Guido van Rossum | f70e43a | 1991-02-19 12:39:46 +0000 | [diff] [blame] | 1 | /*********************************************************** |
Guido van Rossum | fd71b9e | 2000-06-30 23:50:40 +0000 | [diff] [blame] | 2 | Copyright (c) 2000, BeOpen.com. |
| 3 | Copyright (c) 1995-2000, Corporation for National Research Initiatives. |
| 4 | Copyright (c) 1990-1995, Stichting Mathematisch Centrum. |
| 5 | All rights reserved. |
Guido van Rossum | f70e43a | 1991-02-19 12:39:46 +0000 | [diff] [blame] | 6 | |
Guido van Rossum | fd71b9e | 2000-06-30 23:50:40 +0000 | [diff] [blame] | 7 | See the file "Misc/COPYRIGHT" for information on usage and |
| 8 | redistribution of this file, and for a DISCLAIMER OF ALL WARRANTIES. |
Guido van Rossum | f70e43a | 1991-02-19 12:39:46 +0000 | [diff] [blame] | 9 | ******************************************************************/ |
| 10 | |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 11 | /* Tokenizer implementation */ |
| 12 | |
Guido van Rossum | 3f5da24 | 1990-12-20 15:06:42 +0000 | [diff] [blame] | 13 | #include "pgenheaders.h" |
| 14 | |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 15 | #include <ctype.h> |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 16 | |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 17 | #include "tokenizer.h" |
| 18 | #include "errcode.h" |
| 19 | |
Tim Peters | dbd9ba6 | 2000-07-09 03:09:57 +0000 | [diff] [blame] | 20 | extern char *PyOS_Readline(char *); |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 21 | /* Return malloc'ed string including trailing \n; |
| 22 | empty malloc'ed string for EOF; |
| 23 | NULL if interrupted */ |
| 24 | |
Guido van Rossum | 4fe8729 | 1992-02-26 15:24:44 +0000 | [diff] [blame] | 25 | /* Don't ever change this -- it would break the portability of Python code */ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 26 | #define TABSIZE 8 |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 27 | |
Guido van Rossum | cf57d8b | 1998-01-19 22:07:46 +0000 | [diff] [blame] | 28 | /* Convert a possibly signed character to a nonnegative int */ |
| 29 | /* XXX This assumes characters are 8 bits wide */ |
| 30 | #ifdef __CHAR_UNSIGNED__ |
| 31 | #define Py_CHARMASK(c) (c) |
| 32 | #else |
| 33 | #define Py_CHARMASK(c) ((c) & 0xff) |
| 34 | #endif |
| 35 | |
Guido van Rossum | 3f5da24 | 1990-12-20 15:06:42 +0000 | [diff] [blame] | 36 | /* Forward */ |
Tim Peters | dbd9ba6 | 2000-07-09 03:09:57 +0000 | [diff] [blame] | 37 | static struct tok_state *tok_new(void); |
| 38 | static int tok_nextc(struct tok_state *tok); |
| 39 | static void tok_backup(struct tok_state *tok, int c); |
Guido van Rossum | 3f5da24 | 1990-12-20 15:06:42 +0000 | [diff] [blame] | 40 | |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 41 | /* Token names */ |
| 42 | |
Guido van Rossum | 86bea46 | 1997-04-29 21:03:06 +0000 | [diff] [blame] | 43 | char *_PyParser_TokenNames[] = { |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 44 | "ENDMARKER", |
| 45 | "NAME", |
| 46 | "NUMBER", |
| 47 | "STRING", |
| 48 | "NEWLINE", |
| 49 | "INDENT", |
| 50 | "DEDENT", |
| 51 | "LPAR", |
| 52 | "RPAR", |
| 53 | "LSQB", |
| 54 | "RSQB", |
| 55 | "COLON", |
| 56 | "COMMA", |
| 57 | "SEMI", |
| 58 | "PLUS", |
| 59 | "MINUS", |
| 60 | "STAR", |
| 61 | "SLASH", |
| 62 | "VBAR", |
| 63 | "AMPER", |
| 64 | "LESS", |
| 65 | "GREATER", |
| 66 | "EQUAL", |
| 67 | "DOT", |
| 68 | "PERCENT", |
| 69 | "BACKQUOTE", |
| 70 | "LBRACE", |
| 71 | "RBRACE", |
Guido van Rossum | fbab905 | 1991-10-20 20:25:03 +0000 | [diff] [blame] | 72 | "EQEQUAL", |
| 73 | "NOTEQUAL", |
| 74 | "LESSEQUAL", |
| 75 | "GREATEREQUAL", |
Guido van Rossum | baf0ebf | 1991-10-24 14:59:40 +0000 | [diff] [blame] | 76 | "TILDE", |
| 77 | "CIRCUMFLEX", |
| 78 | "LEFTSHIFT", |
| 79 | "RIGHTSHIFT", |
Guido van Rossum | f595fde | 1996-01-12 01:31:58 +0000 | [diff] [blame] | 80 | "DOUBLESTAR", |
Thomas Wouters | 434d082 | 2000-08-24 20:11:32 +0000 | [diff] [blame^] | 81 | "PLUSEQUAL", |
| 82 | "MINEQUAL", |
| 83 | "STAREQUAL", |
| 84 | "SLASHEQUAL", |
| 85 | "PERCENTEQUAL", |
| 86 | "AMPEREQUAL", |
| 87 | "VBAREQUAL", |
| 88 | "CIRCUMFLEXEQUAL", |
| 89 | "LEFTSHIFTEQUAL", |
| 90 | "RIGHTSHIFTEQUAL", |
| 91 | "DOUBLESTAREQUAL", |
Guido van Rossum | fbab905 | 1991-10-20 20:25:03 +0000 | [diff] [blame] | 92 | /* This table must match the #defines in token.h! */ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 93 | "OP", |
| 94 | "<ERRORTOKEN>", |
| 95 | "<N_TOKENS>" |
| 96 | }; |
| 97 | |
| 98 | |
| 99 | /* Create and initialize a new tok_state structure */ |
| 100 | |
| 101 | static struct tok_state * |
Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 102 | tok_new(void) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 103 | { |
Guido van Rossum | 86bea46 | 1997-04-29 21:03:06 +0000 | [diff] [blame] | 104 | struct tok_state *tok = PyMem_NEW(struct tok_state, 1); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 105 | if (tok == NULL) |
| 106 | return NULL; |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 107 | tok->buf = tok->cur = tok->end = tok->inp = tok->start = NULL; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 108 | tok->done = E_OK; |
| 109 | tok->fp = NULL; |
| 110 | tok->tabsize = TABSIZE; |
| 111 | tok->indent = 0; |
| 112 | tok->indstack[0] = 0; |
| 113 | tok->atbol = 1; |
| 114 | tok->pendin = 0; |
| 115 | tok->prompt = tok->nextprompt = NULL; |
