1 | /************************************************* |
---|
2 | * pcregrep program * |
---|
3 | *************************************************/ |
---|
4 | |
---|
5 | /* This is a grep program that uses the PCRE regular expression library to do |
---|
6 | its pattern matching. On a Unix system it can recurse into directories. */ |
---|
7 | |
---|
8 | #include <ctype.h> |
---|
9 | #include <stdio.h> |
---|
10 | #include <string.h> |
---|
11 | #include <stdlib.h> |
---|
12 | #include <errno.h> |
---|
13 | #include "config.h" |
---|
14 | #include "pcre.h" |
---|
15 | |
---|
16 | #define FALSE 0 |
---|
17 | #define TRUE 1 |
---|
18 | |
---|
19 | typedef int BOOL; |
---|
20 | |
---|
21 | #define VERSION "2.0 01-Aug-2001" |
---|
22 | #define MAX_PATTERN_COUNT 100 |
---|
23 | |
---|
24 | |
---|
25 | /************************************************* |
---|
26 | * Global variables * |
---|
27 | *************************************************/ |
---|
28 | |
---|
29 | static char *pattern_filename = NULL; |
---|
30 | static int pattern_count = 0; |
---|
31 | static pcre **pattern_list; |
---|
32 | static pcre_extra **hints_list; |
---|
33 | |
---|
34 | static BOOL count_only = FALSE; |
---|
35 | static BOOL filenames = TRUE; |
---|
36 | static BOOL filenames_only = FALSE; |
---|
37 | static BOOL invert = FALSE; |
---|
38 | static BOOL number = FALSE; |
---|
39 | static BOOL recurse = FALSE; |
---|
40 | static BOOL silent = FALSE; |
---|
41 | static BOOL whole_lines = FALSE; |
---|
42 | |
---|
43 | /* Structure for options and list of them */ |
---|
44 | |
---|
45 | typedef struct option_item { |
---|
46 | int one_char; |
---|
47 | char *long_name; |
---|
48 | char *help_text; |
---|
49 | } option_item; |
---|
50 | |
---|
51 | static option_item optionlist[] = { |
---|
52 | { -1, "help", "display this help and exit" }, |
---|
53 | { 'c', "count", "print only a count of matching lines per FILE" }, |
---|
54 | { 'h', "no-filename", "suppress the prefixing filename on output" }, |
---|
55 | { 'i', "ignore-case", "ignore case distinctions" }, |
---|
56 | { 'l', "files-with-matches", "print only FILE names containing matches" }, |
---|
57 | { 'n', "line-number", "print line number with output lines" }, |
---|
58 | { 'r', "recursive", "recursively scan sub-directories" }, |
---|
59 | { 's', "no-messages", "suppress error messages" }, |
---|
60 | { 'V', "version", "print version information and exit" }, |
---|
61 | { 'v', "invert-match", "select non-matching lines" }, |
---|
62 | { 'x', "line-regex", "force PATTERN to match only whole lines" }, |
---|
63 | { 'x', "line-regexp", "force PATTERN to match only whole lines" }, |
---|
64 | { 0, NULL, NULL } |
---|
65 | }; |
---|
66 | |
---|
67 | |
---|
68 | /************************************************* |
---|
69 | * Functions for directory scanning * |
---|
70 | *************************************************/ |
---|
71 | |
---|
72 | /* These functions are defined so that they can be made system specific, |
---|
73 | although at present the only ones are for Unix, and for "no directory recursion |
---|
74 | support". */ |
---|
75 | |
---|
76 | |
---|
