• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 #include <Python.h>
2 #include "pycore_ast.h"           // _PyAST_Validate(),
3 #include <errcode.h>
4 #include "tokenizer.h"
5 
6 #include "pegen.h"
7 #include "string_parser.h"
8 
9 PyObject *
_PyPegen_new_type_comment(Parser * p,const char * s)10 _PyPegen_new_type_comment(Parser *p, const char *s)
11 {
12     PyObject *res = PyUnicode_DecodeUTF8(s, strlen(s), NULL);
13     if (res == NULL) {
14         return NULL;
15     }
16     if (_PyArena_AddPyObject(p->arena, res) < 0) {
17         Py_DECREF(res);
18         return NULL;
19     }
20     return res;
21 }
22 
23 arg_ty
_PyPegen_add_type_comment_to_arg(Parser * p,arg_ty a,Token * tc)24 _PyPegen_add_type_comment_to_arg(Parser *p, arg_ty a, Token *tc)
25 {
26     if (tc == NULL) {
27         return a;
28     }
29     const char *bytes = PyBytes_AsString(tc->bytes);
30     if (bytes == NULL) {
31         return NULL;
32     }
33     PyObject *tco = _PyPegen_new_type_comment(p, bytes);
34     if (tco == NULL) {
35         return NULL;
36     }
37     return _PyAST_arg(a->arg, a->annotation, tco,
38                       a->lineno, a->col_offset, a->end_lineno, a->end_col_offset,
39                       p->arena);
40 }
41 
42 static int
init_normalization(Parser * p)43 init_normalization(Parser *p)
44 {
45     if (p->normalize) {
46         return 1;
47     }
48     PyObject *m = PyImport_ImportModuleNoBlock("unicodedata");
49     if (!m)
50     {
51         return 0;
52     }
53     p->normalize = PyObject_GetAttrString(m, "normalize");
54     Py_DECREF(m);
55     if (!p->normalize)
56     {
57         return 0;
58     }
59     return 1;
60 }
61 
62 /* Checks if the NOTEQUAL token is valid given the current parser flags
63 0 indicates success and nonzero indicates failure (an exception may be set) */
64 int
_PyPegen_check_barry_as_flufl(Parser * p,Token * t)65 _PyPegen_check_barry_as_flufl(Parser *p, Token* t) {
66     assert(t->bytes != NULL);
67     assert(t->type == NOTEQUAL);
68 
69     const char* tok_str = PyBytes_AS_STRING(t->bytes);
70     if (p->flags & PyPARSE_BARRY_AS_BDFL && strcmp(tok_str, "<>") != 0) {
71         RAISE_SYNTAX_ERROR("with Barry as BDFL, use '<>' instead of '!='");
72         return -1;
73     }
74     if (!(p->flags & PyPARSE_BARRY_AS_BDFL)) {
75         return strcmp(tok_str, "!=");
76     }
77     return 0;
78 }
79 
80 int
_PyPegen_check_legacy_stmt(Parser * p,expr_ty name)81 _PyPegen_check_legacy_stmt(Parser *p, expr_ty name) {
82     if (name->kind != Name_kind) {
83         return 0;
84     }
85     const char* candidates[2] = {"print", "exec"};
86     for (int i=0; i<2; i++) {
87         if (PyUnicode_CompareWithASCIIString(name->v.Name.id, candidates[i]) == 0) {
88             return 1;
89         }
90     }
91     return 0;
92 }
93 
94 PyObject *
_PyPegen_new_identifier(Parser * p,const char * n)95 _PyPegen_new_identifier(Parser *p, const char *n)
96 {
97     PyObject *id = PyUnicode_DecodeUTF8(n, strlen(n), NULL);
98     if (!id) {
99         goto error;
100     }
101     /* PyUnicode_DecodeUTF8 should always return a ready string. */
102     assert(PyUnicode_IS_READY(id));
103     /* Check whether there are non-ASCII characters in the
104        identifier; if so, normalize to NFKC. */
105     if (!PyUnicode_IS_ASCII(id))
106     {
107         PyObject *id2;
108         if (!init_normalization(p))
109         {
110             Py_DECREF(id);
111             goto error;
112         }
113         PyObject *form = PyUnicode_InternFromString("NFKC");
114         if (form == NULL)
115         {
116             Py_DECREF(id);
117             goto error;
118         }
119         PyObject *args[2] = {form, id};
120         id2 = _PyObject_FastCall(p->normalize, args, 2);
121         Py_DECREF(id);
122         Py_DECREF(form);
123         if (!id2) {
124             goto error;
125         }
126         if (!PyUnicode_Check(id2))
127         {
128             PyErr_Format(PyExc_TypeError,
129                          "unicodedata.normalize() must return a string, not "
130                          "%.200s",
131                          _PyType_Name(Py_TYPE(id2)));
132             Py_DECREF(id2);
133             goto error;
134         }
135         id = id2;
136     }
137     PyUnicode_InternInPlace(&id);
138     if (_PyArena_AddPyObject(p->arena, id) < 0)
139     {
140         Py_DECREF(id);
141         goto error;
142     }
143     return id;
144 
145 error:
146     p->error_indicator = 1;
147     return NULL;
148 }
149 
150 static PyObject *
_create_dummy_identifier(Parser * p)151 _create_dummy_identifier(Parser *p)
152 {
153     return _PyPegen_new_identifier(p, "");
154 }
155 
156 static inline Py_ssize_t
byte_offset_to_character_offset(PyObject * line,Py_ssize_t col_offset)157 byte_offset_to_character_offset(PyObject *line, Py_ssize_t col_offset)
158 {
159     const char *str = PyUnicode_AsUTF8(line);
160     if (!str) {
161         return 0;
162     }
163     Py_ssize_t len = strlen(str);
164     if (col_offset > len + 1) {
165         col_offset = len + 1;
166     }
167     assert(col_offset >= 0);
168     PyObject *text = PyUnicode_DecodeUTF8(str, col_offset, "replace");
169     if (!text) {
170         return 0;
171     }
172     Py_ssize_t size = PyUnicode_GET_LENGTH(text);
173     Py_DECREF(text);
174     return size;
175 }
176 
177 const char *
_PyPegen_get_expr_name(expr_ty e)178 _PyPegen_get_expr_name(expr_ty e)
179 {
180     assert(e != NULL);
181     switch (e->kind) {
182         case Attribute_kind:
183             return "attribute";
184         case Subscript_kind:
185             return "subscript";
186         case Starred_kind:
187             return "starred";
188         case Name_kind:
189             return "name";
190         case List_kind:
191             return "list";
192         case Tuple_kind:
193             return "tuple";
194         case Lambda_kind:
195             return "lambda";
196         case Call_kind:
197             return "function call";
198         case BoolOp_kind:
199         case BinOp_kind:
200         case UnaryOp_kind:
201             return "expression";
202         case GeneratorExp_kind:
203             return "generator expression";
204         case Yield_kind:
205         case YieldFrom_kind:
206             return "yield expression";
207         case Await_kind:
208             return "await expression";
209         case ListComp_kind:
210             return "list comprehension";
211         case SetComp_kind:
212             return "set comprehension";
213         case DictComp_kind:
214             return "dict comprehension";
215         case Dict_kind:
216             return "dict literal";
217         case Set_kind:
218             return "set display";
219         case JoinedStr_kind:
220         case FormattedValue_kind:
221             return "f-string expression";
222         case Constant_kind: {
223             PyObject *value = e->v.Constant.value;
224             if (value == Py_None) {
225                 return "None";
226             }
227             if (value == Py_False) {
228                 return "False";
229             }
230             if (value == Py_True) {
231                 return "True";
232             }
233             if (value == Py_Ellipsis) {
234                 return "ellipsis";
235             }
236             return "literal";
237         }
238         case Compare_kind:
239             return "comparison";
240         case IfExp_kind:
241             return "conditional expression";
242         case NamedExpr_kind:
243             return "named expression";
244         default:
245             PyErr_Format(PyExc_SystemError,
246                          "unexpected expression in assignment %d (line %d)",
247                          e->kind, e->lineno);
248             return NULL;
249     }
250 }
251 
252 static int
raise_decode_error(Parser * p)253 raise_decode_error(Parser *p)
254 {
255     assert(PyErr_Occurred());
256     const char *errtype = NULL;
257     if (PyErr_ExceptionMatches(PyExc_UnicodeError)) {
258         errtype = "unicode error";
259     }
260     else if (PyErr_ExceptionMatches(PyExc_ValueError)) {
261         errtype = "value error";
262     }
263     if (errtype) {
264         PyObject *type;
265         PyObject *value;
266         PyObject *tback;
267         PyObject *errstr;
268         PyErr_Fetch(&type, &value, &tback);
269         errstr = PyObject_Str(value);
270         if (errstr) {
271             RAISE_SYNTAX_ERROR("(%s) %U", errtype, errstr);
272             Py_DECREF(errstr);
273         }
274         else {
275             PyErr_Clear();
276             RAISE_SYNTAX_ERROR("(%s) unknown error", errtype);
277         }
278         Py_XDECREF(type);
279         Py_XDECREF(value);
280         Py_XDECREF(tback);
281     }
282 
283     return -1;
284 }
285 
286 static inline void
raise_unclosed_parentheses_error(Parser * p)287 raise_unclosed_parentheses_error(Parser *p) {
288        int error_lineno = p->tok->parenlinenostack[p->tok->level-1];
289        int error_col = p->tok->parencolstack[p->tok->level-1];
290        RAISE_ERROR_KNOWN_LOCATION(p, PyExc_SyntaxError,
291                                   error_lineno, error_col, error_lineno, -1,
292                                   "'%c' was never closed",
293                                   p->tok->parenstack[p->tok->level-1]);
294 }
295 
296 static void
raise_tokenizer_init_error(PyObject * filename)297 raise_tokenizer_init_error(PyObject *filename)
298 {
299     if (!(PyErr_ExceptionMatches(PyExc_LookupError)
300           || PyErr_ExceptionMatches(PyExc_SyntaxError)
301           || PyErr_ExceptionMatches(PyExc_ValueError)
302           || PyErr_ExceptionMatches(PyExc_UnicodeDecodeError))) {
303         return;
304     }
305     PyObject *errstr = NULL;
306     PyObject *tuple = NULL;
307     PyObject *type;
308     PyObject *value;
309     PyObject *tback;
310     PyErr_Fetch(&type, &value, &tback);
311     errstr = PyObject_Str(value);
312     if (!errstr) {
313         goto error;
314     }
315 
316     PyObject *tmp = Py_BuildValue("(OiiO)", filename, 0, -1, Py_None);
317     if (!tmp) {
318         goto error;
319     }
320 
321     tuple = PyTuple_Pack(2, errstr, tmp);
322     Py_DECREF(tmp);
323     if (!value) {
324         goto error;
325     }
326     PyErr_SetObject(PyExc_SyntaxError, tuple);
327 
328 error:
329     Py_XDECREF(type);
330     Py_XDECREF(value);
331     Py_XDECREF(tback);
332     Py_XDECREF(errstr);
333     Py_XDECREF(tuple);
334 }
335 
336 static int
tokenizer_error(Parser * p)337 tokenizer_error(Parser *p)
338 {
339     if (PyErr_Occurred()) {
340         return -1;
341     }
342 
343     const char *msg = NULL;
344     PyObject* errtype = PyExc_SyntaxError;
345     Py_ssize_t col_offset = -1;
346     switch (p->tok->done) {
347         case E_TOKEN:
348             msg = "invalid token";
349             break;
350         case E_EOF:
351             if (p->tok->level) {
352                 raise_unclosed_parentheses_error(p);
353             } else {
354                 RAISE_SYNTAX_ERROR("unexpected EOF while parsing");
355             }
356             return -1;
357         case E_DEDENT:
358             RAISE_INDENTATION_ERROR("unindent does not match any outer indentation level");
359             return -1;
360         case E_INTR:
361             if (!PyErr_Occurred()) {
362                 PyErr_SetNone(PyExc_KeyboardInterrupt);
363             }
364             return -1;
365         case E_NOMEM:
366             PyErr_NoMemory();
367             return -1;
368         case E_TABSPACE:
369             errtype = PyExc_TabError;
370             msg = "inconsistent use of tabs and spaces in indentation";
371             break;
372         case E_TOODEEP:
373             errtype = PyExc_IndentationError;
374             msg = "too many levels of indentation";
375             break;
376         case E_LINECONT: {
377             col_offset = p->tok->cur - p->tok->buf - 1;
378             msg = "unexpected character after line continuation character";
379             break;
380         }
381         default:
382             msg = "unknown parsing error";
383     }
384 
385     RAISE_ERROR_KNOWN_LOCATION(p, errtype, p->tok->lineno,
386                                col_offset >= 0 ? col_offset : 0,
387                                p->tok->lineno, -1, msg);
388     return -1;
389 }
390 
391 void *
_PyPegen_raise_error(Parser * p,PyObject * errtype,const char * errmsg,...)392 _PyPegen_raise_error(Parser *p, PyObject *errtype, const char *errmsg, ...)