| 116 | tok->lineno = 0; |
Guido van Rossum | a849b83 | 1993-05-12 11:35:44 +0000 | [diff] [blame] | 117 | tok->level = 0; |
Guido van Rossum | 926f13a | 1998-04-09 21:38:06 +0000 | [diff] [blame] | 118 | tok->filename = NULL; |
| 119 | tok->altwarning = 0; |
| 120 | tok->alterror = 0; |
| 121 | tok->alttabsize = 1; |
| 122 | tok->altindstack[0] = 0; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 123 | return tok; |
| 124 | } |
| 125 | |
| 126 | |
| 127 | /* Set up tokenizer for string */ |
| 128 | |
| 129 | struct tok_state * |
Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 130 | PyTokenizer_FromString(char *str) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 131 | { |
| 132 | struct tok_state *tok = tok_new(); |
| 133 | if (tok == NULL) |
| 134 | return NULL; |
Guido van Rossum | 1a817c0 | 1994-09-19 08:06:25 +0000 | [diff] [blame] | 135 | tok->buf = tok->cur = tok->end = tok->inp = str; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 136 | return tok; |
| 137 | } |
| 138 | |
| 139 | |
Guido van Rossum | 8c11a5c | 1991-07-27 21:42:56 +0000 | [diff] [blame] | 140 | /* Set up tokenizer for file */ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 141 | |
| 142 | struct tok_state * |
Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 143 | PyTokenizer_FromFile(FILE *fp, char *ps1, char *ps2) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 144 | { |
| 145 | struct tok_state *tok = tok_new(); |
| 146 | if (tok == NULL) |
| 147 | return NULL; |
Guido van Rossum | 86bea46 | 1997-04-29 21:03:06 +0000 | [diff] [blame] | 148 | if ((tok->buf = PyMem_NEW(char, BUFSIZ)) == NULL) { |
| 149 | PyMem_DEL(tok); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 150 | return NULL; |
| 151 | } |
| 152 | tok->cur = tok->inp = tok->buf; |
| 153 | tok->end = tok->buf + BUFSIZ; |
| 154 | tok->fp = fp; |
| 155 | tok->prompt = ps1; |
| 156 | tok->nextprompt = ps2; |
| 157 | return tok; |
| 158 | } |
| 159 | |
| 160 | |
| 161 | /* Free a tok_state structure */ |
| 162 | |
| 163 | void |
Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 164 | PyTokenizer_Free(struct tok_state *tok) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 165 | { |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 166 | if (tok->fp != NULL && tok->buf != NULL) |
Guido van Rossum | 86bea46 | 1997-04-29 21:03:06 +0000 | [diff] [blame] | 167 | PyMem_DEL(tok->buf); |
| 168 | PyMem_DEL(tok); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 169 | } |
| 170 | |
| 171 | |
| 172 | /* Get next char, updating state; error code goes into tok->done */ |
| 173 | |
| 174 | static int |
Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 175 | tok_nextc(register struct tok_state *tok) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 176 | { |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 177 | for (;;) { |
Guido van Rossum | 1a817c0 | 1994-09-19 08:06:25 +0000 | [diff] [blame] | 178 | if (tok->cur != tok->inp) { |
Guido van Rossum | cf57d8b | 1998-01-19 22:07:46 +0000 | [diff] [blame] | 179 | return Py_CHARMASK(*tok->cur++); /* Fast path */ |
Guido van Rossum | 1a817c0 | 1994-09-19 08:06:25 +0000 | [diff] [blame] | 180 | } |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 181 | if (tok->done != E_OK) |
| 182 | return EOF; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 183 | if (tok->fp == NULL) { |
Guido van Rossum | 1a817c0 | 1994-09-19 08:06:25 +0000 | [diff] [blame] | 184 | char *end = strchr(tok->inp, '\n'); |
| 185 | if (end != NULL) |
| 186 | end++; |
| 187 | else { |
| 188 | end = strchr(tok->inp, '\0'); |
| 189 | if (end == tok->inp) { |
| 190 | tok->done = E_EOF; |
| 191 | return EOF; |
| 192 | } |
| 193 | } |
| 194 | if (tok->start == NULL) |
| 195 | tok->buf = tok->cur; |
| 196 | tok->lineno++; |
| 197 | tok->inp = end; |
Guido van Rossum | cf57d8b | 1998-01-19 22:07:46 +0000 | [diff] [blame] | 198 | return Py_CHARMASK(*tok->cur++); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 199 | } |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 200 | if (tok->prompt != NULL) { |
Guido van Rossum | 86bea46 | 1997-04-29 21:03:06 +0000 | [diff] [blame] | 201 | char *new = PyOS_Readline(tok->prompt); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 202 | if (tok->nextprompt != NULL) |
| 203 | tok->prompt = tok->nextprompt; |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 204 | if (new == NULL) |
| 205 | tok->done = E_INTR; |
| 206 | else if (*new == '\0') { |
Guido van Rossum | b18618d | 2000-05-03 23:44:39 +0000 | [diff] [blame] | 207 | PyMem_FREE(new); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 208 | tok->done = E_EOF; |
| 209 | } |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 210 | else if (tok->start != NULL) { |
Guido van Rossum | 6da3434 | 2000-06-28 22:00:02 +0000 | [diff] [blame] | 211 | size_t start = tok->start - tok->buf; |
| 212 | size_t oldlen = tok->cur - tok->buf; |
| 213 | size_t newlen = oldlen + strlen(new); |
Guido van Rossum | b18618d | 2000-05-03 23:44:39 +0000 | [diff] [blame] | 214 | char *buf = tok->buf; |
| 215 | PyMem_RESIZE(buf, char, newlen+1); |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 216 | tok->lineno++; |
| 217 | if (buf == NULL) { |
Guido van Rossum | b18618d | 2000-05-03 23:44:39 +0000 | [diff] [blame] | 218 | PyMem_DEL(tok->buf); |
Guido van Rossum | 588633d | 1994-12-30 15:46:02 +0000 | [diff] [blame] | 219 | tok->buf = NULL; |