77 | /************* Directory scanning in Unix ***********/ |
---|
78 | |
---|
79 | #if IS_UNIX |
---|
80 | #include <sys/types.h> |
---|
81 | #include <sys/stat.h> |
---|
82 | #include <dirent.h> |
---|
83 | |
---|
84 | typedef DIR directory_type; |
---|
85 | |
---|
86 | int |
---|
87 | isdirectory(char *filename) |
---|
88 | { |
---|
89 | struct stat statbuf; |
---|
90 | if (stat(filename, &statbuf) < 0) |
---|
91 | return 0; /* In the expectation that opening as a file will fail */ |
---|
92 | return ((statbuf.st_mode & S_IFMT) == S_IFDIR)? '/' : 0; |
---|
93 | } |
---|
94 | |
---|
95 | directory_type * |
---|
96 | opendirectory(char *filename) |
---|
97 | { |
---|
98 | return opendir(filename); |
---|
99 | } |
---|
100 | |
---|
101 | char * |
---|
102 | readdirectory(directory_type *dir) |
---|
103 | { |
---|
104 | for (;;) |
---|
105 | { |
---|
106 | struct dirent *dent = readdir(dir); |
---|
107 | if (dent == NULL) return NULL; |
---|
108 | if (strcmp(dent->d_name, ".") != 0 && strcmp(dent->d_name, "..") != 0) |
---|
109 | return dent->d_name; |
---|
110 | } |
---|
111 | return NULL; /* Keep compiler happy; never executed */ |
---|
112 | } |
---|
113 | |
---|
114 | void |
---|
115 | closedirectory(directory_type *dir) |
---|
116 | { |
---|
117 | closedir(dir); |
---|
118 | } |
---|
119 | |
---|
120 | |
---|
121 | #else |
---|
122 | |
---|
123 | |
---|
124 | /************* Directory scanning when we can't do it ***********/ |
---|
125 | |
---|
126 | /* The type is void, and apart from isdirectory(), the functions do nothing. */ |
---|
127 | |
---|
128 | typedef void directory_type; |
---|
129 | |
---|
130 | int isdirectory(char *filename) { return FALSE; } |
---|
131 | directory_type * opendirectory(char *filename) {} |
---|
132 | char *readdirectory(directory_type *dir) {} |
---|
133 | void closedirectory(directory_type *dir) {} |
---|
134 | |
---|
135 | #endif |
---|
136 | |
---|
137 | |
---|
138 | |
---|
139 | #if ! HAVE_STRERROR |
---|
140 | /************************************************* |
---|
141 | * Provide strerror() for non-ANSI libraries * |
---|
142 | *************************************************/ |
---|
143 | |
---|
144 | /* Some old-fashioned systems still around (e.g. SunOS4) don't have strerror() |
---|
145 | in their libraries, but can provide the same facility by this simple |
---|
146 | alternative function. */ |
---|
147 | |
---|
148 | extern int sys_nerr; |
---|
149 | extern char *sys_errlist[]; |
---|
150 | |
---|
151 | char * |
---|
152 | strerror(int n) |
---|
153 | { |
---|
154 | if (n < 0 || n >= sys_nerr) return "unknown error number"; |
---|
155 | return sys_errlist[n]; |
---|
156 | } |
---|
157 | #endif /* HAVE_STRERROR */ |
---|
158 | |
---|
159 | |
---|
160 | |
---|
161 | /************************************************* |
---|
162 | * Grep an individual file * |
---|
163 | *************************************************/ |
---|
164 | |
---|
165 | static int |
---|
166 | pcregrep(FILE *in, char *name) |
---|
167 | { |
---|
168 | int rc = 1; |
---|
169 | int linenumber = 0; |
---|
170 | int count = 0; |
---|
171 | int offsets[99]; |
---|
