blob: ed4fe7b760f0b915d8ab8050f67459f1f0abd072 [file] [log] [blame]
Guido van Rossumf70e43a1991-02-19 12:39:46 +00001
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00002/* Parser-tokenizer link implementation */
3
Guido van Rossum3f5da241990-12-20 15:06:42 +00004#include "pgenheaders.h"
Guido van Rossum85a5fbb1990-10-14 12:07:46 +00005#include "tokenizer.h"
6#include "node.h"
7#include "grammar.h"
8#include "parser.h"
Guido van Rossum3f5da241990-12-20 15:06:42 +00009#include "parsetok.h"
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000010#include "errcode.h"
11
Guido van Rossum6135df61998-04-10 19:35:06 +000012int Py_TabcheckFlag;
13
Guido van Rossum3f5da241990-12-20 15:06:42 +000014
15/* Forward */
Tim Petersfe2127d2001-07-16 05:37:24 +000016static node *parsetok(struct tok_state *, grammar *, int, perrdetail *, int);
Martin v. Löwis16eff6f2002-01-05 21:40:08 +000017static void initerr(perrdetail *err_ret, char* filename);
Guido van Rossum3f5da241990-12-20 15:06:42 +000018
19/* Parse input coming from a string. Return error code, print some errors. */
Guido van Rossumbd0389d1994-08-29 12:25:45 +000020node *
Thomas Wouters23c9e002000-07-22 19:20:54 +000021PyParser_ParseString(char *s, grammar *g, int start, perrdetail *err_ret)
Guido van Rossum3f5da241990-12-20 15:06:42 +000022{
Tim Petersfe2127d2001-07-16 05:37:24 +000023 return PyParser_ParseStringFlags(s, g, start, err_ret, 0);
24}
25
26node *
27PyParser_ParseStringFlags(char *s, grammar *g, int start,
28 perrdetail *err_ret, int flags)
29{
Guido van Rossumbd0389d1994-08-29 12:25:45 +000030 struct tok_state *tok;
31
Martin v. Löwis16eff6f2002-01-05 21:40:08 +000032 initerr(err_ret, NULL);
Guido van Rossumbd0389d1994-08-29 12:25:45 +000033
Guido van Rossum86bea461997-04-29 21:03:06 +000034 if ((tok = PyTokenizer_FromString(s)) == NULL) {
Guido van Rossumbd0389d1994-08-29 12:25:45 +000035 err_ret->error = E_NOMEM;
36 return NULL;
Guido van Rossum3f5da241990-12-20 15:06:42 +000037 }
Guido van Rossumbd0389d1994-08-29 12:25:45 +000038
Guido van Rossum89ce4541998-12-21 18:32:40 +000039 if (Py_TabcheckFlag || Py_VerboseFlag) {
40 tok->filename = "<string>";
41 tok->altwarning = (tok->filename != NULL);
42 if (Py_TabcheckFlag >= 2)
43 tok->alterror++;
44 }
45
Tim Petersfe2127d2001-07-16 05:37:24 +000046 return parsetok(tok, g, start, err_ret, flags);
Guido van Rossum3f5da241990-12-20 15:06:42 +000047}
48
49
50/* Parse input coming from a file. Return error code, print some errors. */
51
Guido van Rossumbd0389d1994-08-29 12:25:45 +000052node *
Thomas Wouters23c9e002000-07-22 19:20:54 +000053PyParser_ParseFile(FILE *fp, char *filename, grammar *g, int start,
54 char *ps1, char *ps2, perrdetail *err_ret)
Guido van Rossum3f5da241990-12-20 15:06:42 +000055{
Tim Petersfe2127d2001-07-16 05:37:24 +000056 return PyParser_ParseFileFlags(fp, filename, g, start, ps1, ps2,
57 err_ret, 0);
58}
59
60node *
61PyParser_ParseFileFlags(FILE *fp, char *filename, grammar *g, int start,
62 char *ps1, char *ps2, perrdetail *err_ret, int flags)
63{
Guido van Rossumbd0389d1994-08-29 12:25:45 +000064 struct tok_state *tok;
65
Martin v. Löwis16eff6f2002-01-05 21:40:08 +000066 initerr(err_ret, filename);
Guido van Rossumbd0389d1994-08-29 12:25:45 +000067
Guido van Rossum86bea461997-04-29 21:03:06 +000068 if ((tok = PyTokenizer_FromFile(fp, ps1, ps2)) == NULL) {
Guido van Rossumbd0389d1994-08-29 12:25:45 +000069 err_ret->error = E_NOMEM;
70 return NULL;