Guido van Rossum | b18618d | 2000-05-03 23:44:39 +0000 | [diff] [blame] | 220 | PyMem_FREE(new); |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 221 | tok->done = E_NOMEM; |
| 222 | return EOF; |
| 223 | } |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 224 | tok->buf = buf; |
| 225 | tok->cur = tok->buf + oldlen; |
| 226 | strcpy(tok->buf + oldlen, new); |
Guido van Rossum | b18618d | 2000-05-03 23:44:39 +0000 | [diff] [blame] | 227 | PyMem_FREE(new); |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 228 | tok->inp = tok->buf + newlen; |
| 229 | tok->end = tok->inp + 1; |
| 230 | tok->start = tok->buf + start; |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 231 | } |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 232 | else { |
| 233 | tok->lineno++; |
| 234 | if (tok->buf != NULL) |
Guido van Rossum | b18618d | 2000-05-03 23:44:39 +0000 | [diff] [blame] | 235 | PyMem_DEL(tok->buf); |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 236 | tok->buf = new; |
| 237 | tok->cur = tok->buf; |
| 238 | tok->inp = strchr(tok->buf, '\0'); |
| 239 | tok->end = tok->inp + 1; |
| 240 | } |
| 241 | } |
| 242 | else { |
| 243 | int done = 0; |
| 244 | int cur = 0; |
Guido van Rossum | 2e96eb9 | 1995-06-14 18:26:02 +0000 | [diff] [blame] | 245 | char *pt; |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 246 | if (tok->start == NULL) { |
| 247 | if (tok->buf == NULL) { |
Guido van Rossum | 86bea46 | 1997-04-29 21:03:06 +0000 | [diff] [blame] | 248 | tok->buf = PyMem_NEW(char, BUFSIZ); |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 249 | if (tok->buf == NULL) { |
| 250 | tok->done = E_NOMEM; |
| 251 | return EOF; |
| 252 | } |
| 253 | tok->end = tok->buf + BUFSIZ; |
| 254 | } |
| 255 | if (fgets(tok->buf, (int)(tok->end - tok->buf), |
| 256 | tok->fp) == NULL) { |
| 257 | tok->done = E_EOF; |
| 258 | done = 1; |
| 259 | } |
| 260 | else { |
| 261 | tok->done = E_OK; |
| 262 | tok->inp = strchr(tok->buf, '\0'); |
| 263 | done = tok->inp[-1] == '\n'; |
| 264 | } |
| 265 | } |
| 266 | else { |
| 267 | cur = tok->cur - tok->buf; |
Guido van Rossum | 78c0535 | 1995-01-17 16:12:13 +0000 | [diff] [blame] | 268 | if (feof(tok->fp)) { |
| 269 | tok->done = E_EOF; |
| 270 | done = 1; |
| 271 | } |
| 272 | else |
| 273 | tok->done = E_OK; |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 274 | } |
| 275 | tok->lineno++; |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 276 | /* Read until '\n' or EOF */ |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 277 | while (!done) { |
| 278 | int curstart = tok->start == NULL ? -1 : |
| 279 | tok->start - tok->buf; |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 280 | int curvalid = tok->inp - tok->buf; |
Guido van Rossum | 3f6bb86 | 1995-09-21 20:36:34 +0000 | [diff] [blame] | 281 | int newsize = curvalid + BUFSIZ; |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 282 | char *newbuf = tok->buf; |
Guido van Rossum | 86bea46 | 1997-04-29 21:03:06 +0000 | [diff] [blame] | 283 | PyMem_RESIZE(newbuf, char, newsize); |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 284 | if (newbuf == NULL) { |
| 285 | tok->done = E_NOMEM; |
| 286 | tok->cur = tok->inp; |
| 287 | return EOF; |
| 288 | } |
| 289 | tok->buf = newbuf; |
| 290 | tok->inp = tok->buf + curvalid; |
| 291 | tok->end = tok->buf + newsize; |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 292 | tok->start = curstart < 0 ? NULL : |
| 293 | tok->buf + curstart; |
| 294 | if (fgets(tok->inp, |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 295 | (int)(tok->end - tok->inp), |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 296 | tok->fp) == NULL) { |
| 297 | /* Last line does not end in \n, |
| 298 | fake one */ |
| 299 | strcpy(tok->inp, "\n"); |
| 300 | } |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 301 | tok->inp = strchr(tok->inp, '\0'); |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 302 | done = tok->inp[-1] == '\n'; |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 303 | } |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 304 | tok->cur = tok->buf + cur; |
Guido van Rossum | 2d45be1 | 1997-04-11 19:16:25 +0000 | [diff] [blame] | 305 | #ifndef macintosh |
Guido van Rossum | 2e96eb9 | 1995-06-14 18:26:02 +0000 | [diff] [blame] | 306 | /* replace "\r\n" with "\n" */ |
Guido van Rossum | 2d45be1 | 1997-04-11 19:16:25 +0000 | [diff] [blame] | 307 | /* For Mac we leave the \r, giving a syntax error */ |
Guido van Rossum | 2e96eb9 | 1995-06-14 18:26:02 +0000 | [diff] [blame] | 308 | pt = tok->inp - 2; |
| 309 | if (pt >= tok->buf && *pt == '\r') { |
| 310 | *pt++ = '\n'; |
| 311 | *pt = '\0'; |
| 312 | tok->inp = pt; |
| 313 | } |
Guido van Rossum | 2d45be1 | 1997-04-11 19:16:25 +0000 | [diff] [blame] | 314 | #endif |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 315 | } |
| 316 | if (tok->done != E_OK) { |
| 317 | if (tok->prompt != NULL) |
Guido van Rossum | 6e73bf4 | 1998-08-25 18:13:04 +0000 | [diff] [blame] | 318 | PySys_WriteStderr("\n"); |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 319 | tok->cur = tok->inp; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 320 | return EOF; |
| 321 | } |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 322 | } |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 323 | /*NOTREACHED*/ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 324 | } |
