Annotation of embedaddon/pcre/pcregrep.c, revision 1.1.1.3
1.1 misho 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 or Win32 system it can recurse into
7: directories.
8:
1.1.1.2 misho 9: Copyright (c) 1997-2012 University of Cambridge
1.1 misho 10:
11: -----------------------------------------------------------------------------
12: Redistribution and use in source and binary forms, with or without
13: modification, are permitted provided that the following conditions are met:
14:
15: * Redistributions of source code must retain the above copyright notice,
16: this list of conditions and the following disclaimer.
17:
18: * Redistributions in binary form must reproduce the above copyright
19: notice, this list of conditions and the following disclaimer in the
20: documentation and/or other materials provided with the distribution.
21:
22: * Neither the name of the University of Cambridge nor the names of its
23: contributors may be used to endorse or promote products derived from
24: this software without specific prior written permission.
25:
26: THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
27: AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28: IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29: ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
30: LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
31: CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
32: SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
33: INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
34: CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
35: ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
36: POSSIBILITY OF SUCH DAMAGE.
37: -----------------------------------------------------------------------------
38: */
39:
40: #ifdef HAVE_CONFIG_H
41: #include "config.h"
42: #endif
43:
44: #include <ctype.h>
45: #include <locale.h>
46: #include <stdio.h>
47: #include <string.h>
48: #include <stdlib.h>
49: #include <errno.h>
50:
51: #include <sys/types.h>
52: #include <sys/stat.h>
53:
54: #ifdef HAVE_UNISTD_H
55: #include <unistd.h>
56: #endif
57:
58: #ifdef SUPPORT_LIBZ
59: #include <zlib.h>
60: #endif
61:
62: #ifdef SUPPORT_LIBBZ2
63: #include <bzlib.h>
64: #endif
65:
66: #include "pcre.h"
67:
68: #define FALSE 0
69: #define TRUE 1
70:
71: typedef int BOOL;
72:
73: #define MAX_PATTERN_COUNT 100
74: #define OFFSET_SIZE 99
75:
76: #if BUFSIZ > 8192
77: #define PATBUFSIZE BUFSIZ
78: #else
79: #define PATBUFSIZE 8192
80: #endif
81:
82: /* Values for the "filenames" variable, which specifies options for file name
83: output. The order is important; it is assumed that a file name is wanted for
84: all values greater than FN_DEFAULT. */
85:
86: enum { FN_NONE, FN_DEFAULT, FN_MATCH_ONLY, FN_NOMATCH_ONLY, FN_FORCE };
87:
88: /* File reading styles */
89:
90: enum { FR_PLAIN, FR_LIBZ, FR_LIBBZ2 };
91:
92: /* Actions for the -d and -D options */
93:
94: enum { dee_READ, dee_SKIP, dee_RECURSE };
95: enum { DEE_READ, DEE_SKIP };
96:
97: /* Actions for special processing options (flag bits) */
98:
99: #define PO_WORD_MATCH 0x0001
100: #define PO_LINE_MATCH 0x0002
101: #define PO_FIXED_STRINGS 0x0004
102:
103: /* Line ending types */
104:
105: enum { EL_LF, EL_CR, EL_CRLF, EL_ANY, EL_ANYCRLF };
106:
1.1.1.3 ! misho 107: /* Binary file options */
! 108:
! 109: enum { BIN_BINARY, BIN_NOMATCH, BIN_TEXT };
! 110:
1.1 misho 111: /* In newer versions of gcc, with FORTIFY_SOURCE set (the default in some
112: environments), a warning is issued if the value of fwrite() is ignored.
113: Unfortunately, casting to (void) does not suppress the warning. To get round
114: this, we use a macro that compiles a fudge. Oddly, this does not also seem to
115: apply to fprintf(). */
116:
117: #define FWRITE(a,b,c,d) if (fwrite(a,b,c,d)) {}
118:
119:
120:
121: /*************************************************
122: * Global variables *
123: *************************************************/
124:
125: /* Jeffrey Friedl has some debugging requirements that are not part of the
126: regular code. */
127:
128: #ifdef JFRIEDL_DEBUG
129: static int S_arg = -1;
130: static unsigned int jfriedl_XR = 0; /* repeat regex attempt this many times */
131: static unsigned int jfriedl_XT = 0; /* replicate text this many times */
132: static const char *jfriedl_prefix = "";
133: static const char *jfriedl_postfix = "";
134: #endif
135:
136: static int endlinetype;
137:
138: static char *colour_string = (char *)"1;31";
139: static char *colour_option = NULL;
140: static char *dee_option = NULL;
141: static char *DEE_option = NULL;
142: static char *main_buffer = NULL;
143: static char *newline = NULL;
144: static char *pattern_filename = NULL;
145: static char *stdin_name = (char *)"(standard input)";
146: static char *locale = NULL;
147:
148: static const unsigned char *pcretables = NULL;
149:
150: static int pattern_count = 0;
151: static pcre **pattern_list = NULL;
152: static pcre_extra **hints_list = NULL;
153:
1.1.1.3 ! misho 154: static char *file_list = NULL;
1.1 misho 155: static char *include_pattern = NULL;
156: static char *exclude_pattern = NULL;
157: static char *include_dir_pattern = NULL;
158: static char *exclude_dir_pattern = NULL;
159:
160: static pcre *include_compiled = NULL;
161: static pcre *exclude_compiled = NULL;
162: static pcre *include_dir_compiled = NULL;
163: static pcre *exclude_dir_compiled = NULL;
164:
165: static int after_context = 0;
166: static int before_context = 0;
1.1.1.3 ! misho 167: static int binary_files = BIN_BINARY;
1.1 misho 168: static int both_context = 0;
169: static int bufthird = PCREGREP_BUFSIZE;
170: static int bufsize = 3*PCREGREP_BUFSIZE;
171: static int dee_action = dee_READ;
172: static int DEE_action = DEE_READ;
173: static int error_count = 0;
174: static int filenames = FN_DEFAULT;
175: static int only_matching = -1;
176: static int process_options = 0;
177:
178: #ifdef SUPPORT_PCREGREP_JIT
179: static int study_options = PCRE_STUDY_JIT_COMPILE;
180: #else
181: static int study_options = 0;
182: #endif
183:
184: static unsigned long int match_limit = 0;
185: static unsigned long int match_limit_recursion = 0;
186:
187: static BOOL count_only = FALSE;
188: static BOOL do_colour = FALSE;
189: static BOOL file_offsets = FALSE;
190: static BOOL hyphenpending = FALSE;
191: static BOOL invert = FALSE;
192: static BOOL line_buffered = FALSE;
193: static BOOL line_offsets = FALSE;
194: static BOOL multiline = FALSE;
195: static BOOL number = FALSE;
196: static BOOL omit_zero_count = FALSE;
197: static BOOL resource_error = FALSE;
198: static BOOL quiet = FALSE;
199: static BOOL silent = FALSE;
200: static BOOL utf8 = FALSE;
201:
202: /* Structure for options and list of them */
203:
204: enum { OP_NODATA, OP_STRING, OP_OP_STRING, OP_NUMBER, OP_LONGNUMBER,
1.1.1.3 ! misho 205: OP_OP_NUMBER, OP_PATLIST, OP_BINFILES };
1.1 misho 206:
207: typedef struct option_item {
208: int type;
209: int one_char;
210: void *dataptr;
211: const char *long_name;
212: const char *help_text;
213: } option_item;
214:
215: /* Options without a single-letter equivalent get a negative value. This can be
216: used to identify them. */
217:
218: #define N_COLOUR (-1)
219: #define N_EXCLUDE (-2)
220: #define N_EXCLUDE_DIR (-3)
221: #define N_HELP (-4)
222: #define N_INCLUDE (-5)
223: #define N_INCLUDE_DIR (-6)
224: #define N_LABEL (-7)
225: #define N_LOCALE (-8)
226: #define N_NULL (-9)
227: #define N_LOFFSETS (-10)
228: #define N_FOFFSETS (-11)
229: #define N_LBUFFER (-12)
230: #define N_M_LIMIT (-13)
231: #define N_M_LIMIT_REC (-14)
232: #define N_BUFSIZE (-15)
233: #define N_NOJIT (-16)
1.1.1.3 ! misho 234: #define N_FILE_LIST (-17)
! 235: #define N_BINARY_FILES (-18)
1.1 misho 236:
237: static option_item optionlist[] = {
1.1.1.3 ! misho 238: { OP_NODATA, N_NULL, NULL, "", "terminate options" },
1.1 misho 239: { OP_NODATA, N_HELP, NULL, "help", "display this help and exit" },
240: { OP_NUMBER, 'A', &after_context, "after-context=number", "set number of following context lines" },
1.1.1.3 ! misho 241: { OP_NODATA, 'a', NULL, "text", "treat binary files as text" },
1.1 misho 242: { OP_NUMBER, 'B', &before_context, "before-context=number", "set number of prior context lines" },
1.1.1.3 ! misho 243: { OP_BINFILES, N_BINARY_FILES, NULL, "binary-files=word", "set treatment of binary files" },
1.1 misho 244: { OP_NUMBER, N_BUFSIZE,&bufthird, "buffer-size=number", "set processing buffer size parameter" },
245: { OP_OP_STRING, N_COLOUR, &colour_option, "color=option", "matched text color option" },
246: { OP_OP_STRING, N_COLOUR, &colour_option, "colour=option", "matched text colour option" },
247: { OP_NUMBER, 'C', &both_context, "context=number", "set number of context lines, before & after" },
248: { OP_NODATA, 'c', NULL, "count", "print only a count of matching lines per FILE" },
249: { OP_STRING, 'D', &DEE_option, "devices=action","how to handle devices, FIFOs, and sockets" },
250: { OP_STRING, 'd', &dee_option, "directories=action", "how to handle directories" },
251: { OP_PATLIST, 'e', NULL, "regex(p)=pattern", "specify pattern (may be used more than once)" },
252: { OP_NODATA, 'F', NULL, "fixed-strings", "patterns are sets of newline-separated strings" },
253: { OP_STRING, 'f', &pattern_filename, "file=path", "read patterns from file" },
1.1.1.3 ! misho 254: { OP_STRING, N_FILE_LIST, &file_list, "file-list=path","read files to search from file" },
1.1 misho 255: { OP_NODATA, N_FOFFSETS, NULL, "file-offsets", "output file offsets, not text" },
256: { OP_NODATA, 'H', NULL, "with-filename", "force the prefixing filename on output" },
257: { OP_NODATA, 'h', NULL, "no-filename", "suppress the prefixing filename on output" },
1.1.1.3 ! misho 258: { OP_NODATA, 'I', NULL, "", "treat binary files as not matching (ignore)" },
1.1 misho 259: { OP_NODATA, 'i', NULL, "ignore-case", "ignore case distinctions" },
260: #ifdef SUPPORT_PCREGREP_JIT
261: { OP_NODATA, N_NOJIT, NULL, "no-jit", "do not use just-in-time compiler optimization" },
262: #else
263: { OP_NODATA, N_NOJIT, NULL, "no-jit", "ignored: this pcregrep does not support JIT" },
264: #endif
265: { OP_NODATA, 'l', NULL, "files-with-matches", "print only FILE names containing matches" },
266: { OP_NODATA, 'L', NULL, "files-without-match","print only FILE names not containing matches" },
267: { OP_STRING, N_LABEL, &stdin_name, "label=name", "set name for standard input" },
268: { OP_NODATA, N_LBUFFER, NULL, "line-buffered", "use line buffering" },
269: { OP_NODATA, N_LOFFSETS, NULL, "line-offsets", "output line numbers and offsets, not text" },
270: { OP_STRING, N_LOCALE, &locale, "locale=locale", "use the named locale" },
271: { OP_LONGNUMBER, N_M_LIMIT, &match_limit, "match-limit=number", "set PCRE match limit option" },
272: { OP_LONGNUMBER, N_M_LIMIT_REC, &match_limit_recursion, "recursion-limit=number", "set PCRE match recursion limit option" },
273: { OP_NODATA, 'M', NULL, "multiline", "run in multiline mode" },
274: { OP_STRING, 'N', &newline, "newline=type", "set newline type (CR, LF, CRLF, ANYCRLF or ANY)" },
275: { OP_NODATA, 'n', NULL, "line-number", "print line number with output lines" },
276: { OP_OP_NUMBER, 'o', &only_matching, "only-matching=n", "show only the part of the line that matched" },
277: { OP_NODATA, 'q', NULL, "quiet", "suppress output, just set return code" },
278: { OP_NODATA, 'r', NULL, "recursive", "recursively scan sub-directories" },
279: { OP_STRING, N_EXCLUDE,&exclude_pattern, "exclude=pattern","exclude matching files when recursing" },
280: { OP_STRING, N_INCLUDE,&include_pattern, "include=pattern","include matching files when recursing" },
281: { OP_STRING, N_EXCLUDE_DIR,&exclude_dir_pattern, "exclude-dir=pattern","exclude matching directories when recursing" },
282: { OP_STRING, N_INCLUDE_DIR,&include_dir_pattern, "include-dir=pattern","include matching directories when recursing" },
283:
284: /* These two were accidentally implemented with underscores instead of
285: hyphens in the option names. As this was not discovered for several releases,
286: the incorrect versions are left in the table for compatibility. However, the
287: --help function misses out any option that has an underscore in its name. */
288:
289: { OP_STRING, N_EXCLUDE_DIR,&exclude_dir_pattern, "exclude_dir=pattern","exclude matching directories when recursing" },
290: { OP_STRING, N_INCLUDE_DIR,&include_dir_pattern, "include_dir=pattern","include matching directories when recursing" },
291:
292: #ifdef JFRIEDL_DEBUG
293: { OP_OP_NUMBER, 'S', &S_arg, "jeffS", "replace matched (sub)string with X" },
294: #endif
295: { OP_NODATA, 's', NULL, "no-messages", "suppress error messages" },
296: { OP_NODATA, 'u', NULL, "utf-8", "use UTF-8 mode" },
297: { OP_NODATA, 'V', NULL, "version", "print version information and exit" },
298: { OP_NODATA, 'v', NULL, "invert-match", "select non-matching lines" },
299: { OP_NODATA, 'w', NULL, "word-regex(p)", "force patterns to match only as words" },
300: { OP_NODATA, 'x', NULL, "line-regex(p)", "force patterns to match only whole lines" },
301: { OP_NODATA, 0, NULL, NULL, NULL }
302: };
303:
304: /* Tables for prefixing and suffixing patterns, according to the -w, -x, and -F
305: options. These set the 1, 2, and 4 bits in process_options, respectively. Note
306: that the combination of -w and -x has the same effect as -x on its own, so we
307: can treat them as the same. */
308:
309: static const char *prefix[] = {
310: "", "\\b", "^(?:", "^(?:", "\\Q", "\\b\\Q", "^(?:\\Q", "^(?:\\Q" };
311:
312: static const char *suffix[] = {
313: "", "\\b", ")$", ")$", "\\E", "\\E\\b", "\\E)$", "\\E)$" };
314:
315: /* UTF-8 tables - used only when the newline setting is "any". */
316:
317: const int utf8_table3[] = { 0xff, 0x1f, 0x0f, 0x07, 0x03, 0x01};
318:
319: const char utf8_table4[] = {
320: 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
321: 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
322: 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
323: 3,3,3,3,3,3,3,3,4,4,4,4,5,5,5,5 };
324:
325:
326:
327: /*************************************************
328: * Exit from the program *
329: *************************************************/
330:
331: /* If there has been a resource error, give a suitable message.
