Annotation of embedaddon/pcre/pcregrep.c, revision 1.1.1.4
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 OFFSET_SIZE 99
74:
75: #if BUFSIZ > 8192
1.1.1.4 ! misho 76: #define MAXPATLEN BUFSIZ
1.1 misho 77: #else
1.1.1.4 ! misho 78: #define MAXPATLEN 8192
1.1 misho 79: #endif
80:
1.1.1.4 ! misho 81: #define PATBUFSIZE (MAXPATLEN + 10) /* Allows for prefix+suffix */
! 82:
1.1 misho 83: /* Values for the "filenames" variable, which specifies options for file name
84: output. The order is important; it is assumed that a file name is wanted for
85: all values greater than FN_DEFAULT. */
86:
87: enum { FN_NONE, FN_DEFAULT, FN_MATCH_ONLY, FN_NOMATCH_ONLY, FN_FORCE };
88:
89: /* File reading styles */
90:
91: enum { FR_PLAIN, FR_LIBZ, FR_LIBBZ2 };
92:
93: /* Actions for the -d and -D options */
94:
95: enum { dee_READ, dee_SKIP, dee_RECURSE };
96: enum { DEE_READ, DEE_SKIP };
97:
98: /* Actions for special processing options (flag bits) */
99:
100: #define PO_WORD_MATCH 0x0001
101: #define PO_LINE_MATCH 0x0002
102: #define PO_FIXED_STRINGS 0x0004
103:
104: /* Line ending types */
105:
106: enum { EL_LF, EL_CR, EL_CRLF, EL_ANY, EL_ANYCRLF };
107:
1.1.1.3 misho 108: /* Binary file options */
109:
110: enum { BIN_BINARY, BIN_NOMATCH, BIN_TEXT };
111:
1.1 misho 112: /* In newer versions of gcc, with FORTIFY_SOURCE set (the default in some
113: environments), a warning is issued if the value of fwrite() is ignored.
114: Unfortunately, casting to (void) does not suppress the warning. To get round
115: this, we use a macro that compiles a fudge. Oddly, this does not also seem to
116: apply to fprintf(). */
117:
118: #define FWRITE(a,b,c,d) if (fwrite(a,b,c,d)) {}
119:
120:
121:
122: /*************************************************
123: * Global variables *
124: *************************************************/
125:
126: /* Jeffrey Friedl has some debugging requirements that are not part of the
127: regular code. */
128:
129: #ifdef JFRIEDL_DEBUG
130: static int S_arg = -1;
131: static unsigned int jfriedl_XR = 0; /* repeat regex attempt this many times */
132: static unsigned int jfriedl_XT = 0; /* replicate text this many times */
133: static const char *jfriedl_prefix = "";
134: static const char *jfriedl_postfix = "";
135: #endif
136:
137: static int endlinetype;
138:
139: static char *colour_string = (char *)"1;31";
140: static char *colour_option = NULL;
141: static char *dee_option = NULL;
142: static char *DEE_option = NULL;
1.1.1.4 ! misho 143: static char *locale = NULL;
1.1 misho 144: static char *main_buffer = NULL;
145: static char *newline = NULL;
1.1.1.4 ! misho 146: static char *om_separator = (char *)"";
1.1 misho 147: static char *stdin_name = (char *)"(standard input)";
148:
149: static const unsigned char *pcretables = NULL;
150:
151: static int after_context = 0;
152: static int before_context = 0;
1.1.1.3 misho 153: static int binary_files = BIN_BINARY;
1.1 misho 154: static int both_context = 0;
155: static int bufthird = PCREGREP_BUFSIZE;
156: static int bufsize = 3*PCREGREP_BUFSIZE;
1.1.1.4 ! misho 157:
! 158: #if defined HAVE_WINDOWS_H && HAVE_WINDOWS_H
! 159: static int dee_action = dee_SKIP;
! 160: #else
1.1 misho 161: static int dee_action = dee_READ;
1.1.1.4 ! misho 162: #endif
! 163:
1.1 misho 164: static int DEE_action = DEE_READ;
165: static int error_count = 0;
166: static int filenames = FN_DEFAULT;
1.1.1.4 ! misho 167: static int pcre_options = 0;
1.1 misho 168: static int process_options = 0;
169:
170: #ifdef SUPPORT_PCREGREP_JIT
171: static int study_options = PCRE_STUDY_JIT_COMPILE;
172: #else
173: static int study_options = 0;
174: #endif
175:
176: static unsigned long int match_limit = 0;
177: static unsigned long int match_limit_recursion = 0;
178:
179: static BOOL count_only = FALSE;
180: static BOOL do_colour = FALSE;
181: static BOOL file_offsets = FALSE;
182: static BOOL hyphenpending = FALSE;
183: static BOOL invert = FALSE;
184: static BOOL line_buffered = FALSE;
185: static BOOL line_offsets = FALSE;
186: static BOOL multiline = FALSE;
187: static BOOL number = FALSE;
188: static BOOL omit_zero_count = FALSE;
189: static BOOL resource_error = FALSE;
190: static BOOL quiet = FALSE;
1.1.1.4 ! misho 191: static BOOL show_only_matching = FALSE;
1.1 misho 192: static BOOL silent = FALSE;
193: static BOOL utf8 = FALSE;
194:
1.1.1.4 ! misho 195: /* Structure for list of --only-matching capturing numbers. */
! 196:
! 197: typedef struct omstr {
! 198: struct omstr *next;
! 199: int groupnum;
! 200: } omstr;
! 201:
! 202: static omstr *only_matching = NULL;
! 203: static omstr *only_matching_last = NULL;
! 204:
! 205: /* Structure for holding the two variables that describe a number chain. */
! 206:
! 207: typedef struct omdatastr {
! 208: omstr **anchor;
! 209: omstr **lastptr;
! 210: } omdatastr;
! 211:
! 212: static omdatastr only_matching_data = { &only_matching, &only_matching_last };
! 213:
! 214: /* Structure for list of file names (for -f and --{in,ex}clude-from) */
! 215:
! 216: typedef struct fnstr {
! 217: struct fnstr *next;
! 218: char *name;
! 219: } fnstr;
! 220:
! 221: static fnstr *exclude_from = NULL;
! 222: static fnstr *exclude_from_last = NULL;
! 223: static fnstr *include_from = NULL;
! 224: static fnstr *include_from_last = NULL;
! 225:
! 226: static fnstr *file_lists = NULL;
! 227: static fnstr *file_lists_last = NULL;
! 228: static fnstr *pattern_files = NULL;
! 229: static fnstr *pattern_files_last = NULL;
! 230:
! 231: /* Structure for holding the two variables that describe a file name chain. */
! 232:
! 233: typedef struct fndatastr {
! 234: fnstr **anchor;
! 235: fnstr **lastptr;
! 236: } fndatastr;
! 237:
! 238: static fndatastr exclude_from_data = { &exclude_from, &exclude_from_last };
! 239: static fndatastr include_from_data = { &include_from, &include_from_last };
! 240: static fndatastr file_lists_data = { &file_lists, &file_lists_last };
! 241: static fndatastr pattern_files_data = { &pattern_files, &pattern_files_last };