172 | char buffer[BUFSIZ]; |
---|
173 | |
---|
174 | while (fgets(buffer, sizeof(buffer), in) != NULL) |
---|
175 | { |
---|
176 | BOOL match = FALSE; |
---|
177 | int i; |
---|
178 | int length = (int)strlen(buffer); |
---|
179 | if (length > 0 && buffer[length-1] == '\n') buffer[--length] = 0; |
---|
180 | linenumber++; |
---|
181 | |
---|
182 | for (i = 0; !match && i < pattern_count; i++) |
---|
183 | { |
---|
184 | match = pcre_exec(pattern_list[i], hints_list[i], buffer, length, 0, 0, |
---|
185 | offsets, 99) >= 0; |
---|
186 | if (match && whole_lines && offsets[1] != length) match = FALSE; |
---|
187 | } |
---|
188 | |
---|
189 | if (match != invert) |
---|
190 | { |
---|
191 | if (count_only) count++; |
---|
192 | |
---|
193 | else if (filenames_only) |
---|
194 | { |
---|
195 | fprintf(stdout, "%s\n", (name == NULL)? "<stdin>" : name); |
---|
196 | return 0; |
---|
197 | } |
---|
198 | |
---|
199 | else if (silent) return 0; |
---|
200 | |
---|
201 | else |
---|
202 | { |
---|
203 | if (name != NULL) fprintf(stdout, "%s:", name); |
---|
204 | if (number) fprintf(stdout, "%d:", linenumber); |
---|
205 | fprintf(stdout, "%s\n", buffer); |
---|
206 | } |
---|
207 | |
---|
208 | rc = 0; |
---|
209 | } |
---|
210 | } |
---|
211 | |
---|
212 | if (count_only) |
---|
213 | { |
---|
214 | if (name != NULL) fprintf(stdout, "%s:", name); |
---|
215 | fprintf(stdout, "%d\n", count); |
---|
216 | } |
---|
217 | |
---|
218 | return rc; |
---|
219 | } |
---|
220 | |
---|
221 | |
---|
222 | |
---|
223 | |
---|
224 | /************************************************* |
---|
225 | * Grep a file or recurse into a directory * |
---|
226 | *************************************************/ |
---|
227 | |
---|
228 | static int |
---|
229 | grep_or_recurse(char *filename, BOOL recurse, BOOL show_filenames, |
---|
230 | BOOL only_one_at_top) |
---|
231 | { |
---|
232 | int rc = 1; |
---|
233 | int sep; |
---|
234 | FILE *in; |
---|
235 | |
---|
236 | /* If the file is a directory and we are recursing, scan each file within it. |
---|
237 | The scanning code is localized so it can be made system-specific. */ |
---|
238 | |
---|
239 | if ((sep = isdirectory(filename)) != 0 && recurse) |
---|
240 | { |
---|
241 | char buffer[1024]; |
---|
242 | char *nextfile; |
---|
243 | directory_type *dir = opendirectory(filename); |
---|
244 | |
---|
245 | if (dir == NULL) |
---|
246 | { |
---|
247 | fprintf(stderr, "pcregrep: Failed to open directory %s: %s\n", filename, |
---|
248 | strerror(errno)); |
---|
249 | return 2; |
---|
250 | } |
---|
251 | |
---|
252 | while ((nextfile = readdirectory(dir)) != NULL) |
---|
253 | { |
---|
254 | int frc; |
---|
255 | sprintf(buffer, "%.512s%c%.128s", filename, sep, nextfile); |
---|
256 | frc = grep_or_recurse(buffer, recurse, TRUE, FALSE); |
---|
257 | if (frc == 0 && rc == 1) rc = 0; |
---|
258 | } |
---|
259 | |
---|
260 | closedirectory(dir); |
---|
261 | return rc; |
---|
262 | } |
---|
263 | |
---|
264 | /* If the file is not a directory, or we are not recursing, scan it. If this is |
---|
265 | the first and only argument at top level, we don't show the file name. |
---|
