings); |
1862 | 49.4k | assert(len > 0); |
1863 | | |
1864 | 49.4k | int f_string_found = 0; |
1865 | 49.4k | int unicode_string_found = 0; |
1866 | 49.4k | int bytes_found = 0; |
1867 | | |
1868 | 49.4k | Py_ssize_t i = 0; |
1869 | 135k | for (i = 0; i < len; i++) { |
1870 | 85.7k | expr_ty elem = asdl_seq_GET(strings, i); |
1871 | 85.7k | switch(elem->kind) { |
1872 | 69.5k | case Constant_kind: |
1873 | 69.5k | if (PyBytes_CheckExact(elem->v.Constant.value)) { |
1874 | 6.01k | bytes_found = 1; |
1875 | 63.5k | } else { |
1876 | 63.5k | unicode_string_found = 1; |
1877 | 63.5k | } |
1878 | 69.5k | break; |
1879 | 12.3k | case JoinedStr_kind: |
1880 | 12.3k | f_string_found = 1; |
1881 | 12.3k | break; |
1882 | 0 | case TemplateStr_kind: |
1883 | | // python.gram handles this; we should never get here |
1884 | 0 | assert(0); |
1885 | 0 | break; |
1886 | 3.76k | default: |
1887 | 3.76k | f_string_found = 1; |
1888 | 3.76k | break; |
1889 | 85.7k | } |
1890 | 85.7k | } |
1891 | | |
1892 | | // Cannot mix unicode and bytes |
1893 | 49.4k | if ((unicode_string_found || f_string_found) && bytes_found) { |
1894 | 6 | RAISE_SYNTAX_ERROR("cannot mix bytes and nonbytes literals"); |
1895 | 6 | return NULL; |
1896 | 6 | } |
1897 | | |
1898 | | // If it's only bytes or only unicode string, do a simple concat |
1899 | 49.4k | if (!f_string_found) { |
1900 | 39.3k | if (len == 1) { |
1901 | 36.8k | return asdl_seq_GET(strings, 0); |
1902 | 36.8k | } |
1903 | 2.48k | else if (bytes_found) { |
1904 | 844 | return _build_concatenated_bytes(p, strings, lineno, col_offset, |
1905 | 844 | end_lineno, end_col_offset, arena); |
1906 | 844 | } |
1907 | 1.63k | else { |
1908 | 1.63k | return _build_concatenated_unicode(p, strings, lineno, col_offset, |
1909 | 1.63k | end_lineno, end_col_offset, arena); |
1910 | 1.63k | } |
1911 | 39.3k | } |
1912 | | |
1913 | 10.0k | return _build_concatenated_joined_str(p, strings, lineno, |
1914 | 10.0k | col_offset, end_lineno, end_col_offset, arena); |
1915 | 49.4k | } |
1916 | | |
1917 | | stmt_ty |
1918 | | _PyPegen_checked_future_import(Parser *p, identifier module, asdl_alias_seq * names, int level, |
1919 | | int lineno, int col_offset, int end_lineno, int end_col_offset, |
1920 | 1.91k | PyArena *arena) { |
1921 | 1.91k | if (level == 0 && PyUnicode_CompareWithASCIIString(module, "__future__") == 0) { |
1922 | 1.30k | for (Py_ssize_t i = 0; i < asdl_seq_LEN(names); i++) { |
1923 | 826 | alias_ty alias = asdl_seq_GET(names, i); |
1924 | 826 | if (PyUnicode_CompareWithASCIIString(alias->name, "barry_as_FLUFL") == 0) { |
1925 | 73 | p->flags |= PyPARSE_BARRY_AS_BDFL; |
1926 | 73 | } |
1927 | 826 | } |
1928 | 480 | } |
1929 | 1.91k | return _PyAST_ImportFrom(module, names, level, lineno, col_offset, end_lineno, end_col_offset, arena); |
1930 | 1.91k | } |
1931 | | |
1932 | | asdl_stmt_seq* |
1933 | 37.6k | _PyPegen_register_stmts(Parser *p, asdl_stmt_seq* stmts) { |
1934 | 37.6k | if (!p->call_invalid_rules) { |
1935 | 29.3k | return stmts; |
1936 | 29.3k | } |
1937 | 8.30k | Py_ssize_t len = asdl_seq_LEN(stmts); |
1938 | 8.30k | if (len == 0) { |
1939 | 0 | return stmts; |
1940 | 0 | } |
1941 | 8.30k | stmt_ty last_stmt = asdl_seq_GET(stmts, len - 1); |
1942 | 8.30k | if (p->last_stmt_location.lineno > last_stmt->lineno) { |
1943 | 177 | return stmts; |
1944 | 177 | } |
1945 | 8.12k | p->last_stmt_location.lineno = last_stmt->lineno; |
1946 | 8.12k | p->last_stmt_location.col_offset = last_stmt->col_offset; |
1947 | 8.12k | p->last_stmt_location.end_lineno = last_stmt->end_lineno; |
1948 | 8.12k | p->last_stmt_location.end_col_offset = last_stmt->end_col_offset; |
1949 | 8.12k | return stmts; |
1950 | 8.30k | } |