blob: af160d6a43d6b0236d1481c0d4b58228b57933dd [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>
Victor Stinner94faa072021-03-23 20:47:40 +01007#include <pycore_ast.h>
Pablo Galindoc5fc1562020-04-22 23:29:27 +01008
Pablo Galindo2b74c832020-04-27 18:02:07 +01009#if 0
10#define PyPARSE_YIELD_IS_KEYWORD 0x0001
11#endif
12
13#define PyPARSE_DONT_IMPLY_DEDENT 0x0002
14
15#if 0
16#define PyPARSE_WITH_IS_KEYWORD 0x0003
17#define PyPARSE_PRINT_IS_FUNCTION 0x0004
18#define PyPARSE_UNICODE_LITERALS 0x0008
19#endif
20
21#define PyPARSE_IGNORE_COOKIE 0x0010
22#define PyPARSE_BARRY_AS_BDFL 0x0020
23#define PyPARSE_TYPE_COMMENTS 0x0040
24#define PyPARSE_ASYNC_HACKS 0x0080
25
Pablo Galindoc5fc1562020-04-22 23:29:27 +010026typedef struct _memo {
27 int type;
28 void *node;
29 int mark;
30 struct _memo *next;
31} Memo;
32
33typedef struct {
34 int type;
35 PyObject *bytes;
Lysandros Nikolaou861efc62020-06-20 15:57:27 +030036 int lineno, col_offset, end_lineno, end_col_offset;
Pablo Galindoc5fc1562020-04-22 23:29:27 +010037 Memo *memo;
38} Token;
39
40typedef struct {
41 char *str;
42 int type;
43} KeywordToken;
44
Guido van Rossumc001c092020-04-30 12:12:19 -070045
46typedef struct {
47 struct {
48 int lineno;
49 char *comment; // The " <tag>" in "# type: ignore <tag>"
50 } *items;
51 size_t size;
52 size_t num_items;
53} growable_comment_array;
54
Pablo Galindoc5fc1562020-04-22 23:29:27 +010055typedef struct {
56 struct tok_state *tok;
57 Token **tokens;
58 int mark;
59 int fill, size;
60 PyArena *arena;
61 KeywordToken **keywords;
62 int n_keyword_lists;
63 int start_rule;
64 int *errcode;
65 int parsing_started;
66 PyObject* normalize;
67 int starting_lineno;
68 int starting_col_offset;
69 int error_indicator;
Pablo Galindo2b74c832020-04-27 18:02:07 +010070 int flags;
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +030071 int feature_version;
Guido van Rossumc001c092020-04-30 12:12:19 -070072 growable_comment_array type_ignore_comments;
Lysandros Nikolaou2f37c352020-05-07 13:37:51 +030073 Token *known_err_token;
Pablo Galindo800a35c62020-05-25 18:38:45 +010074 int level;
Lysandros Nikolaoubca70142020-10-27 00:42:04 +020075 int call_invalid_rules;
Pablo Galindoc5fc1562020-04-22 23:29:27 +010076} Parser;
77
78typedef struct {
79 cmpop_ty cmpop;
80 expr_ty expr;
81} CmpopExprPair;
82
83typedef struct {
84 expr_ty key;
85 expr_ty value;
86} KeyValuePair;
87
88typedef struct {
89 arg_ty arg;
90 expr_ty value;
91} NameDefaultPair;
92
93typedef struct {
Pablo Galindoa5634c42020-09-16 19:42:00 +010094 asdl_arg_seq *plain_names;
Pablo Galindoc5fc1562020-04-22 23:29:27 +010095 asdl_seq *names_with_defaults; // asdl_seq* of NameDefaultsPair's
96} SlashWithDefault;
97
98typedef struct {
99 arg_ty vararg;
100 asdl_seq *kwonlyargs; // asdl_seq* of NameDefaultsPair's
101 arg_ty kwarg;
102} StarEtc;
103
Victor Stinner94faa072021-03-23 20:47:40 +0100104typedef struct { operator_ty kind; } AugOperator;
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100105typedef struct {
106 void *element;
107 int is_keyword;
108} KeywordOrStarred;
109
110void _PyPegen_clear_memo_statistics(void);
111PyObject *_PyPegen_get_memo_statistics(void);
112
113int _PyPegen_insert_memo(Parser *p, int mark, int type, void *node);
114int _PyPegen_update_memo(Parser *p, int mark, int type, void *node);
115int _PyPegen_is_memoized(Parser *p, int type, void *pres);
116
Pablo Galindo58fb1562021-02-02 19:54:22 +0000117
Pablo Galindo1df5a9e2020-04-23 12:42:13 +0100118int _PyPegen_lookahead_with_name(int, expr_ty (func)(Parser *), Parser *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100119int _PyPegen_lookahead_with_int(int, Token *(func)(Parser *, int), Parser *, int);