266 | Otherwise, control is via the show_filenames variable. */ |
---|
267 | |
---|
268 | in = fopen(filename, "r"); |
---|
269 | if (in == NULL) |
---|
270 | { |
---|
271 | fprintf(stderr, "pcregrep: Failed to open %s: %s\n", filename, strerror(errno)); |
---|
272 | return 2; |
---|
273 | } |
---|
274 | |
---|
275 | rc = pcregrep(in, (show_filenames && !only_one_at_top)? filename : NULL); |
---|
276 | fclose(in); |
---|
277 | return rc; |
---|
278 | } |
---|
279 | |
---|
280 | |
---|
281 | |
---|
282 | |
---|
283 | /************************************************* |
---|
284 | * Usage function * |
---|
285 | *************************************************/ |
---|
286 | |
---|
287 | static int |
---|
288 | usage(int rc) |
---|
289 | { |
---|
290 | fprintf(stderr, "Usage: pcregrep [-Vcfhilnrsvx] [long-options] pattern [file] ...\n"); |
---|
291 | fprintf(stderr, "Type `pcregrep --help' for more information.\n"); |
---|
292 | return rc; |
---|
293 | } |
---|
294 | |
---|
295 | |
---|
296 | |
---|
297 | |
---|
298 | /************************************************* |
---|
299 | * Help function * |
---|
300 | *************************************************/ |
---|
301 | |
---|
302 | static void |
---|
303 | help(void) |
---|
304 | { |
---|
305 | option_item *op; |
---|
306 | |
---|
307 | printf("Usage: pcregrep [OPTION]... PATTERN [FILE] ...\n"); |
---|
308 | printf("Search for PATTERN in each FILE or standard input.\n"); |
---|
309 | printf("Example: pcregrep -i 'hello.*world' menu.h main.c\n\n"); |
---|
310 | |
---|
311 | printf("Options:\n"); |
---|
312 | |
---|
313 | for (op = optionlist; op->one_char != 0; op++) |
---|
314 | { |
---|
315 | int n; |
---|
316 | char s[4]; |
---|
317 | if (op->one_char > 0) sprintf(s, "-%c,", op->one_char); else strcpy(s, " "); |
---|
318 | printf(" %s --%s%n", s, op->long_name, &n); |
---|
319 | n = 30 - n; |
---|
320 | if (n < 1) n = 1; |
---|
321 | printf("%.*s%s\n", n, " ", op->help_text); |
---|
322 | } |
---|
323 | |
---|
324 | printf("\n -f<filename> or --file=<filename>\n"); |
---|
325 | printf(" Read patterns from <filename> instead of using a command line option.\n"); |
---|
326 | printf(" Trailing white space is removed; blanks lines are ignored.\n"); |
---|
327 | printf(" There is a maximum of %d patterns.\n", MAX_PATTERN_COUNT); |
---|
328 | |
---|
329 | printf("\nWith no FILE, read standard input. If fewer than two FILEs given, assume -h.\n"); |
---|
330 | printf("Exit status is 0 if any matches, 1 if no matches, and 2 if trouble.\n"); |
---|
331 | } |
---|
332 | |
---|
333 | |
---|
334 | |
---|
335 | |
---|
336 | /************************************************* |
---|
337 | * Handle an option * |
---|
338 | *************************************************/ |
---|
339 | |
---|
340 | static int |
---|
341 | handle_option(int letter, int options) |
---|
342 | { |
---|
343 | switch(letter) |
---|
344 | { |
---|
345 | case -1: help(); exit(0); |
---|
346 | case 'c': count_only = TRUE; break; |
---|
347 | case 'h': filenames = FALSE; break; |
---|
348 | case 'i': options |= PCRE_CASELESS; break; |
---|
349 | case 'l': filenames_only = TRUE; |