! 242:
! 243: /* Structure for pattern and its compiled form; used for matching patterns and
! 244: also for include/exclude patterns. */
! 245:
! 246: typedef struct patstr {
! 247: struct patstr *next;
! 248: char *string;
! 249: pcre *compiled;
! 250: pcre_extra *hint;
! 251: } patstr;
! 252:
! 253: static patstr *patterns = NULL;
! 254: static patstr *patterns_last = NULL;
! 255: static patstr *include_patterns = NULL;
! 256: static patstr *include_patterns_last = NULL;
! 257: static patstr *exclude_patterns = NULL;
! 258: static patstr *exclude_patterns_last = NULL;
! 259: static patstr *include_dir_patterns = NULL;
! 260: static patstr *include_dir_patterns_last = NULL;
! 261: static patstr *exclude_dir_patterns = NULL;
! 262: static patstr *exclude_dir_patterns_last = NULL;
! 263:
! 264: /* Structure holding the two variables that describe a pattern chain. A pointer
! 265: to such structures is used for each appropriate option. */
! 266:
! 267: typedef struct patdatastr {
! 268: patstr **anchor;
! 269: patstr **lastptr;
! 270: } patdatastr;
! 271:
! 272: static patdatastr match_patdata = { &patterns, &patterns_last };
! 273: static patdatastr include_patdata = { &include_patterns, &include_patterns_last };
! 274: static patdatastr exclude_patdata = { &exclude_patterns, &exclude_patterns_last };
! 275: static patdatastr include_dir_patdata = { &include_dir_patterns, &include_dir_patterns_last };
! 276: static patdatastr exclude_dir_patdata = { &exclude_dir_patterns, &exclude_dir_patterns_last };
! 277:
! 278: static patstr **incexlist[4] = { &include_patterns, &exclude_patterns,
! 279: &include_dir_patterns, &exclude_dir_patterns };
! 280:
! 281: static const char *incexname[4] = { "--include", "--exclude",
! 282: "--include-dir", "--exclude-dir" };
! 283:
1.1 misho 284: /* Structure for options and list of them */
285:
286: enum { OP_NODATA, OP_STRING, OP_OP_STRING, OP_NUMBER, OP_LONGNUMBER,
1.1.1.4 ! misho 287: OP_OP_NUMBER, OP_OP_NUMBERS, OP_PATLIST, OP_FILELIST, OP_BINFILES };
1.1 misho 288:
289: typedef struct option_item {
290: int type;
291: int one_char;
292: void *dataptr;
293: const char *long_name;
294: const char *help_text;
295: } option_item;
296:
297: /* Options without a single-letter equivalent get a negative value. This can be
298: used to identify them. */
299:
300: #define N_COLOUR (-1)
301: #define N_EXCLUDE (-2)
302: #define N_EXCLUDE_DIR (-3)
303: #define N_HELP (-4)
304: #define N_INCLUDE (-5)
305: #define N_INCLUDE_DIR (-6)
306: #define N_LABEL (-7)
307: #define N_LOCALE (-8)
308: #define N_NULL (-9)
309: #define N_LOFFSETS (-10)
310: #define N_FOFFSETS (-11)
311: #define N_LBUFFER (-12)
312: #define N_M_LIMIT (-13)
313: #define N_M_LIMIT_REC (-14)
314: #define N_BUFSIZE (-15)
315: #define N_NOJIT (-16)
1.1.1.3 misho 316: #define N_FILE_LIST (-17)
317: #define N_BINARY_FILES (-18)
1.1.1.4 ! misho 318: #define N_EXCLUDE_FROM (-19)
! 319: #define N_INCLUDE_FROM (-20)
! 320: #define N_OM_SEPARATOR (-21)
1.1 misho 321:
322: static option_item optionlist[] = {
1.1.1.3 misho 323: { OP_NODATA, N_NULL, NULL, "", "terminate options" },
1.1 misho 324: { OP_NODATA, N_HELP, NULL, "help", "display this help and exit" },
325: { OP_NUMBER, 'A', &after_context, "after-context=number", "set number of following context lines" },
1.1.1.3 misho 326: { OP_NODATA, 'a', NULL, "text", "treat binary files as text" },
1.1 misho 327: { OP_NUMBER, 'B', &before_context, "before-context=number", "set number of prior context lines" },
1.1.1.3 misho 328: { OP_BINFILES, N_BINARY_FILES, NULL, "binary-files=word", "set treatment of binary files" },
1.1 misho 329: { OP_NUMBER, N_BUFSIZE,&bufthird, "buffer-size=number", "set processing buffer size parameter" },
330: { OP_OP_STRING, N_COLOUR, &colour_option, "color=option", "matched text color option" },
331: { OP_OP_STRING, N_COLOUR, &colour_option, "colour=option", "matched text colour option" },
332: { OP_NUMBER, 'C', &both_context, "context=number", "set number of context lines, before & after" },
333: { OP_NODATA, 'c', NULL, "count", "print only a count of matching lines per FILE" },
334: { OP_STRING, 'D', &DEE_option, "devices=action","how to handle devices, FIFOs, and sockets" },
335: { OP_STRING, 'd', &dee_option, "directories=action", "how to handle directories" },
1.1.1.4 ! misho 336: { OP_PATLIST, 'e', &match_patdata, "regex(p)=pattern", "specify pattern (may be used more than once)" },
1.1 misho 337: { OP_NODATA, 'F', NULL, "fixed-strings", "patterns are sets of newline-separated strings" },
1.1.1.4 ! misho 338: { OP_FILELIST, 'f', &pattern_files_data, "file=path", "read patterns from file" },
! 339: { OP_FILELIST, N_FILE_LIST, &file_lists_data, "file-list=path","read files to search from file" },
1.1 misho 340: { OP_NODATA, N_FOFFSETS, NULL, "file-offsets", "output file offsets, not text" },
341: { OP_NODATA, 'H', NULL, "with-filename", "force the prefixing filename on output" },
342: { OP_NODATA, 'h', NULL, "no-filename", "suppress the prefixing filename on output" },
1.1.1.3 misho 343: { OP_NODATA, 'I', NULL, "", "treat binary files as not matching (ignore)" },
1.1 misho 344: { OP_NODATA, 'i', NULL, "ignore-case", "ignore case distinctions" },
345: #ifdef SUPPORT_PCREGREP_JIT
346: { OP_NODATA, N_NOJIT, NULL, "no-jit", "do not use just-in-time compiler optimization" },
347: #else
348: { OP_NODATA, N_NOJIT, NULL, "no-jit", "ignored: this pcregrep does not support JIT" },
349: #endif
350: { OP_NODATA, 'l', NULL, "files-with-matches", "print only FILE names containing matches" },
351: { OP_NODATA, 'L', NULL, "files-without-match","print only FILE names not containing matches" },
352: { OP_STRING, N_LABEL, &stdin_name, "label=name", "set name for standard input" },
353: { OP_NODATA, N_LBUFFER, NULL, "line-buffered", "use line buffering" },
354: { OP_NODATA, N_LOFFSETS, NULL, "line-offsets", "output line numbers and offsets, not text" },
355: { OP_STRING, N_LOCALE, &locale, "locale=locale", "use the named locale" },
