blob: bd3056e6f2b8009cb11e0f489b5c1e27d496bc16 [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;
37 int lineno, col_offset, end_lineno, end_col_offset;
38 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;
Miss Islington (bot)82da2c32020-05-25 10:58:03 -070075 int level;
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 {
94 asdl_seq *plain_names;
95 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
104typedef struct {
105 operator_ty kind;
106} AugOperator;
107
108typedef struct {
109 void *element;
110 int is_keyword;
111} KeywordOrStarred;
112
113void _PyPegen_clear_memo_statistics(void);
114PyObject *_PyPegen_get_memo_statistics(void);
115
116int _PyPegen_insert_memo(Parser *p, int mark, int type, void *node);
117int _PyPegen_update_memo(Parser *p, int mark, int type, void *node);
118int _PyPegen_is_memoized(Parser *p, int type, void *pres);
119
Pablo Galindo1df5a9e2020-04-23 12:42:13 +0100120int _PyPegen_lookahead_with_name(int, expr_ty (func)(Parser *), Parser *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100121int _PyPegen_lookahead_with_int(int, Token *(func)(Parser *, int), Parser *, int);
122int _PyPegen_lookahead(int, void *(func)(Parser *), Parser *);
123
124Token *_PyPegen_expect_token(Parser *p, int type);
125Token *_PyPegen_get_last_nonnwhitespace_token(Parser *);
126int _PyPegen_fill_token(Parser *p);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100127expr_ty _PyPegen_name_token(Parser *p);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100128expr_ty _PyPegen_number_token(Parser *p);
129void *_PyPegen_string_token(Parser *p);
130const char *_PyPegen_get_expr_name(expr_ty);
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300131void *_PyPegen_raise_error(Parser *p, PyObject *errtype, const char *errmsg, ...);
132void *_PyPegen_raise_error_known_location(Parser *p, PyObject *errtype,
133 int lineno, int col_offset,
134 const char *errmsg, va_list va);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100135void *_PyPegen_dummy_name(Parser *p, ...);
136
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300137Py_LOCAL_INLINE(void *)
138RAISE_ERROR_KNOWN_LOCATION(Parser *p, PyObject *errtype, int lineno,
139 int col_offset, const char *errmsg, ...)
140{
141 va_list va;
142 va_start(va, errmsg);
143 _PyPegen_raise_error_known_location(p, errtype, lineno, col_offset + 1,
144 errmsg, va);
145 va_end(va);
146 return NULL;
147}
148
149
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100150#define UNUSED(expr) do { (void)(expr); } while (0)
151#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 +0100152#define EXTRA _start_lineno, _start_col_offset, _end_lineno, _end_col_offset, p->arena
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300153#define RAISE_SYNTAX_ERROR(msg, ...) _PyPegen_raise_error(p, PyExc_SyntaxError, msg, ##__VA_ARGS__)
154#define RAISE_INDENTATION_ERROR(msg, ...) _PyPegen_raise_error(p, PyExc_IndentationError, msg, ##__VA_ARGS__)
155#define RAISE_SYNTAX_ERROR_KNOWN_LOCATION(a, msg, ...) \
Miss Islington (bot)55c89232020-05-21 18:14:55 -0700156 RAISE_ERROR_KNOWN_LOCATION(p, PyExc_SyntaxError, (a)->lineno, (a)->col_offset, msg, ##__VA_ARGS__)
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100157
158Py_LOCAL_INLINE(void *)
159CHECK_CALL(Parser *p, void *result)
160{
161 if (result == NULL) {
162 assert(PyErr_Occurred());
163 p->error_indicator = 1;
164 }
165 return result;
166}
167
168/* This is needed for helper functions that are allowed to
169 return NULL without an error. Example: _PyPegen_seq_extract_starred_exprs */
170Py_LOCAL_INLINE(void *)
171CHECK_CALL_NULL_ALLOWED(Parser *p, void *result)
172{
173 if (result == NULL && PyErr_Occurred()) {
174 p->error_indicator = 1;
175 }
176 return result;
177}
178
179#define CHECK(result) CHECK_CALL(p, result)
180#define CHECK_NULL_ALLOWED(result) CHECK_CALL_NULL_ALLOWED(p, result)
181
Guido van Rossumc001c092020-04-30 12:12:19 -0700182PyObject *_PyPegen_new_type_comment(Parser *, char *);
183
184Py_LOCAL_INLINE(PyObject *)
185NEW_TYPE_COMMENT(Parser *p, Token *tc)
186{
187 if (tc == NULL) {
188 return NULL;
189 }
190 char *bytes = PyBytes_AsString(tc->bytes);
191 if (bytes == NULL) {
192 goto error;
193 }
194 PyObject *tco = _PyPegen_new_type_comment(p, bytes);
195 if (tco == NULL) {
196 goto error;
197 }
198 return tco;
199 error:
200 p->error_indicator = 1; // Inline CHECK_CALL
201 return NULL;
202}
203
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +0300204Py_LOCAL_INLINE(void *)
205INVALID_VERSION_CHECK(Parser *p, int version, char *msg, void *node)