Guido van Rossum3f5da241990-12-20 15:06:42 +000071 }
Guido van Rossum6135df61998-04-10 19:35:06 +000072 if (Py_TabcheckFlag || Py_VerboseFlag) {
73 tok->filename = filename;
74 tok->altwarning = (filename != NULL);
75 if (Py_TabcheckFlag >= 2)
76 tok->alterror++;
77 }
Guido van Rossumbd0389d1994-08-29 12:25:45 +000078
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000079
Tim Petersfe2127d2001-07-16 05:37:24 +000080 return parsetok(tok, g, start, err_ret, flags);
Guido van Rossumbd0389d1994-08-29 12:25:45 +000081}
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000082
83/* Parse input coming from the given tokenizer structure.
84 Return error code. */
85
Guido van Rossumda62ecc2001-07-17 16:53:11 +000086static char yield_msg[] =
87"%s:%d: Warning: 'yield' will become a reserved keyword in the future\n";
88
Guido van Rossumbd0389d1994-08-29 12:25:45 +000089static node *
Tim Petersfe2127d2001-07-16 05:37:24 +000090parsetok(struct tok_state *tok, grammar *g, int start, perrdetail *err_ret,
91 int flags)
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000092{
93 parser_state *ps;
Guido van Rossumbd0389d1994-08-29 12:25:45 +000094 node *n;
Guido van Rossumd8b1d371992-03-04 16:40:44 +000095 int started = 0;
Guido van Rossumbd0389d1994-08-29 12:25:45 +000096
Guido van Rossum86bea461997-04-29 21:03:06 +000097 if ((ps = PyParser_New(g, start)) == NULL) {
Guido van Rossum85a5fbb1990-10-14 12:07:46 +000098 fprintf(stderr, "no mem for new parser\n");
Guido van Rossumbd0389d1994-08-29 12:25:45 +000099 err_ret->error = E_NOMEM;
100 return NULL;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000101 }
Tim Petersfe2127d2001-07-16 05:37:24 +0000102 if (flags & PyPARSE_YIELD_IS_KEYWORD)
103 ps->p_generators = 1;
Guido van Rossumbd0389d1994-08-29 12:25:45 +0000104
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000105 for (;;) {
106 char *a, *b;
107 int type;
Guido van Rossum6da34342000-06-28 22:00:02 +0000108 size_t len;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000109 char *str;
Guido van Rossumbd0389d1994-08-29 12:25:45 +0000110
Guido van Rossum86bea461997-04-29 21:03:06 +0000111 type = PyTokenizer_Get(tok, &a, &b);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000112 if (type == ERRORTOKEN) {
Guido van Rossumbd0389d1994-08-29 12:25:45 +0000113 err_ret->error = tok->done;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000114 break;
115 }
Guido van Rossumd8b1d371992-03-04 16:40:44 +0000116 if (type == ENDMARKER && started) {
117 type = NEWLINE; /* Add an extra newline */
118 started = 0;
119 }
120 else
121 started = 1;
Guido van Rossumbd0389d1994-08-29 12:25:45 +0000122 len = b - a; /* XXX this may compute NULL - NULL */
Guido van Rossum86bea461997-04-29 21:03:06 +0000123 str = PyMem_NEW(char, len + 1);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000124 if (str == NULL) {
125 fprintf(stderr, "no mem for next token\n");
Guido van Rossumbd0389d1994-08-29 12:25:45 +0000126 err_ret->error = E_NOMEM;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000127 break;
128 }
Guido van Rossumbd0389d1994-08-29 12:25:45 +0000129 if (len > 0)
130 strncpy(str, a, len);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000131 str[len] = '\0';
Guido van Rossumda62ecc2001-07-17 16:53:11 +0000132
133 /* Warn about yield as NAME */
134 if (type == NAME && !ps->p_generators &&
135 len == 5 && str[0] == 'y' && strcmp(str, "yield") == 0)
136 PySys_WriteStderr(yield_msg,
137 err_ret->filename==NULL ?