332:
333: Argument: the return code
334: Returns: does not return
335: */
336:
337: static void
338: pcregrep_exit(int rc)
339: {
340: if (resource_error)
341: {
342: fprintf(stderr, "pcregrep: Error %d, %d or %d means that a resource limit "
343: "was exceeded.\n", PCRE_ERROR_MATCHLIMIT, PCRE_ERROR_RECURSIONLIMIT,
344: PCRE_ERROR_JIT_STACKLIMIT);
345: fprintf(stderr, "pcregrep: Check your regex for nested unlimited loops.\n");
346: }
347:
348: exit(rc);
349: }
350:
351:
352: /*************************************************
353: * OS-specific functions *
354: *************************************************/
355:
356: /* These functions are defined so that they can be made system specific,
357: although at present the only ones are for Unix, Win32, and for "no support". */
358:
359:
360: /************* Directory scanning in Unix ***********/
361:
362: #if defined HAVE_SYS_STAT_H && defined HAVE_DIRENT_H && defined HAVE_SYS_TYPES_H
363: #include <sys/types.h>
364: #include <sys/stat.h>
365: #include <dirent.h>
366:
367: typedef DIR directory_type;
368:
369: static int
370: isdirectory(char *filename)
371: {
372: struct stat statbuf;
373: if (stat(filename, &statbuf) < 0)
374: return 0; /* In the expectation that opening as a file will fail */
375: return ((statbuf.st_mode & S_IFMT) == S_IFDIR)? '/' : 0;
376: }
377:
378: static directory_type *
379: opendirectory(char *filename)
380: {
381: return opendir(filename);
382: }
383:
384: static char *
385: readdirectory(directory_type *dir)
386: {
387: for (;;)
388: {
389: struct dirent *dent = readdir(dir);
390: if (dent == NULL) return NULL;
391: if (strcmp(dent->d_name, ".") != 0 && strcmp(dent->d_name, "..") != 0)
392: return dent->d_name;
393: }
394: /* Control never reaches here */
395: }
396:
397: static void
398: closedirectory(directory_type *dir)
399: {
400: closedir(dir);
401: }
402:
403:
404: /************* Test for regular file in Unix **********/
405:
406: static int
407: isregfile(char *filename)
408: {
409: struct stat statbuf;
410: if (stat(filename, &statbuf) < 0)
411: return 1; /* In the expectation that opening as a file will fail */
412: return (statbuf.st_mode & S_IFMT) == S_IFREG;
413: }
414:
415:
416: /************* Test for a terminal in Unix **********/
417:
418: static BOOL
419: is_stdout_tty(void)
420: {
421: return isatty(fileno(stdout));
422: }
423:
424: static BOOL
425: is_file_tty(FILE *f)
426: {
427: return isatty(fileno(f));
428: }
429:
430:
431: /************* Directory scanning in Win32 ***********/
432:
433: /* I (Philip Hazel) have no means of testing this code. It was contributed by
434: Lionel Fourquaux. David Burgess added a patch to define INVALID_FILE_ATTRIBUTES
435: when it did not exist. David Byron added a patch that moved the #include of
436: <windows.h> to before the INVALID_FILE_ATTRIBUTES definition rather than after.
437: The double test below stops gcc 4.4.4 grumbling that HAVE_WINDOWS_H is
438: undefined when it is indeed undefined. */
439:
440: #elif defined HAVE_WINDOWS_H && HAVE_WINDOWS_H
441:
442: #ifndef STRICT
443: # define STRICT
444: #endif
445: #ifndef WIN32_LEAN_AND_MEAN
446: # define WIN32_LEAN_AND_MEAN
447: #endif
448:
449: #include <windows.h>
450:
451: #ifndef INVALID_FILE_ATTRIBUTES
452: #define INVALID_FILE_ATTRIBUTES 0xFFFFFFFF
453: #endif
454:
455: typedef struct directory_type
456: {
457: HANDLE handle;
458: BOOL first;
459: WIN32_FIND_DATA data;
460: } directory_type;
461:
462: int
463: isdirectory(char *filename)
464: {
465: DWORD attr = GetFileAttributes(filename);
466: if (attr == INVALID_FILE_ATTRIBUTES)
467: return 0;
468: return ((attr & FILE_ATTRIBUTE_DIRECTORY) != 0) ? '/' : 0;
469: }
470:
471: directory_type *
472: opendirectory(char *filename)
473: {
474: size_t len;
475: char *pattern;
476: directory_type *dir;
477: DWORD err;
478: len = strlen(filename);
479: pattern = (char *) malloc(len + 3);
480: dir = (directory_type *) malloc(sizeof(*dir));
481: if ((pattern == NULL) || (dir == NULL))
482: {
483: fprintf(stderr, "pcregrep: malloc failed\n");
484: pcregrep_exit(2);
485: }
486: memcpy(pattern, filename, len);
487: memcpy(&(pattern[len]), "\\*", 3);
488: dir->handle = FindFirstFile(pattern, &(dir->data));
489: if (dir->handle != INVALID_HANDLE_VALUE)
490: {
491: free(pattern);
492: dir->first = TRUE;
493: return dir;
494: }
495: err = GetLastError();
496: free(pattern);
497: free(dir);
498: errno = (err == ERROR_ACCESS_DENIED) ? EACCES : ENOENT;
499: return NULL;
500: }
501:
502: char *
503: readdirectory(directory_type *dir)
504: {
505: for (;;)
506: {
507: if (!dir->first)
508: {
509: if (!FindNextFile(dir->handle, &(dir->data)))
510: return NULL;
511: }
512: else
513: {
514: dir->first = FALSE;
515: }
516: if (strcmp(dir->data.cFileName, ".") != 0 && strcmp(dir->data.cFileName, "..") != 0)
517: return dir->data.cFileName;
518: }
519: #ifndef _MSC_VER
520: return NULL; /* Keep compiler happy; never executed */
521: #endif
522: }
523:
524: void
525: closedirectory(directory_type *dir)
526: {
527: FindClose(dir->handle);
528: free(dir);
529: }
530:
531:
532: /************* Test for regular file in Win32 **********/
533:
534: /* I don't know how to do this, or if it can be done; assume all paths are
535: regular if they are not directories. */
536:
537: int isregfile(char *filename)
538: {
539: return !isdirectory(filename);
540: }
541:
542:
543: /************* Test for a terminal in Win32 **********/
544:
545: /* I don't know how to do this; assume never */
546:
547: static BOOL
548: is_stdout_tty(void)
549: {
550: return FALSE;
551: }
552:
553: static BOOL
554: is_file_tty(FILE *f)
555: {
556: return FALSE;
557: }
558:
559:
560: /************* Directory scanning when we can't do it ***********/
561:
562: /* The type is void, and apart from isdirectory(), the functions do nothing. */
563:
564: #else
565:
566: typedef void directory_type;
567:
568: int isdirectory(char *filename) { return 0; }
569: directory_type * opendirectory(char *filename) { return (directory_type*)0;}
570: char *readdirectory(directory_type *dir) { return (char*)0;}
571: void closedirectory(directory_type *dir) {}
572:
573:
574: /************* Test for regular when we can't do it **********/
575:
576: /* Assume all files are regular. */
577:
578: int isregfile(char *filename) { return 1; }
579:
580:
581: /************* Test for a terminal when we can't do it **********/
582:
583: static BOOL
584: is_stdout_tty(void)
585: {
586: return FALSE;
587: }
588:
589: static BOOL
590: is_file_tty(FILE *f)
591: {
592: return FALSE;
593: }
594:
595: #endif
596:
597:
598:
599: #ifndef HAVE_STRERROR
600: /*************************************************
601: * Provide strerror() for non-ANSI libraries *
602: *************************************************/
603:
604: /* Some old-fashioned systems still around (e.g. SunOS4) don't have strerror()
605: in their libraries, but can provide the same facility by this simple
606: alternative function. */
607:
608: extern int sys_nerr;
609: extern char *sys_errlist[];
610:
611: char *
612: strerror(int n)
613: {
614: if (n < 0 || n >= sys_nerr) return "unknown error number";
615: return sys_errlist[n];
616: }
617: #endif /* HAVE_STRERROR */
618:
619:
620:
621: /*************************************************
622: * Read one line of input *
623: *************************************************/
624:
625: /* Normally, input is read using fread() into a large buffer, so many lines may
626: be read at once. However, doing this for tty input means that no output appears
627: until a lot of input has been typed. Instead, tty input is handled line by
628: line. We cannot use fgets() for this, because it does not stop at a binary
629: zero, and therefore there is no way of telling how many characters it has read,
630: because there may be binary zeros embedded in the data.
631:
632: Arguments:
633: buffer the buffer to read into
634: length the maximum number of characters to read
635: f the file
636:
637: Returns: the number of characters read, zero at end of file
638: */
639:
1.1.1.2 misho 640: static unsigned int
1.1 misho 641: read_one_line(char *buffer, int length, FILE *f)
642: {
643: int c;
644: int yield = 0;
645: while ((c = fgetc(f)) != EOF)
646: {
647: buffer[yield++] = c;
648: if (c == '\n' || yield >= length) break;
649: }
650: return yield;
651: }
652:
653:
654:
655: /*************************************************
656: * Find end of line *
657: *************************************************/
658:
659: /* The length of the endline sequence that is found is set via lenptr. This may
660: be zero at the very end of the file if there is no line-ending sequence there.
