Annotation of src/usr.bin/m4/gnum4.c, Revision 1.23
1.23 ! espie 1: /* $OpenBSD: gnum4.c,v 1.22 2003/06/30 21:42:50 espie Exp $ */
1.1 espie 2:
3: /*
4: * Copyright (c) 1999 Marc Espie
5: *
6: * Redistribution and use in source and binary forms, with or without
7: * modification, are permitted provided that the following conditions
8: * are met:
9: * 1. Redistributions of source code must retain the above copyright
10: * notice, this list of conditions and the following disclaimer.
11: * 2. Redistributions in binary form must reproduce the above copyright
12: * notice, this list of conditions and the following disclaimer in the
13: * documentation and/or other materials provided with the distribution.
14: *
15: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
16: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
17: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
18: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
19: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
20: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
21: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
22: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
23: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
24: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
25: * SUCH DAMAGE.
26: */
27:
28: /*
29: * functions needed to support gnu-m4 extensions, including a fake freezing
30: */
31:
32: #include <sys/param.h>
1.5 espie 33: #include <sys/types.h>
1.8 espie 34: #include <sys/wait.h>
1.5 espie 35: #include <ctype.h>
1.8 espie 36: #include <paths.h>
1.5 espie 37: #include <regex.h>
1.1 espie 38: #include <stddef.h>
39: #include <stdlib.h>
40: #include <stdio.h>
41: #include <string.h>
42: #include <err.h>
1.8 espie 43: #include <errno.h>
44: #include <unistd.h>
1.1 espie 45: #include "mdef.h"
46: #include "stdd.h"
47: #include "extern.h"
48:
1.5 espie 49:
50: int mimic_gnu = 0;
51:
1.1 espie 52: /*
53: * Support for include path search
54: * First search in the the current directory.
55: * If not found, and the path is not absolute, include path kicks in.
56: * First, -I options, in the order found on the command line.
57: * Then M4PATH env variable
58: */
59:
60: struct path_entry {
61: char *name;
62: struct path_entry *next;
63: } *first, *last;
64:
1.16 millert 65: static struct path_entry *new_path_entry(const char *);
66: static void ensure_m4path(void);
67: static struct input_file *dopath(struct input_file *, const char *);
1.3 espie 68:
1.1 espie 69: static struct path_entry *
1.18 espie 70: new_path_entry(const char *dirname)
1.1 espie 71: {
72: struct path_entry *n;
73:
74: n = malloc(sizeof(struct path_entry));
75: if (!n)
76: errx(1, "out of memory");
77: n->name = strdup(dirname);
78: if (!n->name)
79: errx(1, "out of memory");
80: n->next = 0;
81: return n;
82: }
83:
84: void
1.18 espie 85: addtoincludepath(const char *dirname)
1.1 espie 86: {
87: struct path_entry *n;
88:
89: n = new_path_entry(dirname);
90:
91: if (last) {
92: last->next = n;
93: last = n;
94: }
95: else
96: last = first = n;
97: }
98:
99: static void
100: ensure_m4path()
101: {
102: static int envpathdone = 0;
103: char *envpath;
104: char *sweep;
105: char *path;
106:
107: if (envpathdone)
108: return;
109: envpathdone = TRUE;
110: envpath = getenv("M4PATH");
111: if (!envpath)
112: return;
113: /* for portability: getenv result is read-only */
114: envpath = strdup(envpath);
115: if (!envpath)
116: errx(1, "out of memory");
117: for (sweep = envpath;
118: (path = strsep(&sweep, ":")) != NULL;)
119: addtoincludepath(path);
120: free(envpath);
121: }
122:
123: static
1.4 espie 124: struct input_file *
1.18 espie 125: dopath(struct input_file *i, const char *filename)
1.1 espie 126: {
127: char path[MAXPATHLEN];
128: struct path_entry *pe;
1.4 espie 129: FILE *f;
1.1 espie 130:
131: for (pe = first; pe; pe = pe->next) {
132: snprintf(path, sizeof(path), "%s/%s", pe->name, filename);
1.4 espie 133: if ((f = fopen(path, "r")) != 0) {