| 325 | |
| 326 | |
| 327 | /* Back-up one character */ |
| 328 | |
| 329 | static void |
Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 330 | tok_backup(register struct tok_state *tok, register int c) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 331 | { |
| 332 | if (c != EOF) { |
Guido van Rossum | 588633d | 1994-12-30 15:46:02 +0000 | [diff] [blame] | 333 | if (--tok->cur < tok->buf) |
Guido van Rossum | 86bea46 | 1997-04-29 21:03:06 +0000 | [diff] [blame] | 334 | Py_FatalError("tok_backup: begin of buffer"); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 335 | if (*tok->cur != c) |
| 336 | *tok->cur = c; |
| 337 | } |
| 338 | } |
| 339 | |
| 340 | |
| 341 | /* Return the token corresponding to a single character */ |
| 342 | |
| 343 | int |
Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 344 | PyToken_OneChar(int c) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 345 | { |
| 346 | switch (c) { |
| 347 | case '(': return LPAR; |
| 348 | case ')': return RPAR; |
| 349 | case '[': return LSQB; |
| 350 | case ']': return RSQB; |
| 351 | case ':': return COLON; |
| 352 | case ',': return COMMA; |
| 353 | case ';': return SEMI; |
| 354 | case '+': return PLUS; |
| 355 | case '-': return MINUS; |
| 356 | case '*': return STAR; |
| 357 | case '/': return SLASH; |
| 358 | case '|': return VBAR; |
| 359 | case '&': return AMPER; |
| 360 | case '<': return LESS; |
| 361 | case '>': return GREATER; |
| 362 | case '=': return EQUAL; |
| 363 | case '.': return DOT; |
| 364 | case '%': return PERCENT; |
| 365 | case '`': return BACKQUOTE; |
| 366 | case '{': return LBRACE; |
| 367 | case '}': return RBRACE; |
Guido van Rossum | baf0ebf | 1991-10-24 14:59:40 +0000 | [diff] [blame] | 368 | case '^': return CIRCUMFLEX; |
| 369 | case '~': return TILDE; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 370 | default: return OP; |
| 371 | } |
| 372 | } |
| 373 | |
| 374 | |
Guido van Rossum | fbab905 | 1991-10-20 20:25:03 +0000 | [diff] [blame] | 375 | int |
Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 376 | PyToken_TwoChars(int c1, int c2) |
Guido van Rossum | fbab905 | 1991-10-20 20:25:03 +0000 | [diff] [blame] | 377 | { |
| 378 | switch (c1) { |
| 379 | case '=': |
| 380 | switch (c2) { |
| 381 | case '=': return EQEQUAL; |
| 382 | } |
| 383 | break; |
| 384 | case '!': |
| 385 | switch (c2) { |
| 386 | case '=': return NOTEQUAL; |
| 387 | } |
| 388 | break; |
| 389 | case '<': |
| 390 | switch (c2) { |
| 391 | case '>': return NOTEQUAL; |
| 392 | case '=': return LESSEQUAL; |
Guido van Rossum | baf0ebf | 1991-10-24 14:59:40 +0000 | [diff] [blame] | 393 | case '<': return LEFTSHIFT; |
Guido van Rossum | fbab905 | 1991-10-20 20:25:03 +0000 | [diff] [blame] | 394 | } |
| 395 | break; |
| 396 | case '>': |
| 397 | switch (c2) { |
| 398 | case '=': return GREATEREQUAL; |
Guido van Rossum | baf0ebf | 1991-10-24 14:59:40 +0000 | [diff] [blame] | 399 | case '>': return RIGHTSHIFT; |
Guido van Rossum | fbab905 | 1991-10-20 20:25:03 +0000 | [diff] [blame] | 400 | } |
| 401 | break; |
Thomas Wouters | 434d082 | 2000-08-24 20:11:32 +0000 | [diff] [blame^] | 402 | case '+': |
| 403 | switch (c2) { |
| 404 | case '=': return PLUSEQUAL; |
| 405 | } |
| 406 | break; |
| 407 | case '-': |
| 408 | switch (c2) { |
| 409 | case '=': return MINEQUAL; |
| 410 | } |
| 411 | break; |
Guido van Rossum | f595fde | 1996-01-12 01:31:58 +0000 | [diff] [blame] | 412 | case '*': |
| 413 | switch (c2) { |
| 414 | case '*': return DOUBLESTAR; |
Thomas Wouters | 434d082 | 2000-08-24 20:11:32 +0000 | [diff] [blame^] | 415 | case '=': return STAREQUAL; |
| 416 | } |
| 417 | break; |
| 418 | case '/': |
| 419 | switch (c2) { |
| 420 | case '=': return SLASHEQUAL; |
| 421 | } |
| 422 | break; |
| 423 | case '|': |
| 424 | switch (c2) { |
| 425 | case '=': return VBAREQUAL; |
| 426 | } |
| 427 | break; |
| 428 | case '%': |
| 429 | switch (c2) { |
| 430 | case '=': return PERCENTEQUAL; |
| 431 | } |
| 432 | break; |
| 433 | case '&': |
| 434 | switch (c2) { |
| 435 | case '=': return AMPEREQUAL; |
| 436 | } |
| 437 | break; |
| 438 | case '^': |
| 439 | switch (c2) { |
| 440 | case '=': return CIRCUMFLEXEQUAL; |
Guido van Rossum | f595fde | 1996-01-12 01:31:58 +0000 | [diff] [blame] | 441 | } |
| 442 | break; |
Guido van Rossum | fbab905 | 1991-10-20 20:25:03 +0000 | [diff] [blame] | 443 | } |
| 444 | return OP; |
| 445 | } |
| 446 | |
Thomas Wouters | 434d082 | 2000-08-24 20:11:32 +0000 | [diff] [blame^] | 447 | int |
| 448 | PyToken_ThreeChars(int c1, int c2, int c3) |
| 449 | { |
| 450 | switch (c1) { |
| 451 | case '<': |
| 452 | switch (c2) { |
| 453 | case '<': |
| 454 | switch (c3) { |
| 455 | case '=': |
| 456 | return LEFTSHIFTEQUAL; |
| 457 | break; |
| 458 | } |
| 459 | break; |
| 460 | } |
| 461 | break; |
| 462 | case '>': |
| 463 | switch (c2) { |
| 464 | case '>': |
| 465 | switch (c3) { |
| 466 | case '=': |
| 467 | return RIGHTSHIFTEQUAL; |
| 468 | break; |
| 469 | } |
| 470 | break; |
| 471 | } |
| 472 | break; |
| 473 | case '*': |
| 474 | switch (c2) { |
| 475 | case '*': |
| 476 | switch (c3) { |
| 477 | case '=': |
| 478 | return DOUBLESTAREQUAL; |
| 479 | break; |
| 480 | } |
| 481 | break; |
| 482 | } |
| 483 | break; |
| 484 | } |
| 485 | return OP; |
| 486 | } |
Guido van Rossum | fbab905 | 1991-10-20 20:25:03 +0000 | [diff] [blame] | 487 | |