Pablo Galindo404b23b2020-05-27 00:15:52 +0100120int _PyPegen_lookahead_with_string(int , expr_ty (func)(Parser *, const char*), Parser *, const char*);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100121int _PyPegen_lookahead(int, void *(func)(Parser *), Parser *);
122
123Token *_PyPegen_expect_token(Parser *p, int type);
Pablo Galindo58fb1562021-02-02 19:54:22 +0000124Token *_PyPegen_expect_forced_token(Parser *p, int type, const char* expected);
Guido van Rossumb45af1a2020-05-26 10:58:44 -0700125expr_ty _PyPegen_expect_soft_keyword(Parser *p, const char *keyword);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100126Token *_PyPegen_get_last_nonnwhitespace_token(Parser *);
127int _PyPegen_fill_token(Parser *p);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100128expr_ty _PyPegen_name_token(Parser *p);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100129expr_ty _PyPegen_number_token(Parser *p);
130void *_PyPegen_string_token(Parser *p);
131const char *_PyPegen_get_expr_name(expr_ty);
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300132void *_PyPegen_raise_error(Parser *p, PyObject *errtype, const char *errmsg, ...);
133void *_PyPegen_raise_error_known_location(Parser *p, PyObject *errtype,
Pablo Galindo51c58962020-06-16 16:49:43 +0100134 Py_ssize_t lineno, Py_ssize_t col_offset,
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300135 const char *errmsg, va_list va);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100136void *_PyPegen_dummy_name(Parser *p, ...);
137
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300138Py_LOCAL_INLINE(void *)
Pablo Galindo96eeff52021-03-22 17:28:11 +0000139RAISE_ERROR_KNOWN_LOCATION(Parser *p, PyObject *errtype,
140 Py_ssize_t lineno, Py_ssize_t col_offset,
141 const char *errmsg, ...)
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300142{
143 va_list va;
144 va_start(va, errmsg);
145 _PyPegen_raise_error_known_location(p, errtype, lineno, col_offset + 1,
146 errmsg, va);
147 va_end(va);
148 return NULL;
149}
150
151
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100152#define UNUSED(expr) do { (void)(expr); } while (0)
153#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 +0100154#define EXTRA _start_lineno, _start_col_offset, _end_lineno, _end_col_offset, p->arena
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300155#define RAISE_SYNTAX_ERROR(msg, ...) _PyPegen_raise_error(p, PyExc_SyntaxError, msg, ##__VA_ARGS__)
156#define RAISE_INDENTATION_ERROR(msg, ...) _PyPegen_raise_error(p, PyExc_IndentationError, msg, ##__VA_ARGS__)
157#define RAISE_SYNTAX_ERROR_KNOWN_LOCATION(a, msg, ...) \
Lysandros Nikolaouae145832020-05-22 03:56:52 +0300158 RAISE_ERROR_KNOWN_LOCATION(p, PyExc_SyntaxError, (a)->lineno, (a)->col_offset, msg, ##__VA_ARGS__)
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100159
160Py_LOCAL_INLINE(void *)
161CHECK_CALL(Parser *p, void *result)
162{
163 if (result == NULL) {
164 assert(PyErr_Occurred());
165 p->error_indicator = 1;
166 }
167 return result;
168}
169
170/* This is needed for helper functions that are allowed to
171 return NULL without an error. Example: _PyPegen_seq_extract_starred_exprs */
172Py_LOCAL_INLINE(void *)
173CHECK_CALL_NULL_ALLOWED(Parser *p, void *result)
174{
175 if (result == NULL && PyErr_Occurred()) {
176 p->error_indicator = 1;
177 }
178 return result;
179}
180
Lysandros Nikolaou2e5ca9e2020-10-21 22:53:14 +0300181#define CHECK(type, result) ((type) CHECK_CALL(p, result))
182#define CHECK_NULL_ALLOWED(type, result) ((type) CHECK_CALL_NULL_ALLOWED(p, result))
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100183