134: set_input(i, f, path);
135: return i;
136: }
1.1 espie 137: }
138: return NULL;
139: }
140:
1.4 espie 141: struct input_file *
1.18 espie 142: fopen_trypath(struct input_file *i, const char *filename)
1.1 espie 143: {
144: FILE *f;
145:
146: f = fopen(filename, "r");
1.4 espie 147: if (f != NULL) {
148: set_input(i, f, filename);
149: return i;
150: }
1.1 espie 151: if (filename[0] == '/')
152: return NULL;
153:
154: ensure_m4path();
155:
1.4 espie 156: return dopath(i, filename);
1.1 espie 157: }
158:
1.5 espie 159: void
1.18 espie 160: doindir(const char *argv[], int argc)
1.5 espie 161: {
1.22 espie 162: struct macro_definition *p;
1.5 espie 163:
1.22 espie 164: p = lookup_macro_definition(argv[2]);
1.5 espie 165: if (p == NULL)
166: errx(1, "undefined macro %s", argv[2]);
167: argv[1] = p->defn;
1.23 ! espie 168:
! 169: eval(argv+1, argc-1, p->type, traced_macros && is_traced(argv[2]));
1.5 espie 170: }
171:
172: void
1.18 espie 173: dobuiltin(const char *argv[], int argc)
1.5 espie 174: {
175: int n;
176: argv[1] = NULL;
177: n = builtin_type(argv[2]);
178: if (n != -1)
1.23 ! espie 179: eval(argv+1, argc-1, n, traced_macros && is_traced(argv[2]));
1.5 espie 180: else
181: errx(1, "unknown builtin %s", argv[2]);
182: }
183:
184:
185: /* We need some temporary buffer space, as pb pushes BACK and substitution
186: * proceeds forward... */
187: static char *buffer;
188: static size_t bufsize = 0;
189: static size_t current = 0;
190:
1.16 millert 191: static void addchars(const char *, size_t);
1.17 espie 192: static void addchar(int);
1.16 millert 193: static char *twiddle(const char *);
194: static char *getstring(void);
195: static void exit_regerror(int, regex_t *);
196: static void do_subst(const char *, regex_t *, const char *, regmatch_t *);
197: static void do_regexpindex(const char *, regex_t *, regmatch_t *);
198: static void do_regexp(const char *, regex_t *, const char *, regmatch_t *);
199: static void add_sub(int, const char *, regex_t *, regmatch_t *);
200: static void add_replace(const char *, regex_t *, const char *, regmatch_t *);
1.14 espie 201: #define addconstantstring(s) addchars((s), sizeof(s)-1)
1.5 espie 202:
203: static void
1.18 espie 204: addchars(const char *c, size_t n)
1.5 espie 205: {
206: if (n == 0)
207: return;
1.15 espie 208: while (current + n > bufsize) {
1.5 espie 209: if (bufsize == 0)
210: bufsize = 1024;
211: else
212: bufsize *= 2;
213: buffer = realloc(buffer, bufsize);
214: if (buffer == NULL)
215: errx(1, "out of memory");
216: }
217: memcpy(buffer+current, c, n);
218: current += n;
219: }
220:
221: static void
1.18 espie 222: addchar(int c)
1.5 espie 223: {
224: if (current +1 > bufsize) {
225: if (bufsize == 0)
226: bufsize = 1024;
227: else
228: bufsize *= 2;
229: buffer = realloc(buffer, bufsize);
230: if (buffer == NULL)
231: errx(1, "out of memory");
232: }
233: buffer[current++] = c;
234: }
235:
236: static char *
237: getstring()
238: {
239: addchar('\0');
240: current = 0;
241: return buffer;
242: }
243:
244:
245: static void
1.18 espie 246: exit_regerror(int er, regex_t *re)
1.5 espie 247: {
248: size_t errlen;
249: char *errbuf;
250:
251: errlen = regerror(er, re, NULL, 0);
252: errbuf = xalloc(errlen);
253: regerror(er, re, errbuf, errlen);
1.19 espie 254: errx(1, "%s at line %lu: regular expression error: %s",
255: CURRENT_NAME, CURRENT_LINE, errbuf);
1.5 espie 256: }
257:
258: static void
1.18 espie 259: add_sub(int n, const char *string, regex_t *re, regmatch_t *pm)
1.5 espie 260: {
261: if (n > re->re_nsub)
262: warnx("No subexpression %d", n);
263: /* Subexpressions that did not match are
264: * not an error. */
265: else if (pm[n].rm_so != -1 &&
266: pm[n].rm_eo != -1) {
267: addchars(string + pm[n].rm_so,
268: pm[n].rm_eo - pm[n].rm_so);
269: }
270: }
271:
272: /* Add replacement string to the output buffer, recognizing special
273: * constructs and replacing them with substrings of the original string.