206{
207 if (node == NULL) {
208 p->error_indicator = 1; // Inline CHECK_CALL
209 return NULL;
210 }
211 if (p->feature_version < version) {
212 p->error_indicator = 1;
Batuhan Taskaya76c1b4d2020-05-01 16:13:43 +0300213 return RAISE_SYNTAX_ERROR("%s only supported in Python 3.%i and greater",
214 msg, version);
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +0300215 }
216 return node;
217}
218
219#define CHECK_VERSION(version, msg, node) INVALID_VERSION_CHECK(p, version, msg, node)
220
Guido van Rossumc001c092020-04-30 12:12:19 -0700221arg_ty _PyPegen_add_type_comment_to_arg(Parser *, arg_ty, Token *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100222PyObject *_PyPegen_new_identifier(Parser *, char *);
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +0300223Parser *_PyPegen_Parser_New(struct tok_state *, int, int, int, int *, PyArena *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100224void _PyPegen_Parser_Free(Parser *);
225mod_ty _PyPegen_run_parser_from_file_pointer(FILE *, int, PyObject *, const char *,
Pablo Galindo2b74c832020-04-27 18:02:07 +0100226 const char *, const char *, PyCompilerFlags *, int *, PyArena *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100227void *_PyPegen_run_parser(Parser *);
Pablo Galindo2b74c832020-04-27 18:02:07 +0100228mod_ty _PyPegen_run_parser_from_file(const char *, int, PyObject *, PyCompilerFlags *, PyArena *);
229mod_ty _PyPegen_run_parser_from_string(const char *, int, PyObject *, PyCompilerFlags *, PyArena *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100230void *_PyPegen_interactive_exit(Parser *);
231asdl_seq *_PyPegen_singleton_seq(Parser *, void *);
232asdl_seq *_PyPegen_seq_insert_in_front(Parser *, void *, asdl_seq *);
Guido van Rossumc001c092020-04-30 12:12:19 -0700233asdl_seq *_PyPegen_seq_append_to_end(Parser *, asdl_seq *, void *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100234asdl_seq *_PyPegen_seq_flatten(Parser *, asdl_seq *);
235expr_ty _PyPegen_join_names_with_dot(Parser *, expr_ty, expr_ty);
236int _PyPegen_seq_count_dots(asdl_seq *);
237alias_ty _PyPegen_alias_for_star(Parser *);
238asdl_seq *_PyPegen_map_names_to_ids(Parser *, asdl_seq *);
239CmpopExprPair *_PyPegen_cmpop_expr_pair(Parser *, cmpop_ty, expr_ty);
240asdl_int_seq *_PyPegen_get_cmpops(Parser *p, asdl_seq *);
241asdl_seq *_PyPegen_get_exprs(Parser *, asdl_seq *);
242expr_ty _PyPegen_set_expr_context(Parser *, expr_ty, expr_context_ty);
243KeyValuePair *_PyPegen_key_value_pair(Parser *, expr_ty, expr_ty);
244asdl_seq *_PyPegen_get_keys(Parser *, asdl_seq *);
245asdl_seq *_PyPegen_get_values(Parser *, asdl_seq *);
Guido van Rossumc001c092020-04-30 12:12:19 -0700246NameDefaultPair *_PyPegen_name_default_pair(Parser *, arg_ty, expr_ty, Token *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100247SlashWithDefault *_PyPegen_slash_with_default(Parser *, asdl_seq *, asdl_seq *);
248StarEtc *_PyPegen_star_etc(Parser *, arg_ty, asdl_seq *, arg_ty);
249arguments_ty _PyPegen_make_arguments(Parser *, asdl_seq *, SlashWithDefault *,
250 asdl_seq *, asdl_seq *, StarEtc *);
251arguments_ty _PyPegen_empty_arguments(Parser *);
252AugOperator *_PyPegen_augoperator(Parser*, operator_ty type);
253stmt_ty _PyPegen_function_def_decorators(Parser *, asdl_seq *, stmt_ty);
254stmt_ty _PyPegen_class_def_decorators(Parser *, asdl_seq *, stmt_ty);
255KeywordOrStarred *_PyPegen_keyword_or_starred(Parser *, void *, int);
256asdl_seq *_PyPegen_seq_extract_starred_exprs(Parser *, asdl_seq *);
257asdl_seq *_PyPegen_seq_delete_starred_exprs(Parser *, asdl_seq *);
258expr_ty _PyPegen_concatenate_strings(Parser *p, asdl_seq *);
259asdl_seq *_PyPegen_join_sequences(Parser *, asdl_seq *, asdl_seq *);
Pablo Galindo2b74c832020-04-27 18:02:07 +0100260int _PyPegen_check_barry_as_flufl(Parser *);
Guido van Rossumc001c092020-04-30 12:12:19 -0700261mod_ty _PyPegen_make_module(Parser *, asdl_seq *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100262
Pablo Galindo16ab0702020-05-15 02:04:52 +0100263// Error reporting helpers
Pablo Galindo16ab0702020-05-15 02:04:52 +0100264expr_ty _PyPegen_get_invalid_target(expr_ty e);
Lysandros Nikolaou75b863a2020-05-18 22:14:47 +0300265void *_PyPegen_arguments_parsing_error(Parser *, expr_ty);
Miss Islington (bot)55c89232020-05-21 18:14:55 -0700266void *_PyPegen_nonparen_genexp_in_call(Parser *p, expr_ty args);
Lysandros Nikolaou75b863a2020-05-18 22:14:47 +0300267
Pablo Galindo16ab0702020-05-15 02:04:52 +0100268
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100269void *_PyPegen_parse(Parser *);
270
271#endif