661:
662: Arguments:
663: p current position in line
664: endptr end of available data
665: lenptr where to put the length of the eol sequence
666:
667: Returns: pointer after the last byte of the line,
668: including the newline byte(s)
669: */
670:
671: static char *
672: end_of_line(char *p, char *endptr, int *lenptr)
673: {
674: switch(endlinetype)
675: {
676: default: /* Just in case */
677: case EL_LF:
678: while (p < endptr && *p != '\n') p++;
679: if (p < endptr)
680: {
681: *lenptr = 1;
682: return p + 1;
683: }
684: *lenptr = 0;
685: return endptr;
686:
687: case EL_CR:
688: while (p < endptr && *p != '\r') p++;
689: if (p < endptr)
690: {
691: *lenptr = 1;
692: return p + 1;
693: }
694: *lenptr = 0;
695: return endptr;
696:
697: case EL_CRLF:
698: for (;;)
699: {
700: while (p < endptr && *p != '\r') p++;
701: if (++p >= endptr)
702: {
703: *lenptr = 0;
704: return endptr;
705: }
706: if (*p == '\n')
707: {
708: *lenptr = 2;
709: return p + 1;
710: }
711: }
712: break;
713:
714: case EL_ANYCRLF:
715: while (p < endptr)
716: {
717: int extra = 0;
718: register int c = *((unsigned char *)p);
719:
720: if (utf8 && c >= 0xc0)
721: {
722: int gcii, gcss;
723: extra = utf8_table4[c & 0x3f]; /* Number of additional bytes */
724: gcss = 6*extra;
725: c = (c & utf8_table3[extra]) << gcss;
726: for (gcii = 1; gcii <= extra; gcii++)
727: {
728: gcss -= 6;
729: c |= (p[gcii] & 0x3f) << gcss;
730: }
731: }
732:
733: p += 1 + extra;
734:
735: switch (c)
736: {
737: case 0x0a: /* LF */
738: *lenptr = 1;
739: return p;
740:
741: case 0x0d: /* CR */
742: if (p < endptr && *p == 0x0a)
743: {
744: *lenptr = 2;
745: p++;
746: }
747: else *lenptr = 1;
748: return p;
749:
750: default:
751: break;
752: }
753: } /* End of loop for ANYCRLF case */
754:
755: *lenptr = 0; /* Must have hit the end */
756: return endptr;
757:
758: case EL_ANY:
759: while (p < endptr)
760: {
761: int extra = 0;
762: register int c = *((unsigned char *)p);
763:
764: if (utf8 && c >= 0xc0)
765: {
766: int gcii, gcss;
767: extra = utf8_table4[c & 0x3f]; /* Number of additional bytes */
768: gcss = 6*extra;
769: c = (c & utf8_table3[extra]) << gcss;
770: for (gcii = 1; gcii <= extra; gcii++)
771: {
772: gcss -= 6;
773: c |= (p[gcii] & 0x3f) << gcss;
774: }
775: }
776:
777: p += 1 + extra;
778:
779: switch (c)
780: {
781: case 0x0a: /* LF */
782: case 0x0b: /* VT */
783: case 0x0c: /* FF */
784: *lenptr = 1;
785: return p;
786:
787: case 0x0d: /* CR */
788: if (p < endptr && *p == 0x0a)
789: {
790: *lenptr = 2;
791: p++;
792: }
793: else *lenptr = 1;
794: return p;
795:
796: case 0x85: /* NEL */
797: *lenptr = utf8? 2 : 1;
798: return p;
799:
800: case 0x2028: /* LS */
801: case 0x2029: /* PS */
802: *lenptr = 3;
803: return p;
804:
805: default:
806: break;
807: }
808: } /* End of loop for ANY case */
809:
810: *lenptr = 0; /* Must have hit the end */
811: return endptr;
812: } /* End of overall switch */
813: }
814:
815:
816:
817: /*************************************************
818: * Find start of previous line *
819: *************************************************/
820:
821: /* This is called when looking back for before lines to print.
822:
823: Arguments:
824: p start of the subsequent line
825: startptr start of available data
826:
827: Returns: pointer to the start of the previous line
828: */
829:
830: static char *
831: previous_line(char *p, char *startptr)
832: {
833: switch(endlinetype)
834: {
835: default: /* Just in case */
836: case EL_LF:
837: p--;
838: while (p > startptr && p[-1] != '\n') p--;
839: return p;
840:
841: case EL_CR:
842: p--;
843: while (p > startptr && p[-1] != '\n') p--;
844: return p;
845:
846: case EL_CRLF:
847: for (;;)
848: {
849: p -= 2;
850: while (p > startptr && p[-1] != '\n') p--;
851: if (p <= startptr + 1 || p[-2] == '\r') return p;
852: }
853: return p; /* But control should never get here */
854:
855: case EL_ANY:
856: case EL_ANYCRLF:
857: if (*(--p) == '\n' && p > startptr && p[-1] == '\r') p--;
858: if (utf8) while ((*p & 0xc0) == 0x80) p--;
859:
860: while (p > startptr)
861: {
862: register int c;
863: char *pp = p - 1;
864:
865: if (utf8)
866: {
867: int extra = 0;
868: while ((*pp & 0xc0) == 0x80) pp--;
869: c = *((unsigned char *)pp);
870: if (c >= 0xc0)
871: {
872: int gcii, gcss;
873: extra = utf8_table4[c & 0x3f]; /* Number of additional bytes */
874: gcss = 6*extra;
875: c = (c & utf8_table3[extra]) << gcss;
876: for (gcii = 1; gcii <= extra; gcii++)
877: {
878: gcss -= 6;
879: c |= (pp[gcii] & 0x3f) << gcss;
880: }
881: }
882: }
883: else c = *((unsigned char *)pp);
884:
885: if (endlinetype == EL_ANYCRLF) switch (c)
886: {
887: case 0x0a: /* LF */
888: case 0x0d: /* CR */
889: return p;
890:
891: default:
892: break;
893: }
894:
895: else switch (c)
896: {
897: case 0x0a: /* LF */
898: case 0x0b: /* VT */
899: case 0x0c: /* FF */
900: case 0x0d: /* CR */
901: case 0x85: /* NEL */
902: case 0x2028: /* LS */
903: case 0x2029: /* PS */
904: return p;
905:
906: default:
907: break;
908: }
909:
910: p = pp; /* Back one character */
911: } /* End of loop for ANY case */
912:
913: return startptr; /* Hit start of data */
914: } /* End of overall switch */
915: }
916:
917:
918:
919:
920:
921: /*************************************************
922: * Print the previous "after" lines *
923: *************************************************/
924:
925: /* This is called if we are about to lose said lines because of buffer filling,
926: and at the end of the file. The data in the line is written using fwrite() so
927: that a binary zero does not terminate it.
928:
929: Arguments:
930: lastmatchnumber the number of the last matching line, plus one
931: lastmatchrestart where we restarted after the last match
932: endptr end of available data
933: printname filename for printing
934:
935: Returns: nothing
936: */
937:
938: static void do_after_lines(int lastmatchnumber, char *lastmatchrestart,
939: char *endptr, char *printname)
940: {
941: if (after_context > 0 && lastmatchnumber > 0)
942: {
943: int count = 0;
944: while (lastmatchrestart < endptr && count++ < after_context)
945: {
946: int ellength;
947: char *pp = lastmatchrestart;
948: if (printname != NULL) fprintf(stdout, "%s-", printname);
949: if (number) fprintf(stdout, "%d-", lastmatchnumber++);
950: pp = end_of_line(pp, endptr, &ellength);
951: FWRITE(lastmatchrestart, 1, pp - lastmatchrestart, stdout);
952: lastmatchrestart = pp;
953: }
954: hyphenpending = TRUE;
955: }
956: }
957:
958:
959:
960: /*************************************************
961: * Apply patterns to subject till one matches *
962: *************************************************/
963:
964: /* This function is called to run through all patterns, looking for a match. It
965: is used multiple times for the same subject when colouring is enabled, in order
966: to find all possible matches.
967:
968: Arguments:
969: matchptr the start of the subject
970: length the length of the subject to match
971: startoffset where to start matching
972: offsets the offets vector to fill in
973: mrc address of where to put the result of pcre_exec()
974:
975: Returns: TRUE if there was a match
976: FALSE if there was no match
977: invert if there was a non-fatal error
978: */
979:
980: static BOOL
981: match_patterns(char *matchptr, size_t length, int startoffset, int *offsets,
982: int *mrc)
983: {
984: int i;
985: size_t slen = length;
986: const char *msg = "this text:\n\n";
987: if (slen > 200)
988: {
989: slen = 200;
990: msg = "text that starts:\n\n";
991: }
992: for (i = 0; i < pattern_count; i++)
993: {
994: *mrc = pcre_exec(pattern_list[i], hints_list[i], matchptr, (int)length,
995: startoffset, PCRE_NOTEMPTY, offsets, OFFSET_SIZE);
996: if (*mrc >= 0) return TRUE;
997: if (*mrc == PCRE_ERROR_NOMATCH) continue;
998: fprintf(stderr, "pcregrep: pcre_exec() gave error %d while matching ", *mrc);
999: if (pattern_count > 1) fprintf(stderr, "pattern number %d to ", i+1);
1000: fprintf(stderr, "%s", msg);
1001: FWRITE(matchptr, 1, slen, stderr); /* In case binary zero included */
1002: fprintf(stderr, "\n\n");
1003: if (*mrc == PCRE_ERROR_MATCHLIMIT || *mrc == PCRE_ERROR_RECURSIONLIMIT ||
1004: *mrc == PCRE_ERROR_JIT_STACKLIMIT)
1005: resource_error = TRUE;
1006: if (error_count++ > 20)
1007: {
1008: fprintf(stderr, "pcregrep: Too many errors - abandoned.\n");
1009: pcregrep_exit(2);
1010: }
1011: return invert; /* No more matching; don't show the line again */
1012: }
1013:
1014: return FALSE; /* No match, no errors */
1015: }
1016:
1017:
1018:
1019: /*************************************************
1020: * Grep an individual file *
1021: *************************************************/
1022:
1023: /* This is called from grep_or_recurse() below. It uses a buffer that is three
1024: times the value of bufthird. The matching point is never allowed to stray into
1025: the top third of the buffer, thus keeping more of the file available for
1026: context printing or for multiline scanning. For large files, the pointer will
1027: be in the middle third most of the time, so the bottom third is available for
1028: "before" context printing.
1029:
1030: Arguments:
1031: handle the fopened FILE stream for a normal file
1032: the gzFile pointer when reading is via libz
1033: the BZFILE pointer when reading is via libbz2
1034: frtype FR_PLAIN, FR_LIBZ, or FR_LIBBZ2
1035: filename the file name or NULL (for errors)
1036: printname the file name if it is to be printed for each match
1037: or NULL if the file name is not to be printed
1038: it cannot be NULL if filenames[_nomatch]_only is set
1039:
1040: Returns: 0 if there was at least one match
1041: 1 otherwise (no matches)
1042: 2 if an overlong line is encountered
1043: 3 if there is a read error on a .bz2 file
1044: */
1045:
1046: static int
1047: pcregrep(void *handle, int frtype, char *filename, char *printname)
1048: {
1049: int rc = 1;
1050: int linenumber = 1;
1051: int lastmatchnumber = 0;
1052: int count = 0;
1053: int filepos = 0;
1054: int offsets[OFFSET_SIZE];
1055: char *lastmatchrestart = NULL;
1056: char *ptr = main_buffer;
1057: char *endptr;
1058: size_t bufflength;
1.1.1.3 ! misho 1059: BOOL binary = FALSE;
1.1 misho 1060: BOOL endhyphenpending = FALSE;
1061: BOOL input_line_buffered = line_buffered;
1062: FILE *in = NULL; /* Ensure initialized */
1063:
1064: #ifdef SUPPORT_LIBZ
1065: gzFile ingz = NULL;
1066: #endif
1067:
1068: #ifdef SUPPORT_LIBBZ2
1069: BZFILE *inbz2 = NULL;
1070: #endif
1071:
1072:
1073: /* Do the first read into the start of the buffer and set up the pointer to end
1074: of what we have. In the case of libz, a non-zipped .gz file will be read as a
1075: plain file. However, if a .bz2 file isn't actually bzipped, the first read will
1076: fail. */
1077:
1078: #ifdef SUPPORT_LIBZ
1079: if (frtype == FR_LIBZ)
1080: {
1081: ingz = (gzFile)handle;
1082: bufflength = gzread (ingz, main_buffer, bufsize);
1083: }
1084: else
1085: #endif
1086:
1087: #ifdef SUPPORT_LIBBZ2
1088: if (frtype == FR_LIBBZ2)
1089: {
1090: inbz2 = (BZFILE *)handle;
1091: bufflength = BZ2_bzread(inbz2, main_buffer, bufsize);
1092: if ((int)bufflength < 0) return 2; /* Gotcha: bufflength is size_t; */
1093: } /* without the cast it is unsigned. */
1094: else
1095: #endif
1096:
1097: {
1098: in = (FILE *)handle;
1099: if (is_file_tty(in)) input_line_buffered = TRUE;
1100: bufflength = input_line_buffered?
