blob: 761e90f06db8e6f343389247aa6978d59a485876 [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;
Pablo Galindoc5fc1562020-04-22 23:29:27 +010075} Parser;
76
77typedef struct {
78 cmpop_ty cmpop;
79 expr_ty expr;
80} CmpopExprPair;
81
82typedef struct {
83 expr_ty key;
84 expr_ty value;
85} KeyValuePair;
86
87typedef struct {
88 arg_ty arg;
89 expr_ty value;
90} NameDefaultPair;
91
92typedef struct {
93 asdl_seq *plain_names;
94 asdl_seq *names_with_defaults; // asdl_seq* of NameDefaultsPair's
95} SlashWithDefault;
96
97typedef struct {
98 arg_ty vararg;
99 asdl_seq *kwonlyargs; // asdl_seq* of NameDefaultsPair's
100 arg_ty kwarg;
101} StarEtc;
102
103typedef struct {
104 operator_ty kind;
105} AugOperator;
106
107typedef struct {
108 void *element;
109 int is_keyword;
110} KeywordOrStarred;
111
112void _PyPegen_clear_memo_statistics(void);
113PyObject *_PyPegen_get_memo_statistics(void);
114
115int _PyPegen_insert_memo(Parser *p, int mark, int type, void *node);
116int _PyPegen_update_memo(Parser *p, int mark, int type, void *node);
117int _PyPegen_is_memoized(Parser *p, int type, void *pres);
118
Pablo Galindo1df5a9e2020-04-23 12:42:13 +0100119int _PyPegen_lookahead_with_name(int, expr_ty (func)(Parser *), Parser *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100120int _PyPegen_lookahead_with_int(int, Token *(func)(Parser *, int), Parser *, int);
121int _PyPegen_lookahead(int, void *(func)(Parser *), Parser *);
122
123Token *_PyPegen_expect_token(Parser *p, int type);
124Token *_PyPegen_get_last_nonnwhitespace_token(Parser *);
125int _PyPegen_fill_token(Parser *p);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100126expr_ty _PyPegen_name_token(Parser *p);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100127expr_ty _PyPegen_number_token(Parser *p);
128void *_PyPegen_string_token(Parser *p);
129const char *_PyPegen_get_expr_name(expr_ty);
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300130void *_PyPegen_raise_error(Parser *p, PyObject *errtype, const char *errmsg, ...);
131void *_PyPegen_raise_error_known_location(Parser *p, PyObject *errtype,
132 int lineno, int col_offset,
133 const char *errmsg, va_list va);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100134void *_PyPegen_dummy_name(Parser *p, ...);
135
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300136Py_LOCAL_INLINE(void *)
137RAISE_ERROR_KNOWN_LOCATION(Parser *p, PyObject *errtype, int lineno,
138 int col_offset, const char *errmsg, ...)
139{
140 va_list va;
141 va_start(va, errmsg);
142 _PyPegen_raise_error_known_location(p, errtype, lineno, col_offset + 1,
143 errmsg, va);
144 va_end(va);
145 return NULL;
146}
147
148
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100149#define UNUSED(expr) do { (void)(expr); } while (0)
150#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 +0100151#define EXTRA _start_lineno, _start_col_offset, _end_lineno, _end_col_offset, p->arena
Lysandros Nikolaoua15c9b32020-05-13 22:36:27 +0300152#define RAISE_SYNTAX_ERROR(msg, ...) _PyPegen_raise_error(p, PyExc_SyntaxError, msg, ##__VA_ARGS__)
153#define RAISE_INDENTATION_ERROR(msg, ...) _PyPegen_raise_error(p, PyExc_IndentationError, msg, ##__VA_ARGS__)
154#define RAISE_SYNTAX_ERROR_KNOWN_LOCATION(a, msg, ...) \
Miss Islington (bot)55c89232020-05-21 18:14:55 -0700155 RAISE_ERROR_KNOWN_LOCATION(p, PyExc_SyntaxError, (a)->lineno, (a)->col_offset, msg, ##__VA_ARGS__)
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100156
157Py_LOCAL_INLINE(void *)
158CHECK_CALL(Parser *p, void *result)
159{
160 if (result == NULL) {
161 assert(PyErr_Occurred());
162 p->error_indicator = 1;
163 }
164 return result;
165}
166
167/* This is needed for helper functions that are allowed to
168 return NULL without an error. Example: _PyPegen_seq_extract_starred_exprs */
169Py_LOCAL_INLINE(void *)
170CHECK_CALL_NULL_ALLOWED(Parser *p, void *result)
171{
172 if (result == NULL && PyErr_Occurred()) {
173 p->error_indicator = 1;
174 }
175 return result;
176}
177
178#define CHECK(result) CHECK_CALL(p, result)
179#define CHECK_NULL_ALLOWED(result) CHECK_CALL_NULL_ALLOWED(p, result)
180
Guido van Rossumc001c092020-04-30 12:12:19 -0700181PyObject *_PyPegen_new_type_comment(Parser *, char *);
182
183Py_LOCAL_INLINE(PyObject *)
184NEW_TYPE_COMMENT(Parser *p, Token *tc)
185{
186 if (tc == NULL) {
187 return NULL;
188 }
189 char *bytes = PyBytes_AsString(tc->bytes);
190 if (bytes == NULL) {
191 goto error;
192 }
193 PyObject *tco = _PyPegen_new_type_comment(p, bytes);
194 if (tco == NULL) {
195 goto error;
196 }
197 return tco;
198 error:
199 p->error_indicator = 1; // Inline CHECK_CALL
200 return NULL;
201}
202
