blob: f82a3a00b2ba01f9ff7e7f27747ddaefe5fd34ed [file] [log] [blame]
Pablo Galindoc5fc1562020-04-22 23:29:27 +01001#ifndef PEGEN_H
2#define PEGEN_H
3
4#define PY_SSIZE_T_CLEAN
5#include <Python.h>
6#include <token.h>
7#include <Python-ast.h>
8#include <pyarena.h>
9
Pablo Galindo2b74c832020-04-27 18:02:07 +010010#if 0
11#define PyPARSE_YIELD_IS_KEYWORD 0x0001
12#endif
13
14#define PyPARSE_DONT_IMPLY_DEDENT 0x0002
15
16#if 0
17#define PyPARSE_WITH_IS_KEYWORD 0x0003
18#define PyPARSE_PRINT_IS_FUNCTION 0x0004
19#define PyPARSE_UNICODE_LITERALS 0x0008
20#endif
21
22#define PyPARSE_IGNORE_COOKIE 0x0010
23#define PyPARSE_BARRY_AS_BDFL 0x0020
24#define PyPARSE_TYPE_COMMENTS 0x0040
25#define PyPARSE_ASYNC_HACKS 0x0080
26
Pablo Galindoc5fc1562020-04-22 23:29:27 +010027typedef struct _memo {
28 int type;
29 void *node;
30 int mark;
31 struct _memo *next;
32} Memo;
33
34typedef struct {
35 int type;
36 PyObject *bytes;
Lysandros Nikolaou861efc62020-06-20 15:57:27 +030037 int lineno, col_offset, end_lineno, end_col_offset;
Pablo Galindoc5fc1562020-04-22 23:29:27 +010038 Memo *memo;
39} Token;
40
41typedef struct {
42 char *str;
43 int type;
44} KeywordToken;
45
Guido van Rossumc001c092020-04-30 12:12:19 -070046
47typedef struct {
48 struct {
49 int lineno;
50 char *comment; // The " <tag>" in "# type: ignore <tag>"
51 } *items;
52 size_t size;
53 size_t num_items;
54} growable_comment_array;
55
Pablo Galindoc5fc1562020-04-22 23:29:27 +010056typedef struct {
57 struct tok_state *tok;
58 Token **tokens;
59 int mark;
60 int fill, size;
61 PyArena *arena;
62 KeywordToken **keywords;
63 int n_keyword_lists;
64 int start_rule;
65 int *errcode;
66 int parsing_started;
67 PyObject* normalize;
68 int starting_lineno;
69 int starting_col_offset;
70 int error_indicator;
Pablo Galindo2b74c832020-04-27 18:02:07 +010071 int flags;
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +030072 int feature_version;
Guido van Rossumc001c092020-04-30 12:12:19 -070073 growable_comment_array type_ignore_comments;
Lysandros Nikolaou2f37c352020-05-07 13:37:51 +030074 Token *known_err_token;
Pablo Galindo800a35c62020-05-25 18:38:45 +010075 int level;
Lysandros Nikolaoubca70142020-10-27 00:42:04 +020076 int call_invalid_rules;
Pablo Galindoc5fc1562020-04-22 23:29:27 +010077} Parser;
78
79typedef struct {
80 cmpop_ty cmpop;
81 expr_ty expr;
82} CmpopExprPair;
83
84typedef struct {
85 expr_ty key;
86 expr_ty value;
87} KeyValuePair;
88
89typedef struct {
90 arg_ty arg;
91 expr_ty value;
92} NameDefaultPair;
93
94typedef struct {
Pablo Galindoa5634c42020-09-16 19:42:00 +010095 asdl_arg_seq *plain_names;
Pablo Galindoc5fc1562020-04-22 23:29:27 +010096 asdl_seq *names_with_defaults; // asdl_seq* of NameDefaultsPair's
97} SlashWithDefault;
98
99typedef struct {
100 arg_ty vararg;
101 asdl_seq *kwonlyargs; // asdl_seq* of NameDefaultsPair's
102 arg_ty kwarg;
103} StarEtc;
104
105typedef struct {
106 operator_ty kind;
107} AugOperator;
108
109typedef struct {
110 void *element;
111 int is_keyword;
112} KeywordOrStarred;
113
114void _PyPegen_clear_memo_statistics(void);
115PyObject *_PyPegen_get_memo_statistics(void);
116
117int _PyPegen_insert_memo(Parser *p, int mark, int type, void *node);
118int _PyPegen_update_memo(Parser *p, int mark, int type, void *node);
119int _PyPegen_is_memoized(Parser *p, int type, void *pres);
120
Pablo Galindo1df5a9e2020-04-23 12:42:13 +0100121int _PyPegen_lookahead_with_name(int, expr_ty (func)(Parser *), Parser *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100122int _PyPegen_lookahead_with_int(int, Token *(func)(Parser *, int), Parser *, int);