393 {
394     if (p->fill == 0) {
395         va_list va;
396         va_start(va, errmsg);
397         _PyPegen_raise_error_known_location(p, errtype, 0, 0, 0, -1, errmsg, va);
398         va_end(va);
399         return NULL;
400     }
401 
402     Token *t = p->known_err_token != NULL ? p->known_err_token : p->tokens[p->fill - 1];
403     Py_ssize_t col_offset;
404     Py_ssize_t end_col_offset = -1;
405     if (t->col_offset == -1) {
406         if (p->tok->cur == p->tok->buf) {
407             col_offset = 0;
408         } else {
409             const char* start = p->tok->buf  ? p->tok->line_start : p->tok->buf;
410             col_offset = Py_SAFE_DOWNCAST(p->tok->cur - start, intptr_t, int);
411         }
412     } else {
413         col_offset = t->col_offset + 1;
414     }
415 
416     if (t->end_col_offset != -1) {
417         end_col_offset = t->end_col_offset + 1;
418     }
419 
420     va_list va;
421     va_start(va, errmsg);
422     _PyPegen_raise_error_known_location(p, errtype, t->lineno, col_offset, t->end_lineno, end_col_offset, errmsg, va);
423     va_end(va);
424 
425     return NULL;
426 }
427 
428 static PyObject *
get_error_line(Parser * p,Py_ssize_t lineno)429 get_error_line(Parser *p, Py_ssize_t lineno)
430 {
431     /* If the file descriptor is interactive, the source lines of the current
432      * (multi-line) statement are stored in p->tok->interactive_src_start.
433      * If not, we're parsing from a string, which means that the whole source
434      * is stored in p->tok->str. */
435     assert((p->tok->fp == NULL && p->tok->str != NULL) || p->tok->fp == stdin);
436 
437     char *cur_line = p->tok->fp_interactive ? p->tok->interactive_src_start : p->tok->str;
438     assert(cur_line != NULL);
439 
440     for (int i = 0; i < lineno - 1; i++) {
441         cur_line = strchr(cur_line, '\n') + 1;
442     }
443 
444     char *next_newline;
445     if ((next_newline = strchr(cur_line, '\n')) == NULL) { // This is the last line
446         next_newline = cur_line + strlen(cur_line);
447     }
448     return PyUnicode_DecodeUTF8(cur_line, next_newline - cur_line, "replace");
449 }
450 
451 void *
_PyPegen_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,va_list va)452 _PyPegen_raise_error_known_location(Parser *p, PyObject *errtype,
453                                     Py_ssize_t lineno, Py_ssize_t col_offset,
454                                     Py_ssize_t end_lineno, Py_ssize_t end_col_offset,
455                                     const char *errmsg, va_list va)
456 {
457     PyObject *value = NULL;
458     PyObject *errstr = NULL;
459     PyObject *error_line = NULL;
460     PyObject *tmp = NULL;
461     p->error_indicator = 1;
462 
463     if (end_lineno == CURRENT_POS) {
464         end_lineno = p->tok->lineno;
465     }
466     if (end_col_offset == CURRENT_POS) {
467         end_col_offset = p->tok->cur - p->tok->line_start;
468     }
469 
470     if (p->start_rule == Py_fstring_input) {
471         const char *fstring_msg = "f-string: ";
472         Py_ssize_t len = strlen(fstring_msg) + strlen(errmsg);
473 
474         char *new_errmsg = PyMem_Malloc(len + 1); // Lengths of both strings plus NULL character
475         if (!new_errmsg) {
476             return (void *) PyErr_NoMemory();
477         }
478 
479         // Copy both strings into new buffer
480         memcpy(new_errmsg, fstring_msg, strlen(fstring_msg));
481         memcpy(new_errmsg + strlen(fstring_msg), errmsg, strlen(errmsg));
482         new_errmsg[len] = 0;
483         errmsg = new_errmsg;
484     }
485     errstr = PyUnicode_FromFormatV(errmsg, va);
486     if (!errstr) {
487         goto error;
488     }
489 
490     if (p->tok->fp_interactive) {
491         error_line = get_error_line(p, lineno);
492     }
493     else if (p->start_rule == Py_file_input) {
494         error_line = _PyErr_ProgramDecodedTextObject(p->tok->filename,
495                                                      (int) lineno, p->tok->encoding);
496     }
497 
498     if (!error_line) {
499         /* PyErr_ProgramTextObject was not called or returned NULL. If it was not called,
500            then we need to find the error line from some other source, because
501            p->start_rule != Py_file_input. If it returned NULL, then it either unexpectedly
502            failed or we're parsing from a string or the REPL. There's a third edge case where
503            we're actually parsing from a file, which has an E_EOF SyntaxError and in that case
504            `PyErr_ProgramTextObject` fails because lineno points to last_file_line + 1, which
505            does not physically exist */
506         assert(p->tok->fp == NULL || p->tok->fp == stdin || p->tok->done == E_EOF);
507 
508         if (p->tok->lineno <= lineno && p->tok->inp > p->tok->buf) {
509             Py_ssize_t size = p->tok->inp - p->tok->buf;
510             error_line = PyUnicode_DecodeUTF8(p->tok->buf, size, "replace");
511         }
512         else if (p->tok->fp == NULL || p->tok->fp == stdin) {
513             error_line = get_error_line(p, lineno);
514         }
515         else {
516             error_line = PyUnicode_FromStringAndSize("", 0);
517         }
518         if (!error_line) {
519             goto error;
520         }
521     }
522 
523     if (p->start_rule == Py_fstring_input) {
524         col_offset -= p->starting_col_offset;
525         end_col_offset -= p->starting_col_offset;
526     }
527 
528     Py_ssize_t col_number = col_offset;
529     Py_ssize_t end_col_number = end_col_offset;
530 
531     if (p->tok->encoding != NULL) {
532         col_number = byte_offset_to_character_offset(error_line, col_offset);
533         end_col_number = end_col_number > 0 ?
534                          byte_offset_to_character_offset(error_line, end_col_offset) :
535                          end_col_number;
536     }
537     tmp = Py_BuildValue("(OiiNii)", p->tok->filename, lineno, col_number, error_line, end_lineno, end_col_number);
538     if (!tmp) {
539         goto error;
540     }
541     value = PyTuple_Pack(2, errstr, tmp);
542     Py_DECREF(tmp);
543     if (!value) {
544         goto error;
545     }
546     PyErr_SetObject(errtype, value);
547 
548     Py_DECREF(errstr);
549     Py_DECREF(value);
550     if (p->start_rule == Py_fstring_input) {
551         PyMem_Free((void *)errmsg);
552     }
553     return NULL;
554 
555 error:
556     Py_XDECREF(errstr);
557     Py_XDECREF(error_line);
558     if (p->start_rule == Py_fstring_input) {
559         PyMem_Free((void *)errmsg);
560     }
561     return NULL;
562 }
563 
564 #if 0
565 static const char *
566 token_name(int type)
567 {
568     if (0 <= type && type <= N_TOKENS) {
569         return _PyParser_TokenNames[type];
570     }
571     return "<Huh?>";
572 }
573 #endif
574 
575 // Here, mark is the start of the node, while p->mark is the end.
576 // If node==NULL, they should be the same.
577 int
_PyPegen_insert_memo(Parser * p,int mark,int type,void * node)578 _PyPegen_insert_memo(Parser *p, int mark, int type, void *node)
579 {
580     // Insert in front
581     Memo *m = _PyArena_Malloc(p->arena, sizeof(Memo));
582     if (m == NULL) {
583         return -1;
584     }
585     m->type = type;
586     m->node = node;
587     m->mark = p->mark;
588     m->next = p->tokens[mark]->memo;
589     p->tokens[mark]->memo = m;
590     return 0;
591 }
592 
593 // Like _PyPegen_insert_memo(), but updates an existing node if found.
594 int
_PyPegen_update_memo(Parser * p,int mark,int type,void * node)595 _PyPegen_update_memo(Parser *p, int mark, int type, void *node)
596 {
597     for (Memo *m = p->tokens[mark]->memo; m != NULL; m = m->next) {
598         if (m->type == type) {
599             // Update existing node.
600             m->node = node;
601             m->mark = p->mark;
602             return 0;
603         }
604     }
605     // Insert new node.
606     return _PyPegen_insert_memo(p, mark, type, node);
607 }
608 
609 // Return dummy NAME.
610 void *
_PyPegen_dummy_name(Parser * p,...)611 _PyPegen_dummy_name(Parser *p, ...)
612 {
613     static void *cache = NULL;
614 
615     if (cache != NULL) {
616         return cache;
617     }
618 
619     PyObject *id = _create_dummy_identifier(p);
620     if (!id) {
621         return NULL;
622     }
623     cache = _PyAST_Name(id, Load, 1, 0, 1, 0, p->arena);
624     return cache;
625 }
626 
627 static int
_get_keyword_or_name_type(Parser * p,const char * name,int name_len)628 _get_keyword_or_name_type(Parser *p, const char *name, int name_len)
629 {
630     assert(name_len > 0);
631     if (name_len >= p->n_keyword_lists ||
632         p->keywords[name_len] == NULL ||
633         p->keywords[name_len]->type == -1) {
634         return NAME;
635     }
636     for (KeywordToken *k = p->keywords[name_len]; k != NULL && k->type != -1; k++) {
637         if (strncmp(k->str, name, name_len) == 0) {
638             return k->type;
639         }
640     }
641     return NAME;
642 }
643 
644 static int
growable_comment_array_init(growable_comment_array * arr,size_t initial_size)645 growable_comment_array_init(growable_comment_array *arr, size_t initial_size) {
646     assert(initial_size > 0);
647     arr->items = PyMem_Malloc(initial_size * sizeof(*arr->items));
648     arr->size = initial_size;
649     arr->num_items = 0;
650 
651     return arr->items != NULL;
652 }
653 
654 static int
growable_comment_array_add(growable_comment_array * arr,int lineno,char * comment)655 growable_comment_array_add(growable_comment_array *arr, int lineno, char *comment) {
656     if (arr->num_items >= arr->size) {
657         size_t new_size = arr->size * 2;
658         void *new_items_array = PyMem_Realloc(arr->items, new_size * sizeof(*arr->items));
659         if (!new_items_array) {
660             return 0;
661         }
662         arr->items = new_items_array;
663         arr->size = new_size;
664     }
665 
666     arr->items[arr->num_items].lineno = lineno;
667     arr->items[arr->num_items].comment = comment;  // Take ownership
668     arr->num_items++;
669     return 1;
670 }
671 
672 static void
growable_comment_array_deallocate(growable_comment_array * arr)673 growable_comment_array_deallocate(growable_comment_array *arr) {
674     for (unsigned i = 0; i < arr->num_items; i++) {
675         PyMem_Free(arr->items[i].comment);
676     }
677     PyMem_Free(arr->items);
678 }
679 
680 static int
initialize_token(Parser * p,Token * token,const char * start,const char * end,int token_type)681 initialize_token(Parser *p, Token *token, const char *start, const char *end, int token_type) {
682     assert(token != NULL);
683 
684     token->type = (token_type == NAME) ? _get_keyword_or_name_type(p, start, (int)(end - start)) : token_type;
685     token->bytes = PyBytes_FromStringAndSize(start, end - start);
686     if (token->bytes == NULL) {
687         return -1;
688     }
689 
690     if (_PyArena_AddPyObject(p->arena, token->bytes) < 0) {
691         Py_DECREF(token->bytes);
692         return -1;
693     }
694 
695     token->level = p->tok->level;
696 
697     const char *line_start = token_type == STRING ? p->tok->multi_line_start : p->tok->line_start;
698     int lineno = token_type == STRING ? p->tok->first_lineno : p->tok->lineno;
699     int end_lineno = p->tok->lineno;
700 
701     int col_offset = (start != NULL && start >= line_start) ? (int)(start - line_start) : -1;
702     int end_col_offset = (end != NULL && end >= p->tok->line_start) ? (int)(end - p->tok->line_start) : -1;
703 
704     token->lineno = lineno;
705     token->col_offset = p->tok->lineno == p->starting_lineno ? p->starting_col_offset + col_offset : col_offset;
706     token->end_lineno = end_lineno;
707     token->end_col_offset = p->tok->lineno == p->starting_lineno ? p->starting_col_offset + end_col_offset : end_col_offset;
708 
709     p->fill += 1;
710 
711     if (token_type == ERRORTOKEN && p->tok->done == E_DECODE) {
712         return raise_decode_error(p);
713     }
714 
715     return (token_type == ERRORTOKEN ? tokenizer_error(p) : 0);
716 }
717 
718 static int
_resize_tokens_array(Parser * p)719 _resize_tokens_array(Parser *p) {
720     int newsize = p->size * 2;
721     Token **new_tokens = PyMem_Realloc(p->tokens, newsize * sizeof(Token *));
722     if (new_tokens == NULL) {
723         PyErr_NoMemory();
724         return -1;
725     }
726     p->tokens = new_tokens;
727 
728     for (int i = p->size; i < newsize; i++) {
729         p->tokens[i] = PyMem_Calloc(1, sizeof(Token));
730         if (p->tokens[i] == NULL) {
731             p->size = i; // Needed, in order to cleanup correctly after parser fails
732             PyErr_NoMemory();
733             return -1;
734         }
735     }
736     p->size = newsize;
737     return 0;
738 }
739 
740 int
_PyPegen_fill_token(Parser * p)741 _PyPegen_fill_token(Parser *p)
742 {
743     const char *start;
744     const char *end;
745     int type = PyTokenizer_Get(p->tok, &start, &end);
746 
747     // Record and skip '# type: ignore' comments
748     while (type == TYPE_IGNORE) {
749         Py_ssize_t len = end - start;
750         char *tag = PyMem_Malloc(len + 1);
751         if (tag == NULL) {
752             PyErr_NoMemory();
753             return -1;
754         }
755         strncpy(tag, start, len);
756         tag[len] = '\0';
757         // Ownership of tag passes to the growable array
758         if (!growable_comment_array_add(&p->type_ignore_comments, p->tok->lineno, tag)) {
759             PyErr_NoMemory();
760             return -1;
761         }
762         type = PyTokenizer_Get(p->tok, &start, &end);
763     }
764 
765     // If we have reached the end and we are in single input mode we need to insert a newline and reset the parsing
766     if (p->start_rule == Py_single_input && type == ENDMARKER && p->parsing_started) {
767         type = NEWLINE; /* Add an extra newline */
768         p->parsing_started = 0;
769 
770         if (p->tok->indent && !(p->flags & PyPARSE_DONT_IMPLY_DEDENT)) {
771             p->tok->pendin = -p->tok->indent;
772             p->tok->indent = 0;
773         }
774     }
775     else {
776         p->parsing_started = 1;
777     }
778 
779     // Check if we are at the limit of the token array capacity and resize if needed
780     if ((p->fill == p->size) && (_resize_tokens_array(p) != 0)) {
781         return -1;
782     }
783 
784     Token *t = p->tokens[p->fill];
785     return initialize_token(p, t, start, end, type);
786 }
787 
788 
789 #if defined(Py_DEBUG)
790 // Instrumentation to count the effectiveness of memoization.