274: */
275: static void
1.18 espie 276: add_replace(const char *string, regex_t *re, const char *replace, regmatch_t *pm)
1.5 espie 277: {
278: const char *p;
279:
280: for (p = replace; *p != '\0'; p++) {
281: if (*p == '&' && !mimic_gnu) {
282: add_sub(0, string, re, pm);
283: continue;
284: }
285: if (*p == '\\') {
286: if (p[1] == '\\') {
287: addchar(p[1]);
1.10 espie 288: p++;
1.5 espie 289: continue;
290: }
291: if (p[1] == '&') {
292: if (mimic_gnu)
293: add_sub(0, string, re, pm);
294: else
295: addchar(p[1]);
296: p++;
297: continue;
298: }
299: if (isdigit(p[1])) {
300: add_sub(*(++p) - '0', string, re, pm);
301: continue;
302: }
303: }
304: addchar(*p);
305: }
306: }
307:
308: static void
1.18 espie 309: do_subst(const char *string, regex_t *re, const char *replace, regmatch_t *pm)
1.5 espie 310: {
311: int error;
1.11 espie 312: int flags = 0;
313: const char *last_match = NULL;
1.5 espie 314:
1.11 espie 315: while ((error = regexec(re, string, re->re_nsub+1, pm, flags)) == 0) {
1.13 espie 316: if (pm[0].rm_eo != 0) {
317: if (string[pm[0].rm_eo-1] == '\n')
318: flags = 0;
319: else
320: flags = REG_NOTBOL;
321: }
1.5 espie 322:
323: /* NULL length matches are special... We use the `vi-mode'
324: * rule: don't allow a NULL-match at the last match
325: * position.
326: */
1.11 espie 327: if (pm[0].rm_so == pm[0].rm_eo &&
328: string + pm[0].rm_so == last_match) {
1.5 espie 329: if (*string == '\0')
330: return;
331: addchar(*string);
1.13 espie 332: if (*string++ == '\n')
333: flags = 0;
334: else
335: flags = REG_NOTBOL;
1.5 espie 336: continue;
337: }
1.11 espie 338: last_match = string + pm[0].rm_so;
339: addchars(string, pm[0].rm_so);
1.5 espie 340: add_replace(string, re, replace, pm);
341: string += pm[0].rm_eo;
342: }
343: if (error != REG_NOMATCH)
344: exit_regerror(error, re);
345: pbstr(string);
346: }
347:
348: static void
1.18 espie 349: do_regexp(const char *string, regex_t *re, const char *replace, regmatch_t *pm)
1.5 espie 350: {
351: int error;
352:
353: switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
354: case 0:
355: add_replace(string, re, replace, pm);
356: pbstr(getstring());
357: break;
358: case REG_NOMATCH:
359: break;
360: default:
361: exit_regerror(error, re);
362: }
363: }
364:
365: static void
1.18 espie 366: do_regexpindex(const char *string, regex_t *re, regmatch_t *pm)
1.5 espie 367: {
368: int error;
369:
370: switch(error = regexec(re, string, re->re_nsub+1, pm, 0)) {
371: case 0:
372: pbunsigned(pm[0].rm_so);
373: break;
374: case REG_NOMATCH:
375: pbnum(-1);
376: break;
377: default:
378: exit_regerror(error, re);
379: }
380: }
381:
382: /* In Gnu m4 mode, parentheses for backmatch don't work like POSIX 1003.2
383: * says. So we twiddle with the regexp before passing it to regcomp.