Pablo Galindo404b23b2020-05-27 00:15:52 +0100123int _PyPegen_lookahead_with_string(int , expr_ty (func)(Parser *, const char*), Parser *, const char*);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100124int _PyPegen_lookahead(int, void *(func)(Parser *), Parser *);
125
126Token *_PyPegen_expect_token(Parser *p, int type);
Guido van Rossumb45af1a2020-05-26 10:58:44 -0700127expr_ty _PyPegen_expect_soft_keyword(Parser *p, const char *keyword);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100128Token *_PyPegen_get_last_nonnwhitespace_token(Parser *);
129int _PyPegen_fill_token(Parser *p);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100130expr_ty _PyPegen_name_token(Parser *p);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100131expr_ty _PyPegen_number_token(Parser *p);
132void *_PyPegen_string_token(Parser *p);
133const char *_PyPegen_get_expr_name(expr_ty);
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300134void *_PyPegen_raise_error(Parser *p, PyObject *errtype, const char *errmsg, ...);
135void *_PyPegen_raise_error_known_location(Parser *p, PyObject *errtype,
Pablo Galindo51c58962020-06-16 16:49:43 +0100136 Py_ssize_t lineno, Py_ssize_t col_offset,
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300137 const char *errmsg, va_list va);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100138void *_PyPegen_dummy_name(Parser *p, ...);
139
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300140Py_LOCAL_INLINE(void *)
141RAISE_ERROR_KNOWN_LOCATION(Parser *p, PyObject *errtype, int lineno,
142 int col_offset, const char *errmsg, ...)
143{
144 va_list va;
145 va_start(va, errmsg);
146 _PyPegen_raise_error_known_location(p, errtype, lineno, col_offset + 1,
147 errmsg, va);
148 va_end(va);
149 return NULL;
150}
151
152
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100153#define UNUSED(expr) do { (void)(expr); } while (0)
154#define EXTRA_EXPR(head, tail) head->lineno, head->col_offset, tail->end_lineno, tail->end_col_offset, p->arena
Pablo Galindoac7a92c2020-05-10 05:34:50 +0100155#define EXTRA _start_lineno, _start_col_offset, _end_lineno, _end_col_offset, p->arena
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300156#define RAISE_SYNTAX_ERROR(msg, ...) _PyPegen_raise_error(p, PyExc_SyntaxError, msg, ##__VA_ARGS__)
157#define RAISE_INDENTATION_ERROR(msg, ...) _PyPegen_raise_error(p, PyExc_IndentationError, msg, ##__VA_ARGS__)
158#define RAISE_SYNTAX_ERROR_KNOWN_LOCATION(a, msg, ...) \
Lysandros Nikolaouae145832020-05-22 03:56:52 +0300159 RAISE_ERROR_KNOWN_LOCATION(p, PyExc_SyntaxError, (a)->lineno, (a)->col_offset, msg, ##__VA_ARGS__)
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100160
161Py_LOCAL_INLINE(void *)
162CHECK_CALL(Parser *p, void *result)
163{
164 if (result == NULL) {
165 assert(PyErr_Occurred());
166 p->error_indicator = 1;
167 }
168 return result;
169}
170
171/* This is needed for helper functions that are allowed to
172 return NULL without an error. Example: _PyPegen_seq_extract_starred_exprs */
173Py_LOCAL_INLINE(void *)
174CHECK_CALL_NULL_ALLOWED(Parser *p, void *result)
175{
176 if (result == NULL && PyErr_Occurred()) {
177 p->error_indicator = 1;
178 }
179 return result;
180}
181
Lysandros Nikolaou2e5ca9e2020-10-21 22:53:14 +0300182#define CHECK(type, result) ((type) CHECK_CALL(p, result))
183#define CHECK_NULL_ALLOWED(type, result) ((type) CHECK_CALL_NULL_ALLOWED(p, result))
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100184
Guido van Rossumc001c092020-04-30 12:12:19 -0700185PyObject *_PyPegen_new_type_comment(Parser *, char *);
186
187Py_LOCAL_INLINE(PyObject *)
188NEW_TYPE_COMMENT(Parser *p, Token *tc)
189{
190 if (tc == NULL) {
191 return NULL;