791 // The array counts the number of tokens skipped by memoization,
792 // indexed by type.
793 
794 #define NSTATISTICS 2000
795 static long memo_statistics[NSTATISTICS];
796 
797 void
_PyPegen_clear_memo_statistics()798 _PyPegen_clear_memo_statistics()
799 {
800     for (int i = 0; i < NSTATISTICS; i++) {
801         memo_statistics[i] = 0;
802     }
803 }
804 
805 PyObject *
_PyPegen_get_memo_statistics()806 _PyPegen_get_memo_statistics()
807 {
808     PyObject *ret = PyList_New(NSTATISTICS);
809     if (ret == NULL) {
810         return NULL;
811     }
812     for (int i = 0; i < NSTATISTICS; i++) {
813         PyObject *value = PyLong_FromLong(memo_statistics[i]);
814         if (value == NULL) {
815             Py_DECREF(ret);
816             return NULL;
817         }
818         // PyList_SetItem borrows a reference to value.
819         if (PyList_SetItem(ret, i, value) < 0) {
820             Py_DECREF(ret);
821             return NULL;
822         }
823     }
824     return ret;
825 }
826 #endif
827 
828 int  // bool
_PyPegen_is_memoized(Parser * p,int type,void * pres)829 _PyPegen_is_memoized(Parser *p, int type, void *pres)
830 {
831     if (p->mark == p->fill) {
832         if (_PyPegen_fill_token(p) < 0) {
833             p->error_indicator = 1;
834             return -1;
835         }
836     }
837 
838     Token *t = p->tokens[p->mark];
839 
840     for (Memo *m = t->memo; m != NULL; m = m->next) {
841         if (m->type == type) {
842 #if defined(PY_DEBUG)
843             if (0 <= type && type < NSTATISTICS) {
844                 long count = m->mark - p->mark;
845                 // A memoized negative result counts for one.
846                 if (count <= 0) {
847                     count = 1;
848                 }
849                 memo_statistics[type] += count;
850             }
851 #endif
852             p->mark = m->mark;
853             *(void **)(pres) = m->node;
854             return 1;
855         }
856     }
857     return 0;
858 }
859 
860 int
_PyPegen_lookahead_with_name(int positive,expr_ty (func)(Parser *),Parser * p)861 _PyPegen_lookahead_with_name(int positive, expr_ty (func)(Parser *), Parser *p)
862 {
863     int mark = p->mark;
864     void *res = func(p);
865     p->mark = mark;
866     return (res != NULL) == positive;
867 }
868 
869 int
_PyPegen_lookahead_with_string(int positive,expr_ty (func)(Parser *,const char *),Parser * p,const char * arg)870 _PyPegen_lookahead_with_string(int positive, expr_ty (func)(Parser *, const char*), Parser *p, const char* arg)
871 {
872     int mark = p->mark;
873     void *res = func(p, arg);
874     p->mark = mark;
875     return (res != NULL) == positive;
876 }
877 
878 int
_PyPegen_lookahead_with_int(int positive,Token * (func)(Parser *,int),Parser * p,int arg)879 _PyPegen_lookahead_with_int(int positive, Token *(func)(Parser *, int), Parser *p, int arg)
880 {
881     int mark = p->mark;
882     void *res = func(p, arg);
883     p->mark = mark;
884     return (res != NULL) == positive;
885 }
886 
887 int
_PyPegen_lookahead(int positive,void * (func)(Parser *),Parser * p)888 _PyPegen_lookahead(int positive, void *(func)(Parser *), Parser *p)
889 {
890     int mark = p->mark;
891     void *res = (void*)func(p);
892     p->mark = mark;
893     return (res != NULL) == positive;
894 }
895 
896 Token *
_PyPegen_expect_token(Parser * p,int type)897 _PyPegen_expect_token(Parser *p, int type)
898 {
899     if (p->mark == p->fill) {
900         if (_PyPegen_fill_token(p) < 0) {
901             p->error_indicator = 1;
902             return NULL;
903         }
904     }
905     Token *t = p->tokens[p->mark];
906     if (t->type != type) {
907         return NULL;
908     }
909     p->mark += 1;
910     return t;
911 }
912 
913 Token *
_PyPegen_expect_forced_token(Parser * p,int type,const char * expected)914 _PyPegen_expect_forced_token(Parser *p, int type, const char* expected) {
915 
916     if (p->error_indicator == 1) {
917         return NULL;
918     }
919 
920     if (p->mark == p->fill) {
921         if (_PyPegen_fill_token(p) < 0) {
922             p->error_indicator = 1;
923             return NULL;
924         }
925     }
926     Token *t = p->tokens[p->mark];
927     if (t->type != type) {
928         RAISE_SYNTAX_ERROR_KNOWN_LOCATION(t, "expected '%s'", expected);
929         return NULL;
930     }
931     p->mark += 1;
932     return t;
933 }
934 
935 expr_ty
_PyPegen_expect_soft_keyword(Parser * p,const char * keyword)936 _PyPegen_expect_soft_keyword(Parser *p, const char *keyword)
937 {
938     if (p->mark == p->fill) {
939         if (_PyPegen_fill_token(p) < 0) {
940             p->error_indicator = 1;
941             return NULL;
942         }
943     }
944     Token *t = p->tokens[p->mark];
945     if (t->type != NAME) {
946         return NULL;
947     }
948     const char *s = PyBytes_AsString(t->bytes);
949     if (!s) {
950         p->error_indicator = 1;
951         return NULL;
952     }
953     if (strcmp(s, keyword) != 0) {
954         return NULL;
955     }
956     return _PyPegen_name_token(p);
957 }
958 
959 Token *
_PyPegen_get_last_nonnwhitespace_token(Parser * p)960 _PyPegen_get_last_nonnwhitespace_token(Parser *p)
961 {
962     assert(p->mark >= 0);
963     Token *token = NULL;
964     for (int m = p->mark - 1; m >= 0; m--) {
965         token = p->tokens[m];
966         if (token->type != ENDMARKER && (token->type < NEWLINE || token->type > DEDENT)) {
967             break;
968         }
969     }
970     return token;
971 }
972 
973 static expr_ty
_PyPegen_name_from_token(Parser * p,Token * t)974 _PyPegen_name_from_token(Parser *p, Token* t)
975 {
976     if (t == NULL) {
977         return NULL;
978     }
979     const char *s = PyBytes_AsString(t->bytes);
980     if (!s) {
981         p->error_indicator = 1;
982         return NULL;
983     }
984     PyObject *id = _PyPegen_new_identifier(p, s);
985     if (id == NULL) {
986         p->error_indicator = 1;
987         return NULL;
988     }
989     return _PyAST_Name(id, Load, t->lineno, t->col_offset, t->end_lineno,
990                        t->end_col_offset, p->arena);
991 }
992 
993 
994 expr_ty
_PyPegen_name_token(Parser * p)995 _PyPegen_name_token(Parser *p)
996 {
997     Token *t = _PyPegen_expect_token(p, NAME);
998     return _PyPegen_name_from_token(p, t);
999 }
1000 
1001 void *
_PyPegen_string_token(Parser * p)1002 _PyPegen_string_token(Parser *p)
1003 {
1004     return _PyPegen_expect_token(p, STRING);
1005 }
1006 
1007 
_PyPegen_soft_keyword_token(Parser * p)1008 expr_ty _PyPegen_soft_keyword_token(Parser *p) {
1009     Token *t = _PyPegen_expect_token(p, NAME);
1010     if (t == NULL) {
1011         return NULL;
1012     }
1013     char *the_token;
1014     Py_ssize_t size;
1015     PyBytes_AsStringAndSize(t->bytes, &the_token, &size);
1016     for (char **keyword = p->soft_keywords; *keyword != NULL; keyword++) {
1017         if (strncmp(*keyword, the_token, size) == 0) {
1018             return _PyPegen_name_from_token(p, t);
1019         }
1020     }
1021     return NULL;
1022 }
1023 
1024 static PyObject *
parsenumber_raw(const char * s)1025 parsenumber_raw(const char *s)
1026 {
1027     const char *end;
1028     long x;
1029     double dx;
1030     Py_complex compl;
1031     int imflag;
1032 
1033     assert(s != NULL);
1034     errno = 0;
1035     end = s + strlen(s) - 1;
1036     imflag = *end == 'j' || *end == 'J';
1037     if (s[0] == '0') {
1038         x = (long)PyOS_strtoul(s, (char **)&end, 0);
1039         if (x < 0 && errno == 0) {
1040             return PyLong_FromString(s, (char **)0, 0);
1041         }
1042     }
1043     else {
1044         x = PyOS_strtol(s, (char **)&end, 0);
1045     }
1046     if (*end == '\0') {
1047         if (errno != 0) {
1048             return PyLong_FromString(s, (char **)0, 0);
1049         }
1050         return PyLong_FromLong(x);
1051     }
1052     /* XXX Huge floats may silently fail */
1053     if (imflag) {
1054         compl.real = 0.;
1055         compl.imag = PyOS_string_to_double(s, (char **)&end, NULL);
1056         if (compl.imag == -1.0 && PyErr_Occurred()) {
1057             return NULL;
1058         }
1059         return PyComplex_FromCComplex(compl);
1060     }
1061     dx = PyOS_string_to_double(s, NULL, NULL);
1062     if (dx == -1.0 && PyErr_Occurred()) {
1063         return NULL;
1064     }
1065     return PyFloat_FromDouble(dx);
1066 }
1067 
1068 static PyObject *
parsenumber(const char * s)1069 parsenumber(const char *s)
1070 {
1071     char *dup;
1072     char *end;
1073     PyObject *res = NULL;
1074 
1075     assert(s != NULL);
1076 
1077     if (strchr(s, '_') == NULL) {
1078         return parsenumber_raw(s);
1079     }
1080     /* Create a duplicate without underscores. */
1081     dup = PyMem_Malloc(strlen(s) + 1);
1082     if (dup == NULL) {
1083         return PyErr_NoMemory();
1084     }
1085     end = dup;
1086     for (; *s; s++) {
1087         if (*s != '_') {
1088             *end++ = *s;
1089         }
1090     }
1091     *end = '\0';
1092     res = parsenumber_raw(dup);
1093     PyMem_Free(dup);
1094     return res;
1095 }
1096 
1097 expr_ty
_PyPegen_number_token(Parser * p)1098 _PyPegen_number_token(Parser *p)
1099 {
1100     Token *t = _PyPegen_expect_token(p, NUMBER);
1101     if (t == NULL) {
1102         return NULL;
1103     }
1104 
1105     const char *num_raw = PyBytes_AsString(t->bytes);
1106     if (num_raw == NULL) {
1107         p->error_indicator = 1;
1108         return NULL;
1109     }
1110 
1111     if (p->feature_version < 6 && strchr(num_raw, '_') != NULL) {
1112         p->error_indicator = 1;
1113         return RAISE_SYNTAX_ERROR("Underscores in numeric literals are only supported "
1114                                   "in Python 3.6 and greater");
1115     }
1116 
1117     PyObject *c = parsenumber(num_raw);
1118 
1119     if (c == NULL) {
1120         p->error_indicator = 1;
1121         return NULL;
1122     }
1123 
1124     if (_PyArena_AddPyObject(p->arena, c) < 0) {
1125         Py_DECREF(c);
1126         p->error_indicator = 1;
1127         return NULL;
1128     }
1129 
1130     return _PyAST_Constant(c, NULL, t->lineno, t->col_offset, t->end_lineno,
1131                            t->end_col_offset, p->arena);
1132 }
1133 
1134 /* Check that the source for a single input statement really is a single
1135    statement by looking at what is left in the buffer after parsing.