1101: read_one_line(main_buffer, bufsize, in) :
1102: fread(main_buffer, 1, bufsize, in);
1103: }
1104:
1105: endptr = main_buffer + bufflength;
1106:
1.1.1.3 ! misho 1107: /* Unless binary-files=text, see if we have a binary file. This uses the same
! 1108: rule as GNU grep, namely, a search for a binary zero byte near the start of the
! 1109: file. */
! 1110:
! 1111: if (binary_files != BIN_TEXT)
! 1112: {
! 1113: binary =
! 1114: memchr(main_buffer, 0, (bufflength > 1024)? 1024 : bufflength) != NULL;
! 1115: if (binary && binary_files == BIN_NOMATCH) return 1;
! 1116: }
! 1117:
1.1 misho 1118: /* Loop while the current pointer is not at the end of the file. For large
1119: files, endptr will be at the end of the buffer when we are in the middle of the
1120: file, but ptr will never get there, because as soon as it gets over 2/3 of the
1121: way, the buffer is shifted left and re-filled. */
1122:
1123: while (ptr < endptr)
1124: {
1125: int endlinelength;
1126: int mrc = 0;
1127: int startoffset = 0;
1128: BOOL match;
1129: char *matchptr = ptr;
1130: char *t = ptr;
1131: size_t length, linelength;
1132:
1133: /* At this point, ptr is at the start of a line. We need to find the length
1134: of the subject string to pass to pcre_exec(). In multiline mode, it is the
1135: length remainder of the data in the buffer. Otherwise, it is the length of
1136: the next line, excluding the terminating newline. After matching, we always
1137: advance by the length of the next line. In multiline mode the PCRE_FIRSTLINE
1138: option is used for compiling, so that any match is constrained to be in the
1139: first line. */
1140:
1141: t = end_of_line(t, endptr, &endlinelength);
1142: linelength = t - ptr - endlinelength;
1143: length = multiline? (size_t)(endptr - ptr) : linelength;
1144:
1145: /* Check to see if the line we are looking at extends right to the very end
1146: of the buffer without a line terminator. This means the line is too long to
1147: handle. */
1148:
1149: if (endlinelength == 0 && t == main_buffer + bufsize)
1150: {
1151: fprintf(stderr, "pcregrep: line %d%s%s is too long for the internal buffer\n"
1152: "pcregrep: check the --buffer-size option\n",
1153: linenumber,
1154: (filename == NULL)? "" : " of file ",
1155: (filename == NULL)? "" : filename);
1156: return 2;
1157: }
1158:
1159: /* Extra processing for Jeffrey Friedl's debugging. */
1160:
1161: #ifdef JFRIEDL_DEBUG
1162: if (jfriedl_XT || jfriedl_XR)
1163: {
1164: #include <sys/time.h>
1165: #include <time.h>
1166: struct timeval start_time, end_time;
1167: struct timezone dummy;
1168: int i;
1169:
1170: if (jfriedl_XT)
1171: {
1172: unsigned long newlen = length * jfriedl_XT + strlen(jfriedl_prefix) + strlen(jfriedl_postfix);
1173: const char *orig = ptr;
1174: ptr = malloc(newlen + 1);
1175: if (!ptr) {
1176: printf("out of memory");
1177: pcregrep_exit(2);
1178: }
1179: endptr = ptr;
1180: strcpy(endptr, jfriedl_prefix); endptr += strlen(jfriedl_prefix);
1181: for (i = 0; i < jfriedl_XT; i++) {
1182: strncpy(endptr, orig, length);
1183: endptr += length;
1184: }
1185: strcpy(endptr, jfriedl_postfix); endptr += strlen(jfriedl_postfix);
1186: length = newlen;
1187: }
1188:
1189: if (gettimeofday(&start_time, &dummy) != 0)
1190: perror("bad gettimeofday");
1191:
1192:
1193: for (i = 0; i < jfriedl_XR; i++)
1194: match = (pcre_exec(pattern_list[0], hints_list[0], ptr, length, 0,
1195: PCRE_NOTEMPTY, offsets, OFFSET_SIZE) >= 0);
1196:
1197: if (gettimeofday(&end_time, &dummy) != 0)
1198: perror("bad gettimeofday");
1199:
1200: double delta = ((end_time.tv_sec + (end_time.tv_usec / 1000000.0))
1201: -
1202: (start_time.tv_sec + (start_time.tv_usec / 1000000.0)));
1203:
1204: printf("%s TIMER[%.4f]\n", match ? "MATCH" : "FAIL", delta);
1205: return 0;
1206: }
1207: #endif
1208:
1209: /* We come back here after a match when the -o option (only_matching) is set,
1210: in order to find any further matches in the same line. */
1211:
1212: ONLY_MATCHING_RESTART:
1213:
1214: /* Run through all the patterns until one matches or there is an error other
1215: than NOMATCH. This code is in a subroutine so that it can be re-used for
1216: finding subsequent matches when colouring matched lines. */
1217:
1218: match = match_patterns(matchptr, length, startoffset, offsets, &mrc);
1219:
1220: /* If it's a match or a not-match (as required), do what's wanted. */
1221:
1222: if (match != invert)
1223: {
1224: BOOL hyphenprinted = FALSE;
1225:
1226: /* We've failed if we want a file that doesn't have any matches. */
1227:
1228: if (filenames == FN_NOMATCH_ONLY) return 1;
1229:
1230: /* Just count if just counting is wanted. */
1231:
1232: if (count_only) count++;
1233:
1.1.1.3 ! misho 1234: /* When handling a binary file and binary-files==binary, the "binary"
! 1235: variable will be set true (it's false in all other cases). In this
! 1236: situation we just want to output the file name. No need to scan further. */
! 1237:
! 1238: else if (binary)
! 1239: {
! 1240: fprintf(stdout, "Binary file %s matches\n", filename);
! 1241: return 0;
! 1242: }
! 1243:
1.1 misho 1244: /* If all we want is a file name, there is no need to scan any more lines
1245: in the file. */
1246:
1247: else if (filenames == FN_MATCH_ONLY)
1248: {
1249: fprintf(stdout, "%s\n", printname);
1250: return 0;
1251: }
1252:
1253: /* Likewise, if all we want is a yes/no answer. */
1254:
1255: else if (quiet) return 0;
1256:
1257: /* The --only-matching option prints just the substring that matched, or a
1258: captured portion of it, as long as this string is not empty, and the
1259: --file-offsets and --line-offsets options output offsets for the matching
1260: substring (they both force --only-matching = 0). None of these options
1261: prints any context. Afterwards, adjust the start and then jump back to look
1262: for further matches in the same line. If we are in invert mode, however,
1263: nothing is printed and we do not restart - this could still be useful
1264: because the return code is set. */
1265:
1266: else if (only_matching >= 0)
1267: {
1268: if (!invert)
1269: {
1270: if (printname != NULL) fprintf(stdout, "%s:", printname);
1271: if (number) fprintf(stdout, "%d:", linenumber);
1272: if (line_offsets)
1273: fprintf(stdout, "%d,%d\n", (int)(matchptr + offsets[0] - ptr),
1274: offsets[1] - offsets[0]);
1275: else if (file_offsets)
1276: fprintf(stdout, "%d,%d\n",
1277: (int)(filepos + matchptr + offsets[0] - ptr),
1278: offsets[1] - offsets[0]);
1279: else if (only_matching < mrc)
1280: {
1281: int plen = offsets[2*only_matching + 1] - offsets[2*only_matching];
1282: if (plen > 0)
1283: {
1284: if (do_colour) fprintf(stdout, "%c[%sm", 0x1b, colour_string);
1285: FWRITE(matchptr + offsets[only_matching*2], 1, plen, stdout);
1286: if (do_colour) fprintf(stdout, "%c[00m", 0x1b);
1287: fprintf(stdout, "\n");
1288: }
1289: }
1290: else if (printname != NULL || number) fprintf(stdout, "\n");
1291: match = FALSE;
1292: if (line_buffered) fflush(stdout);
1293: rc = 0; /* Had some success */
1294: startoffset = offsets[1]; /* Restart after the match */
1295: goto ONLY_MATCHING_RESTART;
1296: }
1297: }
1298:
1299: /* This is the default case when none of the above options is set. We print
1300: the matching lines(s), possibly preceded and/or followed by other lines of
1301: context. */
1302:
1303: else
1304: {
1305: /* See if there is a requirement to print some "after" lines from a
1306: previous match. We never print any overlaps. */
1307:
1308: if (after_context > 0 && lastmatchnumber > 0)
1309: {
1310: int ellength;
1311: int linecount = 0;
1312: char *p = lastmatchrestart;
1313:
1314: while (p < ptr && linecount < after_context)
1315: {
1316: p = end_of_line(p, ptr, &ellength);
1317: linecount++;
1318: }
1319:
1320: /* It is important to advance lastmatchrestart during this printing so
1321: that it interacts correctly with any "before" printing below. Print
1322: each line's data using fwrite() in case there are binary zeroes. */
1323:
1324: while (lastmatchrestart < p)
1325: {
1326: char *pp = lastmatchrestart;
1327: if (printname != NULL) fprintf(stdout, "%s-", printname);
1328: if (number) fprintf(stdout, "%d-", lastmatchnumber++);
1329: pp = end_of_line(pp, endptr, &ellength);
1330: FWRITE(lastmatchrestart, 1, pp - lastmatchrestart, stdout);
1331: lastmatchrestart = pp;
1332: }
1333: if (lastmatchrestart != ptr) hyphenpending = TRUE;
1334: }
1335:
1336: /* If there were non-contiguous lines printed above, insert hyphens. */
1337:
1338: if (hyphenpending)
1339: {
1340: fprintf(stdout, "--\n");
1341: hyphenpending = FALSE;
1342: hyphenprinted = TRUE;
1343: }
1344:
1345: /* See if there is a requirement to print some "before" lines for this
1346: match. Again, don't print overlaps. */
1347:
1348: if (before_context > 0)
1349: {
1350: int linecount = 0;
1351: char *p = ptr;
1352:
1353: while (p > main_buffer && (lastmatchnumber == 0 || p > lastmatchrestart) &&
1354: linecount < before_context)
1355: {
1356: linecount++;
1357: p = previous_line(p, main_buffer);
1358: }
1359:
1360: if (lastmatchnumber > 0 && p > lastmatchrestart && !hyphenprinted)
1361: fprintf(stdout, "--\n");
1362:
1363: while (p < ptr)
1364: {
1365: int ellength;
1366: char *pp = p;
1367: if (printname != NULL) fprintf(stdout, "%s-", printname);
1368: if (number) fprintf(stdout, "%d-", linenumber - linecount--);
1369: pp = end_of_line(pp, endptr, &ellength);
1370: FWRITE(p, 1, pp - p, stdout);
1371: p = pp;
1372: }
1373: }
1374:
1375: /* Now print the matching line(s); ensure we set hyphenpending at the end
1376: of the file if any context lines are being output. */
1377:
1378: if (after_context > 0 || before_context > 0)
1379: endhyphenpending = TRUE;
1380:
1381: if (printname != NULL) fprintf(stdout, "%s:", printname);
1382: if (number) fprintf(stdout, "%d:", linenumber);
1383:
1384: /* In multiline mode, we want to print to the end of the line in which
1385: the end of the matched string is found, so we adjust linelength and the
1386: line number appropriately, but only when there actually was a match
1387: (invert not set). Because the PCRE_FIRSTLINE option is set, the start of
1388: the match will always be before the first newline sequence. */
1389:
1390: if (multiline & !invert)
1391: {
1392: char *endmatch = ptr + offsets[1];
1393: t = ptr;
1394: while (t < endmatch)
1395: {
1396: t = end_of_line(t, endptr, &endlinelength);
1397: if (t < endmatch) linenumber++; else break;
1398: }
1399: linelength = t - ptr - endlinelength;
1400: }
1401:
1402: /*** NOTE: Use only fwrite() to output the data line, so that binary
1403: zeroes are treated as just another data character. */
1404:
1405: /* This extra option, for Jeffrey Friedl's debugging requirements,
1406: replaces the matched string, or a specific captured string if it exists,
1407: with X. When this happens, colouring is ignored. */
1408:
1409: #ifdef JFRIEDL_DEBUG
1410: if (S_arg >= 0 && S_arg < mrc)
1411: {
1412: int first = S_arg * 2;
1413: int last = first + 1;
1414: FWRITE(ptr, 1, offsets[first], stdout);
1415: fprintf(stdout, "X");
1416: FWRITE(ptr + offsets[last], 1, linelength - offsets[last], stdout);
1417: }
1418: else
1419: #endif
1420:
1421: /* We have to split the line(s) up if colouring, and search for further
1422: matches, but not of course if the line is a non-match. */
1423:
1424: if (do_colour && !invert)
1425: {
1426: int plength;
1427: FWRITE(ptr, 1, offsets[0], stdout);
1428: fprintf(stdout, "%c[%sm", 0x1b, colour_string);
1429: FWRITE(ptr + offsets[0], 1, offsets[1] - offsets[0], stdout);
1430: fprintf(stdout, "%c[00m", 0x1b);
1431: for (;;)
1432: {
1433: startoffset = offsets[1];
1434: if (startoffset >= (int)linelength + endlinelength ||
1435: !match_patterns(matchptr, length, startoffset, offsets, &mrc))
1436: break;
1437: FWRITE(matchptr + startoffset, 1, offsets[0] - startoffset, stdout);
1438: fprintf(stdout, "%c[%sm", 0x1b, colour_string);
1439: FWRITE(matchptr + offsets[0], 1, offsets[1] - offsets[0], stdout);
1440: fprintf(stdout, "%c[00m", 0x1b);
1441: }
1442:
1443: /* In multiline mode, we may have already printed the complete line
1444: and its line-ending characters (if they matched the pattern), so there
1445: may be no more to print. */
1446:
1447: plength = (int)((linelength + endlinelength) - startoffset);
1448: if (plength > 0) FWRITE(ptr + startoffset, 1, plength, stdout);
1449: }
1450:
1451: /* Not colouring; no need to search for further matches */
1452:
1453: else FWRITE(ptr, 1, linelength + endlinelength, stdout);
1454: }
1455:
1456: /* End of doing what has to be done for a match. If --line-buffered was
1457: given, flush the output. */
1458:
1459: if (line_buffered) fflush(stdout);
1460: rc = 0; /* Had some success */
1461:
1462: /* Remember where the last match happened for after_context. We remember
1463: where we are about to restart, and that line's number. */
1464:
1465: lastmatchrestart = ptr + linelength + endlinelength;
1466: lastmatchnumber = linenumber + 1;
1467: }
1468:
1469: /* For a match in multiline inverted mode (which of course did not cause
1470: anything to be printed), we have to move on to the end of the match before
1471: proceeding. */
1472:
1473: if (multiline && invert && match)
1474: {
1475: int ellength;
1476: char *endmatch = ptr + offsets[1];
1477: t = ptr;
1478: while (t < endmatch)
1479: {
1480: t = end_of_line(t, endptr, &ellength);
1481: if (t <= endmatch) linenumber++; else break;
1482: }
1483: endmatch = end_of_line(endmatch, endptr, &ellength);
1484: linelength = endmatch - ptr - ellength;
1485: }
1486:
1487: /* Advance to after the newline and increment the line number. The file
1488: offset to the current line is maintained in filepos. */
1489:
1490: ptr += linelength + endlinelength;
1491: filepos += (int)(linelength + endlinelength);
1492: linenumber++;
1493:
1494: /* If input is line buffered, and the buffer is not yet full, read another
1495: line and add it into the buffer. */
1496:
1497: if (input_line_buffered && bufflength < (size_t)bufsize)
1498: {
1499: int add = read_one_line(ptr, bufsize - (int)(ptr - main_buffer), in);
1500: bufflength += add;
1501: endptr += add;
1502: }
1503:
1504: /* If we haven't yet reached the end of the file (the buffer is full), and
1505: the current point is in the top 1/3 of the buffer, slide the buffer down by
1506: 1/3 and refill it. Before we do this, if some unprinted "after" lines are
1507: about to be lost, print them. */
1508:
1509: if (bufflength >= (size_t)bufsize && ptr > main_buffer + 2*bufthird)
1510: {
1511: if (after_context > 0 &&
1512: lastmatchnumber > 0 &&
1513: lastmatchrestart < main_buffer + bufthird)
1514: {
1515: do_after_lines(lastmatchnumber, lastmatchrestart, endptr, printname);
1516: lastmatchnumber = 0;
1517: }
1518:
1519: /* Now do the shuffle */
1520:
1521: memmove(main_buffer, main_buffer + bufthird, 2*bufthird);
1522: ptr -= bufthird;
1523:
1524: #ifdef SUPPORT_LIBZ
1525: if (frtype == FR_LIBZ)
1526: bufflength = 2*bufthird +
1527: gzread (ingz, main_buffer + 2*bufthird, bufthird);
1528: else
1529: #endif
1530:
1531: #ifdef SUPPORT_LIBBZ2
1532: if (frtype == FR_LIBBZ2)
1533: bufflength = 2*bufthird +
1534: BZ2_bzread(inbz2, main_buffer + 2*bufthird, bufthird);
1535: else
1536: #endif
1537:
1538: bufflength = 2*bufthird +
1539: (input_line_buffered?