192 }
193 char *bytes = PyBytes_AsString(tc->bytes);
194 if (bytes == NULL) {
195 goto error;
196 }
197 PyObject *tco = _PyPegen_new_type_comment(p, bytes);
198 if (tco == NULL) {
199 goto error;
200 }
201 return tco;
202 error:
203 p->error_indicator = 1; // Inline CHECK_CALL
204 return NULL;
205}
206
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +0300207Py_LOCAL_INLINE(void *)
208INVALID_VERSION_CHECK(Parser *p, int version, char *msg, void *node)
209{
210 if (node == NULL) {
211 p->error_indicator = 1; // Inline CHECK_CALL
212 return NULL;
213 }
214 if (p->feature_version < version) {
215 p->error_indicator = 1;
Batuhan Taskaya76c1b4d2020-05-01 16:13:43 +0300216 return RAISE_SYNTAX_ERROR("%s only supported in Python 3.%i and greater",
217 msg, version);
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +0300218 }
219 return node;
220}
221
Lysandros Nikolaou2e5ca9e2020-10-21 22:53:14 +0300222#define CHECK_VERSION(type, version, msg, node) ((type) INVALID_VERSION_CHECK(p, version, msg, node))
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +0300223
Guido van Rossumc001c092020-04-30 12:12:19 -0700224arg_ty _PyPegen_add_type_comment_to_arg(Parser *, arg_ty, Token *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100225PyObject *_PyPegen_new_identifier(Parser *, char *);
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +0300226Parser *_PyPegen_Parser_New(struct tok_state *, int, int, int, int *, PyArena *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100227void _PyPegen_Parser_Free(Parser *);
228mod_ty _PyPegen_run_parser_from_file_pointer(FILE *, int, PyObject *, const char *,
Pablo Galindo2b74c832020-04-27 18:02:07 +0100229 const char *, const char *, PyCompilerFlags *, int *, PyArena *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100230void *_PyPegen_run_parser(Parser *);
Pablo Galindo2b74c832020-04-27 18:02:07 +0100231mod_ty _PyPegen_run_parser_from_file(const char *, int, PyObject *, PyCompilerFlags *, PyArena *);
232mod_ty _PyPegen_run_parser_from_string(const char *, int, PyObject *, PyCompilerFlags *, PyArena *);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100233asdl_stmt_seq *_PyPegen_interactive_exit(Parser *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100234asdl_seq *_PyPegen_singleton_seq(Parser *, void *);
235asdl_seq *_PyPegen_seq_insert_in_front(Parser *, void *, asdl_seq *);
Guido van Rossumc001c092020-04-30 12:12:19 -0700236asdl_seq *_PyPegen_seq_append_to_end(Parser *, asdl_seq *, void *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100237asdl_seq *_PyPegen_seq_flatten(Parser *, asdl_seq *);
238expr_ty _PyPegen_join_names_with_dot(Parser *, expr_ty, expr_ty);
239int _PyPegen_seq_count_dots(asdl_seq *);
240alias_ty _PyPegen_alias_for_star(Parser *);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100241asdl_identifier_seq *_PyPegen_map_names_to_ids(Parser *, asdl_expr_seq *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100242CmpopExprPair *_PyPegen_cmpop_expr_pair(Parser *, cmpop_ty, expr_ty);
243asdl_int_seq *_PyPegen_get_cmpops(Parser *p, asdl_seq *);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100244asdl_expr_seq *_PyPegen_get_exprs(Parser *, asdl_seq *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100245expr_ty _PyPegen_set_expr_context(Parser *, expr_ty, expr_context_ty);
246KeyValuePair *_PyPegen_key_value_pair(Parser *, expr_ty, expr_ty);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100247asdl_expr_seq *_PyPegen_get_keys(Parser *, asdl_seq *);
248asdl_expr_seq *_PyPegen_get_values(Parser *, asdl_seq *);