1136    Trailing whitespace and comments are OK. */
1137 static int // bool
bad_single_statement(Parser * p)1138 bad_single_statement(Parser *p)
1139 {
1140     char *cur = p->tok->cur;
1141     char c = *cur;
1142 
1143     for (;;) {
1144         while (c == ' ' || c == '\t' || c == '\n' || c == '\014') {
1145             c = *++cur;
1146         }
1147 
1148         if (!c) {
1149             return 0;
1150         }
1151 
1152         if (c != '#') {
1153             return 1;
1154         }
1155 
1156         /* Suck up comment. */
1157         while (c && c != '\n') {
1158             c = *++cur;
1159         }
1160     }
1161 }
1162 
1163 void
_PyPegen_Parser_Free(Parser * p)1164 _PyPegen_Parser_Free(Parser *p)
1165 {
1166     Py_XDECREF(p->normalize);
1167     for (int i = 0; i < p->size; i++) {
1168         PyMem_Free(p->tokens[i]);
1169     }
1170     PyMem_Free(p->tokens);
1171     growable_comment_array_deallocate(&p->type_ignore_comments);
1172     PyMem_Free(p);
1173 }
1174 
1175 static int
compute_parser_flags(PyCompilerFlags * flags)1176 compute_parser_flags(PyCompilerFlags *flags)
1177 {
1178     int parser_flags = 0;
1179     if (!flags) {
1180         return 0;
1181     }
1182     if (flags->cf_flags & PyCF_DONT_IMPLY_DEDENT) {
1183         parser_flags |= PyPARSE_DONT_IMPLY_DEDENT;
1184     }
1185     if (flags->cf_flags & PyCF_IGNORE_COOKIE) {
1186         parser_flags |= PyPARSE_IGNORE_COOKIE;
1187     }
1188     if (flags->cf_flags & CO_FUTURE_BARRY_AS_BDFL) {
1189         parser_flags |= PyPARSE_BARRY_AS_BDFL;
1190     }
1191     if (flags->cf_flags & PyCF_TYPE_COMMENTS) {
1192         parser_flags |= PyPARSE_TYPE_COMMENTS;
1193     }
1194     if ((flags->cf_flags & PyCF_ONLY_AST) && flags->cf_feature_version < 7) {
1195         parser_flags |= PyPARSE_ASYNC_HACKS;
1196     }
1197     return parser_flags;
1198 }
1199 
1200 Parser *
_PyPegen_Parser_New(struct tok_state * tok,int start_rule,int flags,int feature_version,int * errcode,PyArena * arena)1201 _PyPegen_Parser_New(struct tok_state *tok, int start_rule, int flags,
1202                     int feature_version, int *errcode, PyArena *arena)
1203 {
1204     Parser *p = PyMem_Malloc(sizeof(Parser));
1205     if (p == NULL) {
1206         return (Parser *) PyErr_NoMemory();
1207     }
1208     assert(tok != NULL);
1209     tok->type_comments = (flags & PyPARSE_TYPE_COMMENTS) > 0;
1210     tok->async_hacks = (flags & PyPARSE_ASYNC_HACKS) > 0;
1211     p->tok = tok;
1212     p->keywords = NULL;
1213     p->n_keyword_lists = -1;
1214     p->soft_keywords = NULL;
1215     p->tokens = PyMem_Malloc(sizeof(Token *));
1216     if (!p->tokens) {
1217         PyMem_Free(p);
1218         return (Parser *) PyErr_NoMemory();
1219     }
1220     p->tokens[0] = PyMem_Calloc(1, sizeof(Token));
1221     if (!p->tokens) {
1222         PyMem_Free(p->tokens);
1223         PyMem_Free(p);
1224         return (Parser *) PyErr_NoMemory();
1225     }
1226     if (!growable_comment_array_init(&p->type_ignore_comments, 10)) {
1227         PyMem_Free(p->tokens[0]);
1228         PyMem_Free(p->tokens);
1229         PyMem_Free(p);
1230         return (Parser *) PyErr_NoMemory();
1231     }
1232 
1233     p->mark = 0;
1234     p->fill = 0;
1235     p->size = 1;
1236 
1237     p->errcode = errcode;
1238     p->arena = arena;
1239     p->start_rule = start_rule;
1240     p->parsing_started = 0;
1241     p->normalize = NULL;
1242     p->error_indicator = 0;
1243 
1244     p->starting_lineno = 0;
1245     p->starting_col_offset = 0;
1246     p->flags = flags;
1247     p->feature_version = feature_version;
1248     p->known_err_token = NULL;
1249     p->level = 0;
1250     p->call_invalid_rules = 0;
1251     p->in_raw_rule = 0;
1252     return p;
1253 }
1254 
1255 static void
reset_parser_state(Parser * p)1256 reset_parser_state(Parser *p)
1257 {
1258     for (int i = 0; i < p->fill; i++) {
1259         p->tokens[i]->memo = NULL;
1260     }
1261     p->mark = 0;
1262     p->call_invalid_rules = 1;
1263     // Don't try to get extra tokens in interactive mode when trying to
1264     // raise specialized errors in the second pass.
1265     p->tok->interactive_underflow = IUNDERFLOW_STOP;
1266 }
1267 
1268 static int
_PyPegen_check_tokenizer_errors(Parser * p)1269 _PyPegen_check_tokenizer_errors(Parser *p) {
1270     // Tokenize the whole input to see if there are any tokenization
1271     // errors such as mistmatching parentheses. These will get priority
1272     // over generic syntax errors only if the line number of the error is
1273     // before the one that we had for the generic error.
1274 
1275     // We don't want to tokenize to the end for interactive input
1276     if (p->tok->prompt != NULL) {
1277         return 0;
1278     }
1279 
1280     PyObject *type, *value, *traceback;
1281     PyErr_Fetch(&type, &value, &traceback);
1282 
1283     Token *current_token = p->known_err_token != NULL ? p->known_err_token : p->tokens[p->fill - 1];
1284     Py_ssize_t current_err_line = current_token->lineno;
1285 
1286     int ret = 0;
1287 
1288     for (;;) {
1289         const char *start;
1290         const char *end;
1291         switch (PyTokenizer_Get(p->tok, &start, &end)) {
1292             case ERRORTOKEN:
1293                 if (p->tok->level != 0) {
1294                     int error_lineno = p->tok->parenlinenostack[p->tok->level-1];
1295                     if (current_err_line > error_lineno) {
1296                         raise_unclosed_parentheses_error(p);
1297                         ret = -1;
1298                         goto exit;
1299                     }
1300                 }
1301                 break;
1302             case ENDMARKER:
1303                 break;
1304             default:
1305                 continue;
1306         }
1307         break;
1308     }
1309 
1310 
1311 exit:
1312     if (PyErr_Occurred()) {
1313         Py_XDECREF(value);
1314         Py_XDECREF(type);
1315         Py_XDECREF(traceback);
1316     } else {
1317         PyErr_Restore(type, value, traceback);
1318     }
1319     return ret;
1320 }
1321 
1322 void *
_PyPegen_run_parser(Parser * p)1323 _PyPegen_run_parser(Parser *p)
1324 {
1325     void *res = _PyPegen_parse(p);
1326     assert(p->level == 0);
1327     if (res == NULL) {
1328         if (PyErr_Occurred() && !PyErr_ExceptionMatches(PyExc_SyntaxError)) {
1329             return NULL;
1330         }
1331         Token *last_token = p->tokens[p->fill - 1];
1332         reset_parser_state(p);
1333         _PyPegen_parse(p);
1334         if (PyErr_Occurred()) {
1335             // Prioritize tokenizer errors to custom syntax errors raised
1336             // on the second phase only if the errors come from the parser.
1337             if (p->tok->done == E_DONE && PyErr_ExceptionMatches(PyExc_SyntaxError)) {
1338                 _PyPegen_check_tokenizer_errors(p);
1339             }
1340             return NULL;
1341         }
1342         if (p->fill == 0) {
1343             RAISE_SYNTAX_ERROR("error at start before reading any input");
1344         }
1345         else if (last_token->type == ERRORTOKEN && p->tok->done == E_EOF) {
1346             if (p->tok->level) {
1347                 raise_unclosed_parentheses_error(p);
1348             } else {
1349                 RAISE_SYNTAX_ERROR("unexpected EOF while parsing");
1350             }
1351         }
1352         else {
1353             if (p->tokens[p->fill-1]->type == INDENT) {
1354                 RAISE_INDENTATION_ERROR("unexpected indent");
1355             }
1356             else if (p->tokens[p->fill-1]->type == DEDENT) {
1357                 RAISE_INDENTATION_ERROR("unexpected unindent");
1358             }
1359             else {
1360                 // Use the last token we found on the first pass to avoid reporting
1361                 // incorrect locations for generic syntax errors just because we reached
1362                 // further away when trying to find specific syntax errors in the second
1363                 // pass.
1364                 RAISE_SYNTAX_ERROR_KNOWN_LOCATION(last_token, "invalid syntax");
1365                 // _PyPegen_check_tokenizer_errors will override the existing
1366                 // generic SyntaxError we just raised if errors are found.
1367                 _PyPegen_check_tokenizer_errors(p);
1368             }
1369         }
1370         return NULL;
1371     }
1372 
1373     if (p->start_rule == Py_single_input && bad_single_statement(p)) {
1374         p->tok->done = E_BADSINGLE; // This is not necessary for now, but might be in the future
1375         return RAISE_SYNTAX_ERROR("multiple statements found while compiling a single statement");
1376     }
1377 
1378     // test_peg_generator defines _Py_TEST_PEGEN to not call PyAST_Validate()
1379 #if defined(Py_DEBUG) && !defined(_Py_TEST_PEGEN)
1380     if (p->start_rule == Py_single_input ||
1381         p->start_rule == Py_file_input ||
1382         p->start_rule == Py_eval_input)
1383     {
1384         if (!_PyAST_Validate(res)) {
1385             return NULL;
1386         }
1387     }
1388 #endif
1389     return res;
1390 }
1391 
1392 mod_ty
_PyPegen_run_parser_from_file_pointer(FILE * fp,int start_rule,PyObject * filename_ob,const char * enc,const char * ps1,const char * ps2,PyCompilerFlags * flags,int * errcode,PyArena * arena)1393 _PyPegen_run_parser_from_file_pointer(FILE *fp, int start_rule, PyObject *filename_ob,
1394                              const char *enc, const char *ps1, const char *ps2,
1395                              PyCompilerFlags *flags, int *errcode, PyArena *arena)
1396 {
1397     struct tok_state *tok = PyTokenizer_FromFile(fp, enc, ps1, ps2);
1398     if (tok == NULL) {
1399         if (PyErr_Occurred()) {
1400             raise_tokenizer_init_error(filename_ob);
1401             return NULL;
1402         }
1403         return NULL;
1404     }
1405     if (!tok->fp || ps1 != NULL || ps2 != NULL ||
1406         PyUnicode_CompareWithASCIIString(filename_ob, "<stdin>") == 0) {
1407         tok->fp_interactive = 1;
1408     }
1409     // This transfers the ownership to the tokenizer
1410     tok->filename = filename_ob;
1411     Py_INCREF(filename_ob);
1412 
1413     // From here on we need to clean up even if there's an error
1414     mod_ty result = NULL;
1415 
1416     int parser_flags = compute_parser_flags(flags);
1417     Parser *p = _PyPegen_Parser_New(tok, start_rule, parser_flags, PY_MINOR_VERSION,
1418                                     errcode, arena);
1419     if (p == NULL) {
1420         goto error;
1421     }
1422 
1423     result = _PyPegen_run_parser(p);
1424     _PyPegen_Parser_Free(p);
1425 
1426 error:
1427     PyTokenizer_Free(tok);
1428     return result;
1429 }
1430 
1431 mod_ty
_PyPegen_run_parser_from_string(const char * str,int start_rule,PyObject * filename_ob,PyCompilerFlags * flags,PyArena * arena)1432 _PyPegen_run_parser_from_string(const char *str, int start_rule, PyObject *filename_ob,
1433                        PyCompilerFlags *flags, PyArena *arena)
1434 {
1435     int exec_input = start_rule == Py_file_input;
1436 
1437     struct tok_state *tok;
1438     if (flags != NULL && flags->cf_flags & PyCF_IGNORE_COOKIE) {
1439         tok = PyTokenizer_FromUTF8(str, exec_input);
1440     } else {
1441         tok = PyTokenizer_FromString(str, exec_input);
1442     }
1443     if (tok == NULL) {
1444         if (PyErr_Occurred()) {
1445             raise_tokenizer_init_error(filename_ob);
1446         }
1447         return NULL;
1448     }
1449     // This transfers the ownership to the tokenizer
1450     tok->filename = filename_ob;
1451     Py_INCREF(filename_ob);
1452 
1453     // We need to clear up from here on
1454     mod_ty result = NULL;
1455 
1456     int parser_flags = compute_parser_flags(flags);
1457     int feature_version = flags && (flags->cf_flags & PyCF_ONLY_AST) ?