356: { OP_LONGNUMBER, N_M_LIMIT, &match_limit, "match-limit=number", "set PCRE match limit option" },
357: { OP_LONGNUMBER, N_M_LIMIT_REC, &match_limit_recursion, "recursion-limit=number", "set PCRE match recursion limit option" },
358: { OP_NODATA, 'M', NULL, "multiline", "run in multiline mode" },
359: { OP_STRING, 'N', &newline, "newline=type", "set newline type (CR, LF, CRLF, ANYCRLF or ANY)" },
360: { OP_NODATA, 'n', NULL, "line-number", "print line number with output lines" },
1.1.1.4 ! misho 361: { OP_OP_NUMBERS, 'o', &only_matching_data, "only-matching=n", "show only the part of the line that matched" },
! 362: { OP_STRING, N_OM_SEPARATOR, &om_separator, "om-separator=text", "set separator for multiple -o output" },
1.1 misho 363: { OP_NODATA, 'q', NULL, "quiet", "suppress output, just set return code" },
364: { OP_NODATA, 'r', NULL, "recursive", "recursively scan sub-directories" },
1.1.1.4 ! misho 365: { OP_PATLIST, N_EXCLUDE,&exclude_patdata, "exclude=pattern","exclude matching files when recursing" },
! 366: { OP_PATLIST, N_INCLUDE,&include_patdata, "include=pattern","include matching files when recursing" },
! 367: { OP_PATLIST, N_EXCLUDE_DIR,&exclude_dir_patdata, "exclude-dir=pattern","exclude matching directories when recursing" },
! 368: { OP_PATLIST, N_INCLUDE_DIR,&include_dir_patdata, "include-dir=pattern","include matching directories when recursing" },
! 369: { OP_FILELIST, N_EXCLUDE_FROM,&exclude_from_data, "exclude-from=path", "read exclude list from file" },
! 370: { OP_FILELIST, N_INCLUDE_FROM,&include_from_data, "include-from=path", "read include list from file" },
1.1 misho 371:
372: /* These two were accidentally implemented with underscores instead of
373: hyphens in the option names. As this was not discovered for several releases,
374: the incorrect versions are left in the table for compatibility. However, the
375: --help function misses out any option that has an underscore in its name. */
376:
1.1.1.4 ! misho 377: { OP_PATLIST, N_EXCLUDE_DIR,&exclude_dir_patdata, "exclude_dir=pattern","exclude matching directories when recursing" },
! 378: { OP_PATLIST, N_INCLUDE_DIR,&include_dir_patdata, "include_dir=pattern","include matching directories when recursing" },
1.1 misho 379:
380: #ifdef JFRIEDL_DEBUG
381: { OP_OP_NUMBER, 'S', &S_arg, "jeffS", "replace matched (sub)string with X" },
382: #endif
383: { OP_NODATA, 's', NULL, "no-messages", "suppress error messages" },
384: { OP_NODATA, 'u', NULL, "utf-8", "use UTF-8 mode" },
385: { OP_NODATA, 'V', NULL, "version", "print version information and exit" },
386: { OP_NODATA, 'v', NULL, "invert-match", "select non-matching lines" },
387: { OP_NODATA, 'w', NULL, "word-regex(p)", "force patterns to match only as words" },
388: { OP_NODATA, 'x', NULL, "line-regex(p)", "force patterns to match only whole lines" },
389: { OP_NODATA, 0, NULL, NULL, NULL }
390: };
391:
392: /* Tables for prefixing and suffixing patterns, according to the -w, -x, and -F
393: options. These set the 1, 2, and 4 bits in process_options, respectively. Note
394: that the combination of -w and -x has the same effect as -x on its own, so we
1.1.1.4 ! misho 395: can treat them as the same. Note that the MAXPATLEN macro assumes the longest
! 396: prefix+suffix is 10 characters; if anything longer is added, it must be
! 397: adjusted. */
1.1 misho 398:
399: static const char *prefix[] = {
400: "", "\\b", "^(?:", "^(?:", "\\Q", "\\b\\Q", "^(?:\\Q", "^(?:\\Q" };
401:
402: static const char *suffix[] = {
403: "", "\\b", ")$", ")$", "\\E", "\\E\\b", "\\E)$", "\\E)$" };
404:
405: /* UTF-8 tables - used only when the newline setting is "any". */
406:
407: const int utf8_table3[] = { 0xff, 0x1f, 0x0f, 0x07, 0x03, 0x01};
408:
409: const char utf8_table4[] = {
410: 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
411: 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,
412: 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,
413: 3,3,3,3,3,3,3,3,4,4,4,4,5,5,5,5 };
414:
415:
416:
417: /*************************************************
418: * Exit from the program *
419: *************************************************/
420:
421: /* If there has been a resource error, give a suitable message.
422:
423: Argument: the return code
424: Returns: does not return
425: */
426:
427: static void
428: pcregrep_exit(int rc)
429: {
430: if (resource_error)
431: {
432: fprintf(stderr, "pcregrep: Error %d, %d or %d means that a resource limit "
433: "was exceeded.\n", PCRE_ERROR_MATCHLIMIT, PCRE_ERROR_RECURSIONLIMIT,
434: PCRE_ERROR_JIT_STACKLIMIT);
435: fprintf(stderr, "pcregrep: Check your regex for nested unlimited loops.\n");
436: }
437: exit(rc);
438: }
439:
440:
441: /*************************************************
1.1.1.4 ! misho 442: * Add item to chain of patterns *
! 443: *************************************************/
! 444:
! 445: /* Used to add an item onto a chain, or just return an unconnected item if the
! 446: "after" argument is NULL.
! 447:
! 448: Arguments:
! 449: s pattern string to add
! 450: after if not NULL points to item to insert after
! 451:
! 452: Returns: new pattern block
! 453: */
! 454:
! 455: static patstr *
! 456: add_pattern(char *s, patstr *after)
! 457: {
! 458: patstr *p = (patstr *)malloc(sizeof(patstr));
! 459: if (p == NULL)
! 460: {
! 461: fprintf(stderr, "pcregrep: malloc failed\n");
! 462: pcregrep_exit(2);
! 463: }
! 464: if (strlen(s) > MAXPATLEN)
! 465: {
! 466: fprintf(stderr, "pcregrep: pattern is too long (limit is %d bytes)\n",
! 467: MAXPATLEN);
! 468: return NULL;
! 469: }
! 470: p->next = NULL;
! 471: p->string = s;
! 472: p->compiled = NULL;
! 473: p->hint = NULL;
! 474:
! 475: if (after != NULL)
! 476: {
! 477: p->next = after->next;
! 478: after->next = p;
! 479: }
! 480: return p;
! 481: }
! 482:
! 483:
! 484: /*************************************************
! 485: * Free chain of patterns *
! 486: *************************************************/
! 487:
! 488: /* Used for several chains of patterns.