138 "<string>" : err_ret->filename,
139 tok->lineno);
140
Guido van Rossumbd0389d1994-08-29 12:25:45 +0000141 if ((err_ret->error =
Fred Drake85f36392000-07-11 17:53:00 +0000142 PyParser_AddToken(ps, (int)type, str, tok->lineno,
143 &(err_ret->expected))) != E_OK) {
Guido van Rossumff0ec521997-07-27 01:52:50 +0000144 if (err_ret->error != E_DONE)
145 PyMem_DEL(str);
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000146 break;
Guido van Rossumff0ec521997-07-27 01:52:50 +0000147 }
Guido van Rossumbd0389d1994-08-29 12:25:45 +0000148 }
149
150 if (err_ret->error == E_DONE) {
151 n = ps->p_tree;
152 ps->p_tree = NULL;
153 }
154 else
155 n = NULL;
156
Guido van Rossum86bea461997-04-29 21:03:06 +0000157 PyParser_Delete(ps);
Guido van Rossumbd0389d1994-08-29 12:25:45 +0000158
159 if (n == NULL) {
160 if (tok->lineno <= 1 && tok->done == E_EOF)
161 err_ret->error = E_EOF;
162 err_ret->lineno = tok->lineno;
163 err_ret->offset = tok->cur - tok->buf;
164 if (tok->buf != NULL) {
Guido van Rossum6da34342000-06-28 22:00:02 +0000165 size_t len = tok->inp - tok->buf;
Guido van Rossumb18618d2000-05-03 23:44:39 +0000166 err_ret->text = PyMem_NEW(char, len + 1);
Guido van Rossumbd0389d1994-08-29 12:25:45 +0000167 if (err_ret->text != NULL) {
Guido van Rossumec498271995-01-20 16:59:12 +0000168 if (len > 0)
169 strncpy(err_ret->text, tok->buf, len);
Guido van Rossumbd0389d1994-08-29 12:25:45 +0000170 err_ret->text[len] = '\0';
171 }
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000172 }
173 }
Guido van Rossumbd0389d1994-08-29 12:25:45 +0000174
Guido van Rossum86bea461997-04-29 21:03:06 +0000175 PyTokenizer_Free(tok);
Guido van Rossumbd0389d1994-08-29 12:25:45 +0000176
177 return n;
Guido van Rossum85a5fbb1990-10-14 12:07:46 +0000178}
Guido van Rossum0c156a52001-10-20 14:27:56 +0000179
180static void
Martin v. Löwis16eff6f2002-01-05 21:40:08 +0000181initerr(perrdetail *err_ret, char* filename)
Guido van Rossum0c156a52001-10-20 14:27:56 +0000182{
183 err_ret->error = E_OK;
Martin v. Löwis16eff6f2002-01-05 21:40:08 +0000184 err_ret->filename = filename;
Guido van Rossum0c156a52001-10-20 14:27:56 +0000185 err_ret->lineno = 0;
186 err_ret->offset = 0;
187 err_ret->text = NULL;
188 err_ret->token = -1;
189 err_ret->expected = -1;
190}