1458         flags->cf_feature_version : PY_MINOR_VERSION;
1459     Parser *p = _PyPegen_Parser_New(tok, start_rule, parser_flags, feature_version,
1460                                     NULL, arena);
1461     if (p == NULL) {
1462         goto error;
1463     }
1464 
1465     result = _PyPegen_run_parser(p);
1466     _PyPegen_Parser_Free(p);
1467 
1468 error:
1469     PyTokenizer_Free(tok);
1470     return result;
1471 }
1472 
1473 asdl_stmt_seq*
_PyPegen_interactive_exit(Parser * p)1474 _PyPegen_interactive_exit(Parser *p)
1475 {
1476     if (p->errcode) {
1477         *(p->errcode) = E_EOF;
1478     }
1479     return NULL;
1480 }
1481 
1482 /* Creates a single-element asdl_seq* that contains a */
1483 asdl_seq *
_PyPegen_singleton_seq(Parser * p,void * a)1484 _PyPegen_singleton_seq(Parser *p, void *a)
1485 {
1486     assert(a != NULL);
1487     asdl_seq *seq = (asdl_seq*)_Py_asdl_generic_seq_new(1, p->arena);
1488     if (!seq) {
1489         return NULL;
1490     }
1491     asdl_seq_SET_UNTYPED(seq, 0, a);
1492     return seq;
1493 }
1494 
1495 /* Creates a copy of seq and prepends a to it */
1496 asdl_seq *
_PyPegen_seq_insert_in_front(Parser * p,void * a,asdl_seq * seq)1497 _PyPegen_seq_insert_in_front(Parser *p, void *a, asdl_seq *seq)
1498 {
1499     assert(a != NULL);
1500     if (!seq) {
1501         return _PyPegen_singleton_seq(p, a);
1502     }
1503 
1504     asdl_seq *new_seq = (asdl_seq*)_Py_asdl_generic_seq_new(asdl_seq_LEN(seq) + 1, p->arena);
1505     if (!new_seq) {
1506         return NULL;
1507     }
1508 
1509     asdl_seq_SET_UNTYPED(new_seq, 0, a);
1510     for (Py_ssize_t i = 1, l = asdl_seq_LEN(new_seq); i < l; i++) {
1511         asdl_seq_SET_UNTYPED(new_seq, i, asdl_seq_GET_UNTYPED(seq, i - 1));
1512     }
1513     return new_seq;
1514 }
1515 
1516 /* Creates a copy of seq and appends a to it */
1517 asdl_seq *
_PyPegen_seq_append_to_end(Parser * p,asdl_seq * seq,void * a)1518 _PyPegen_seq_append_to_end(Parser *p, asdl_seq *seq, void *a)
1519 {
1520     assert(a != NULL);
1521     if (!seq) {
1522         return _PyPegen_singleton_seq(p, a);
1523     }
1524 
1525     asdl_seq *new_seq = (asdl_seq*)_Py_asdl_generic_seq_new(asdl_seq_LEN(seq) + 1, p->arena);
1526     if (!new_seq) {
1527         return NULL;
1528     }
1529 
1530     for (Py_ssize_t i = 0, l = asdl_seq_LEN(new_seq); i + 1 < l; i++) {
1531         asdl_seq_SET_UNTYPED(new_seq, i, asdl_seq_GET_UNTYPED(seq, i));
1532     }
1533     asdl_seq_SET_UNTYPED(new_seq, asdl_seq_LEN(new_seq) - 1, a);
1534     return new_seq;
1535 }
1536 
1537 static Py_ssize_t
_get_flattened_seq_size(asdl_seq * seqs)1538 _get_flattened_seq_size(asdl_seq *seqs)
1539 {
1540     Py_ssize_t size = 0;
1541     for (Py_ssize_t i = 0, l = asdl_seq_LEN(seqs); i < l; i++) {
1542         asdl_seq *inner_seq = asdl_seq_GET_UNTYPED(seqs, i);
1543         size += asdl_seq_LEN(inner_seq);
1544     }
1545     return size;
1546 }
1547 
1548 /* Flattens an asdl_seq* of asdl_seq*s */
1549 asdl_seq *
_PyPegen_seq_flatten(Parser * p,asdl_seq * seqs)1550 _PyPegen_seq_flatten(Parser *p, asdl_seq *seqs)
1551 {
1552     Py_ssize_t flattened_seq_size = _get_flattened_seq_size(seqs);
1553     assert(flattened_seq_size > 0);
1554 
1555     asdl_seq *flattened_seq = (asdl_seq*)_Py_asdl_generic_seq_new(flattened_seq_size, p->arena);
1556     if (!flattened_seq) {
1557         return NULL;
1558     }
1559 
1560     int flattened_seq_idx = 0;
1561     for (Py_ssize_t i = 0, l = asdl_seq_LEN(seqs); i < l; i++) {
1562         asdl_seq *inner_seq = asdl_seq_GET_UNTYPED(seqs, i);
1563         for (Py_ssize_t j = 0, li = asdl_seq_LEN(inner_seq); j < li; j++) {
1564             asdl_seq_SET_UNTYPED(flattened_seq, flattened_seq_idx++, asdl_seq_GET_UNTYPED(inner_seq, j));
1565         }
1566     }
1567     assert(flattened_seq_idx == flattened_seq_size);
1568 
1569     return flattened_seq;
1570 }
1571 
1572 void *
_PyPegen_seq_last_item(asdl_seq * seq)1573 _PyPegen_seq_last_item(asdl_seq *seq)
1574 {
1575     Py_ssize_t len = asdl_seq_LEN(seq);
1576     return asdl_seq_GET_UNTYPED(seq, len - 1);
1577 }
1578 
1579 void *
_PyPegen_seq_first_item(asdl_seq * seq)1580 _PyPegen_seq_first_item(asdl_seq *seq)
1581 {
1582     return asdl_seq_GET_UNTYPED(seq, 0);
1583 }
1584 
1585 
1586 /* Creates a new name of the form <first_name>.<second_name> */
1587 expr_ty
_PyPegen_join_names_with_dot(Parser * p,expr_ty first_name,expr_ty second_name)1588 _PyPegen_join_names_with_dot(Parser *p, expr_ty first_name, expr_ty second_name)
1589 {
1590     assert(first_name != NULL && second_name != NULL);
1591     PyObject *first_identifier = first_name->v.Name.id;
1592     PyObject *second_identifier = second_name->v.Name.id;
1593 
1594     if (PyUnicode_READY(first_identifier) == -1) {
1595         return NULL;
1596     }
1597     if (PyUnicode_READY(second_identifier) == -1) {
1598         return NULL;
1599     }
1600     const char *first_str = PyUnicode_AsUTF8(first_identifier);
1601     if (!first_str) {
1602         return NULL;
1603     }
1604     const char *second_str = PyUnicode_AsUTF8(second_identifier);
1605     if (!second_str) {
1606         return NULL;
1607     }
1608     Py_ssize_t len = strlen(first_str) + strlen(second_str) + 1;  // +1 for the dot
1609 
1610     PyObject *str = PyBytes_FromStringAndSize(NULL, len);
1611     if (!str) {
1612         return NULL;
1613     }
1614 
1615     char *s = PyBytes_AS_STRING(str);
1616     if (!s) {
1617         return NULL;
1618     }
1619 
1620     strcpy(s, first_str);
1621     s += strlen(first_str);
1622     *s++ = '.';
1623     strcpy(s, second_str);
1624     s += strlen(second_str);
1625     *s = '\0';
1626 
1627     PyObject *uni = PyUnicode_DecodeUTF8(PyBytes_AS_STRING(str), PyBytes_GET_SIZE(str), NULL);
1628     Py_DECREF(str);
1629     if (!uni) {
1630         return NULL;
1631     }
1632     PyUnicode_InternInPlace(&uni);
1633     if (_PyArena_AddPyObject(p->arena, uni) < 0) {
1634         Py_DECREF(uni);
1635         return NULL;
1636     }
1637 
1638     return _PyAST_Name(uni, Load, EXTRA_EXPR(first_name, second_name));
1639 }
1640 
1641 /* Counts the total number of dots in seq's tokens */
1642 int
_PyPegen_seq_count_dots(asdl_seq * seq)1643 _PyPegen_seq_count_dots(asdl_seq *seq)
1644 {
1645     int number_of_dots = 0;
1646     for (Py_ssize_t i = 0, l = asdl_seq_LEN(seq); i < l; i++) {
1647         Token *current_expr = asdl_seq_GET_UNTYPED(seq, i);
1648         switch (current_expr->type) {
1649             case ELLIPSIS:
1650                 number_of_dots += 3;
1651                 break;
1652             case DOT:
1653                 number_of_dots += 1;
1654                 break;
1655             default:
1656                 Py_UNREACHABLE();
1657         }
1658     }
1659 
1660     return number_of_dots;
1661 }
1662 
1663 /* Creates an alias with '*' as the identifier name */
1664 alias_ty
_PyPegen_alias_for_star(Parser * p,int lineno,int col_offset,int end_lineno,int end_col_offset,PyArena * arena)1665 _PyPegen_alias_for_star(Parser *p, int lineno, int col_offset, int end_lineno,
1666                         int end_col_offset, PyArena *arena) {
1667     PyObject *str = PyUnicode_InternFromString("*");
1668     if (!str) {
1669         return NULL;
1670     }
1671     if (_PyArena_AddPyObject(p->arena, str) < 0) {
1672         Py_DECREF(str);
1673         return NULL;
1674     }
1675     return _PyAST_alias(str, NULL, lineno, col_offset, end_lineno, end_col_offset, arena);
1676 }
1677 
1678 /* Creates a new asdl_seq* with the identifiers of all the names in seq */
1679 asdl_identifier_seq *
_PyPegen_map_names_to_ids(Parser * p,asdl_expr_seq * seq)1680 _PyPegen_map_names_to_ids(Parser *p, asdl_expr_seq *seq)
1681 {
1682     Py_ssize_t len = asdl_seq_LEN(seq);
1683     assert(len > 0);
1684 
1685     asdl_identifier_seq *new_seq = _Py_asdl_identifier_seq_new(len, p->arena);
1686     if (!new_seq) {
1687         return NULL;
1688     }
1689     for (Py_ssize_t i = 0; i < len; i++) {
1690         expr_ty e = asdl_seq_GET(seq, i);
1691         asdl_seq_SET(new_seq, i, e->v.Name.id);
1692     }
1693     return new_seq;
1694 }
1695 
1696 /* Constructs a CmpopExprPair */
1697 CmpopExprPair *
_PyPegen_cmpop_expr_pair(Parser * p,cmpop_ty cmpop,expr_ty expr)1698 _PyPegen_cmpop_expr_pair(Parser *p, cmpop_ty cmpop, expr_ty expr)
1699 {
1700     assert(expr != NULL);
1701     CmpopExprPair *a = _PyArena_Malloc(p->arena, sizeof(CmpopExprPair));
1702     if (!a) {
1703         return NULL;
1704     }
1705     a->cmpop = cmpop;
1706     a->expr = expr;
1707     return a;
1708 }
1709 
1710 asdl_int_seq *
_PyPegen_get_cmpops(Parser * p,asdl_seq * seq)1711 _PyPegen_get_cmpops(Parser *p, asdl_seq *seq)
1712 {
1713     Py_ssize_t len = asdl_seq_LEN(seq);
1714     assert(len > 0);
1715 
1716     asdl_int_seq *new_seq = _Py_asdl_int_seq_new(len, p->arena);
1717     if (!new_seq) {
1718         return NULL;
1719     }
1720     for (Py_ssize_t i = 0; i < len; i++) {
1721         CmpopExprPair *pair = asdl_seq_GET_UNTYPED(seq, i);
1722         asdl_seq_SET(new_seq, i, pair->cmpop);
1723     }
1724     return new_seq;
1725 }
1726 
1727 asdl_expr_seq *
_PyPegen_get_exprs(Parser * p,asdl_seq * seq)1728 _PyPegen_get_exprs(Parser *p, asdl_seq *seq)
1729 {
1730     Py_ssize_t len = asdl_seq_LEN(seq);
1731     assert(len > 0);
1732 
1733     asdl_expr_seq *new_seq = _Py_asdl_expr_seq_new(len, p->arena);
1734     if (!new_seq) {
1735         return NULL;
1736     }
1737     for (Py_ssize_t i = 0; i < len; i++) {
1738         CmpopExprPair *pair = asdl_seq_GET_UNTYPED(seq, i);
1739         asdl_seq_SET(new_seq, i, pair->expr);
1740     }
1741     return new_seq;
1742 }
1743 
1744 /* Creates an asdl_seq* where all the elements have been changed to have ctx as context */
1745 static asdl_expr_seq *
_set_seq_context(Parser * p,asdl_expr_seq * seq,expr_context_ty ctx)1746 _set_seq_context(Parser *p, asdl_expr_seq *seq, expr_context_ty ctx)
1747 {
1748     Py_ssize_t len = asdl_seq_LEN(seq);
1749     if (len == 0) {
1750         return NULL;
1751     }
1752 
1753     asdl_expr_seq *new_seq = _Py_asdl_expr_seq_new(len, p->arena);
1754     if (!new_seq) {
1755         return NULL;
1756     }
1757     for (Py_ssize_t i = 0; i < len; i++) {
1758         expr_ty e = asdl_seq_GET(seq, i);
1759         asdl_seq_SET(new_seq, i, _PyPegen_set_expr_context(p, e, ctx));
1760     }
1761     return new_seq;
1762 }
1763 
1764 static expr_ty
_set_name_context(Parser * p,expr_ty e,expr_context_ty ctx)1765 _set_name_context(Parser *p, expr_ty e, expr_context_ty ctx)
1766 {
1767     return _PyAST_Name(e->v.Name.id, ctx, EXTRA_EXPR(e, e));
1768 }
1769 
1770 static expr_ty
_set_tuple_context(Parser * p,expr_ty e,expr_context_ty ctx)1771 _set_tuple_context(Parser *p, expr_ty e, expr_context_ty ctx)
1772 {
1773     return _PyAST_Tuple(