384: */
385: static char *
1.18 espie 386: twiddle(const char *p)
1.5 espie 387: {
388: /* This could use strcspn for speed... */
389: while (*p != '\0') {
1.14 espie 390: if (*p == '\\') {
391: switch(p[1]) {
392: case '(':
393: case ')':
394: case '|':
395: addchar(p[1]);
396: break;
397: case 'w':
398: addconstantstring("[_a-zA-Z0-9]");
399: break;
400: case 'W':
401: addconstantstring("[^_a-zA-Z0-9]");
402: break;
403: case '<':
404: addconstantstring("[[:<:]]");
405: break;
406: case '>':
407: addconstantstring("[[:>:]]");
408: break;
409: default:
410: addchars(p, 2);
411: break;
412: }
1.5 espie 413: p+=2;
414: continue;
415: }
1.14 espie 416: if (*p == '(' || *p == ')' || *p == '|')
1.5 espie 417: addchar('\\');
418:
419: addchar(*p);
420: p++;
421: }
422: return getstring();
423: }
424:
425: /* patsubst(string, regexp, opt replacement) */
426: /* argv[2]: string
427: * argv[3]: regexp
428: * argv[4]: opt rep
429: */
430: void
1.18 espie 431: dopatsubst(const char *argv[], int argc)
1.5 espie 432: {
433: if (argc <= 3) {
434: warnx("Too few arguments to patsubst");
435: return;
436: }
1.20 espie 437: /* special case: empty regexp */
438: if (argv[3][0] == '\0') {
439: const char *s;
1.21 espie 440: size_t len;
441: if (argv[4] && argc > 4)
442: len = strlen(argv[4]);
443: else
444: len = 0;
1.20 espie 445: for (s = argv[2]; *s != '\0'; s++) {
446: addchars(argv[4], len);
447: addchar(*s);
448: }
449: } else {
450: int error;
451: regex_t re;
452: regmatch_t *pmatch;
453:
454: error = regcomp(&re, mimic_gnu ? twiddle(argv[3]) : argv[3],
455: REG_NEWLINE | REG_EXTENDED);
456: if (error != 0)
457: exit_regerror(error, &re);
458:
459: pmatch = xalloc(sizeof(regmatch_t) * (re.re_nsub+1));
460: do_subst(argv[2], &re,
1.21 espie 461: argc > 4 && argv[4] != NULL ? argv[4] : "", pmatch);
1.20 espie 462: free(pmatch);
463: regfree(&re);
464: }
1.5 espie 465: pbstr(getstring());
466: }
467:
468: void
1.18 espie 469: doregexp(const char *argv[], int argc)
1.5 espie 470: {
471: int error;
472: regex_t re;
473: regmatch_t *pmatch;
474:
475: if (argc <= 3) {
1.7 espie 476: warnx("Too few arguments to regexp");
1.5 espie 477: return;
478: }
479: error = regcomp(&re, mimic_gnu ? twiddle(argv[3]) : argv[3],
480: REG_EXTENDED);
481: if (error != 0)
482: exit_regerror(error, &re);
483:
484: pmatch = xalloc(sizeof(regmatch_t) * (re.re_nsub+1));
1.7 espie 485: if (argv[4] == NULL || argc == 4)
1.5 espie 486: do_regexpindex(argv[2], &re, pmatch);
487: else
488: do_regexp(argv[2], &re, argv[4], pmatch);
489: free(pmatch);
490: regfree(&re);
1.8 espie 491: }
492:
493: void
1.18 espie 494: doesyscmd(const char *cmd)
1.8 espie 495: {
496: int p[2];
497: pid_t pid, cpid;
498: char *argv[4];
499: int cc;
500: int status;
501:
502: /* Follow gnu m4 documentation: first flush buffers. */
503: fflush(NULL);
504:
505: argv[0] = "sh";
506: argv[1] = "-c";
507: argv[2] = (char *)cmd;
508: argv[3] = NULL;
509:
510: /* Just set up standard output, share stderr and stdin with m4 */
511: if (pipe(p) == -1)
512: err(1, "bad pipe");
513: switch(cpid = fork()) {
514: case -1:
515: err(1, "bad fork");
516: /* NOTREACHED */
517: case 0:
518: (void) close(p[0]);
519: (void) dup2(p[1], 1);
520: (void) close(p[1]);
521: execv(_PATH_BSHELL, argv);
522: exit(1);
523: default:
524: /* Read result in two stages, since m4's buffer is
525: * pushback-only. */
526: (void) close(p[1]);
527: do {
528: char result[BUFSIZE];
529: cc = read(p[0], result, sizeof result);
530: if (cc > 0)
531: addchars(result, cc);
532: } while (cc > 0 || (cc == -1 && errno == EINTR));
533:
534: (void) close(p[0]);
535: while ((pid = wait(&status)) != cpid && pid >= 0)
536: continue;
537: pbstr(getstring());
538: }
1.5 espie 539: }