Guido van Rossumc001c092020-04-30 12:12:19 -0700184PyObject *_PyPegen_new_type_comment(Parser *, char *);
185
186Py_LOCAL_INLINE(PyObject *)
187NEW_TYPE_COMMENT(Parser *p, Token *tc)
188{
189 if (tc == NULL) {
190 return NULL;
191 }
192 char *bytes = PyBytes_AsString(tc->bytes);
193 if (bytes == NULL) {
194 goto error;
195 }
196 PyObject *tco = _PyPegen_new_type_comment(p, bytes);
197 if (tco == NULL) {
198 goto error;
199 }
200 return tco;
201 error:
202 p->error_indicator = 1; // Inline CHECK_CALL
203 return NULL;
204}
205
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +0300206Py_LOCAL_INLINE(void *)
207INVALID_VERSION_CHECK(Parser *p, int version, char *msg, void *node)
208{
209 if (node == NULL) {
210 p->error_indicator = 1; // Inline CHECK_CALL
211 return NULL;
212 }
213 if (p->feature_version < version) {
214 p->error_indicator = 1;
Batuhan Taskaya76c1b4d2020-05-01 16:13:43 +0300215 return RAISE_SYNTAX_ERROR("%s only supported in Python 3.%i and greater",
216 msg, version);
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +0300217 }
218 return node;
219}
220
Lysandros Nikolaou2e5ca9e2020-10-21 22:53:14 +0300221#define CHECK_VERSION(type, version, msg, node) ((type) INVALID_VERSION_CHECK(p, version, msg, node))
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +0300222
Guido van Rossumc001c092020-04-30 12:12:19 -0700223arg_ty _PyPegen_add_type_comment_to_arg(Parser *, arg_ty, Token *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100224PyObject *_PyPegen_new_identifier(Parser *, char *);
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +0300225Parser *_PyPegen_Parser_New(struct tok_state *, int, int, int, int *, PyArena *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100226void _PyPegen_Parser_Free(Parser *);
227mod_ty _PyPegen_run_parser_from_file_pointer(FILE *, int, PyObject *, const char *,
Pablo Galindo2b74c832020-04-27 18:02:07 +0100228 const char *, const char *, PyCompilerFlags *, int *, PyArena *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100229void *_PyPegen_run_parser(Parser *);
Pablo Galindo2b74c832020-04-27 18:02:07 +0100230mod_ty _PyPegen_run_parser_from_string(const char *, int, PyObject *, PyCompilerFlags *, PyArena *);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100231asdl_stmt_seq *_PyPegen_interactive_exit(Parser *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100232asdl_seq *_PyPegen_singleton_seq(Parser *, void *);
233asdl_seq *_PyPegen_seq_insert_in_front(Parser *, void *, asdl_seq *);
Guido van Rossumc001c092020-04-30 12:12:19 -0700234asdl_seq *_PyPegen_seq_append_to_end(Parser *, asdl_seq *, void *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100235asdl_seq *_PyPegen_seq_flatten(Parser *, asdl_seq *);
236expr_ty _PyPegen_join_names_with_dot(Parser *, expr_ty, expr_ty);
237int _PyPegen_seq_count_dots(asdl_seq *);
238alias_ty _PyPegen_alias_for_star(Parser *);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100239asdl_identifier_seq *_PyPegen_map_names_to_ids(Parser *, asdl_expr_seq *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100240CmpopExprPair *_PyPegen_cmpop_expr_pair(Parser *, cmpop_ty, expr_ty);
241asdl_int_seq *_PyPegen_get_cmpops(Parser *p, asdl_seq *);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100242asdl_expr_seq *_PyPegen_get_exprs(Parser *, asdl_seq *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100243expr_ty _PyPegen_set_expr_context(Parser *, expr_ty, expr_context_ty);
244KeyValuePair *_PyPegen_key_value_pair(Parser *, expr_ty, expr_ty);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100245asdl_expr_seq *_PyPegen_get_keys(Parser *, asdl_seq *);
246asdl_expr_seq *_PyPegen_get_values(Parser *, asdl_seq *);