1774             _set_seq_context(p, e->v.Tuple.elts, ctx),
1775             ctx,
1776             EXTRA_EXPR(e, e));
1777 }
1778 
1779 static expr_ty
_set_list_context(Parser * p,expr_ty e,expr_context_ty ctx)1780 _set_list_context(Parser *p, expr_ty e, expr_context_ty ctx)
1781 {
1782     return _PyAST_List(
1783             _set_seq_context(p, e->v.List.elts, ctx),
1784             ctx,
1785             EXTRA_EXPR(e, e));
1786 }
1787 
1788 static expr_ty
_set_subscript_context(Parser * p,expr_ty e,expr_context_ty ctx)1789 _set_subscript_context(Parser *p, expr_ty e, expr_context_ty ctx)
1790 {
1791     return _PyAST_Subscript(e->v.Subscript.value, e->v.Subscript.slice,
1792                             ctx, EXTRA_EXPR(e, e));
1793 }
1794 
1795 static expr_ty
_set_attribute_context(Parser * p,expr_ty e,expr_context_ty ctx)1796 _set_attribute_context(Parser *p, expr_ty e, expr_context_ty ctx)
1797 {
1798     return _PyAST_Attribute(e->v.Attribute.value, e->v.Attribute.attr,
1799                             ctx, EXTRA_EXPR(e, e));
1800 }
1801 
1802 static expr_ty
_set_starred_context(Parser * p,expr_ty e,expr_context_ty ctx)1803 _set_starred_context(Parser *p, expr_ty e, expr_context_ty ctx)
1804 {
1805     return _PyAST_Starred(_PyPegen_set_expr_context(p, e->v.Starred.value, ctx),
1806                           ctx, EXTRA_EXPR(e, e));
1807 }
1808 
1809 /* Creates an `expr_ty` equivalent to `expr` but with `ctx` as context */
1810 expr_ty
_PyPegen_set_expr_context(Parser * p,expr_ty expr,expr_context_ty ctx)1811 _PyPegen_set_expr_context(Parser *p, expr_ty expr, expr_context_ty ctx)
1812 {
1813     assert(expr != NULL);
1814 
1815     expr_ty new = NULL;
1816     switch (expr->kind) {
1817         case Name_kind:
1818             new = _set_name_context(p, expr, ctx);
1819             break;
1820         case Tuple_kind:
1821             new = _set_tuple_context(p, expr, ctx);
1822             break;
1823         case List_kind:
1824             new = _set_list_context(p, expr, ctx);
1825             break;
1826         case Subscript_kind:
1827             new = _set_subscript_context(p, expr, ctx);
1828             break;
1829         case Attribute_kind:
1830             new = _set_attribute_context(p, expr, ctx);
1831             break;
1832         case Starred_kind:
1833             new = _set_starred_context(p, expr, ctx);
1834             break;
1835         default:
1836             new = expr;
1837     }
1838     return new;
1839 }
1840 
1841 /* Constructs a KeyValuePair that is used when parsing a dict's key value pairs */
1842 KeyValuePair *
_PyPegen_key_value_pair(Parser * p,expr_ty key,expr_ty value)1843 _PyPegen_key_value_pair(Parser *p, expr_ty key, expr_ty value)
1844 {
1845     KeyValuePair *a = _PyArena_Malloc(p->arena, sizeof(KeyValuePair));
1846     if (!a) {
1847         return NULL;
1848     }
1849     a->key = key;
1850     a->value = value;
1851     return a;
1852 }
1853 
1854 /* Extracts all keys from an asdl_seq* of KeyValuePair*'s */
1855 asdl_expr_seq *
_PyPegen_get_keys(Parser * p,asdl_seq * seq)1856 _PyPegen_get_keys(Parser *p, asdl_seq *seq)
1857 {
1858     Py_ssize_t len = asdl_seq_LEN(seq);
1859     asdl_expr_seq *new_seq = _Py_asdl_expr_seq_new(len, p->arena);
1860     if (!new_seq) {
1861         return NULL;
1862     }
1863     for (Py_ssize_t i = 0; i < len; i++) {
1864         KeyValuePair *pair = asdl_seq_GET_UNTYPED(seq, i);
1865         asdl_seq_SET(new_seq, i, pair->key);
1866     }
1867     return new_seq;
1868 }
1869 
1870 /* Extracts all values from an asdl_seq* of KeyValuePair*'s */
1871 asdl_expr_seq *
_PyPegen_get_values(Parser * p,asdl_seq * seq)1872 _PyPegen_get_values(Parser *p, asdl_seq *seq)
1873 {
1874     Py_ssize_t len = asdl_seq_LEN(seq);
1875     asdl_expr_seq *new_seq = _Py_asdl_expr_seq_new(len, p->arena);
1876     if (!new_seq) {
1877         return NULL;
1878     }
1879     for (Py_ssize_t i = 0; i < len; i++) {
1880         KeyValuePair *pair = asdl_seq_GET_UNTYPED(seq, i);
1881         asdl_seq_SET(new_seq, i, pair->value);
1882     }
1883     return new_seq;
1884 }
1885 
1886 /* Constructs a KeyPatternPair that is used when parsing mapping & class patterns */
1887 KeyPatternPair *
_PyPegen_key_pattern_pair(Parser * p,expr_ty key,pattern_ty pattern)1888 _PyPegen_key_pattern_pair(Parser *p, expr_ty key, pattern_ty pattern)
1889 {
1890     KeyPatternPair *a = _PyArena_Malloc(p->arena, sizeof(KeyPatternPair));
1891     if (!a) {
1892         return NULL;
1893     }
1894     a->key = key;
1895     a->pattern = pattern;
1896     return a;
1897 }
1898 
1899 /* Extracts all keys from an asdl_seq* of KeyPatternPair*'s */
1900 asdl_expr_seq *
_PyPegen_get_pattern_keys(Parser * p,asdl_seq * seq)1901 _PyPegen_get_pattern_keys(Parser *p, asdl_seq *seq)
1902 {
1903     Py_ssize_t len = asdl_seq_LEN(seq);
1904     asdl_expr_seq *new_seq = _Py_asdl_expr_seq_new(len, p->arena);
1905     if (!new_seq) {
1906         return NULL;
1907     }
1908     for (Py_ssize_t i = 0; i < len; i++) {
1909         KeyPatternPair *pair = asdl_seq_GET_UNTYPED(seq, i);
1910         asdl_seq_SET(new_seq, i, pair->key);
1911     }
1912     return new_seq;
1913 }
1914 
1915 /* Extracts all patterns from an asdl_seq* of KeyPatternPair*'s */
1916 asdl_pattern_seq *
_PyPegen_get_patterns(Parser * p,asdl_seq * seq)1917 _PyPegen_get_patterns(Parser *p, asdl_seq *seq)
1918 {
1919     Py_ssize_t len = asdl_seq_LEN(seq);
1920     asdl_pattern_seq *new_seq = _Py_asdl_pattern_seq_new(len, p->arena);
1921     if (!new_seq) {
1922         return NULL;
1923     }
1924     for (Py_ssize_t i = 0; i < len; i++) {
1925         KeyPatternPair *pair = asdl_seq_GET_UNTYPED(seq, i);
1926         asdl_seq_SET(new_seq, i, pair->pattern);
1927     }
1928     return new_seq;
1929 }
1930 
1931 /* Constructs a NameDefaultPair */
1932 NameDefaultPair *
_PyPegen_name_default_pair(Parser * p,arg_ty arg,expr_ty value,Token * tc)1933 _PyPegen_name_default_pair(Parser *p, arg_ty arg, expr_ty value, Token *tc)
1934 {
1935     NameDefaultPair *a = _PyArena_Malloc(p->arena, sizeof(NameDefaultPair));
1936     if (!a) {
1937         return NULL;
1938     }
1939     a->arg = _PyPegen_add_type_comment_to_arg(p, arg, tc);
1940     a->value = value;
1941     return a;
1942 }
1943 
1944 /* Constructs a SlashWithDefault */
1945 SlashWithDefault *
_PyPegen_slash_with_default(Parser * p,asdl_arg_seq * plain_names,asdl_seq * names_with_defaults)1946 _PyPegen_slash_with_default(Parser *p, asdl_arg_seq *plain_names, asdl_seq *names_with_defaults)
1947 {
1948     SlashWithDefault *a = _PyArena_Malloc(p->arena, sizeof(SlashWithDefault));
1949     if (!a) {
1950         return NULL;
1951     }
1952     a->plain_names = plain_names;
1953     a->names_with_defaults = names_with_defaults;
1954     return a;
1955 }
1956 
1957 /* Constructs a StarEtc */
1958 StarEtc *
_PyPegen_star_etc(Parser * p,arg_ty vararg,asdl_seq * kwonlyargs,arg_ty kwarg)1959 _PyPegen_star_etc(Parser *p, arg_ty vararg, asdl_seq *kwonlyargs, arg_ty kwarg)
1960 {
1961     StarEtc *a = _PyArena_Malloc(p->arena, sizeof(StarEtc));
1962     if (!a) {
1963         return NULL;
1964     }
1965     a->vararg = vararg;
1966     a->kwonlyargs = kwonlyargs;
1967     a->kwarg = kwarg;
1968     return a;
1969 }
1970 
1971 asdl_seq *
_PyPegen_join_sequences(Parser * p,asdl_seq * a,asdl_seq * b)1972 _PyPegen_join_sequences(Parser *p, asdl_seq *a, asdl_seq *b)
1973 {
1974     Py_ssize_t first_len = asdl_seq_LEN(a);
1975     Py_ssize_t second_len = asdl_seq_LEN(b);
1976     asdl_seq *new_seq = (asdl_seq*)_Py_asdl_generic_seq_new(first_len + second_len, p->arena);
1977     if (!new_seq) {
1978         return NULL;
1979     }
1980 
1981     int k = 0;
1982     for (Py_ssize_t i = 0; i < first_len; i++) {
1983         asdl_seq_SET_UNTYPED(new_seq, k++, asdl_seq_GET_UNTYPED(a, i));
1984     }
1985     for (Py_ssize_t i = 0; i < second_len; i++) {
1986         asdl_seq_SET_UNTYPED(new_seq, k++, asdl_seq_GET_UNTYPED(b, i));
1987     }
1988 
1989     return new_seq;
1990 }
1991 
1992 static asdl_arg_seq*
_get_names(Parser * p,asdl_seq * names_with_defaults)1993 _get_names(Parser *p, asdl_seq *names_with_defaults)
1994 {
1995     Py_ssize_t len = asdl_seq_LEN(names_with_defaults);
1996     asdl_arg_seq *seq = _Py_asdl_arg_seq_new(len, p->arena);
1997     if (!seq) {
1998         return NULL;
1999     }
2000     for (Py_ssize_t i = 0; i < len; i++) {
2001         NameDefaultPair *pair = asdl_seq_GET_UNTYPED(names_with_defaults, i);
2002         asdl_seq_SET(seq, i, pair->arg);
2003     }
2004     return seq;
2005 }
2006 
2007 static asdl_expr_seq *
_get_defaults(Parser * p,asdl_seq * names_with_defaults)2008 _get_defaults(Parser *p, asdl_seq *names_with_defaults)
2009 {
2010     Py_ssize_t len = asdl_seq_LEN(names_with_defaults);
2011     asdl_expr_seq *seq = _Py_asdl_expr_seq_new(len, p->arena);
2012     if (!seq) {
2013         return NULL;
2014     }
2015     for (Py_ssize_t i = 0; i < len; i++) {
2016         NameDefaultPair *pair = asdl_seq_GET_UNTYPED(names_with_defaults, i);
2017         asdl_seq_SET(seq, i, pair->value);
2018     }
2019     return seq;
2020 }
2021 
2022 static int
_make_posonlyargs(Parser * p,asdl_arg_seq * slash_without_default,SlashWithDefault * slash_with_default,asdl_arg_seq ** posonlyargs)2023 _make_posonlyargs(Parser *p,
2024                   asdl_arg_seq *slash_without_default,
2025                   SlashWithDefault *slash_with_default,
2026                   asdl_arg_seq **posonlyargs) {
2027     if (slash_without_default != NULL) {
2028         *posonlyargs = slash_without_default;
2029     }
2030     else if (slash_with_default != NULL) {
2031         asdl_arg_seq *slash_with_default_names =
2032                 _get_names(p, slash_with_default->names_with_defaults);
2033         if (!slash_with_default_names) {
2034             return -1;
2035         }
2036         *posonlyargs = (asdl_arg_seq*)_PyPegen_join_sequences(
2037                 p,
2038                 (asdl_seq*)slash_with_default->plain_names,
2039                 (asdl_seq*)slash_with_default_names);
2040     }
2041     else {
2042         *posonlyargs = _Py_asdl_arg_seq_new(0, p->arena);
2043     }
2044     return *posonlyargs == NULL ? -1 : 0;
2045 }
2046 
2047 static int
_make_posargs(Parser * p,asdl_arg_seq * plain_names,asdl_seq * names_with_default,asdl_arg_seq ** posargs)2048 _make_posargs(Parser *p,
2049               asdl_arg_seq *plain_names,
2050               asdl_seq *names_with_default,
2051               asdl_arg_seq **posargs) {
2052     if (plain_names != NULL && names_with_default != NULL) {
2053         asdl_arg_seq *names_with_default_names = _get_names(p, names_with_default);
2054         if (!names_with_default_names) {
2055             return -1;
2056         }
2057         *posargs = (asdl_arg_seq*)_PyPegen_join_sequences(