! 489:
! 490: Argument: pointer to start of chain
! 491: Returns: nothing
! 492: */
! 493:
! 494: static void
! 495: free_pattern_chain(patstr *pc)
! 496: {
! 497: while (pc != NULL)
! 498: {
! 499: patstr *p = pc;
! 500: pc = p->next;
! 501: if (p->hint != NULL) pcre_free_study(p->hint);
! 502: if (p->compiled != NULL) pcre_free(p->compiled);
! 503: free(p);
! 504: }
! 505: }
! 506:
! 507:
! 508: /*************************************************
! 509: * Free chain of file names *
! 510: *************************************************/
! 511:
! 512: /*
! 513: Argument: pointer to start of chain
! 514: Returns: nothing
! 515: */
! 516:
! 517: static void
! 518: free_file_chain(fnstr *fn)
! 519: {
! 520: while (fn != NULL)
! 521: {
! 522: fnstr *f = fn;
! 523: fn = f->next;
! 524: free(f);
! 525: }
! 526: }
! 527:
! 528:
! 529: /*************************************************
1.1 misho 530: * OS-specific functions *
531: *************************************************/
532:
533: /* These functions are defined so that they can be made system specific,
534: although at present the only ones are for Unix, Win32, and for "no support". */
535:
536:
537: /************* Directory scanning in Unix ***********/
538:
539: #if defined HAVE_SYS_STAT_H && defined HAVE_DIRENT_H && defined HAVE_SYS_TYPES_H
540: #include <sys/types.h>
541: #include <sys/stat.h>
542: #include <dirent.h>
543:
544: typedef DIR directory_type;
1.1.1.4 ! misho 545: #define FILESEP '/'
1.1 misho 546:
547: static int
548: isdirectory(char *filename)
549: {
550: struct stat statbuf;
551: if (stat(filename, &statbuf) < 0)
552: return 0; /* In the expectation that opening as a file will fail */
1.1.1.4 ! misho 553: return (statbuf.st_mode & S_IFMT) == S_IFDIR;
1.1 misho 554: }
555:
556: static directory_type *
557: opendirectory(char *filename)
558: {
559: return opendir(filename);
560: }
561:
562: static char *
563: readdirectory(directory_type *dir)
564: {
565: for (;;)
566: {
567: struct dirent *dent = readdir(dir);
568: if (dent == NULL) return NULL;
569: if (strcmp(dent->d_name, ".") != 0 && strcmp(dent->d_name, "..") != 0)
570: return dent->d_name;
571: }
572: /* Control never reaches here */
573: }
574:
575: static void
576: closedirectory(directory_type *dir)
577: {
578: closedir(dir);
579: }
580:
581:
582: /************* Test for regular file in Unix **********/
583:
584: static int
585: isregfile(char *filename)
586: {
587: struct stat statbuf;
588: if (stat(filename, &statbuf) < 0)
589: return 1; /* In the expectation that opening as a file will fail */
590: return (statbuf.st_mode & S_IFMT) == S_IFREG;
591: }
592:
593:
594: /************* Test for a terminal in Unix **********/
595:
596: static BOOL
597: is_stdout_tty(void)
598: {
599: return isatty(fileno(stdout));
600: }
601:
602: static BOOL
603: is_file_tty(FILE *f)
604: {
605: return isatty(fileno(f));
606: }
607:
608:
609: /************* Directory scanning in Win32 ***********/
610:
611: /* I (Philip Hazel) have no means of testing this code. It was contributed by
612: Lionel Fourquaux. David Burgess added a patch to define INVALID_FILE_ATTRIBUTES
613: when it did not exist. David Byron added a patch that moved the #include of
614: <windows.h> to before the INVALID_FILE_ATTRIBUTES definition rather than after.
615: The double test below stops gcc 4.4.4 grumbling that HAVE_WINDOWS_H is
616: undefined when it is indeed undefined. */
617:
618: #elif defined HAVE_WINDOWS_H && HAVE_WINDOWS_H
619:
620: #ifndef STRICT
621: # define STRICT
622: #endif
623: #ifndef WIN32_LEAN_AND_MEAN
624: # define WIN32_LEAN_AND_MEAN
625: #endif
626:
627: #include <windows.h>
628:
629: #ifndef INVALID_FILE_ATTRIBUTES
630: #define INVALID_FILE_ATTRIBUTES 0xFFFFFFFF
631: #endif
632:
633: typedef struct directory_type
634: {
635: HANDLE handle;
636: BOOL first;
637: WIN32_FIND_DATA data;
638: } directory_type;
639:
1.1.1.4 ! misho 640: #define FILESEP '/'
! 641:
1.1 misho 642: int
643: isdirectory(char *filename)
644: {
645: DWORD attr = GetFileAttributes(filename);
646: if (attr == INVALID_FILE_ATTRIBUTES)
647: return 0;
1.1.1.4 ! misho 648: return (attr & FILE_ATTRIBUTE_DIRECTORY) != 0;
1.1 misho 649: }
650:
651: directory_type *
652: opendirectory(char *filename)
653: {
654: size_t len;
655: char *pattern;
656: directory_type *dir;
657: DWORD err;
658: len = strlen(filename);
1.1.1.4 ! misho 659: pattern = (char *)malloc(len + 3);
! 660: dir = (directory_type *)malloc(sizeof(*dir));
1.1 misho 661: if ((pattern == NULL) || (dir == NULL))
662: {
663: fprintf(stderr, "pcregrep: malloc failed\n");
664: pcregrep_exit(2);
665: }
666: memcpy(pattern, filename, len);
667: memcpy(&(pattern[len]), "\\*", 3);
668: dir->handle = FindFirstFile(pattern, &(dir->data));
669: if (dir->handle != INVALID_HANDLE_VALUE)
670: {
671: free(pattern);
672: dir->first = TRUE;
673: return dir;
674: }
675: err = GetLastError();
676: free(pattern);
677: free(dir);
678: errno = (err == ERROR_ACCESS_DENIED) ? EACCES : ENOENT;
679: return NULL;
680: }
681:
682: char *
683: readdirectory(directory_type *dir)
684: {
685: for (;;)
686: {
687: if (!dir->first)
688: {
689: if (!FindNextFile(dir->handle, &(dir->data)))
690: return NULL;
691: }
692: else
693: {
694: dir->first = FALSE;
695: }
696: if (strcmp(dir->data.cFileName, ".") != 0 && strcmp(dir->data.cFileName, "..") != 0)
697: return dir->data.cFileName;
698: }
699: #ifndef _MSC_VER
700: return NULL; /* Keep compiler happy; never executed */
701: #endif
702: }
703:
704: void
705: closedirectory(directory_type *dir)
706: {
707: FindClose(dir->handle);
708: free(dir);
709: }
710:
711:
712: /************* Test for regular file in Win32 **********/
713:
714: /* I don't know how to do this, or if it can be done; assume all paths are
715: regular if they are not directories. */
716:
717: int isregfile(char *filename)
718: {
719: return !isdirectory(filename);
720: }
721:
722:
723: /************* Test for a terminal in Win32 **********/
724:
725: /* I don't know how to do this; assume never */
726:
727: static BOOL
728: is_stdout_tty(void)
729: {
730: return FALSE;
731: }
732:
733: static BOOL
734: is_file_tty(FILE *f)
735: {
736: return FALSE;
737: }
738:
739:
740: /************* Directory scanning when we can't do it ***********/
741:
742: /* The type is void, and apart from isdirectory(), the functions do nothing. */