---|
350 | case 'n': number = TRUE; break; |
---|
351 | case 'r': recurse = TRUE; break; |
---|
352 | case 's': silent = TRUE; break; |
---|
353 | case 'v': invert = TRUE; break; |
---|
354 | case 'x': whole_lines = TRUE; options |= PCRE_ANCHORED; break; |
---|
355 | |
---|
356 | case 'V': |
---|
357 | fprintf(stderr, "pcregrep version %s using ", VERSION); |
---|
358 | fprintf(stderr, "PCRE version %s\n", pcre_version()); |
---|
359 | exit(0); |
---|
360 | break; |
---|
361 | |
---|
362 | default: |
---|
363 | fprintf(stderr, "pcregrep: Unknown option -%c\n", letter); |
---|
364 | exit(usage(2)); |
---|
365 | } |
---|
366 | |
---|
367 | return options; |
---|
368 | } |
---|
369 | |
---|
370 | |
---|
371 | |
---|
372 | |
---|
373 | /************************************************* |
---|
374 | * Main program * |
---|
375 | *************************************************/ |
---|
376 | |
---|
377 | int |
---|
378 | main(int argc, char **argv) |
---|
379 | { |
---|
380 | int i, j; |
---|
381 | int rc = 1; |
---|
382 | int options = 0; |
---|
383 | int errptr; |
---|
384 | const char *error; |
---|
385 | BOOL only_one_at_top; |
---|
386 | |
---|
387 | /* Process the options */ |
---|
388 | |
---|
389 | for (i = 1; i < argc; i++) |
---|
390 | { |
---|
391 | if (argv[i][0] != '-') break; |
---|
392 | |
---|
393 | /* Long name options */ |
---|
394 | |
---|
395 | if (argv[i][1] == '-') |
---|
396 | { |
---|
397 | option_item *op; |
---|
398 | |
---|
399 | if (strncmp(argv[i]+2, "file=", 5) == 0) |
---|
400 | { |
---|
401 | pattern_filename = argv[i] + 7; |
---|
402 | continue; |
---|
403 | } |
---|
404 | |
---|
405 | for (op = optionlist; op->one_char != 0; op++) |
---|
406 | { |
---|
407 | if (strcmp(argv[i]+2, op->long_name) == 0) |
---|
408 | { |
---|
409 | options = handle_option(op->one_char, options); |
---|
410 | break; |
---|
411 | } |
---|
412 | } |
---|
413 | if (op->one_char == 0) |
---|
414 | { |
---|
415 | fprintf(stderr, "pcregrep: Unknown option %s\n", argv[i]); |
---|
416 | exit(usage(2)); |
---|
417 | } |
---|
418 | } |
---|
419 | |
---|
420 | /* One-char options */ |
---|
421 | |
---|
422 | else |
---|
423 | { |
---|
424 | char *s = argv[i] + 1; |
---|
425 | while (*s != 0) |
---|
426 | { |
---|
427 | if (*s == 'f') |
---|
428 | { |
---|
429 | pattern_filename = s + 1; |
---|
430 | if (pattern_filename[0] == 0) |
---|
431 | { |
---|
432 | if (i >= argc - 1) |
---|
433 | { |
---|
434 | fprintf(stderr, "pcregrep: File name missing after -f\n"); |
---|
435 | exit(usage(2)); |
---|
436 | } |
---|
437 | pattern_filename = argv[++i]; |
---|
438 | } |
---|
439 | break; |
---|
440 | } |
---|
441 | else options = handle_option(*s++, options); |
---|
442 | } |
---|
443 | } |
---|
444 | } |
---|
445 | |
---|
446 | pattern_list = malloc(MAX_PATTERN_COUNT * sizeof(pcre *)); |
---|
447 | hints_list = malloc(MAX_PATTERN_COUNT * sizeof(pcre_extra *)); |
---|
448 | |
---|
449 | if (pattern_list == NULL || hints_list == NULL) |
---|
450 | { |
---|
451 | fprintf(stderr, "pcregrep: malloc failed\n"); |
---|
452 | return 2; |
---|
453 | } |
---|
454 | |
---|
455 | /* Compile the regular expression(s). */ |