1540: read_one_line(main_buffer + 2*bufthird, bufthird, in) :
1541: fread(main_buffer + 2*bufthird, 1, bufthird, in));
1542: endptr = main_buffer + bufflength;
1543:
1544: /* Adjust any last match point */
1545:
1546: if (lastmatchnumber > 0) lastmatchrestart -= bufthird;
1547: }
1548: } /* Loop through the whole file */
1549:
1550: /* End of file; print final "after" lines if wanted; do_after_lines sets
1551: hyphenpending if it prints something. */
1552:
1553: if (only_matching < 0 && !count_only)
1554: {
1555: do_after_lines(lastmatchnumber, lastmatchrestart, endptr, printname);
1556: hyphenpending |= endhyphenpending;
1557: }
1558:
1559: /* Print the file name if we are looking for those without matches and there
1560: were none. If we found a match, we won't have got this far. */
1561:
1562: if (filenames == FN_NOMATCH_ONLY)
1563: {
1564: fprintf(stdout, "%s\n", printname);
1565: return 0;
1566: }
1567:
1568: /* Print the match count if wanted */
1569:
1570: if (count_only)
1571: {
1572: if (count > 0 || !omit_zero_count)
1573: {
1574: if (printname != NULL && filenames != FN_NONE)
1575: fprintf(stdout, "%s:", printname);
1576: fprintf(stdout, "%d\n", count);
1577: }
1578: }
1579:
1580: return rc;
1581: }
1582:
1583:
1584:
1585: /*************************************************
1586: * Grep a file or recurse into a directory *
1587: *************************************************/
1588:
1589: /* Given a path name, if it's a directory, scan all the files if we are
1590: recursing; if it's a file, grep it.
1591:
1592: Arguments:
1593: pathname the path to investigate
1594: dir_recurse TRUE if recursing is wanted (-r or -drecurse)
1595: only_one_at_top TRUE if the path is the only one at toplevel
1596:
1597: Returns: 0 if there was at least one match
1598: 1 if there were no matches
1599: 2 there was some kind of error
1600:
1601: However, file opening failures are suppressed if "silent" is set.
1602: */
1603:
1604: static int
1605: grep_or_recurse(char *pathname, BOOL dir_recurse, BOOL only_one_at_top)
1606: {
1607: int rc = 1;
1608: int sep;
1609: int frtype;
1610: void *handle;
1611: FILE *in = NULL; /* Ensure initialized */
1612:
1613: #ifdef SUPPORT_LIBZ
1614: gzFile ingz = NULL;
1615: #endif
1616:
1617: #ifdef SUPPORT_LIBBZ2
1618: BZFILE *inbz2 = NULL;
1619: #endif
1620:
1.1.1.3 ! misho 1621: #if defined SUPPORT_LIBZ || defined SUPPORT_LIBBZ2
1.1.1.2 misho 1622: int pathlen;
1623: #endif
1624:
1.1 misho 1625: /* If the file name is "-" we scan stdin */
1626:
1627: if (strcmp(pathname, "-") == 0)
1628: {
1629: return pcregrep(stdin, FR_PLAIN, stdin_name,
1630: (filenames > FN_DEFAULT || (filenames == FN_DEFAULT && !only_one_at_top))?
1631: stdin_name : NULL);
1632: }
1633:
1634: /* If the file is a directory, skip if skipping or if we are recursing, scan
1635: each file and directory within it, subject to any include or exclude patterns
1636: that were set. The scanning code is localized so it can be made
1637: system-specific. */
1638:
1639: if ((sep = isdirectory(pathname)) != 0)
1640: {
1641: if (dee_action == dee_SKIP) return 1;
1642: if (dee_action == dee_RECURSE)
1643: {
1644: char buffer[1024];
1645: char *nextfile;
1646: directory_type *dir = opendirectory(pathname);
1647:
1648: if (dir == NULL)
1649: {
1650: if (!silent)
1651: fprintf(stderr, "pcregrep: Failed to open directory %s: %s\n", pathname,
1652: strerror(errno));
1653: return 2;
1654: }
1655:
1656: while ((nextfile = readdirectory(dir)) != NULL)
1657: {
1658: int frc, nflen;
1659: sprintf(buffer, "%.512s%c%.128s", pathname, sep, nextfile);
1660: nflen = (int)(strlen(nextfile));
1661:
1662: if (isdirectory(buffer))
1663: {
1664: if (exclude_dir_compiled != NULL &&
1665: pcre_exec(exclude_dir_compiled, NULL, nextfile, nflen, 0, 0, NULL, 0) >= 0)
1666: continue;
1667:
1668: if (include_dir_compiled != NULL &&
1669: pcre_exec(include_dir_compiled, NULL, nextfile, nflen, 0, 0, NULL, 0) < 0)
1670: continue;
1671: }
1672: else
1673: {
1674: if (exclude_compiled != NULL &&
1675: pcre_exec(exclude_compiled, NULL, nextfile, nflen, 0, 0, NULL, 0) >= 0)
1676: continue;
1677:
1678: if (include_compiled != NULL &&
1679: pcre_exec(include_compiled, NULL, nextfile, nflen, 0, 0, NULL, 0) < 0)
1680: continue;
1681: }
1682:
1683: frc = grep_or_recurse(buffer, dir_recurse, FALSE);
1684: if (frc > 1) rc = frc;
1685: else if (frc == 0 && rc == 1) rc = 0;
1686: }
1687:
1688: closedirectory(dir);
1689: return rc;
1690: }
1691: }
1692:
1693: /* If the file is not a directory and not a regular file, skip it if that's
1694: been requested. */
1695:
1696: else if (!isregfile(pathname) && DEE_action == DEE_SKIP) return 1;
1697:
1698: /* Control reaches here if we have a regular file, or if we have a directory
1699: and recursion or skipping was not requested, or if we have anything else and
1700: skipping was not requested. The scan proceeds. If this is the first and only
1701: argument at top level, we don't show the file name, unless we are only showing
1702: the file name, or the filename was forced (-H). */
1703:
1.1.1.3 ! misho 1704: #if defined SUPPORT_LIBZ || defined SUPPORT_LIBBZ2
1.1 misho 1705: pathlen = (int)(strlen(pathname));
1.1.1.2 misho 1706: #endif
1.1 misho 1707:
1708: /* Open using zlib if it is supported and the file name ends with .gz. */
1709:
1710: #ifdef SUPPORT_LIBZ
1711: if (pathlen > 3 && strcmp(pathname + pathlen - 3, ".gz") == 0)
1712: {
1713: ingz = gzopen(pathname, "rb");
1714: if (ingz == NULL)
1715: {
1716: if (!silent)
1717: fprintf(stderr, "pcregrep: Failed to open %s: %s\n", pathname,
1718: strerror(errno));
1719: return 2;
1720: }
1721: handle = (void *)ingz;
1722: frtype = FR_LIBZ;
1723: }
1724: else
1725: #endif
1726:
1727: /* Otherwise open with bz2lib if it is supported and the name ends with .bz2. */
1728:
1729: #ifdef SUPPORT_LIBBZ2
1730: if (pathlen > 4 && strcmp(pathname + pathlen - 4, ".bz2") == 0)
1731: {
1732: inbz2 = BZ2_bzopen(pathname, "rb");
1733: handle = (void *)inbz2;
1734: frtype = FR_LIBBZ2;
1735: }
1736: else
1737: #endif
1738:
1739: /* Otherwise use plain fopen(). The label is so that we can come back here if
1740: an attempt to read a .bz2 file indicates that it really is a plain file. */
1741:
1742: #ifdef SUPPORT_LIBBZ2
1743: PLAIN_FILE:
1744: #endif
1745: {
1746: in = fopen(pathname, "rb");
1747: handle = (void *)in;
1748: frtype = FR_PLAIN;
1749: }
1750:
1751: /* All the opening methods return errno when they fail. */
1752:
1753: if (handle == NULL)
1754: {
1755: if (!silent)
1756: fprintf(stderr, "pcregrep: Failed to open %s: %s\n", pathname,
1757: strerror(errno));
1758: return 2;
1759: }
1760:
1761: /* Now grep the file */
1762:
1763: rc = pcregrep(handle, frtype, pathname, (filenames > FN_DEFAULT ||
1764: (filenames == FN_DEFAULT && !only_one_at_top))? pathname : NULL);
1765:
1766: /* Close in an appropriate manner. */
1767:
1768: #ifdef SUPPORT_LIBZ
1769: if (frtype == FR_LIBZ)
1770: gzclose(ingz);
1771: else
1772: #endif
1773:
1774: /* If it is a .bz2 file and the result is 3, it means that the first attempt to
1775: read failed. If the error indicates that the file isn't in fact bzipped, try
1776: again as a normal file. */
1777:
1778: #ifdef SUPPORT_LIBBZ2
1779: if (frtype == FR_LIBBZ2)
1780: {
1781: if (rc == 3)
1782: {
1783: int errnum;
1784: const char *err = BZ2_bzerror(inbz2, &errnum);
1785: if (errnum == BZ_DATA_ERROR_MAGIC)
1786: {
1787: BZ2_bzclose(inbz2);
1788: goto PLAIN_FILE;
1789: }
1790: else if (!silent)
1791: fprintf(stderr, "pcregrep: Failed to read %s using bzlib: %s\n",
1792: pathname, err);
1793: rc = 2; /* The normal "something went wrong" code */
1794: }
1795: BZ2_bzclose(inbz2);
1796: }
1797: else
1798: #endif
1799:
1800: /* Normal file close */
1801:
1802: fclose(in);
1803:
1804: /* Pass back the yield from pcregrep(). */
1805:
1806: return rc;
1807: }
1808:
1809:
1810:
1811:
1812: /*************************************************
1813: * Usage function *
1814: *************************************************/
1815:
1816: static int
1817: usage(int rc)
1818: {
1819: option_item *op;
1820: fprintf(stderr, "Usage: pcregrep [-");
1821: for (op = optionlist; op->one_char != 0; op++)
1822: {
1823: if (op->one_char > 0) fprintf(stderr, "%c", op->one_char);
1824: }
1825: fprintf(stderr, "] [long options] [pattern] [files]\n");
1826: fprintf(stderr, "Type `pcregrep --help' for more information and the long "
1827: "options.\n");
1828: return rc;
1829: }
1830:
1831:
1832:
1833:
1834: /*************************************************
1835: * Help function *
1836: *************************************************/
1837:
1838: static void
1839: help(void)
1840: {
1841: option_item *op;
1842:
1843: printf("Usage: pcregrep [OPTION]... [PATTERN] [FILE1 FILE2 ...]\n");
1844: printf("Search for PATTERN in each FILE or standard input.\n");
1845: printf("PATTERN must be present if neither -e nor -f is used.\n");
1846: printf("\"-\" can be used as a file name to mean STDIN.\n");
1847:
1848: #ifdef SUPPORT_LIBZ
1849: printf("Files whose names end in .gz are read using zlib.\n");
1850: #endif
1851:
1852: #ifdef SUPPORT_LIBBZ2
1853: printf("Files whose names end in .bz2 are read using bzlib2.\n");
1854: #endif
1855:
1856: #if defined SUPPORT_LIBZ || defined SUPPORT_LIBBZ2
1857: printf("Other files and the standard input are read as plain files.\n\n");
1858: #else
1859: printf("All files are read as plain files, without any interpretation.\n\n");
1860: #endif
1861:
1862: printf("Example: pcregrep -i 'hello.*world' menu.h main.c\n\n");
1863: printf("Options:\n");
1864:
1865: for (op = optionlist; op->one_char != 0; op++)
1866: {
1867: int n;
1868: char s[4];
1869:
1870: /* Two options were accidentally implemented and documented with underscores
1871: instead of hyphens in their names, something that was not noticed for quite a
1872: few releases. When fixing this, I left the underscored versions in the list