743:
744: #else
745:
1.1.1.4 ! misho 746: #define FILESEP 0
1.1 misho 747: typedef void directory_type;
748:
749: int isdirectory(char *filename) { return 0; }
750: directory_type * opendirectory(char *filename) { return (directory_type*)0;}
751: char *readdirectory(directory_type *dir) { return (char*)0;}
752: void closedirectory(directory_type *dir) {}
753:
754:
755: /************* Test for regular when we can't do it **********/
756:
757: /* Assume all files are regular. */
758:
759: int isregfile(char *filename) { return 1; }
760:
761:
762: /************* Test for a terminal when we can't do it **********/
763:
764: static BOOL
765: is_stdout_tty(void)
766: {
767: return FALSE;
768: }
769:
770: static BOOL
771: is_file_tty(FILE *f)
772: {
773: return FALSE;
774: }
775:
776: #endif
777:
778:
779:
780: #ifndef HAVE_STRERROR
781: /*************************************************
782: * Provide strerror() for non-ANSI libraries *
783: *************************************************/
784:
785: /* Some old-fashioned systems still around (e.g. SunOS4) don't have strerror()
786: in their libraries, but can provide the same facility by this simple
787: alternative function. */
788:
789: extern int sys_nerr;
790: extern char *sys_errlist[];
791:
792: char *
793: strerror(int n)
794: {
795: if (n < 0 || n >= sys_nerr) return "unknown error number";
796: return sys_errlist[n];
797: }
798: #endif /* HAVE_STRERROR */
799:
800:
801:
802: /*************************************************
1.1.1.4 ! misho 803: * Usage function *
! 804: *************************************************/
! 805:
! 806: static int
! 807: usage(int rc)
! 808: {
! 809: option_item *op;
! 810: fprintf(stderr, "Usage: pcregrep [-");
! 811: for (op = optionlist; op->one_char != 0; op++)
! 812: {
! 813: if (op->one_char > 0) fprintf(stderr, "%c", op->one_char);
! 814: }
! 815: fprintf(stderr, "] [long options] [pattern] [files]\n");
! 816: fprintf(stderr, "Type `pcregrep --help' for more information and the long "
! 817: "options.\n");
! 818: return rc;
! 819: }
! 820:
! 821:
! 822:
! 823: /*************************************************
! 824: * Help function *
! 825: *************************************************/
! 826:
! 827: static void
! 828: help(void)
! 829: {
! 830: option_item *op;
! 831:
! 832: printf("Usage: pcregrep [OPTION]... [PATTERN] [FILE1 FILE2 ...]\n");
! 833: printf("Search for PATTERN in each FILE or standard input.\n");
! 834: printf("PATTERN must be present if neither -e nor -f is used.\n");
! 835: printf("\"-\" can be used as a file name to mean STDIN.\n");
! 836:
! 837: #ifdef SUPPORT_LIBZ
! 838: printf("Files whose names end in .gz are read using zlib.\n");
! 839: #endif
! 840:
! 841: #ifdef SUPPORT_LIBBZ2
! 842: printf("Files whose names end in .bz2 are read using bzlib2.\n");
! 843: #endif
! 844:
! 845: #if defined SUPPORT_LIBZ || defined SUPPORT_LIBBZ2
! 846: printf("Other files and the standard input are read as plain files.\n\n");
! 847: #else
! 848: printf("All files are read as plain files, without any interpretation.\n\n");
! 849: #endif
! 850:
! 851: printf("Example: pcregrep -i 'hello.*world' menu.h main.c\n\n");
! 852: printf("Options:\n");
! 853:
! 854: for (op = optionlist; op->one_char != 0; op++)
! 855: {
! 856: int n;
! 857: char s[4];
! 858:
! 859: /* Two options were accidentally implemented and documented with underscores
! 860: instead of hyphens in their names, something that was not noticed for quite a
! 861: few releases. When fixing this, I left the underscored versions in the list
! 862: in case people were using them. However, we don't want to display them in the
! 863: help data. There are no other options that contain underscores, and we do not
! 864: expect ever to implement such options. Therefore, just omit any option that
! 865: contains an underscore. */
! 866:
! 867: if (strchr(op->long_name, '_') != NULL) continue;
! 868:
! 869: if (op->one_char > 0 && (op->long_name)[0] == 0)
! 870: n = 31 - printf(" -%c", op->one_char);
! 871: else
! 872: {
! 873: if (op->one_char > 0) sprintf(s, "-%c,", op->one_char);
! 874: else strcpy(s, " ");
! 875: n = 31 - printf(" %s --%s", s, op->long_name);
! 876: }
! 877:
! 878: if (n < 1) n = 1;
! 879: printf("%.*s%s\n", n, " ", op->help_text);
! 880: }
! 881:
! 882: printf("\nNumbers may be followed by K or M, e.g. --buffer-size=100K.\n");
! 883: printf("The default value for --buffer-size is %d.\n", PCREGREP_BUFSIZE);
! 884: printf("When reading patterns or file names from a file, trailing white\n");
! 885: printf("space is removed and blank lines are ignored.\n");
! 886: printf("The maximum size of any pattern is %d bytes.\n", MAXPATLEN);
! 887:
! 888: printf("\nWith no FILEs, read standard input. If fewer than two FILEs given, assume -h.\n");
! 889: printf("Exit status is 0 if any matches, 1 if no matches, and 2 if trouble.\n");
! 890: }
! 891:
! 892:
! 893:
! 894: /*************************************************
! 895: * Test exclude/includes *
! 896: *************************************************/
! 897:
! 898: /* If any exclude pattern matches, the path is excluded. Otherwise, unless
! 899: there are no includes, the path must match an include pattern.
! 900:
! 901: Arguments:
! 902: path the path to be matched
! 903: ip the chain of include patterns
! 904: ep the chain of exclude patterns
! 905:
! 906: Returns: TRUE if the path is not excluded
! 907: */
! 908:
! 909: static BOOL
! 910: test_incexc(char *path, patstr *ip, patstr *ep)
! 911: {
! 912: int plen = strlen(path);
! 913:
! 914: for (; ep != NULL; ep = ep->next)
! 915: {
! 916: if (pcre_exec(ep->compiled, NULL, path, plen, 0, 0, NULL, 0) >= 0)
! 917: return FALSE;
! 918: }
! 919:
! 920: if (ip == NULL) return TRUE;
! 921:
! 922: for (; ip != NULL; ip = ip->next)
! 923: {
! 924: if (pcre_exec(ip->compiled, NULL, path, plen, 0, 0, NULL, 0) >= 0)
! 925: return TRUE;
! 926: }
! 927:
! 928: return FALSE;
! 929: }
! 930:
! 931:
! 932:
! 933: /*************************************************
! 934: * Decode integer argument value *
! 935: *************************************************/
! 936:
! 937: /* Integer arguments can be followed by K or M. Avoid the use of strtoul()
! 938: because SunOS4 doesn't have it. This is used only for unpicking arguments, so
! 939: just keep it simple.