Guido van Rossum | 926f13a | 1998-04-09 21:38:06 +0000 | [diff] [blame] | 488 | static int |
Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 489 | indenterror(struct tok_state *tok) |
Guido van Rossum | 926f13a | 1998-04-09 21:38:06 +0000 | [diff] [blame] | 490 | { |
| 491 | if (tok->alterror) { |
Fred Drake | 85f3639 | 2000-07-11 17:53:00 +0000 | [diff] [blame] | 492 | tok->done = E_TABSPACE; |
Guido van Rossum | 926f13a | 1998-04-09 21:38:06 +0000 | [diff] [blame] | 493 | tok->cur = tok->inp; |
| 494 | return 1; |
| 495 | } |
| 496 | if (tok->altwarning) { |
Fred Drake | 85f3639 | 2000-07-11 17:53:00 +0000 | [diff] [blame] | 497 | PySys_WriteStderr("%s: inconsistent use of tabs and spaces " |
| 498 | "in indentation\n", tok->filename); |
Guido van Rossum | 926f13a | 1998-04-09 21:38:06 +0000 | [diff] [blame] | 499 | tok->altwarning = 0; |
| 500 | } |
| 501 | return 0; |
| 502 | } |
| 503 | |
| 504 | |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 505 | /* Get next token, after space stripping etc. */ |
| 506 | |
| 507 | int |
Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 508 | PyTokenizer_Get(register struct tok_state *tok, char **p_start, |
| 509 | char **p_end) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 510 | { |
| 511 | register int c; |
Guido van Rossum | 8c11a5c | 1991-07-27 21:42:56 +0000 | [diff] [blame] | 512 | int blankline; |
| 513 | |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 514 | *p_start = *p_end = NULL; |
Guido van Rossum | 8c11a5c | 1991-07-27 21:42:56 +0000 | [diff] [blame] | 515 | nextline: |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 516 | tok->start = NULL; |
Guido van Rossum | 8c11a5c | 1991-07-27 21:42:56 +0000 | [diff] [blame] | 517 | blankline = 0; |
| 518 | |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 519 | /* Get indentation level */ |
| 520 | if (tok->atbol) { |
| 521 | register int col = 0; |
Guido van Rossum | 926f13a | 1998-04-09 21:38:06 +0000 | [diff] [blame] | 522 | register int altcol = 0; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 523 | tok->atbol = 0; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 524 | for (;;) { |
| 525 | c = tok_nextc(tok); |
| 526 | if (c == ' ') |
Guido van Rossum | 926f13a | 1998-04-09 21:38:06 +0000 | [diff] [blame] | 527 | col++, altcol++; |
| 528 | else if (c == '\t') { |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 529 | col = (col/tok->tabsize + 1) * tok->tabsize; |
Guido van Rossum | 926f13a | 1998-04-09 21:38:06 +0000 | [diff] [blame] | 530 | altcol = (altcol/tok->alttabsize + 1) |
| 531 | * tok->alttabsize; |
| 532 | } |
Guido van Rossum | 94d32b1 | 1995-07-07 22:27:27 +0000 | [diff] [blame] | 533 | else if (c == '\014') /* Control-L (formfeed) */ |
Guido van Rossum | 926f13a | 1998-04-09 21:38:06 +0000 | [diff] [blame] | 534 | col = altcol = 0; /* For Emacs users */ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 535 | else |
| 536 | break; |
| 537 | } |
| 538 | tok_backup(tok, c); |
Guido van Rossum | 8c11a5c | 1991-07-27 21:42:56 +0000 | [diff] [blame] | 539 | if (c == '#' || c == '\n') { |
| 540 | /* Lines with only whitespace and/or comments |
| 541 | shouldn't affect the indentation and are |
| 542 | not passed to the parser as NEWLINE tokens, |
| 543 | except *totally* empty lines in interactive |
| 544 | mode, which signal the end of a command group. */ |
| 545 | if (col == 0 && c == '\n' && tok->prompt != NULL) |
| 546 | blankline = 0; /* Let it through */ |
| 547 | else |
| 548 | blankline = 1; /* Ignore completely */ |
| 549 | /* We can't jump back right here since we still |
| 550 | may need to skip to the end of a comment */ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 551 | } |
Guido van Rossum | a849b83 | 1993-05-12 11:35:44 +0000 | [diff] [blame] | 552 | if (!blankline && tok->level == 0) { |
Guido van Rossum | 8c11a5c | 1991-07-27 21:42:56 +0000 | [diff] [blame] | 553 | if (col == tok->indstack[tok->indent]) { |
| 554 | /* No change */ |
Guido van Rossum | 926f13a | 1998-04-09 21:38:06 +0000 | [diff] [blame] | 555 | if (altcol != tok->altindstack[tok->indent]) { |
| 556 | if (indenterror(tok)) |
| 557 | return ERRORTOKEN; |
| 558 | } |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 559 | } |
Guido van Rossum | 8c11a5c | 1991-07-27 21:42:56 +0000 | [diff] [blame] | 560 | else if (col > tok->indstack[tok->indent]) { |
| 561 | /* Indent -- always one */ |
| 562 | if (tok->indent+1 >= MAXINDENT) { |
Fred Drake | 85f3639 | 2000-07-11 17:53:00 +0000 | [diff] [blame] | 563 | tok->done = E_TOODEEP; |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 564 | tok->cur = tok->inp; |
Guido van Rossum | 8c11a5c | 1991-07-27 21:42:56 +0000 | [diff] [blame] | 565 | return ERRORTOKEN; |
| 566 | } |
Guido van Rossum | 926f13a | 1998-04-09 21:38:06 +0000 | [diff] [blame] | 567 | if (altcol <= tok->altindstack[tok->indent]) { |
| 568 | if (indenterror(tok)) |
| 569 | return ERRORTOKEN; |
| 570 | } |
Guido van Rossum | 8c11a5c | 1991-07-27 21:42:56 +0000 | [diff] [blame] | 571 | tok->pendin++; |
| 572 | tok->indstack[++tok->indent] = col; |
Guido van Rossum | 926f13a | 1998-04-09 21:38:06 +0000 | [diff] [blame] | 573 | tok->altindstack[tok->indent] = altcol; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 574 | } |
Guido van Rossum | 8c11a5c | 1991-07-27 21:42:56 +0000 | [diff] [blame] | 575 | else /* col < tok->indstack[tok->indent] */ { |