2058                 p,(asdl_seq*)plain_names, (asdl_seq*)names_with_default_names);
2059     }
2060     else if (plain_names == NULL && names_with_default != NULL) {
2061         *posargs = _get_names(p, names_with_default);
2062     }
2063     else if (plain_names != NULL && names_with_default == NULL) {
2064         *posargs = plain_names;
2065     }
2066     else {
2067         *posargs = _Py_asdl_arg_seq_new(0, p->arena);
2068     }
2069     return *posargs == NULL ? -1 : 0;
2070 }
2071 
2072 static int
_make_posdefaults(Parser * p,SlashWithDefault * slash_with_default,asdl_seq * names_with_default,asdl_expr_seq ** posdefaults)2073 _make_posdefaults(Parser *p,
2074                   SlashWithDefault *slash_with_default,
2075                   asdl_seq *names_with_default,
2076                   asdl_expr_seq **posdefaults) {
2077     if (slash_with_default != NULL && names_with_default != NULL) {
2078         asdl_expr_seq *slash_with_default_values =
2079                 _get_defaults(p, slash_with_default->names_with_defaults);
2080         if (!slash_with_default_values) {
2081             return -1;
2082         }
2083         asdl_expr_seq *names_with_default_values = _get_defaults(p, names_with_default);
2084         if (!names_with_default_values) {
2085             return -1;
2086         }
2087         *posdefaults = (asdl_expr_seq*)_PyPegen_join_sequences(
2088                 p,
2089                 (asdl_seq*)slash_with_default_values,
2090                 (asdl_seq*)names_with_default_values);
2091     }
2092     else if (slash_with_default == NULL && names_with_default != NULL) {
2093         *posdefaults = _get_defaults(p, names_with_default);
2094     }
2095     else if (slash_with_default != NULL && names_with_default == NULL) {
2096         *posdefaults = _get_defaults(p, slash_with_default->names_with_defaults);
2097     }
2098     else {
2099         *posdefaults = _Py_asdl_expr_seq_new(0, p->arena);
2100     }
2101     return *posdefaults == NULL ? -1 : 0;
2102 }
2103 
2104 static int
_make_kwargs(Parser * p,StarEtc * star_etc,asdl_arg_seq ** kwonlyargs,asdl_expr_seq ** kwdefaults)2105 _make_kwargs(Parser *p, StarEtc *star_etc,
2106              asdl_arg_seq **kwonlyargs,
2107              asdl_expr_seq **kwdefaults) {
2108     if (star_etc != NULL && star_etc->kwonlyargs != NULL) {
2109         *kwonlyargs = _get_names(p, star_etc->kwonlyargs);
2110     }
2111     else {
2112         *kwonlyargs = _Py_asdl_arg_seq_new(0, p->arena);
2113     }
2114 
2115     if (*kwonlyargs == NULL) {
2116         return -1;
2117     }
2118 
2119     if (star_etc != NULL && star_etc->kwonlyargs != NULL) {
2120         *kwdefaults = _get_defaults(p, star_etc->kwonlyargs);
2121     }
2122     else {
2123         *kwdefaults = _Py_asdl_expr_seq_new(0, p->arena);
2124     }
2125 
2126     if (*kwdefaults == NULL) {
2127         return -1;
2128     }
2129 
2130     return 0;
2131 }
2132 
2133 /* Constructs an arguments_ty object out of all the parsed constructs in the parameters rule */
2134 arguments_ty
_PyPegen_make_arguments(Parser * p,asdl_arg_seq * slash_without_default,SlashWithDefault * slash_with_default,asdl_arg_seq * plain_names,asdl_seq * names_with_default,StarEtc * star_etc)2135 _PyPegen_make_arguments(Parser *p, asdl_arg_seq *slash_without_default,
2136                         SlashWithDefault *slash_with_default, asdl_arg_seq *plain_names,
2137                         asdl_seq *names_with_default, StarEtc *star_etc)
2138 {
2139     asdl_arg_seq *posonlyargs;
2140     if (_make_posonlyargs(p, slash_without_default, slash_with_default, &posonlyargs) == -1) {
2141         return NULL;
2142     }
2143 
2144     asdl_arg_seq *posargs;
2145     if (_make_posargs(p, plain_names, names_with_default, &posargs) == -1) {
2146         return NULL;
2147     }
2148 
2149     asdl_expr_seq *posdefaults;
2150     if (_make_posdefaults(p,slash_with_default, names_with_default, &posdefaults) == -1) {
2151         return NULL;
2152     }
2153 
2154     arg_ty vararg = NULL;
2155     if (star_etc != NULL && star_etc->vararg != NULL) {
2156         vararg = star_etc->vararg;
2157     }
2158 
2159     asdl_arg_seq *kwonlyargs;
2160     asdl_expr_seq *kwdefaults;
2161     if (_make_kwargs(p, star_etc, &kwonlyargs, &kwdefaults) == -1) {
2162         return NULL;
2163     }
2164 
2165     arg_ty kwarg = NULL;
2166     if (star_etc != NULL && star_etc->kwarg != NULL) {
2167         kwarg = star_etc->kwarg;
2168     }
2169 
2170     return _PyAST_arguments(posonlyargs, posargs, vararg, kwonlyargs,
2171                             kwdefaults, kwarg, posdefaults, p->arena);
2172 }
2173 
2174 
2175 /* Constructs an empty arguments_ty object, that gets used when a function accepts no
2176  * arguments. */
2177 arguments_ty
_PyPegen_empty_arguments(Parser * p)2178 _PyPegen_empty_arguments(Parser *p)
2179 {
2180     asdl_arg_seq *posonlyargs = _Py_asdl_arg_seq_new(0, p->arena);
2181     if (!posonlyargs) {
2182         return NULL;
2183     }
2184     asdl_arg_seq *posargs = _Py_asdl_arg_seq_new(0, p->arena);
2185     if (!posargs) {
2186         return NULL;
2187     }
2188     asdl_expr_seq *posdefaults = _Py_asdl_expr_seq_new(0, p->arena);
2189     if (!posdefaults) {
2190         return NULL;
2191     }
2192     asdl_arg_seq *kwonlyargs = _Py_asdl_arg_seq_new(0, p->arena);
2193     if (!kwonlyargs) {
2194         return NULL;
2195     }
2196     asdl_expr_seq *kwdefaults = _Py_asdl_expr_seq_new(0, p->arena);
2197     if (!kwdefaults) {
2198         return NULL;
2199     }
2200 
2201     return _PyAST_arguments(posonlyargs, posargs, NULL, kwonlyargs,
2202                             kwdefaults, NULL, posdefaults, p->arena);
2203 }
2204 
2205 /* Encapsulates the value of an operator_ty into an AugOperator struct */
2206 AugOperator *
_PyPegen_augoperator(Parser * p,operator_ty kind)2207 _PyPegen_augoperator(Parser *p, operator_ty kind)
2208 {
2209     AugOperator *a = _PyArena_Malloc(p->arena, sizeof(AugOperator));
2210     if (!a) {
2211         return NULL;
2212     }
2213     a->kind = kind;
2214     return a;
2215 }
2216 
2217 /* Construct a FunctionDef equivalent to function_def, but with decorators */
2218 stmt_ty
_PyPegen_function_def_decorators(Parser * p,asdl_expr_seq * decorators,stmt_ty function_def)2219 _PyPegen_function_def_decorators(Parser *p, asdl_expr_seq *decorators, stmt_ty function_def)
2220 {
2221     assert(function_def != NULL);
2222     if (function_def->kind == AsyncFunctionDef_kind) {
2223         return _PyAST_AsyncFunctionDef(
2224             function_def->v.FunctionDef.name, function_def->v.FunctionDef.args,
2225             function_def->v.FunctionDef.body, decorators, function_def->v.FunctionDef.returns,
2226             function_def->v.FunctionDef.type_comment, function_def->lineno,
2227             function_def->col_offset, function_def->end_lineno, function_def->end_col_offset,
2228             p->arena);
2229     }
2230 
2231     return _PyAST_FunctionDef(
2232         function_def->v.FunctionDef.name, function_def->v.FunctionDef.args,
2233         function_def->v.FunctionDef.body, decorators,
2234         function_def->v.FunctionDef.returns,
2235         function_def->v.FunctionDef.type_comment, function_def->lineno,
2236         function_def->col_offset, function_def->end_lineno,
2237         function_def->end_col_offset, p->arena);
2238 }
2239 
2240 /* Construct a ClassDef equivalent to class_def, but with decorators */
2241 stmt_ty
_PyPegen_class_def_decorators(Parser * p,asdl_expr_seq * decorators,stmt_ty class_def)2242 _PyPegen_class_def_decorators(Parser *p, asdl_expr_seq *decorators, stmt_ty class_def)
2243 {
2244     assert(class_def != NULL);
2245     return _PyAST_ClassDef(
2246         class_def->v.ClassDef.name, class_def->v.ClassDef.bases,
2247         class_def->v.ClassDef.keywords, class_def->v.ClassDef.body, decorators,
2248         class_def->lineno, class_def->col_offset, class_def->end_lineno,
2249         class_def->end_col_offset, p->arena);
2250 }
2251 
2252 /* Construct a KeywordOrStarred */
2253 KeywordOrStarred *
_PyPegen_keyword_or_starred(Parser * p,void * element,int is_keyword)2254 _PyPegen_keyword_or_starred(Parser *p, void *element, int is_keyword)
2255 {
2256     KeywordOrStarred *a = _PyArena_Malloc(p->arena, sizeof(KeywordOrStarred));
2257     if (!a) {
2258         return NULL;
2259     }
2260     a->element = element;
2261     a->is_keyword = is_keyword;
2262     return a;
2263 }
2264 
2265 /* Get the number of starred expressions in an asdl_seq* of KeywordOrStarred*s */
2266 static int
_seq_number_of_starred_exprs(asdl_seq * seq)2267 _seq_number_of_starred_exprs(asdl_seq *seq)
2268 {
2269     int n = 0;
2270     for (Py_ssize_t i = 0, l = asdl_seq_LEN(seq); i < l; i++) {
2271         KeywordOrStarred *k = asdl_seq_GET_UNTYPED(seq, i);
2272         if (!k->is_keyword) {
2273             n++;
2274         }
2275     }
2276     return n;
2277 }
2278 
2279 /* Extract the starred expressions of an asdl_seq* of KeywordOrStarred*s */
2280 asdl_expr_seq *
_PyPegen_seq_extract_starred_exprs(Parser * p,asdl_seq * kwargs)2281 _PyPegen_seq_extract_starred_exprs(Parser *p, asdl_seq *kwargs)
2282 {
2283     int new_len = _seq_number_of_starred_exprs(kwargs);
2284     if (new_len == 0) {
2285         return NULL;
2286     }
2287     asdl_expr_seq *new_seq = _Py_asdl_expr_seq_new(new_len, p->arena);
2288     if (!new_seq) {
2289         return NULL;
2290     }
2291 
2292     int idx = 0;
2293     for (Py_ssize_t i = 0, len = asdl_seq_LEN(kwargs); i < len; i++) {
2294         KeywordOrStarred *k = asdl_seq_GET_UNTYPED(kwargs, i);
2295         if (!k->is_keyword) {
2296             asdl_seq_SET(new_seq, idx++, k->element);
2297         }
2298     }
2299     return new_seq;
2300 }
2301 
2302 /* Return a new asdl_seq* with only the keywords in kwargs */
2303 asdl_keyword_seq*
_PyPegen_seq_delete_starred_exprs(Parser * p,asdl_seq * kwargs)2304 _PyPegen_seq_delete_starred_exprs(Parser *p, asdl_seq *kwargs)
2305 {
2306     Py_ssize_t len = asdl_seq_LEN(kwargs);
2307     Py_ssize_t new_len = len - _seq_number_of_starred_exprs(kwargs);
2308     if (new_len == 0) {
2309         return NULL;
2310     }
2311     asdl_keyword_seq *new_seq = _Py_asdl_keyword_seq_new(new_len, p->arena);
2312     if (!new_seq) {
2313         return NULL;
2314     }
2315 
2316     int idx = 0;
2317     for (Py_ssize_t i = 0; i < len; i++) {
2318         KeywordOrStarred *k = asdl_seq_GET_UNTYPED(kwargs, i);
2319         if (k->is_keyword) {
2320             asdl_seq_SET(new_seq, idx++, k->element);
2321         }
2322     }
2323     return new_seq;
2324 }
2325 
2326 expr_ty
_PyPegen_concatenate_strings(Parser * p,asdl_seq * strings)2327 _PyPegen_concatenate_strings(Parser *p, asdl_seq *strings)
2328 {
2329     Py_ssize_t len = asdl_seq_LEN(strings);
2330     assert(len > 0);
2331 