! 940:
! 941: Arguments:
! 942: option_data the option data string
! 943: op the option item (for error messages)
! 944: longop TRUE if option given in long form
! 945:
! 946: Returns: a long integer
! 947: */
! 948:
! 949: static long int
! 950: decode_number(char *option_data, option_item *op, BOOL longop)
! 951: {
! 952: unsigned long int n = 0;
! 953: char *endptr = option_data;
! 954: while (*endptr != 0 && isspace((unsigned char)(*endptr))) endptr++;
! 955: while (isdigit((unsigned char)(*endptr)))
! 956: n = n * 10 + (int)(*endptr++ - '0');
! 957: if (toupper(*endptr) == 'K')
! 958: {
! 959: n *= 1024;
! 960: endptr++;
! 961: }
! 962: else if (toupper(*endptr) == 'M')
! 963: {
! 964: n *= 1024*1024;
! 965: endptr++;
! 966: }
! 967:
! 968: if (*endptr != 0) /* Error */
! 969: {
! 970: if (longop)
! 971: {
! 972: char *equals = strchr(op->long_name, '=');
! 973: int nlen = (equals == NULL)? (int)strlen(op->long_name) :
! 974: (int)(equals - op->long_name);
! 975: fprintf(stderr, "pcregrep: Malformed number \"%s\" after --%.*s\n",
! 976: option_data, nlen, op->long_name);
! 977: }
! 978: else
! 979: fprintf(stderr, "pcregrep: Malformed number \"%s\" after -%c\n",
! 980: option_data, op->one_char);
! 981: pcregrep_exit(usage(2));
! 982: }
! 983:
! 984: return n;
! 985: }
! 986:
! 987:
! 988:
! 989: /*************************************************
! 990: * Add item to a chain of numbers *
! 991: *************************************************/
! 992:
! 993: /* Used to add an item onto a chain, or just return an unconnected item if the
! 994: "after" argument is NULL.
! 995:
! 996: Arguments:
! 997: n the number to add
! 998: after if not NULL points to item to insert after
! 999:
! 1000: Returns: new number block
! 1001: */
! 1002:
! 1003: static omstr *
! 1004: add_number(int n, omstr *after)
! 1005: {
! 1006: omstr *om = (omstr *)malloc(sizeof(omstr));
! 1007:
! 1008: if (om == NULL)
! 1009: {
! 1010: fprintf(stderr, "pcregrep: malloc failed\n");
! 1011: pcregrep_exit(2);
! 1012: }
! 1013: om->next = NULL;
! 1014: om->groupnum = n;
! 1015:
! 1016: if (after != NULL)
! 1017: {
! 1018: om->next = after->next;
! 1019: after->next = om;
! 1020: }
! 1021: return om;
! 1022: }
! 1023:
! 1024:
! 1025:
! 1026: /*************************************************
1.1 misho 1027: * Read one line of input *
1028: *************************************************/
1029:
1030: /* Normally, input is read using fread() into a large buffer, so many lines may
1031: be read at once. However, doing this for tty input means that no output appears
1032: until a lot of input has been typed. Instead, tty input is handled line by
1033: line. We cannot use fgets() for this, because it does not stop at a binary
1034: zero, and therefore there is no way of telling how many characters it has read,
1035: because there may be binary zeros embedded in the data.
1036:
1037: Arguments:
1038: buffer the buffer to read into
1039: length the maximum number of characters to read
1040: f the file
1041:
1042: Returns: the number of characters read, zero at end of file
1043: */
1044:
1.1.1.2 misho 1045: static unsigned int
1.1 misho 1046: read_one_line(char *buffer, int length, FILE *f)
1047: {
1048: int c;
1049: int yield = 0;
1050: while ((c = fgetc(f)) != EOF)
1051: {
1052: buffer[yield++] = c;
1053: if (c == '\n' || yield >= length) break;
1054: }
1055: return yield;
1056: }
1057:
1058:
1059:
1060: /*************************************************
1061: * Find end of line *
1062: *************************************************/
1063:
1064: /* The length of the endline sequence that is found is set via lenptr. This may
1065: be zero at the very end of the file if there is no line-ending sequence there.
1066:
1067: Arguments:
1068: p current position in line
1069: endptr end of available data
1070: lenptr where to put the length of the eol sequence
1071:
1072: Returns: pointer after the last byte of the line,
1073: including the newline byte(s)
1074: */
1075:
1076: static char *
1077: end_of_line(char *p, char *endptr, int *lenptr)
1078: {
1079: switch(endlinetype)
1080: {
1081: default: /* Just in case */
1082: case EL_LF:
1083: while (p < endptr && *p != '\n') p++;
1084: if (p < endptr)
1085: {
1086: *lenptr = 1;
1087: return p + 1;
1088: }
1089: *lenptr = 0;
1090: return endptr;
1091:
1092: case EL_CR:
1093: while (p < endptr && *p != '\r') p++;
1094: if (p < endptr)
1095: {
1096: *lenptr = 1;
1097: return p + 1;
1098: }
1099: *lenptr = 0;
1100: return endptr;
1101:
1102: case EL_CRLF:
1103: for (;;)
1104: {
1105: while (p < endptr && *p != '\r') p++;
1106: if (++p >= endptr)
1107: {
1108: *lenptr = 0;
1109: return endptr;
1110: }
1111: if (*p == '\n')
1112: {
1113: *lenptr = 2;
1114: return p + 1;
1115: }
1116: }
1117: break;
1118:
1119: case EL_ANYCRLF:
1120: while (p < endptr)
1121: {
1122: int extra = 0;
1123: register int c = *((unsigned char *)p);
1124:
1125: if (utf8 && c >= 0xc0)
1126: {
1127: int gcii, gcss;
1128: extra = utf8_table4[c & 0x3f]; /* Number of additional bytes */
1129: gcss = 6*extra;
1130: c = (c & utf8_table3[extra]) << gcss;
1131: for (gcii = 1; gcii <= extra; gcii++)
1132: {
1133: gcss -= 6;
1134: c |= (p[gcii] & 0x3f) << gcss;
1135: }
1136: }
1137:
1138: p += 1 + extra;
1139:
1140: switch (c)
1141: {
1.1.1.4 ! misho 1142: case '\n':
1.1 misho 1143: *lenptr = 1;
1144: return p;
1145:
1.1.1.4 ! misho 1146: case '\r':
! 1147: if (p < endptr && *p == '\n')
1.1 misho 1148: {
1149: *lenptr = 2;
1150: p++;
1151: }
1152: else *lenptr = 1;
1153: return p;
1154:
1155: default:
1156: break;
1157: }
1158: } /* End of loop for ANYCRLF case */
1159:
1160: *lenptr = 0; /* Must have hit the end */
1161: return endptr;
1162:
1163: case EL_ANY:
1164: while (p < endptr)
1165: {
1166: int extra = 0;
1167: register int c = *((unsigned char *)p);
1168:
1169: if (utf8 && c >= 0xc0)
1170: {
1171: int gcii, gcss;
1172: extra = utf8_table4[c & 0x3f]; /* Number of additional bytes */
1173: gcss = 6*extra;
1174: c = (c & utf8_table3[extra]) << gcss;
1175: for (gcii = 1; gcii <= extra; gcii++)
1176: {
1177: gcss -= 6;
1178: c |= (p[gcii] & 0x3f) << gcss;
1179: }
1180: }
1181:
1182: p += 1 + extra;
1183:
1184: switch (c)
1185: {
1.1.1.4 ! misho 1186: case '\n': /* LF */
! 1187: case '\v': /* VT */
! 1188: case '\f': /* FF */
1.1 misho 1189: *lenptr = 1;
1190: return p;
1191:
1.1.1.4 ! misho 1192: case '\r': /* CR */
! 1193: if (p < endptr && *p == '\n')
1.1 misho 1194: {
1195: *lenptr = 2;
1196: p++;
1197: }
1198: else *lenptr = 1;
1199: return p;
1200:
1.1.1.4 ! misho 1201: #ifndef EBCDIC
! 1202: case 0x85: /* Unicode NEL */
1.1 misho 1203: *lenptr = utf8? 2 : 1;
1204: return p;
1205:
1.1.1.4 ! misho 1206: case 0x2028: /* Unicode LS */
! 1207: case 0x2029: /* Unicode PS */
1.1 misho 1208: *lenptr = 3;
1209: return p;
1.1.1.4 ! misho 1210: #endif /* Not EBCDIC */
1.1 misho 1211:
1212: default:
1213: break;
1214: }
1215: } /* End of loop for ANY case */
1216:
1217: *lenptr = 0; /* Must have hit the end */
1218: return endptr;
1219: } /* End of overall switch */
1220: }
1221:
1222:
1223:
1224: /*************************************************
1225: * Find start of previous line *
1226: *************************************************/
1227:
1228: /* This is called when looking back for before lines to print.