| 576 | /* Dedent -- any number, must be consistent */ |
| 577 | while (tok->indent > 0 && |
| 578 | col < tok->indstack[tok->indent]) { |
Guido van Rossum | 8c11a5c | 1991-07-27 21:42:56 +0000 | [diff] [blame] | 579 | tok->pendin--; |
Guido van Rossum | 54758fa | 1998-02-16 22:18:00 +0000 | [diff] [blame] | 580 | tok->indent--; |
Guido van Rossum | 8c11a5c | 1991-07-27 21:42:56 +0000 | [diff] [blame] | 581 | } |
| 582 | if (col != tok->indstack[tok->indent]) { |
Fred Drake | 85f3639 | 2000-07-11 17:53:00 +0000 | [diff] [blame] | 583 | tok->done = E_DEDENT; |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 584 | tok->cur = tok->inp; |
Guido van Rossum | 8c11a5c | 1991-07-27 21:42:56 +0000 | [diff] [blame] | 585 | return ERRORTOKEN; |
| 586 | } |
Guido van Rossum | 926f13a | 1998-04-09 21:38:06 +0000 | [diff] [blame] | 587 | if (altcol != tok->altindstack[tok->indent]) { |
| 588 | if (indenterror(tok)) |
| 589 | return ERRORTOKEN; |
| 590 | } |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 591 | } |
| 592 | } |
| 593 | } |
| 594 | |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 595 | tok->start = tok->cur; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 596 | |
| 597 | /* Return pending indents/dedents */ |
| 598 | if (tok->pendin != 0) { |
| 599 | if (tok->pendin < 0) { |
| 600 | tok->pendin++; |
| 601 | return DEDENT; |
| 602 | } |
| 603 | else { |
| 604 | tok->pendin--; |
| 605 | return INDENT; |
| 606 | } |
| 607 | } |
| 608 | |
| 609 | again: |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 610 | tok->start = NULL; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 611 | /* Skip spaces */ |
| 612 | do { |
| 613 | c = tok_nextc(tok); |
Guido van Rossum | 94d32b1 | 1995-07-07 22:27:27 +0000 | [diff] [blame] | 614 | } while (c == ' ' || c == '\t' || c == '\014'); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 615 | |
| 616 | /* Set start of current token */ |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 617 | tok->start = tok->cur - 1; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 618 | |
Guido van Rossum | ab5ca15 | 2000-03-31 00:52:27 +0000 | [diff] [blame] | 619 | /* Skip comment, while looking for tab-setting magic */ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 620 | if (c == '#') { |
Guido van Rossum | ab5ca15 | 2000-03-31 00:52:27 +0000 | [diff] [blame] | 621 | static char *tabforms[] = { |
| 622 | "tab-width:", /* Emacs */ |
| 623 | ":tabstop=", /* vim, full form */ |
| 624 | ":ts=", /* vim, abbreviated form */ |
| 625 | "set tabsize=", /* will vi never die? */ |
| 626 | /* more templates can be added here to support other editors */ |
| 627 | }; |
| 628 | char cbuf[80]; |
| 629 | char *tp, **cp; |
| 630 | tp = cbuf; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 631 | do { |
Guido van Rossum | ab5ca15 | 2000-03-31 00:52:27 +0000 | [diff] [blame] | 632 | *tp++ = c = tok_nextc(tok); |
| 633 | } while (c != EOF && c != '\n' && |
| 634 | tp - cbuf + 1 < sizeof(cbuf)); |
| 635 | *tp = '\0'; |
| 636 | for (cp = tabforms; |
| 637 | cp < tabforms + sizeof(tabforms)/sizeof(tabforms[0]); |
| 638 | cp++) { |
| 639 | if ((tp = strstr(cbuf, *cp))) { |
| 640 | int newsize = atoi(tp + strlen(*cp)); |
| 641 | |
| 642 | if (newsize >= 1 && newsize <= 40) { |
| 643 | tok->tabsize = newsize; |
Guido van Rossum | 6c981ad | 2000-04-03 23:02:17 +0000 | [diff] [blame] | 644 | if (Py_VerboseFlag) |
| 645 | PySys_WriteStderr( |
Guido van Rossum | ab5ca15 | 2000-03-31 00:52:27 +0000 | [diff] [blame] | 646 | "Tab size set to %d\n", |
| 647 | newsize); |
| 648 | } |
| 649 | } |
| 650 | } |
| 651 | while (c != EOF && c != '\n') |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 652 | c = tok_nextc(tok); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 653 | } |
| 654 | |
| 655 | /* Check for EOF and errors now */ |
Guido van Rossum | baf0ebf | 1991-10-24 14:59:40 +0000 | [diff] [blame] | 656 | if (c == EOF) { |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 657 | return tok->done == E_EOF ? ENDMARKER : ERRORTOKEN; |
Guido van Rossum | baf0ebf | 1991-10-24 14:59:40 +0000 | [diff] [blame] | 658 | } |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 659 | |
| 660 | /* Identifier (most frequent token!) */ |
| 661 | if (isalpha(c) || c == '_') { |
Guido van Rossum | 86016cb | 2000-03-10 22:56:54 +0000 | [diff] [blame] | 662 | /* Process r"", u"" and ur"" */ |
Guido van Rossum | 5026cb4 | 1997-04-25 17:32:00 +0000 | [diff] [blame] | 663 | switch (c) { |
| 664 | case 'r': |
| 665 | case 'R': |
| 666 | c = tok_nextc(tok); |
| 667 | if (c == '"' || c == '\'') |
| 668 | goto letter_quote; |
Guido van Rossum | 86016cb | 2000-03-10 22:56:54 +0000 | [diff] [blame] | 669 | break; |
| 670 | case 'u': |
| 671 | case 'U': |
| 672 | c = tok_nextc(tok); |
| 673 | if (c == 'r' || c == 'R') |
| 674 | c = tok_nextc(tok); |
| 675 | if (c == '"' || c == '\'') |
| 676 | goto letter_quote; |
| 677 | break; |
Guido van Rossum | 5026cb4 | 1997-04-25 17:32:00 +0000 | [diff] [blame] | 678 | } |
Guido van Rossum | 24dacb3 | 1997-04-06 03:46:20 +0000 | [diff] [blame] | 679 | while (isalnum(c) || c == '_') { |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 680 | c = tok_nextc(tok); |
Guido van Rossum | 24dacb3 | 1997-04-06 03:46:20 +0000 | [diff] [blame] | 681 | } |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 682 | tok_backup(tok, c); |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 683 | *p_start = tok->start; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 684 | *p_end = tok->cur; |