Guido van Rossumc001c092020-04-30 12:12:19 -0700249NameDefaultPair *_PyPegen_name_default_pair(Parser *, arg_ty, expr_ty, Token *);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100250SlashWithDefault *_PyPegen_slash_with_default(Parser *, asdl_arg_seq *, asdl_seq *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100251StarEtc *_PyPegen_star_etc(Parser *, arg_ty, asdl_seq *, arg_ty);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100252arguments_ty _PyPegen_make_arguments(Parser *, asdl_arg_seq *, SlashWithDefault *,
253 asdl_arg_seq *, asdl_seq *, StarEtc *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100254arguments_ty _PyPegen_empty_arguments(Parser *);
255AugOperator *_PyPegen_augoperator(Parser*, operator_ty type);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100256stmt_ty _PyPegen_function_def_decorators(Parser *, asdl_expr_seq *, stmt_ty);
257stmt_ty _PyPegen_class_def_decorators(Parser *, asdl_expr_seq *, stmt_ty);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100258KeywordOrStarred *_PyPegen_keyword_or_starred(Parser *, void *, int);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100259asdl_expr_seq *_PyPegen_seq_extract_starred_exprs(Parser *, asdl_seq *);
260asdl_keyword_seq *_PyPegen_seq_delete_starred_exprs(Parser *, asdl_seq *);
261expr_ty _PyPegen_collect_call_seqs(Parser *, asdl_expr_seq *, asdl_seq *,
Pablo Galindo315a61f2020-09-03 15:29:32 +0100262 int lineno, int col_offset, int end_lineno,
263 int end_col_offset, PyArena *arena);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100264expr_ty _PyPegen_concatenate_strings(Parser *p, asdl_seq *);
265asdl_seq *_PyPegen_join_sequences(Parser *, asdl_seq *, asdl_seq *);
Pablo Galindo06f8c332020-10-30 23:48:42 +0000266int _PyPegen_check_barry_as_flufl(Parser *, Token *);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100267mod_ty _PyPegen_make_module(Parser *, asdl_stmt_seq *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100268
Pablo Galindo16ab0702020-05-15 02:04:52 +0100269// Error reporting helpers
Lysandros Nikolaou01ece632020-06-19 02:10:43 +0300270typedef enum {
271 STAR_TARGETS,
272 DEL_TARGETS,
273 FOR_TARGETS
274} TARGETS_TYPE;
275expr_ty _PyPegen_get_invalid_target(expr_ty e, TARGETS_TYPE targets_type);
Lysandros Nikolaou6c4e0bd2020-06-21 05:18:01 +0300276#define RAISE_SYNTAX_ERROR_INVALID_TARGET(type, e) _RAISE_SYNTAX_ERROR_INVALID_TARGET(p, type, e)
277
278Py_LOCAL_INLINE(void *)
279_RAISE_SYNTAX_ERROR_INVALID_TARGET(Parser *p, TARGETS_TYPE type, void *e)
280{
Lysandros Nikolaou2e5ca9e2020-10-21 22:53:14 +0300281 expr_ty invalid_target = CHECK_NULL_ALLOWED(expr_ty, _PyPegen_get_invalid_target(e, type));
Lysandros Nikolaou6c4e0bd2020-06-21 05:18:01 +0300282 if (invalid_target != NULL) {
283 const char *msg;
284 if (type == STAR_TARGETS || type == FOR_TARGETS) {
285 msg = "cannot assign to %s";
286 }
287 else {
288 msg = "cannot delete %s";
289 }
290 return RAISE_SYNTAX_ERROR_KNOWN_LOCATION(
291 invalid_target,
292 msg,
293 _PyPegen_get_expr_name(invalid_target)
294 );
295 }
296 return RAISE_SYNTAX_ERROR("invalid syntax");
297}
Lysandros Nikolaou01ece632020-06-19 02:10:43 +0300298
Lysandros Nikolaou75b863a2020-05-18 22:14:47 +0300299void *_PyPegen_arguments_parsing_error(Parser *, expr_ty);
Lysandros Nikolaouae145832020-05-22 03:56:52 +0300300void *_PyPegen_nonparen_genexp_in_call(Parser *p, expr_ty args);
Lysandros Nikolaou75b863a2020-05-18 22:14:47 +0300301
Pablo Galindo16ab0702020-05-15 02:04:52 +0100302
Lysandros Nikolaou01ece632020-06-19 02:10:43 +0300303// Generated function in parse.c - function definition in python.gram
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100304void *_PyPegen_parse(Parser *);
305
306#endif