1229:
1230: Arguments:
1231: p start of the subsequent line
1232: startptr start of available data
1233:
1234: Returns: pointer to the start of the previous line
1235: */
1236:
1237: static char *
1238: previous_line(char *p, char *startptr)
1239: {
1240: switch(endlinetype)
1241: {
1242: default: /* Just in case */
1243: case EL_LF:
1244: p--;
1245: while (p > startptr && p[-1] != '\n') p--;
1246: return p;
1247:
1248: case EL_CR:
1249: p--;
1250: while (p > startptr && p[-1] != '\n') p--;
1251: return p;
1252:
1253: case EL_CRLF:
1254: for (;;)
1255: {
1256: p -= 2;
1257: while (p > startptr && p[-1] != '\n') p--;
1258: if (p <= startptr + 1 || p[-2] == '\r') return p;
1259: }
1260: return p; /* But control should never get here */
1261:
1262: case EL_ANY:
1263: case EL_ANYCRLF:
1264: if (*(--p) == '\n' && p > startptr && p[-1] == '\r') p--;
1265: if (utf8) while ((*p & 0xc0) == 0x80) p--;
1266:
1267: while (p > startptr)
1268: {
1.1.1.4 ! misho 1269: register unsigned int c;
1.1 misho 1270: char *pp = p - 1;
1271:
1272: if (utf8)
1273: {
1274: int extra = 0;
1275: while ((*pp & 0xc0) == 0x80) pp--;
1276: c = *((unsigned char *)pp);
1277: if (c >= 0xc0)
1278: {
1279: int gcii, gcss;
1280: extra = utf8_table4[c & 0x3f]; /* Number of additional bytes */
1281: gcss = 6*extra;
1282: c = (c & utf8_table3[extra]) << gcss;
1283: for (gcii = 1; gcii <= extra; gcii++)
1284: {
1285: gcss -= 6;
1286: c |= (pp[gcii] & 0x3f) << gcss;
1287: }
1288: }
1289: }
1290: else c = *((unsigned char *)pp);
1291:
1292: if (endlinetype == EL_ANYCRLF) switch (c)
1293: {
1.1.1.4 ! misho 1294: case '\n': /* LF */
! 1295: case '\r': /* CR */
1.1 misho 1296: return p;
1297:
1298: default:
1299: break;
1300: }
1301:
1302: else switch (c)
1303: {
1.1.1.4 ! misho 1304: case '\n': /* LF */
! 1305: case '\v': /* VT */
! 1306: case '\f': /* FF */
! 1307: case '\r': /* CR */
! 1308: #ifndef EBCDIE
! 1309: case 0x85: /* Unicode NEL */
! 1310: case 0x2028: /* Unicode LS */
! 1311: case 0x2029: /* Unicode PS */
! 1312: #endif /* Not EBCDIC */
1.1 misho 1313: return p;
1314:
1315: default:
1316: break;
1317: }
1318:
1319: p = pp; /* Back one character */
1320: } /* End of loop for ANY case */
1321:
1322: return startptr; /* Hit start of data */
1323: } /* End of overall switch */
1324: }
1325:
1326:
1327:
1328:
1329:
1330: /*************************************************
1331: * Print the previous "after" lines *
1332: *************************************************/
1333:
1334: /* This is called if we are about to lose said lines because of buffer filling,
1335: and at the end of the file. The data in the line is written using fwrite() so
1336: that a binary zero does not terminate it.
1337:
1338: Arguments:
1339: lastmatchnumber the number of the last matching line, plus one
1340: lastmatchrestart where we restarted after the last match
1341: endptr end of available data
1342: printname filename for printing
1343:
1344: Returns: nothing
1345: */
1346:
1.1.1.4 ! misho 1347: static void
! 1348: do_after_lines(int lastmatchnumber, char *lastmatchrestart, char *endptr,
! 1349: char *printname)
1.1 misho 1350: {
1351: if (after_context > 0 && lastmatchnumber > 0)
1352: {
1353: int count = 0;
1354: while (lastmatchrestart < endptr && count++ < after_context)
1355: {
1356: int ellength;
1357: char *pp = lastmatchrestart;
1358: if (printname != NULL) fprintf(stdout, "%s-", printname);
1359: if (number) fprintf(stdout, "%d-", lastmatchnumber++);
1360: pp = end_of_line(pp, endptr, &ellength);
1361: FWRITE(lastmatchrestart, 1, pp - lastmatchrestart, stdout);
1362: lastmatchrestart = pp;
1363: }
1364: hyphenpending = TRUE;
1365: }
1366: }
1367:
1368:
1369:
1370: /*************************************************
1371: * Apply patterns to subject till one matches *
1372: *************************************************/
1373:
1374: /* This function is called to run through all patterns, looking for a match. It
1375: is used multiple times for the same subject when colouring is enabled, in order
1376: to find all possible matches.