1873: in case people were using them. However, we don't want to display them in the
1874: help data. There are no other options that contain underscores, and we do not
1875: expect ever to implement such options. Therefore, just omit any option that
1876: contains an underscore. */
1877:
1878: if (strchr(op->long_name, '_') != NULL) continue;
1879:
1.1.1.3 ! misho 1880: if (op->one_char > 0 && (op->long_name)[0] == 0)
! 1881: n = 31 - printf(" -%c", op->one_char);
! 1882: else
! 1883: {
! 1884: if (op->one_char > 0) sprintf(s, "-%c,", op->one_char);
! 1885: else strcpy(s, " ");
! 1886: n = 31 - printf(" %s --%s", s, op->long_name);
! 1887: }
! 1888:
1.1 misho 1889: if (n < 1) n = 1;
1.1.1.3 ! misho 1890: printf("%.*s%s\n", n, " ", op->help_text);
1.1 misho 1891: }
1892:
1893: printf("\nNumbers may be followed by K or M, e.g. --buffer-size=100K.\n");
1894: printf("The default value for --buffer-size is %d.\n", PCREGREP_BUFSIZE);
1.1.1.3 ! misho 1895: printf("When reading patterns or file names from a file, trailing white\n");
! 1896: printf("space is removed and blank lines are ignored.\n");
1.1 misho 1897: printf("There is a maximum of %d patterns, each of maximum size %d bytes.\n",
1898: MAX_PATTERN_COUNT, PATBUFSIZE);
1899:
1900: printf("\nWith no FILEs, read standard input. If fewer than two FILEs given, assume -h.\n");
1901: printf("Exit status is 0 if any matches, 1 if no matches, and 2 if trouble.\n");
1902: }
1903:
1904:
1905:
1906:
1907: /*************************************************
1908: * Handle a single-letter, no data option *
1909: *************************************************/
1910:
1911: static int
1912: handle_option(int letter, int options)
1913: {
1914: switch(letter)
1915: {
1916: case N_FOFFSETS: file_offsets = TRUE; break;
1917: case N_HELP: help(); pcregrep_exit(0);
1918: case N_LBUFFER: line_buffered = TRUE; break;
1919: case N_LOFFSETS: line_offsets = number = TRUE; break;
1920: case N_NOJIT: study_options &= ~PCRE_STUDY_JIT_COMPILE; break;
1.1.1.3 ! misho 1921: case 'a': binary_files = BIN_TEXT; break;
1.1 misho 1922: case 'c': count_only = TRUE; break;
1923: case 'F': process_options |= PO_FIXED_STRINGS; break;
1924: case 'H': filenames = FN_FORCE; break;
1.1.1.3 ! misho 1925: case 'I': binary_files = BIN_NOMATCH; break;
1.1 misho 1926: case 'h': filenames = FN_NONE; break;
1927: case 'i': options |= PCRE_CASELESS; break;
1928: case 'l': omit_zero_count = TRUE; filenames = FN_MATCH_ONLY; break;
1929: case 'L': filenames = FN_NOMATCH_ONLY; break;
1930: case 'M': multiline = TRUE; options |= PCRE_MULTILINE|PCRE_FIRSTLINE; break;
1931: case 'n': number = TRUE; break;
1932: case 'o': only_matching = 0; break;
1933: case 'q': quiet = TRUE; break;
1934: case 'r': dee_action = dee_RECURSE; break;
1935: case 's': silent = TRUE; break;
1936: case 'u': options |= PCRE_UTF8; utf8 = TRUE; break;
1937: case 'v': invert = TRUE; break;
1938: case 'w': process_options |= PO_WORD_MATCH; break;
1939: case 'x': process_options |= PO_LINE_MATCH; break;
1940:
1941: case 'V':
1942: fprintf(stderr, "pcregrep version %s\n", pcre_version());
1943: pcregrep_exit(0);
1944: break;
1945:
1946: default:
1947: fprintf(stderr, "pcregrep: Unknown option -%c\n", letter);
1948: pcregrep_exit(usage(2));
1949: }
1950:
1951: return options;
1952: }
1953:
1954:
1955:
1956:
1957: /*************************************************
1958: * Construct printed ordinal *
1959: *************************************************/
1960:
1961: /* This turns a number into "1st", "3rd", etc. */
1962:
1963: static char *
1964: ordin(int n)
1965: {
1966: static char buffer[8];
1967: char *p = buffer;
1968: sprintf(p, "%d", n);
1969: while (*p != 0) p++;
1970: switch (n%10)
1971: {
1972: case 1: strcpy(p, "st"); break;
1973: case 2: strcpy(p, "nd"); break;
1974: case 3: strcpy(p, "rd"); break;
1975: default: strcpy(p, "th"); break;
1976: }
1977: return buffer;
1978: }
1979:
1980:
1981:
1982: /*************************************************
1983: * Compile a single pattern *
1984: *************************************************/
1985:
1986: /* When the -F option has been used, this is called for each substring.
1987: Otherwise it's called for each supplied pattern.
1988:
1989: Arguments:
1990: pattern the pattern string
1991: options the PCRE options
1992: filename the file name, or NULL for a command-line pattern
1993: count 0 if this is the only command line pattern, or
1994: number of the command line pattern, or
1995: linenumber for a pattern from a file
1996:
1997: Returns: TRUE on success, FALSE after an error
1998: */
1999:
2000: static BOOL
2001: compile_single_pattern(char *pattern, int options, char *filename, int count)
2002: {
2003: char buffer[PATBUFSIZE];
2004: const char *error;
2005: int errptr;
2006:
2007: if (pattern_count >= MAX_PATTERN_COUNT)
2008: {
2009: fprintf(stderr, "pcregrep: Too many %spatterns (max %d)\n",
2010: (filename == NULL)? "command-line " : "", MAX_PATTERN_COUNT);
2011: return FALSE;
2012: }
2013:
2014: sprintf(buffer, "%s%.*s%s", prefix[process_options], bufthird, pattern,
2015: suffix[process_options]);
2016: pattern_list[pattern_count] =
2017: pcre_compile(buffer, options, &error, &errptr, pcretables);
2018: if (pattern_list[pattern_count] != NULL)
2019: {
2020: pattern_count++;
2021: return TRUE;
2022: }
2023:
2024: /* Handle compile errors */
2025:
2026: errptr -= (int)strlen(prefix[process_options]);
2027: if (errptr > (int)strlen(pattern)) errptr = (int)strlen(pattern);
2028:
2029: if (filename == NULL)
2030: {
2031: if (count == 0)
2032: fprintf(stderr, "pcregrep: Error in command-line regex "
2033: "at offset %d: %s\n", errptr, error);
2034: else
2035: fprintf(stderr, "pcregrep: Error in %s command-line regex "
2036: "at offset %d: %s\n", ordin(count), errptr, error);
2037: }
2038: else
2039: {
2040: fprintf(stderr, "pcregrep: Error in regex in line %d of %s "
2041: "at offset %d: %s\n", count, filename, errptr, error);
2042: }
2043:
2044: return FALSE;
2045: }
2046:
2047:
2048:
2049: /*************************************************
2050: * Compile one supplied pattern *
2051: *************************************************/
2052:
2053: /* When the -F option has been used, each string may be a list of strings,
2054: separated by line breaks. They will be matched literally.
2055:
2056: Arguments:
2057: pattern the pattern string
2058: options the PCRE options
2059: filename the file name, or NULL for a command-line pattern
2060: count 0 if this is the only command line pattern, or
2061: number of the command line pattern, or
2062: linenumber for a pattern from a file
2063:
2064: Returns: TRUE on success, FALSE after an error
2065: */
2066:
2067: static BOOL
2068: compile_pattern(char *pattern, int options, char *filename, int count)
2069: {
2070: if ((process_options & PO_FIXED_STRINGS) != 0)
2071: {
2072: char *eop = pattern + strlen(pattern);
2073: char buffer[PATBUFSIZE];
2074: for(;;)
2075: {
2076: int ellength;
2077: char *p = end_of_line(pattern, eop, &ellength);
2078: if (ellength == 0)
2079: return compile_single_pattern(pattern, options, filename, count);
2080: sprintf(buffer, "%.*s", (int)(p - pattern - ellength), pattern);
2081: pattern = p;
2082: if (!compile_single_pattern(buffer, options, filename, count))
2083: return FALSE;
2084: }
2085: }
2086: else return compile_single_pattern(pattern, options, filename, count);
2087: }
2088:
2089:
2090:
2091: /*************************************************
2092: * Main program *
2093: *************************************************/
2094:
2095: /* Returns 0 if something matched, 1 if nothing matched, 2 after an error. */
2096:
2097: int
2098: main(int argc, char **argv)
2099: {
2100: int i, j;
2101: int rc = 1;
2102: int pcre_options = 0;
2103: int cmd_pattern_count = 0;
2104: int hint_count = 0;
2105: int errptr;
2106: BOOL only_one_at_top;
2107: char *patterns[MAX_PATTERN_COUNT];
2108: const char *locale_from = "--locale";
2109: const char *error;
2110:
2111: #ifdef SUPPORT_PCREGREP_JIT
2112: pcre_jit_stack *jit_stack = NULL;
2113: #endif
2114:
2115: /* Set the default line ending value from the default in the PCRE library;
2116: "lf", "cr", "crlf", and "any" are supported. Anything else is treated as "lf".
2117: Note that the return values from pcre_config(), though derived from the ASCII
2118: codes, are the same in EBCDIC environments, so we must use the actual values
2119: rather than escapes such as as '\r'. */
2120:
2121: (void)pcre_config(PCRE_CONFIG_NEWLINE, &i);
2122: switch(i)
2123: {
2124: default: newline = (char *)"lf"; break;
2125: case 13: newline = (char *)"cr"; break;
2126: case (13 << 8) | 10: newline = (char *)"crlf"; break;
2127: case -1: newline = (char *)"any"; break;
2128: case -2: newline = (char *)"anycrlf"; break;
2129: }
2130:
2131: /* Process the options */
2132:
2133: for (i = 1; i < argc; i++)
2134: {
2135: option_item *op = NULL;
2136: char *option_data = (char *)""; /* default to keep compiler happy */
2137: BOOL longop;
2138: BOOL longopwasequals = FALSE;
2139:
2140: if (argv[i][0] != '-') break;
2141:
2142: /* If we hit an argument that is just "-", it may be a reference to STDIN,
2143: but only if we have previously had -e or -f to define the patterns. */
2144:
2145: if (argv[i][1] == 0)
2146: {
2147: if (pattern_filename != NULL || pattern_count > 0) break;
2148: else pcregrep_exit(usage(2));
2149: }
2150:
2151: /* Handle a long name option, or -- to terminate the options */
2152:
2153: if (argv[i][1] == '-')
2154: {
2155: char *arg = argv[i] + 2;
2156: char *argequals = strchr(arg, '=');
2157:
2158: if (*arg == 0) /* -- terminates options */
2159: {
2160: i++;
2161: break; /* out of the options-handling loop */
2162: }
2163:
2164: longop = TRUE;
2165:
2166: /* Some long options have data that follows after =, for example file=name.