---|
456 | |
---|
457 | if (pattern_filename != NULL) |
---|
458 | { |
---|
459 | FILE *f = fopen(pattern_filename, "r"); |
---|
460 | char buffer[BUFSIZ]; |
---|
461 | if (f == NULL) |
---|
462 | { |
---|
463 | fprintf(stderr, "pcregrep: Failed to open %s: %s\n", pattern_filename, |
---|
464 | strerror(errno)); |
---|
465 | return 2; |
---|
466 | } |
---|
467 | while (fgets(buffer, sizeof(buffer), f) != NULL) |
---|
468 | { |
---|
469 | char *s = buffer + (int)strlen(buffer); |
---|
470 | if (pattern_count >= MAX_PATTERN_COUNT) |
---|
471 | { |
---|
472 | fprintf(stderr, "pcregrep: Too many patterns in file (max %d)\n", |
---|
473 | MAX_PATTERN_COUNT); |
---|
474 | return 2; |
---|
475 | } |
---|
476 | while (s > buffer && isspace((unsigned char)(s[-1]))) s--; |
---|
477 | if (s == buffer) continue; |
---|
478 | *s = 0; |
---|
479 | pattern_list[pattern_count] = pcre_compile(buffer, options, &error, |
---|
480 | &errptr, NULL); |
---|
481 | if (pattern_list[pattern_count++] == NULL) |
---|
482 | { |
---|
483 | fprintf(stderr, "pcregrep: Error in regex number %d at offset %d: %s\n", |
---|
484 | pattern_count, errptr, error); |
---|
485 | return 2; |
---|
486 | } |
---|
487 | } |
---|
488 | fclose(f); |
---|
489 | } |
---|
490 | |
---|
491 | /* If no file name, a single regex must be given inline */ |
---|
492 | |
---|
493 | else |
---|
494 | { |
---|
495 | if (i >= argc) return usage(0); |
---|
496 | pattern_list[0] = pcre_compile(argv[i++], options, &error, &errptr, NULL); |
---|
497 | if (pattern_list[0] == NULL) |
---|
498 | { |
---|
499 | fprintf(stderr, "pcregrep: Error in regex at offset %d: %s\n", errptr, |
---|
500 | error); |
---|
501 | return 2; |
---|
502 | } |
---|
503 | pattern_count++; |
---|
504 | } |
---|
505 | |
---|
506 | /* Study the regular expressions, as we will be running them may times */ |
---|
507 | |
---|
508 | for (j = 0; j < pattern_count; j++) |
---|
509 | { |
---|
510 | hints_list[j] = pcre_study(pattern_list[j], 0, &error); |
---|
511 | if (error != NULL) |
---|
512 | { |
---|
513 | char s[16]; |
---|
514 | if (pattern_count == 1) s[0] = 0; else sprintf(s, " number %d", j); |
---|
515 | fprintf(stderr, "pcregrep: Error while studying regex%s: %s\n", s, error); |
---|
516 | return 2; |
---|
517 | } |
---|
518 | } |
---|
519 | |
---|
520 | /* If there are no further arguments, do the business on stdin and exit */ |
---|
521 | |
---|
522 | if (i >= argc) return pcregrep(stdin, NULL); |
---|
523 | |
---|
524 | /* Otherwise, work through the remaining arguments as files or directories. |
---|
525 | Pass in the fact that there is only one argument at top level - this suppresses |
---|
526 | the file name if the argument is not a directory. */ |
---|
527 | |
---|
528 | only_one_at_top = (i == argc - 1); |
---|
529 | if (filenames_only) filenames = TRUE; |
---|
530 | |
---|
531 | for (; i < argc; i++) |
---|
532 | { |
---|
533 | int frc = grep_or_recurse(argv[i], recurse, filenames, only_one_at_top); |
---|
534 | if (frc == 0 && rc == 1) rc = 0; |
---|
535 | } |
---|
536 | |
---|
537 | return rc; |
---|
538 | } |
---|
539 | |
---|
540 | /* End */ |
---|