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +0300203Py_LOCAL_INLINE(void *)
204INVALID_VERSION_CHECK(Parser *p, int version, char *msg, void *node)
205{
206 if (node == NULL) {
207 p->error_indicator = 1; // Inline CHECK_CALL
208 return NULL;
209 }
210 if (p->feature_version < version) {
211 p->error_indicator = 1;
Batuhan Taskaya76c1b4d2020-05-01 16:13:43 +0300212 return RAISE_SYNTAX_ERROR("%s only supported in Python 3.%i and greater",
213 msg, version);
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +0300214 }
215 return node;
216}
217
218#define CHECK_VERSION(version, msg, node) INVALID_VERSION_CHECK(p, version, msg, node)
219
Guido van Rossumc001c092020-04-30 12:12:19 -0700220arg_ty _PyPegen_add_type_comment_to_arg(Parser *, arg_ty, Token *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100221PyObject *_PyPegen_new_identifier(Parser *, char *);
Lysandros Nikolaou3e0a6f32020-05-01 06:27:52 +0300222Parser *_PyPegen_Parser_New(struct tok_state *, int, int, int, int *, PyArena *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100223void _PyPegen_Parser_Free(Parser *);
224mod_ty _PyPegen_run_parser_from_file_pointer(FILE *, int, PyObject *, const char *,
Pablo Galindo2b74c832020-04-27 18:02:07 +0100225 const char *, const char *, PyCompilerFlags *, int *, PyArena *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100226void *_PyPegen_run_parser(Parser *);
Pablo Galindo2b74c832020-04-27 18:02:07 +0100227mod_ty _PyPegen_run_parser_from_file(const char *, int, PyObject *, PyCompilerFlags *, PyArena *);
228mod_ty _PyPegen_run_parser_from_string(const char *, int, PyObject *, PyCompilerFlags *, PyArena *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100229void *_PyPegen_interactive_exit(Parser *);
230asdl_seq *_PyPegen_singleton_seq(Parser *, void *);
231asdl_seq *_PyPegen_seq_insert_in_front(Parser *, void *, asdl_seq *);
Guido van Rossumc001c092020-04-30 12:12:19 -0700232asdl_seq *_PyPegen_seq_append_to_end(Parser *, asdl_seq *, void *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100233asdl_seq *_PyPegen_seq_flatten(Parser *, asdl_seq *);
234expr_ty _PyPegen_join_names_with_dot(Parser *, expr_ty, expr_ty);
235int _PyPegen_seq_count_dots(asdl_seq *);
236alias_ty _PyPegen_alias_for_star(Parser *);
237asdl_seq *_PyPegen_map_names_to_ids(Parser *, asdl_seq *);
238CmpopExprPair *_PyPegen_cmpop_expr_pair(Parser *, cmpop_ty, expr_ty);
239asdl_int_seq *_PyPegen_get_cmpops(Parser *p, asdl_seq *);
240asdl_seq *_PyPegen_get_exprs(Parser *, asdl_seq *);
241expr_ty _PyPegen_set_expr_context(Parser *, expr_ty, expr_context_ty);
242KeyValuePair *_PyPegen_key_value_pair(Parser *, expr_ty, expr_ty);
243asdl_seq *_PyPegen_get_keys(Parser *, asdl_seq *);
244asdl_seq *_PyPegen_get_values(Parser *, asdl_seq *);
Guido van Rossumc001c092020-04-30 12:12:19 -0700245NameDefaultPair *_PyPegen_name_default_pair(Parser *, arg_ty, expr_ty, Token *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100246SlashWithDefault *_PyPegen_slash_with_default(Parser *, asdl_seq *, asdl_seq *);
247StarEtc *_PyPegen_star_etc(Parser *, arg_ty, asdl_seq *, arg_ty);
248arguments_ty _PyPegen_make_arguments(Parser *, asdl_seq *, SlashWithDefault *,
249 asdl_seq *, asdl_seq *, StarEtc *);
250arguments_ty _PyPegen_empty_arguments(Parser *);
251AugOperator *_PyPegen_augoperator(Parser*, operator_ty type);
252stmt_ty _PyPegen_function_def_decorators(Parser *, asdl_seq *, stmt_ty);
253stmt_ty _PyPegen_class_def_decorators(Parser *, asdl_seq *, stmt_ty);
254KeywordOrStarred *_PyPegen_keyword_or_starred(Parser *, void *, int);
255asdl_seq *_PyPegen_seq_extract_starred_exprs(Parser *, asdl_seq *);
256asdl_seq *_PyPegen_seq_delete_starred_exprs(Parser *, asdl_seq *);
257expr_ty _PyPegen_concatenate_strings(Parser *p, asdl_seq *);
258asdl_seq *_PyPegen_join_sequences(Parser *, asdl_seq *, asdl_seq *);
Pablo Galindo2b74c832020-04-27 18:02:07 +0100259int _PyPegen_check_barry_as_flufl(Parser *);
Guido van Rossumc001c092020-04-30 12:12:19 -0700260mod_ty _PyPegen_make_module(Parser *, asdl_seq *);
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100261
Pablo Galindo16ab0702020-05-15 02:04:52 +0100262// Error reporting helpers
Pablo Galindo16ab0702020-05-15 02:04:52 +0100263expr_ty _PyPegen_get_invalid_target(expr_ty e);
Lysandros Nikolaou75b863a2020-05-18 22:14:47 +0300264void *_PyPegen_arguments_parsing_error(Parser *, expr_ty);
Miss Islington (bot)55c89232020-05-21 18:14:55 -0700265void *_PyPegen_nonparen_genexp_in_call(Parser *p, expr_ty args);
Lysandros Nikolaou75b863a2020-05-18 22:14:47 +0300266
Pablo Galindo16ab0702020-05-15 02:04:52 +0100267
Pablo Galindoc5fc1562020-04-22 23:29:27 +0100268void *_PyPegen_parse(Parser *);
269
270#endif