2167: Some options have variations in the long name spelling: specifically, we
2168: allow "regexp" because GNU grep allows it, though I personally go along
2169: with Jeffrey Friedl and Larry Wall in preferring "regex" without the "p".
2170: These options are entered in the table as "regex(p)". Options can be in
2171: both these categories. */
2172:
2173: for (op = optionlist; op->one_char != 0; op++)
2174: {
2175: char *opbra = strchr(op->long_name, '(');
2176: char *equals = strchr(op->long_name, '=');
2177:
2178: /* Handle options with only one spelling of the name */
2179:
2180: if (opbra == NULL) /* Does not contain '(' */
2181: {
2182: if (equals == NULL) /* Not thing=data case */
2183: {
2184: if (strcmp(arg, op->long_name) == 0) break;
2185: }
2186: else /* Special case xxx=data */
2187: {
2188: int oplen = (int)(equals - op->long_name);
2189: int arglen = (argequals == NULL)?
2190: (int)strlen(arg) : (int)(argequals - arg);
2191: if (oplen == arglen && strncmp(arg, op->long_name, oplen) == 0)
2192: {
2193: option_data = arg + arglen;
2194: if (*option_data == '=')
2195: {
2196: option_data++;
2197: longopwasequals = TRUE;
2198: }
2199: break;
2200: }
2201: }
2202: }
2203:
2204: /* Handle options with an alternate spelling of the name */
2205:
2206: else
2207: {
2208: char buff1[24];
2209: char buff2[24];
2210:
2211: int baselen = (int)(opbra - op->long_name);
2212: int fulllen = (int)(strchr(op->long_name, ')') - op->long_name + 1);
2213: int arglen = (argequals == NULL || equals == NULL)?
2214: (int)strlen(arg) : (int)(argequals - arg);
2215:
2216: sprintf(buff1, "%.*s", baselen, op->long_name);
2217: sprintf(buff2, "%s%.*s", buff1, fulllen - baselen - 2, opbra + 1);
2218:
2219: if (strncmp(arg, buff1, arglen) == 0 ||
2220: strncmp(arg, buff2, arglen) == 0)
2221: {
2222: if (equals != NULL && argequals != NULL)
2223: {
2224: option_data = argequals;
2225: if (*option_data == '=')
2226: {
2227: option_data++;
2228: longopwasequals = TRUE;
2229: }
2230: }
2231: break;
2232: }
2233: }
2234: }
2235:
2236: if (op->one_char == 0)
2237: {
2238: fprintf(stderr, "pcregrep: Unknown option %s\n", argv[i]);
2239: pcregrep_exit(usage(2));
2240: }
2241: }
2242:
2243: /* Jeffrey Friedl's debugging harness uses these additional options which
2244: are not in the right form for putting in the option table because they use
2245: only one hyphen, yet are more than one character long. By putting them
2246: separately here, they will not get displayed as part of the help() output,
2247: but I don't think Jeffrey will care about that. */
2248:
2249: #ifdef JFRIEDL_DEBUG
2250: else if (strcmp(argv[i], "-pre") == 0) {
2251: jfriedl_prefix = argv[++i];
2252: continue;
2253: } else if (strcmp(argv[i], "-post") == 0) {
2254: jfriedl_postfix = argv[++i];
2255: continue;
2256: } else if (strcmp(argv[i], "-XT") == 0) {
2257: sscanf(argv[++i], "%d", &jfriedl_XT);
2258: continue;
2259: } else if (strcmp(argv[i], "-XR") == 0) {
2260: sscanf(argv[++i], "%d", &jfriedl_XR);
2261: continue;
2262: }
2263: #endif
2264:
2265:
2266: /* One-char options; many that have no data may be in a single argument; we
2267: continue till we hit the last one or one that needs data. */
2268:
2269: else
2270: {
2271: char *s = argv[i] + 1;
2272: longop = FALSE;
2273: while (*s != 0)
2274: {
2275: for (op = optionlist; op->one_char != 0; op++)
2276: {
2277: if (*s == op->one_char) break;
2278: }
2279: if (op->one_char == 0)
2280: {
2281: fprintf(stderr, "pcregrep: Unknown option letter '%c' in \"%s\"\n",
2282: *s, argv[i]);
2283: pcregrep_exit(usage(2));
2284: }
2285:
2286: /* Check for a single-character option that has data: OP_OP_NUMBER
2287: is used for one that either has a numerical number or defaults, i.e. the
2288: data is optional. If a digit follows, there is data; if not, carry on
2289: with other single-character options in the same string. */
2290:
2291: option_data = s+1;
2292: if (op->type == OP_OP_NUMBER)
2293: {
2294: if (isdigit((unsigned char)s[1])) break;
2295: }
2296: else /* Check for end or a dataless option */
2297: {
2298: if (op->type != OP_NODATA || s[1] == 0) break;
2299: }
2300:
2301: /* Handle a single-character option with no data, then loop for the
2302: next character in the string. */
2303:
2304: pcre_options = handle_option(*s++, pcre_options);
2305: }
2306: }
2307:
2308: /* At this point we should have op pointing to a matched option. If the type
2309: is NO_DATA, it means that there is no data, and the option might set
2310: something in the PCRE options. */
2311:
2312: if (op->type == OP_NODATA)
2313: {
2314: pcre_options = handle_option(op->one_char, pcre_options);
2315: continue;
2316: }
2317:
2318: /* If the option type is OP_OP_STRING or OP_OP_NUMBER, it's an option that
2319: either has a value or defaults to something. It cannot have data in a
2320: separate item. At the moment, the only such options are "colo(u)r",
2321: "only-matching", and Jeffrey Friedl's special -S debugging option. */
2322:
2323: if (*option_data == 0 &&
2324: (op->type == OP_OP_STRING || op->type == OP_OP_NUMBER))
2325: {
2326: switch (op->one_char)
2327: {
2328: case N_COLOUR:
2329: colour_option = (char *)"auto";
2330: break;
2331:
2332: case 'o':
2333: only_matching = 0;
2334: break;
2335:
2336: #ifdef JFRIEDL_DEBUG
2337: case 'S':
2338: S_arg = 0;
2339: break;
2340: #endif
2341: }
2342: continue;
2343: }
2344:
2345: /* Otherwise, find the data string for the option. */
2346:
2347: if (*option_data == 0)
2348: {
2349: if (i >= argc - 1 || longopwasequals)
2350: {
2351: fprintf(stderr, "pcregrep: Data missing after %s\n", argv[i]);
2352: pcregrep_exit(usage(2));
2353: }
2354: option_data = argv[++i];
2355: }
2356:
2357: /* If the option type is OP_PATLIST, it's the -e option, which can be called
2358: multiple times to create a list of patterns. */
2359:
2360: if (op->type == OP_PATLIST)
2361: {
2362: if (cmd_pattern_count >= MAX_PATTERN_COUNT)
2363: {
2364: fprintf(stderr, "pcregrep: Too many command-line patterns (max %d)\n",
2365: MAX_PATTERN_COUNT);
2366: return 2;
2367: }
2368: patterns[cmd_pattern_count++] = option_data;
2369: }
2370:
1.1.1.3 ! misho 2371: /* Handle OP_BINARY_FILES */
! 2372:
! 2373: else if (op->type == OP_BINFILES)
! 2374: {
! 2375: if (strcmp(option_data, "binary") == 0)
! 2376: binary_files = BIN_BINARY;
! 2377: else if (strcmp(option_data, "without-match") == 0)
! 2378: binary_files = BIN_NOMATCH;
! 2379: else if (strcmp(option_data, "text") == 0)
! 2380: binary_files = BIN_TEXT;
! 2381: else
! 2382: {
! 2383: fprintf(stderr, "pcregrep: unknown value \"%s\" for binary-files\n",
! 2384: option_data);
! 2385: pcregrep_exit(usage(2));
! 2386: }
! 2387: }
! 2388:
1.1 misho 2389: /* Otherwise, deal with single string or numeric data values. */
2390:
2391: else if (op->type != OP_NUMBER && op->type != OP_LONGNUMBER &&
2392: op->type != OP_OP_NUMBER)
2393: {
2394: *((char **)op->dataptr) = option_data;
2395: }
2396:
2397: /* Avoid the use of strtoul() because SunOS4 doesn't have it. This is used
2398: only for unpicking arguments, so just keep it simple. */
2399:
2400: else
2401: {
2402: unsigned long int n = 0;
2403: char *endptr = option_data;
2404: while (*endptr != 0 && isspace((unsigned char)(*endptr))) endptr++;
2405: while (isdigit((unsigned char)(*endptr)))
2406: n = n * 10 + (int)(*endptr++ - '0');
2407: if (toupper(*endptr) == 'K')
2408: {
2409: n *= 1024;
2410: endptr++;
2411: }
2412: else if (toupper(*endptr) == 'M')
2413: {
2414: n *= 1024*1024;
2415: endptr++;
2416: }
2417: if (*endptr != 0)
2418: {
2419: if (longop)
2420: {
2421: char *equals = strchr(op->long_name, '=');
2422: int nlen = (equals == NULL)? (int)strlen(op->long_name) :
2423: (int)(equals - op->long_name);
2424: fprintf(stderr, "pcregrep: Malformed number \"%s\" after --%.*s\n",
2425: option_data, nlen, op->long_name);
2426: }
2427: else
2428: fprintf(stderr, "pcregrep: Malformed number \"%s\" after -%c\n",
2429: option_data, op->one_char);
2430: pcregrep_exit(usage(2));
2431: }
2432: if (op->type == OP_LONGNUMBER)
2433: *((unsigned long int *)op->dataptr) = n;
2434: else
2435: *((int *)op->dataptr) = n;
2436: }
2437: }
2438:
2439: /* Options have been decoded. If -C was used, its value is used as a default
2440: for -A and -B. */
2441:
2442: if (both_context > 0)
2443: {
2444: if (after_context == 0) after_context = both_context;
2445: if (before_context == 0) before_context = both_context;
2446: }
2447:
2448: /* Only one of --only-matching, --file-offsets, or --line-offsets is permitted.