2332     Token *first = asdl_seq_GET_UNTYPED(strings, 0);
2333     Token *last = asdl_seq_GET_UNTYPED(strings, len - 1);
2334 
2335     int bytesmode = 0;
2336     PyObject *bytes_str = NULL;
2337 
2338     FstringParser state;
2339     _PyPegen_FstringParser_Init(&state);
2340 
2341     for (Py_ssize_t i = 0; i < len; i++) {
2342         Token *t = asdl_seq_GET_UNTYPED(strings, i);
2343 
2344         int this_bytesmode;
2345         int this_rawmode;
2346         PyObject *s;
2347         const char *fstr;
2348         Py_ssize_t fstrlen = -1;
2349 
2350         if (_PyPegen_parsestr(p, &this_bytesmode, &this_rawmode, &s, &fstr, &fstrlen, t) != 0) {
2351             goto error;
2352         }
2353 
2354         /* Check that we are not mixing bytes with unicode. */
2355         if (i != 0 && bytesmode != this_bytesmode) {
2356             RAISE_SYNTAX_ERROR("cannot mix bytes and nonbytes literals");
2357             Py_XDECREF(s);
2358             goto error;
2359         }
2360         bytesmode = this_bytesmode;
2361 
2362         if (fstr != NULL) {
2363             assert(s == NULL && !bytesmode);
2364 
2365             int result = _PyPegen_FstringParser_ConcatFstring(p, &state, &fstr, fstr + fstrlen,
2366                                                      this_rawmode, 0, first, t, last);
2367             if (result < 0) {
2368                 goto error;
2369             }
2370         }
2371         else {
2372             /* String or byte string. */
2373             assert(s != NULL && fstr == NULL);
2374             assert(bytesmode ? PyBytes_CheckExact(s) : PyUnicode_CheckExact(s));
2375 
2376             if (bytesmode) {
2377                 if (i == 0) {
2378                     bytes_str = s;
2379                 }
2380                 else {
2381                     PyBytes_ConcatAndDel(&bytes_str, s);
2382                     if (!bytes_str) {
2383                         goto error;
2384                     }
2385                 }
2386             }
2387             else {
2388                 /* This is a regular string. Concatenate it. */
2389                 if (_PyPegen_FstringParser_ConcatAndDel(&state, s) < 0) {
2390                     goto error;
2391                 }
2392             }
2393         }
2394     }
2395 
2396     if (bytesmode) {
2397         if (_PyArena_AddPyObject(p->arena, bytes_str) < 0) {
2398             goto error;
2399         }
2400         return _PyAST_Constant(bytes_str, NULL, first->lineno,
2401                                first->col_offset, last->end_lineno,
2402                                last->end_col_offset, p->arena);
2403     }
2404 
2405     return _PyPegen_FstringParser_Finish(p, &state, first, last);
2406 
2407 error:
2408     Py_XDECREF(bytes_str);
2409     _PyPegen_FstringParser_Dealloc(&state);
2410     if (PyErr_Occurred()) {
2411         raise_decode_error(p);
2412     }
2413     return NULL;
2414 }
2415 
2416 expr_ty
_PyPegen_ensure_imaginary(Parser * p,expr_ty exp)2417 _PyPegen_ensure_imaginary(Parser *p, expr_ty exp)
2418 {
2419     if (exp->kind != Constant_kind || !PyComplex_CheckExact(exp->v.Constant.value)) {
2420         RAISE_SYNTAX_ERROR_KNOWN_LOCATION(exp, "imaginary number required in complex literal");
2421         return NULL;
2422     }
2423     return exp;
2424 }
2425 
2426 expr_ty
_PyPegen_ensure_real(Parser * p,expr_ty exp)2427 _PyPegen_ensure_real(Parser *p, expr_ty exp)
2428 {
2429     if (exp->kind != Constant_kind || PyComplex_CheckExact(exp->v.Constant.value)) {
2430         RAISE_SYNTAX_ERROR_KNOWN_LOCATION(exp, "real number required in complex literal");
2431         return NULL;
2432     }
2433     return exp;
2434 }
2435 
2436 mod_ty
_PyPegen_make_module(Parser * p,asdl_stmt_seq * a)2437 _PyPegen_make_module(Parser *p, asdl_stmt_seq *a) {
2438     asdl_type_ignore_seq *type_ignores = NULL;
2439     Py_ssize_t num = p->type_ignore_comments.num_items;
2440     if (num > 0) {
2441         // Turn the raw (comment, lineno) pairs into TypeIgnore objects in the arena
2442         type_ignores = _Py_asdl_type_ignore_seq_new(num, p->arena);
2443         if (type_ignores == NULL) {
2444             return NULL;
2445         }
2446         for (int i = 0; i < num; i++) {
2447             PyObject *tag = _PyPegen_new_type_comment(p, p->type_ignore_comments.items[i].comment);
2448             if (tag == NULL) {
2449                 return NULL;
2450             }
2451             type_ignore_ty ti = _PyAST_TypeIgnore(p->type_ignore_comments.items[i].lineno,
2452                                                   tag, p->arena);
2453             if (ti == NULL) {
2454                 return NULL;
2455             }
2456             asdl_seq_SET(type_ignores, i, ti);
2457         }
2458     }
2459     return _PyAST_Module(a, type_ignores, p->arena);
2460 }
2461 
2462 // Error reporting helpers
2463 
2464 expr_ty
_PyPegen_get_invalid_target(expr_ty e,TARGETS_TYPE targets_type)2465 _PyPegen_get_invalid_target(expr_ty e, TARGETS_TYPE targets_type)
2466 {
2467     if (e == NULL) {
2468         return NULL;
2469     }
2470 
2471 #define VISIT_CONTAINER(CONTAINER, TYPE) do { \
2472         Py_ssize_t len = asdl_seq_LEN((CONTAINER)->v.TYPE.elts);\
2473         for (Py_ssize_t i = 0; i < len; i++) {\
2474             expr_ty other = asdl_seq_GET((CONTAINER)->v.TYPE.elts, i);\
2475             expr_ty child = _PyPegen_get_invalid_target(other, targets_type);\
2476             if (child != NULL) {\
2477                 return child;\
2478             }\
2479         }\
2480     } while (0)
2481 
2482     // We only need to visit List and Tuple nodes recursively as those
2483     // are the only ones that can contain valid names in targets when
2484     // they are parsed as expressions. Any other kind of expression
2485     // that is a container (like Sets or Dicts) is directly invalid and
2486     // we don't need to visit it recursively.
2487 
2488     switch (e->kind) {
2489         case List_kind:
2490             VISIT_CONTAINER(e, List);
2491             return NULL;
2492         case Tuple_kind:
2493             VISIT_CONTAINER(e, Tuple);
2494             return NULL;
2495         case Starred_kind:
2496             if (targets_type == DEL_TARGETS) {
2497                 return e;
2498             }
2499             return _PyPegen_get_invalid_target(e->v.Starred.value, targets_type);
2500         case Compare_kind:
2501             // This is needed, because the `a in b` in `for a in b` gets parsed
2502             // as a comparison, and so we need to search the left side of the comparison
2503             // for invalid targets.
2504             if (targets_type == FOR_TARGETS) {
2505                 cmpop_ty cmpop = (cmpop_ty) asdl_seq_GET(e->v.Compare.ops, 0);
2506                 if (cmpop == In) {
2507                     return _PyPegen_get_invalid_target(e->v.Compare.left, targets_type);
2508                 }
2509                 return NULL;
2510             }
2511             return e;
2512         case Name_kind:
2513         case Subscript_kind:
2514         case Attribute_kind:
2515             return NULL;
2516         default:
2517             return e;
2518     }
2519 }
2520 
_PyPegen_arguments_parsing_error(Parser * p,expr_ty e)2521 void *_PyPegen_arguments_parsing_error(Parser *p, expr_ty e) {
2522     int kwarg_unpacking = 0;
2523     for (Py_ssize_t i = 0, l = asdl_seq_LEN(e->v.Call.keywords); i < l; i++) {
2524         keyword_ty keyword = asdl_seq_GET(e->v.Call.keywords, i);
2525         if (!keyword->arg) {
2526             kwarg_unpacking = 1;
2527         }
2528     }
2529 
2530     const char *msg = NULL;
2531     if (kwarg_unpacking) {
2532         msg = "positional argument follows keyword argument unpacking";
2533     } else {
2534         msg = "positional argument follows keyword argument";
2535     }
2536 
2537     return RAISE_SYNTAX_ERROR(msg);
2538 }
2539 
2540 
2541 static inline expr_ty
_PyPegen_get_last_comprehension_item(comprehension_ty comprehension)2542 _PyPegen_get_last_comprehension_item(comprehension_ty comprehension) {
2543     if (comprehension->ifs == NULL || asdl_seq_LEN(comprehension->ifs) == 0) {
2544         return comprehension->iter;
2545     }
2546     return PyPegen_last_item(comprehension->ifs, expr_ty);
2547 }
2548 
2549 void *
_PyPegen_nonparen_genexp_in_call(Parser * p,expr_ty args,asdl_comprehension_seq * comprehensions)2550 _PyPegen_nonparen_genexp_in_call(Parser *p, expr_ty args, asdl_comprehension_seq *comprehensions)
2551 {
2552     /* The rule that calls this function is 'args for_if_clauses'.
2553        For the input f(L, x for x in y), L and x are in args and
2554        the for is parsed as a for_if_clause. We have to check if
2555        len <= 1, so that input like dict((a, b) for a, b in x)
2556        gets successfully parsed and then we pass the last
2557        argument (x in the above example) as the location of the
2558        error */
2559     Py_ssize_t len = asdl_seq_LEN(args->v.Call.args);
2560     if (len <= 1) {
2561         return NULL;
2562     }
2563 
2564     comprehension_ty last_comprehension = PyPegen_last_item(comprehensions, comprehension_ty);
2565 
2566     return RAISE_SYNTAX_ERROR_KNOWN_RANGE(
2567         (expr_ty) asdl_seq_GET(args->v.Call.args, len - 1),
2568         _PyPegen_get_last_comprehension_item(last_comprehension),
2569         "Generator expression must be parenthesized"
2570     );
2571 }
2572 
2573 
_PyPegen_collect_call_seqs(Parser * p,asdl_expr_seq * a,asdl_seq * b,int lineno,int col_offset,int end_lineno,int end_col_offset,PyArena * arena)2574 expr_ty _PyPegen_collect_call_seqs(Parser *p, asdl_expr_seq *a, asdl_seq *b,
2575                      int lineno, int col_offset, int end_lineno,
2576                      int end_col_offset, PyArena *arena) {
2577     Py_ssize_t args_len = asdl_seq_LEN(a);
2578     Py_ssize_t total_len = args_len;
2579 
2580     if (b == NULL) {
2581         return _PyAST_Call(_PyPegen_dummy_name(p), a, NULL, lineno, col_offset,
2582                         end_lineno, end_col_offset, arena);
2583 
2584     }
2585 
2586     asdl_expr_seq *starreds = _PyPegen_seq_extract_starred_exprs(p, b);
2587     asdl_keyword_seq *keywords = _PyPegen_seq_delete_starred_exprs(p, b);
2588 
2589     if (starreds) {
2590         total_len += asdl_seq_LEN(starreds);
2591     }
2592 
2593     asdl_expr_seq *args = _Py_asdl_expr_seq_new(total_len, arena);
2594 
2595     Py_ssize_t i = 0;
2596     for (i = 0; i < args_len; i++) {
2597         asdl_seq_SET(args, i, asdl_seq_GET(a, i));
2598     }
2599     for (; i < total_len; i++) {
2600         asdl_seq_SET(args, i, asdl_seq_GET(starreds, i - args_len));
2601     }
2602 
2603     return _PyAST_Call(_PyPegen_dummy_name(p), args, keywords, lineno,
2604                        col_offset, end_lineno, end_col_offset, arena);
2605 }
2606