Guido van Rossumc001c092020-04-30 12:12:19 -0700247NameDefaultPair *_PyPegen_name_default_pair(Parser *, arg_ty, expr_ty, Token *);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100248SlashWithDefault *_PyPegen_slash_with_default(Parser *, asdl_arg_seq *, asdl_seq *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100249StarEtc *_PyPegen_star_etc(Parser *, arg_ty, asdl_seq *, arg_ty);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100250arguments_ty _PyPegen_make_arguments(Parser *, asdl_arg_seq *, SlashWithDefault *,
251 asdl_arg_seq *, asdl_seq *, StarEtc *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100252arguments_ty _PyPegen_empty_arguments(Parser *);
253AugOperator *_PyPegen_augoperator(Parser*, operator_ty type);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100254stmt_ty _PyPegen_function_def_decorators(Parser *, asdl_expr_seq *, stmt_ty);
255stmt_ty _PyPegen_class_def_decorators(Parser *, asdl_expr_seq *, stmt_ty);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100256KeywordOrStarred *_PyPegen_keyword_or_starred(Parser *, void *, int);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100257asdl_expr_seq *_PyPegen_seq_extract_starred_exprs(Parser *, asdl_seq *);
258asdl_keyword_seq *_PyPegen_seq_delete_starred_exprs(Parser *, asdl_seq *);
259expr_ty _PyPegen_collect_call_seqs(Parser *, asdl_expr_seq *, asdl_seq *,
Pablo Galindo315a61f2020-09-03 15:29:32 +0100260 int lineno, int col_offset, int end_lineno,
261 int end_col_offset, PyArena *arena);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100262expr_ty _PyPegen_concatenate_strings(Parser *p, asdl_seq *);
263asdl_seq *_PyPegen_join_sequences(Parser *, asdl_seq *, asdl_seq *);
Pablo Galindo06f8c332020-10-30 23:48:42 +0000264int _PyPegen_check_barry_as_flufl(Parser *, Token *);
Pablo Galindoa5634c42020-09-16 19:42:00 +0100265mod_ty _PyPegen_make_module(Parser *, asdl_stmt_seq *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100266
Pablo Galindo16ab0702020-05-15 02:04:52 +0100267// Error reporting helpers
Lysandros Nikolaou01ece632020-06-19 02:10:43 +0300268typedef enum {
269 STAR_TARGETS,
270 DEL_TARGETS,
271 FOR_TARGETS
272} TARGETS_TYPE;
273expr_ty _PyPegen_get_invalid_target(expr_ty e, TARGETS_TYPE targets_type);
Lysandros Nikolaou6c4e0bd2020-06-21 05:18:01 +0300274#define RAISE_SYNTAX_ERROR_INVALID_TARGET(type, e) _RAISE_SYNTAX_ERROR_INVALID_TARGET(p, type, e)
275
276Py_LOCAL_INLINE(void *)
277_RAISE_SYNTAX_ERROR_INVALID_TARGET(Parser *p, TARGETS_TYPE type, void *e)
278{
Lysandros Nikolaou2e5ca9e2020-10-21 22:53:14 +0300279 expr_ty invalid_target = CHECK_NULL_ALLOWED(expr_ty, _PyPegen_get_invalid_target(e, type));
Lysandros Nikolaou6c4e0bd2020-06-21 05:18:01 +0300280 if (invalid_target != NULL) {
281 const char *msg;
282 if (type == STAR_TARGETS || type == FOR_TARGETS) {
283 msg = "cannot assign to %s";
284 }
285 else {
286 msg = "cannot delete %s";
287 }
288 return RAISE_SYNTAX_ERROR_KNOWN_LOCATION(
289 invalid_target,
290 msg,
291 _PyPegen_get_expr_name(invalid_target)
292 );
293 }
294 return RAISE_SYNTAX_ERROR("invalid syntax");
295}
Lysandros Nikolaou01ece632020-06-19 02:10:43 +0300296
Lysandros Nikolaou75b863a2020-05-18 22:14:47 +0300297void *_PyPegen_arguments_parsing_error(Parser *, expr_ty);
Lysandros Nikolaouae145832020-05-22 03:56:52 +0300298void *_PyPegen_nonparen_genexp_in_call(Parser *p, expr_ty args);
Lysandros Nikolaou75b863a2020-05-18 22:14:47 +0300299
Pablo Galindo16ab0702020-05-15 02:04:52 +0100300
Lysandros Nikolaou01ece632020-06-19 02:10:43 +0300301// Generated function in parse.c - function definition in python.gram
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100302void *_PyPegen_parse(Parser *);
303
304#endif