| 685 | return NAME; |
| 686 | } |
| 687 | |
| 688 | /* Newline */ |
| 689 | if (c == '\n') { |
| 690 | tok->atbol = 1; |
Guido van Rossum | a849b83 | 1993-05-12 11:35:44 +0000 | [diff] [blame] | 691 | if (blankline || tok->level > 0) |
Guido van Rossum | 8c11a5c | 1991-07-27 21:42:56 +0000 | [diff] [blame] | 692 | goto nextline; |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 693 | *p_start = tok->start; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 694 | *p_end = tok->cur - 1; /* Leave '\n' out of the string */ |
| 695 | return NEWLINE; |
| 696 | } |
| 697 | |
Guido van Rossum | 2d45be1 | 1997-04-11 19:16:25 +0000 | [diff] [blame] | 698 | #ifdef macintosh |
| 699 | if (c == '\r') { |
Guido van Rossum | 6e73bf4 | 1998-08-25 18:13:04 +0000 | [diff] [blame] | 700 | PySys_WriteStderr( |
Guido van Rossum | 86bea46 | 1997-04-29 21:03:06 +0000 | [diff] [blame] | 701 | "File contains \\r characters (incorrect line endings?)\n"); |
Guido van Rossum | 2d45be1 | 1997-04-11 19:16:25 +0000 | [diff] [blame] | 702 | tok->done = E_TOKEN; |
| 703 | tok->cur = tok->inp; |
| 704 | return ERRORTOKEN; |
| 705 | } |
| 706 | #endif |
Guido van Rossum | baf0ebf | 1991-10-24 14:59:40 +0000 | [diff] [blame] | 707 | /* Period or number starting with period? */ |
| 708 | if (c == '.') { |
| 709 | c = tok_nextc(tok); |
| 710 | if (isdigit(c)) { |
| 711 | goto fraction; |
| 712 | } |
| 713 | else { |
| 714 | tok_backup(tok, c); |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 715 | *p_start = tok->start; |
Guido van Rossum | baf0ebf | 1991-10-24 14:59:40 +0000 | [diff] [blame] | 716 | *p_end = tok->cur; |
| 717 | return DOT; |
| 718 | } |
| 719 | } |
Guido van Rossum | f595fde | 1996-01-12 01:31:58 +0000 | [diff] [blame] | 720 | |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 721 | /* Number */ |
| 722 | if (isdigit(c)) { |
| 723 | if (c == '0') { |
| 724 | /* Hex or octal */ |
| 725 | c = tok_nextc(tok); |
| 726 | if (c == '.') |
| 727 | goto fraction; |
Guido van Rossum | f595fde | 1996-01-12 01:31:58 +0000 | [diff] [blame] | 728 | #ifndef WITHOUT_COMPLEX |
Guido van Rossum | faa436c | 1996-01-26 18:59:07 +0000 | [diff] [blame] | 729 | if (c == 'j' || c == 'J') |
Guido van Rossum | f595fde | 1996-01-12 01:31:58 +0000 | [diff] [blame] | 730 | goto imaginary; |
| 731 | #endif |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 732 | if (c == 'x' || c == 'X') { |
| 733 | /* Hex */ |
| 734 | do { |
| 735 | c = tok_nextc(tok); |
| 736 | } while (isxdigit(c)); |
| 737 | } |
| 738 | else { |
Guido van Rossum | 9430945 | 1991-12-10 14:01:05 +0000 | [diff] [blame] | 739 | /* XXX This is broken! E.g., |
| 740 | 09.9 should be accepted as float! */ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 741 | /* Octal; c is first char of it */ |
| 742 | /* There's no 'isoctdigit' macro, sigh */ |
| 743 | while ('0' <= c && c < '8') { |
| 744 | c = tok_nextc(tok); |
| 745 | } |
| 746 | } |
Guido van Rossum | f023c46 | 1991-05-05 20:16:20 +0000 | [diff] [blame] | 747 | if (c == 'l' || c == 'L') |
| 748 | c = tok_nextc(tok); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 749 | } |
| 750 | else { |
| 751 | /* Decimal */ |
| 752 | do { |
| 753 | c = tok_nextc(tok); |
| 754 | } while (isdigit(c)); |
Guido van Rossum | f023c46 | 1991-05-05 20:16:20 +0000 | [diff] [blame] | 755 | if (c == 'l' || c == 'L') |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 756 | c = tok_nextc(tok); |
Guido van Rossum | f023c46 | 1991-05-05 20:16:20 +0000 | [diff] [blame] | 757 | else { |
| 758 | /* Accept floating point numbers. |
| 759 | XXX This accepts incomplete things like |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 760 | XXX 12e or 1e+; worry run-time */ |
Guido van Rossum | f023c46 | 1991-05-05 20:16:20 +0000 | [diff] [blame] | 761 | if (c == '.') { |
| 762 | fraction: |
| 763 | /* Fraction */ |
| 764 | do { |
| 765 | c = tok_nextc(tok); |
| 766 | } while (isdigit(c)); |
| 767 | } |
| 768 | if (c == 'e' || c == 'E') { |
| 769 | /* Exponent part */ |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 770 | c = tok_nextc(tok); |
Guido van Rossum | f023c46 | 1991-05-05 20:16:20 +0000 | [diff] [blame] | 771 | if (c == '+' || c == '-') |
| 772 | c = tok_nextc(tok); |
| 773 | while (isdigit(c)) { |
| 774 | c = tok_nextc(tok); |
| 775 | } |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 776 | } |
Guido van Rossum | f595fde | 1996-01-12 01:31:58 +0000 | [diff] [blame] | 777 | #ifndef WITHOUT_COMPLEX |
Guido van Rossum | faa436c | 1996-01-26 18:59:07 +0000 | [diff] [blame] | 778 | if (c == 'j' || c == 'J') |
Guido van Rossum | f595fde | 1996-01-12 01:31:58 +0000 | [diff] [blame] | 779 | /* Imaginary part */ |
| 780 | imaginary: |
| 781 | c = tok_nextc(tok); |
| 782 | #endif |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 783 | } |
| 784 | } |
| 785 | tok_backup(tok, c); |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 786 | *p_start = tok->start; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 787 | *p_end = tok->cur; |
| 788 | return NUMBER; |
| 789 | } |
Guido van Rossum | 24dacb3 | 1997-04-06 03:46:20 +0000 | [diff] [blame] | 790 | |
| 791 | letter_quote: |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 792 | /* String */ |