2449: However, the latter two set only_matching. */
2450:
2451: if ((only_matching >= 0 && (file_offsets || line_offsets)) ||
2452: (file_offsets && line_offsets))
2453: {
2454: fprintf(stderr, "pcregrep: Cannot mix --only-matching, --file-offsets "
2455: "and/or --line-offsets\n");
2456: pcregrep_exit(usage(2));
2457: }
2458:
2459: if (file_offsets || line_offsets) only_matching = 0;
2460:
2461: /* If a locale has not been provided as an option, see if the LC_CTYPE or
2462: LC_ALL environment variable is set, and if so, use it. */
2463:
2464: if (locale == NULL)
2465: {
2466: locale = getenv("LC_ALL");
2467: locale_from = "LCC_ALL";
2468: }
2469:
2470: if (locale == NULL)
2471: {
2472: locale = getenv("LC_CTYPE");
2473: locale_from = "LC_CTYPE";
2474: }
2475:
2476: /* If a locale has been provided, set it, and generate the tables the PCRE
2477: needs. Otherwise, pcretables==NULL, which causes the use of default tables. */
2478:
2479: if (locale != NULL)
2480: {
2481: if (setlocale(LC_CTYPE, locale) == NULL)
2482: {
2483: fprintf(stderr, "pcregrep: Failed to set locale %s (obtained from %s)\n",
2484: locale, locale_from);
2485: return 2;
2486: }
2487: pcretables = pcre_maketables();
2488: }
2489:
2490: /* Sort out colouring */
2491:
2492: if (colour_option != NULL && strcmp(colour_option, "never") != 0)
2493: {
2494: if (strcmp(colour_option, "always") == 0) do_colour = TRUE;
2495: else if (strcmp(colour_option, "auto") == 0) do_colour = is_stdout_tty();
2496: else
2497: {
2498: fprintf(stderr, "pcregrep: Unknown colour setting \"%s\"\n",
2499: colour_option);
2500: return 2;
2501: }
2502: if (do_colour)
2503: {
2504: char *cs = getenv("PCREGREP_COLOUR");
2505: if (cs == NULL) cs = getenv("PCREGREP_COLOR");
2506: if (cs != NULL) colour_string = cs;
2507: }
2508: }
2509:
2510: /* Interpret the newline type; the default settings are Unix-like. */
2511:
2512: if (strcmp(newline, "cr") == 0 || strcmp(newline, "CR") == 0)
2513: {
2514: pcre_options |= PCRE_NEWLINE_CR;
2515: endlinetype = EL_CR;
2516: }
2517: else if (strcmp(newline, "lf") == 0 || strcmp(newline, "LF") == 0)
2518: {
2519: pcre_options |= PCRE_NEWLINE_LF;
2520: endlinetype = EL_LF;
2521: }
2522: else if (strcmp(newline, "crlf") == 0 || strcmp(newline, "CRLF") == 0)
2523: {
2524: pcre_options |= PCRE_NEWLINE_CRLF;
2525: endlinetype = EL_CRLF;
2526: }
2527: else if (strcmp(newline, "any") == 0 || strcmp(newline, "ANY") == 0)
2528: {
2529: pcre_options |= PCRE_NEWLINE_ANY;
2530: endlinetype = EL_ANY;
2531: }
2532: else if (strcmp(newline, "anycrlf") == 0 || strcmp(newline, "ANYCRLF") == 0)
2533: {
2534: pcre_options |= PCRE_NEWLINE_ANYCRLF;
2535: endlinetype = EL_ANYCRLF;
2536: }
2537: else
2538: {
2539: fprintf(stderr, "pcregrep: Invalid newline specifier \"%s\"\n", newline);
2540: return 2;
2541: }
2542:
2543: /* Interpret the text values for -d and -D */
2544:
2545: if (dee_option != NULL)
2546: {
2547: if (strcmp(dee_option, "read") == 0) dee_action = dee_READ;
2548: else if (strcmp(dee_option, "recurse") == 0) dee_action = dee_RECURSE;
2549: else if (strcmp(dee_option, "skip") == 0) dee_action = dee_SKIP;
2550: else
2551: {
2552: fprintf(stderr, "pcregrep: Invalid value \"%s\" for -d\n", dee_option);
2553: return 2;
2554: }
2555: }
2556:
2557: if (DEE_option != NULL)
2558: {
2559: if (strcmp(DEE_option, "read") == 0) DEE_action = DEE_READ;
2560: else if (strcmp(DEE_option, "skip") == 0) DEE_action = DEE_SKIP;
2561: else
2562: {
2563: fprintf(stderr, "pcregrep: Invalid value \"%s\" for -D\n", DEE_option);
2564: return 2;
2565: }
2566: }
2567:
2568: /* Check the values for Jeffrey Friedl's debugging options. */
2569:
2570: #ifdef JFRIEDL_DEBUG
2571: if (S_arg > 9)
2572: {
2573: fprintf(stderr, "pcregrep: bad value for -S option\n");
2574: return 2;
2575: }
2576: if (jfriedl_XT != 0 || jfriedl_XR != 0)
2577: {
2578: if (jfriedl_XT == 0) jfriedl_XT = 1;
2579: if (jfriedl_XR == 0) jfriedl_XR = 1;
2580: }
2581: #endif
2582:
2583: /* Get memory for the main buffer, and to store the pattern and hints lists. */
2584:
2585: bufsize = 3*bufthird;
2586: main_buffer = (char *)malloc(bufsize);
2587: pattern_list = (pcre **)malloc(MAX_PATTERN_COUNT * sizeof(pcre *));
2588: hints_list = (pcre_extra **)malloc(MAX_PATTERN_COUNT * sizeof(pcre_extra *));
2589:
2590: if (main_buffer == NULL || pattern_list == NULL || hints_list == NULL)
2591: {
2592: fprintf(stderr, "pcregrep: malloc failed\n");
2593: goto EXIT2;
2594: }
2595:
2596: /* If no patterns were provided by -e, and there is no file provided by -f,
2597: the first argument is the one and only pattern, and it must exist. */
2598:
2599: if (cmd_pattern_count == 0 && pattern_filename == NULL)
2600: {
2601: if (i >= argc) return usage(2);
2602: patterns[cmd_pattern_count++] = argv[i++];
2603: }
2604:
2605: /* Compile the patterns that were provided on the command line, either by
2606: multiple uses of -e or as a single unkeyed pattern. */
2607:
2608: for (j = 0; j < cmd_pattern_count; j++)
2609: {
2610: if (!compile_pattern(patterns[j], pcre_options, NULL,
2611: (j == 0 && cmd_pattern_count == 1)? 0 : j + 1))
2612: goto EXIT2;
2613: }
2614:
2615: /* Compile the regular expressions that are provided in a file. */
2616:
2617: if (pattern_filename != NULL)
2618: {
2619: int linenumber = 0;
2620: FILE *f;
2621: char *filename;
2622: char buffer[PATBUFSIZE];
2623:
2624: if (strcmp(pattern_filename, "-") == 0)
2625: {
2626: f = stdin;
2627: filename = stdin_name;
2628: }
2629: else
2630: {
2631: f = fopen(pattern_filename, "r");
2632: if (f == NULL)
2633: {
2634: fprintf(stderr, "pcregrep: Failed to open %s: %s\n", pattern_filename,
2635: strerror(errno));
2636: goto EXIT2;
2637: }
2638: filename = pattern_filename;
2639: }
2640:
2641: while (fgets(buffer, PATBUFSIZE, f) != NULL)
2642: {
2643: char *s = buffer + (int)strlen(buffer);
2644: while (s > buffer && isspace((unsigned char)(s[-1]))) s--;
2645: *s = 0;
2646: linenumber++;
2647: if (buffer[0] == 0) continue; /* Skip blank lines */
2648: if (!compile_pattern(buffer, pcre_options, filename, linenumber))
2649: goto EXIT2;
2650: }
2651:
2652: if (f != stdin) fclose(f);
2653: }
2654:
2655: /* Study the regular expressions, as we will be running them many times. Unless
2656: JIT has been explicitly disabled, arrange a stack for it to use. */
2657:
2658: #ifdef SUPPORT_PCREGREP_JIT
2659: if ((study_options & PCRE_STUDY_JIT_COMPILE) != 0)
2660: jit_stack = pcre_jit_stack_alloc(32*1024, 1024*1024);
2661: #endif
2662:
2663: for (j = 0; j < pattern_count; j++)
2664: {
2665: hints_list[j] = pcre_study(pattern_list[j], study_options, &error);
2666: if (error != NULL)
2667: {
2668: char s[16];
2669: if (pattern_count == 1) s[0] = 0; else sprintf(s, " number %d", j);
2670: fprintf(stderr, "pcregrep: Error while studying regex%s: %s\n", s, error);
2671: goto EXIT2;
2672: }
2673: hint_count++;
2674: #ifdef SUPPORT_PCREGREP_JIT
2675: if (jit_stack != NULL && hints_list[j] != NULL)
2676: pcre_assign_jit_stack(hints_list[j], NULL, jit_stack);
2677: #endif
2678: }
2679:
2680: /* If --match-limit or --recursion-limit was set, put the value(s) into the
2681: pcre_extra block for each pattern. */
2682:
2683: if (match_limit > 0 || match_limit_recursion > 0)
2684: {
2685: for (j = 0; j < pattern_count; j++)
2686: {
2687: if (hints_list[j] == NULL)
2688: {
2689: hints_list[j] = malloc(sizeof(pcre_extra));
2690: if (hints_list[j] == NULL)
2691: {
2692: fprintf(stderr, "pcregrep: malloc failed\n");
2693: pcregrep_exit(2);
2694: }
2695: }
2696: if (match_limit > 0)
2697: {
2698: hints_list[j]->flags |= PCRE_EXTRA_MATCH_LIMIT;
2699: hints_list[j]->match_limit = match_limit;
2700: }
2701: if (match_limit_recursion > 0)
2702: {
2703: hints_list[j]->flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION;
2704: hints_list[j]->match_limit_recursion = match_limit_recursion;
2705: }
2706: }
2707: }
2708:
2709: /* If there are include or exclude patterns, compile them. */
2710:
2711: if (exclude_pattern != NULL)
2712: {
2713: exclude_compiled = pcre_compile(exclude_pattern, 0, &error, &errptr,
2714: pcretables);
2715: if (exclude_compiled == NULL)
2716: {
2717: fprintf(stderr, "pcregrep: Error in 'exclude' regex at offset %d: %s\n",
2718: errptr, error);
2719: goto EXIT2;
2720: }
2721: }
2722:
2723: if (include_pattern != NULL)
2724: {
2725: include_compiled = pcre_compile(include_pattern, 0, &error, &errptr,
2726: pcretables);
2727: if (include_compiled == NULL)
2728: {
2729: fprintf(stderr, "pcregrep: Error in 'include' regex at offset %d: %s\n",
2730: errptr, error);
2731: goto EXIT2;
2732: }
2733: }
2734:
2735: if (exclude_dir_pattern != NULL)
2736: {
2737: exclude_dir_compiled = pcre_compile(exclude_dir_pattern, 0, &error, &errptr,
2738: pcretables);
2739: if (exclude_dir_compiled == NULL)
2740: {
2741: fprintf(stderr, "pcregrep: Error in 'exclude_dir' regex at offset %d: %s\n",
2742: errptr, error);
2743: goto EXIT2;
2744: }
2745: }
2746:
2747: if (include_dir_pattern != NULL)
2748: {
2749: include_dir_compiled = pcre_compile(include_dir_pattern, 0, &error, &errptr,
2750: pcretables);
2751: if (include_dir_compiled == NULL)
2752: {
2753: fprintf(stderr, "pcregrep: Error in 'include_dir' regex at offset %d: %s\n",
2754: errptr, error);
2755: goto EXIT2;
2756: }
2757: }
2758:
1.1.1.3 ! misho 2759: /* If a file that contains a list of files to search has been specified, read
! 2760: it line by line and search the given files. Otherwise, if there are no further
! 2761: arguments, do the business on stdin and exit. */
! 2762:
! 2763: if (file_list != NULL)
! 2764: {
! 2765: char buffer[PATBUFSIZE];
! 2766: FILE *fl;
! 2767: if (strcmp(file_list, "-") == 0) fl = stdin; else
! 2768: {
! 2769: fl = fopen(file_list, "rb");
! 2770: if (fl == NULL)
! 2771: {
! 2772: fprintf(stderr, "pcregrep: Failed to open %s: %s\n", file_list,
! 2773: strerror(errno));
! 2774: goto EXIT2;
! 2775: }
! 2776: }
! 2777: while (fgets(buffer, PATBUFSIZE, fl) != NULL)
! 2778: {
! 2779: int frc;
! 2780: char *end = buffer + (int)strlen(buffer);
! 2781: while (end > buffer && isspace(end[-1])) end--;
! 2782: *end = 0;
! 2783: if (*buffer != 0)
! 2784: {
! 2785: frc = grep_or_recurse(buffer, dee_action == dee_RECURSE, FALSE);
! 2786: if (frc > 1) rc = frc;
! 2787: else if (frc == 0 && rc == 1) rc = 0;
! 2788: }
! 2789: }
! 2790: if (fl != stdin) fclose (fl);
! 2791: }
! 2792:
! 2793: /* Do this only if there was no file list (and no file arguments). */
1.1 misho 2794:
1.1.1.3 ! misho 2795: else if (i >= argc)
1.1 misho 2796: {
2797: rc = pcregrep(stdin, FR_PLAIN, stdin_name,
2798: (filenames > FN_DEFAULT)? stdin_name : NULL);
2799: goto EXIT;
2800: }
2801:
1.1.1.3 ! misho 2802: /* After handling file-list or if there are remaining arguments, work through
! 2803: them as files or directories. Pass in the fact that there is only one argument
! 2804: at top level - this suppresses the file name if the argument is not a directory
! 2805: and filenames are not otherwise forced. */
1.1 misho 2806:
1.1.1.3 ! misho 2807: only_one_at_top = i == argc - 1 && file_list == NULL;
1.1 misho 2808:
2809: for (; i < argc; i++)
2810: {
2811: int frc = grep_or_recurse(argv[i], dee_action == dee_RECURSE,
2812: only_one_at_top);
2813: if (frc > 1) rc = frc;
2814: else if (frc == 0 && rc == 1) rc = 0;
2815: }
2816:
2817: EXIT:
2818: #ifdef SUPPORT_PCREGREP_JIT
2819: if (jit_stack != NULL) pcre_jit_stack_free(jit_stack);
2820: #endif
2821: if (main_buffer != NULL) free(main_buffer);
2822: if (pattern_list != NULL)
2823: {
2824: for (i = 0; i < pattern_count; i++) free(pattern_list[i]);
2825: free(pattern_list);
2826: }
2827: if (hints_list != NULL)
2828: {
2829: for (i = 0; i < hint_count; i++)
2830: {
2831: if (hints_list[i] != NULL) pcre_free_study(hints_list[i]);
2832: }
2833: free(hints_list);
2834: }
2835: pcregrep_exit(rc);
2836:
2837: EXIT2:
2838: rc = 2;
2839: goto EXIT;
2840: }
2841:
2842: /* End of pcregrep */
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>