| 793 | if (c == '\'' || c == '"') { |
Guido van Rossum | 3568524 | 1998-02-16 15:42:50 +0000 | [diff] [blame] | 794 | int quote2 = tok->cur - tok->start + 1; |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 795 | int quote = c; |
| 796 | int triple = 0; |
| 797 | int tripcount = 0; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 798 | for (;;) { |
| 799 | c = tok_nextc(tok); |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 800 | if (c == '\n') { |
| 801 | if (!triple) { |
| 802 | tok->done = E_TOKEN; |
| 803 | tok_backup(tok, c); |
| 804 | return ERRORTOKEN; |
| 805 | } |
| 806 | tripcount = 0; |
| 807 | } |
| 808 | else if (c == EOF) { |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 809 | tok->done = E_TOKEN; |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 810 | tok->cur = tok->inp; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 811 | return ERRORTOKEN; |
| 812 | } |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 813 | else if (c == quote) { |
| 814 | tripcount++; |
Guido van Rossum | 3568524 | 1998-02-16 15:42:50 +0000 | [diff] [blame] | 815 | if (tok->cur - tok->start == quote2) { |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 816 | c = tok_nextc(tok); |
| 817 | if (c == quote) { |
| 818 | triple = 1; |
| 819 | tripcount = 0; |
| 820 | continue; |
| 821 | } |
| 822 | tok_backup(tok, c); |
| 823 | } |
| 824 | if (!triple || tripcount == 3) |
| 825 | break; |
| 826 | } |
| 827 | else if (c == '\\') { |
| 828 | tripcount = 0; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 829 | c = tok_nextc(tok); |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 830 | if (c == EOF) { |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 831 | tok->done = E_TOKEN; |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 832 | tok->cur = tok->inp; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 833 | return ERRORTOKEN; |
| 834 | } |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 835 | } |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 836 | else |
| 837 | tripcount = 0; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 838 | } |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 839 | *p_start = tok->start; |
Guido van Rossum | 8054fad | 1993-10-26 15:19:44 +0000 | [diff] [blame] | 840 | *p_end = tok->cur; |
| 841 | return STRING; |
| 842 | } |
| 843 | |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 844 | /* Line continuation */ |
| 845 | if (c == '\\') { |
| 846 | c = tok_nextc(tok); |
| 847 | if (c != '\n') { |
| 848 | tok->done = E_TOKEN; |
Guido van Rossum | 6ac258d | 1993-05-12 08:24:20 +0000 | [diff] [blame] | 849 | tok->cur = tok->inp; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 850 | return ERRORTOKEN; |
| 851 | } |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 852 | goto again; /* Read next line */ |
| 853 | } |
| 854 | |
Guido van Rossum | fbab905 | 1991-10-20 20:25:03 +0000 | [diff] [blame] | 855 | /* Check for two-character token */ |
| 856 | { |
| 857 | int c2 = tok_nextc(tok); |
Guido van Rossum | 86bea46 | 1997-04-29 21:03:06 +0000 | [diff] [blame] | 858 | int token = PyToken_TwoChars(c, c2); |
Guido van Rossum | fbab905 | 1991-10-20 20:25:03 +0000 | [diff] [blame] | 859 | if (token != OP) { |
Thomas Wouters | 434d082 | 2000-08-24 20:11:32 +0000 | [diff] [blame^] | 860 | int c3 = tok_nextc(tok); |
| 861 | int token3 = PyToken_ThreeChars(c, c2, c3); |
| 862 | if (token3 != OP) { |
| 863 | token = token3; |
| 864 | } else { |
| 865 | tok_backup(tok, c3); |
| 866 | } |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 867 | *p_start = tok->start; |
Guido van Rossum | fbab905 | 1991-10-20 20:25:03 +0000 | [diff] [blame] | 868 | *p_end = tok->cur; |
| 869 | return token; |
| 870 | } |
| 871 | tok_backup(tok, c2); |
| 872 | } |
| 873 | |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 874 | /* Keep track of parentheses nesting level */ |
Guido van Rossum | a849b83 | 1993-05-12 11:35:44 +0000 | [diff] [blame] | 875 | switch (c) { |
| 876 | case '(': |
| 877 | case '[': |
| 878 | case '{': |
| 879 | tok->level++; |
| 880 | break; |
| 881 | case ')': |
| 882 | case ']': |
| 883 | case '}': |
| 884 | tok->level--; |
| 885 | break; |
| 886 | } |
| 887 | |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 888 | /* Punctuation character */ |
Guido van Rossum | f4b1a64 | 1994-08-29 12:43:07 +0000 | [diff] [blame] | 889 | *p_start = tok->start; |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 890 | *p_end = tok->cur; |
Guido van Rossum | 86bea46 | 1997-04-29 21:03:06 +0000 | [diff] [blame] | 891 | return PyToken_OneChar(c); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 892 | } |
| 893 | |
| 894 | |
Guido van Rossum | 408027e | 1996-12-30 16:17:54 +0000 | [diff] [blame] | 895 | #ifdef Py_DEBUG |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 896 | |
| 897 | void |
Thomas Wouters | 23c9e00 | 2000-07-22 19:20:54 +0000 | [diff] [blame] | 898 | tok_dump(int type, char *start, char *end) |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 899 | { |
Guido van Rossum | 86bea46 | 1997-04-29 21:03:06 +0000 | [diff] [blame] | 900 | printf("%s", _PyParser_TokenNames[type]); |
Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 901 | if (type == NAME || type == NUMBER || type == STRING || type == OP) |
| 902 | printf("(%.*s)", (int)(end - start), start); |
| 903 | } |
| 904 | |
| 905 | #endif |