1377:
1378: Arguments:
1379: matchptr the start of the subject
1380: length the length of the subject to match
1.1.1.4 ! misho 1381: options options for pcre_exec
1.1 misho 1382: startoffset where to start matching
1383: offsets the offets vector to fill in
1384: mrc address of where to put the result of pcre_exec()
1385:
1386: Returns: TRUE if there was a match
1387: FALSE if there was no match
1388: invert if there was a non-fatal error
1389: */
1390:
1391: static BOOL
1.1.1.4 ! misho 1392: match_patterns(char *matchptr, size_t length, unsigned int options,
! 1393: int startoffset, int *offsets, int *mrc)
1.1 misho 1394: {
1395: int i;
1396: size_t slen = length;
1.1.1.4 ! misho 1397: patstr *p = patterns;
1.1 misho 1398: const char *msg = "this text:\n\n";
1.1.1.4 ! misho 1399:
1.1 misho 1400: if (slen > 200)
1401: {
1402: slen = 200;
1403: msg = "text that starts:\n\n";
1404: }
1.1.1.4 ! misho 1405: for (i = 1; p != NULL; p = p->next, i++)
1.1 misho 1406: {
1.1.1.4 ! misho 1407: *mrc = pcre_exec(p->compiled, p->hint, matchptr, (int)length,
! 1408: startoffset, options, offsets, OFFSET_SIZE);
1.1 misho 1409: if (*mrc >= 0) return TRUE;
1410: if (*mrc == PCRE_ERROR_NOMATCH) continue;
1411: fprintf(stderr, "pcregrep: pcre_exec() gave error %d while matching ", *mrc);
1.1.1.4 ! misho 1412: if (patterns->next != NULL) fprintf(stderr, "pattern number %d to ", i);
1.1 misho 1413: fprintf(stderr, "%s", msg);
1414: FWRITE(matchptr, 1, slen, stderr); /* In case binary zero included */
1415: fprintf(stderr, "\n\n");
1416: if (*mrc == PCRE_ERROR_MATCHLIMIT || *mrc == PCRE_ERROR_RECURSIONLIMIT ||
1417: *mrc == PCRE_ERROR_JIT_STACKLIMIT)
1418: resource_error = TRUE;
1419: if (error_count++ > 20)
1420: {
1421: fprintf(stderr, "pcregrep: Too many errors - abandoned.\n");
1422: pcregrep_exit(2);
1423: }
1424: return invert; /* No more matching; don't show the line again */
1425: }
1426:
1427: return FALSE; /* No match, no errors */
1428: }
1429:
1430:
1431:
1432: /*************************************************
1433: * Grep an individual file *
1434: *************************************************/
1435:
1436: /* This is called from grep_or_recurse() below. It uses a buffer that is three
1437: times the value of bufthird. The matching point is never allowed to stray into
1438: the top third of the buffer, thus keeping more of the file available for
1439: context printing or for multiline scanning. For large files, the pointer will
1440: be in the middle third most of the time, so the bottom third is available for
1441: "before" context printing.
1442:
1443: Arguments:
1444: handle the fopened FILE stream for a normal file
1445: the gzFile pointer when reading is via libz
1446: the BZFILE pointer when reading is via libbz2
1447: frtype FR_PLAIN, FR_LIBZ, or FR_LIBBZ2
1448: filename the file name or NULL (for errors)
1449: printname the file name if it is to be printed for each match
1450: or NULL if the file name is not to be printed
1451: it cannot be NULL if filenames[_nomatch]_only is set
1452:
1453: Returns: 0 if there was at least one match
1454: 1 otherwise (no matches)
1455: 2 if an overlong line is encountered
1456: 3 if there is a read error on a .bz2 file
1457: */
1458:
1459: static int
1460: pcregrep(void *handle, int frtype, char *filename, char *printname)
1461: {
1462: int rc = 1;
1463: int linenumber = 1;
1464: int lastmatchnumber = 0;
1465: int count = 0;
1466: int filepos = 0;
1467: int offsets[OFFSET_SIZE];
1468: char *lastmatchrestart = NULL;
1469: char *ptr = main_buffer;
1470: char *endptr;
1471: size_t bufflength;
1.1.1.3 misho 1472: BOOL binary = FALSE;
1.1 misho 1473: BOOL endhyphenpending = FALSE;
1474: BOOL input_line_buffered = line_buffered;
1475: FILE *in = NULL; /* Ensure initialized */
1476:
1477: #ifdef SUPPORT_LIBZ
1478: gzFile ingz = NULL;
1479: #endif
1480:
1481: #ifdef SUPPORT_LIBBZ2
1482: BZFILE *inbz2 = NULL;
1483: #endif
1484:
1485:
1486: /* Do the first read into the start of the buffer and set up the pointer to end
1487: of what we have. In the case of libz, a non-zipped .gz file will be read as a
1488: plain file. However, if a .bz2 file isn't actually bzipped, the first read will
1489: fail. */
1490:
1.1.1.4 ! misho 1491: (void)frtype;
! 1492:
1.1 misho 1493: #ifdef SUPPORT_LIBZ
1494: if (frtype == FR_LIBZ)
1495: {
1496: ingz = (gzFile)handle;
1497: bufflength = gzread (ingz, main_buffer, bufsize);
1498: }
1499: else
1500: #endif
1501:
1502: #ifdef SUPPORT_LIBBZ2
1503: if (frtype == FR_LIBBZ2)
1504: {
1505: inbz2 = (BZFILE *)handle;
1506: bufflength = BZ2_bzread(inbz2, main_buffer, bufsize);
1507: if ((int)bufflength < 0) return 2; /* Gotcha: bufflength is size_t; */
1508: } /* without the cast it is unsigned. */
1509: else
1510: #endif
1511:
1512: {
1513: in = (FILE *)handle;
1514: if (is_file_tty(in)) input_line_buffered = TRUE;
1515: bufflength = input_line_buffered?
1516: read_one_line(main_buffer, bufsize, in) :
1517: fread(main_buffer, 1, bufsize, in);
1518: }
1519:
1520: endptr = main_buffer + bufflength;
1521:
1.1.1.3 misho 1522: /* Unless binary-files=text, see if we have a binary file. This uses the same
1523: rule as GNU grep, namely, a search for a binary zero byte near the start of the
1524: file. */
1525:
1526: if (binary_files != BIN_TEXT)
1527: {
1528: binary =
1529: memchr(main_buffer, 0, (bufflength > 1024)? 1024 : bufflength) != NULL;
1530: if (binary && binary_files == BIN_NOMATCH) return 1;
1531: }
1532:
1.1 misho 1533: /* Loop while the current pointer is not at the end of the file. For large
1534: files, endptr will be at the end of the buffer when we are in the middle of the
1535: file, but ptr will never get there, because as soon as it gets over 2/3 of the
1536: way, the buffer is shifted left and re-filled. */
1537:
1538: while (ptr < endptr)
1539: {
1540: int endlinelength;
1541: int mrc = 0;
1542: int startoffset = 0;
1.1.1.4 ! misho 1543: unsigned int options = 0;
1.1 misho 1544: BOOL match;
1545: char *matchptr = ptr;
1546: char *t = ptr;
1547: size_t length, linelength;
1548:
1549: /* At this point, ptr is at the start of a line. We need to find the length
1550: of the subject string to pass to pcre_exec(). In multiline mode, it is the
1551: length remainder of the data in the buffer. Otherwise, it is the length of
1552: the next line, excluding the terminating newline. After matching, we always
1553: advance by the length of the next line. In multiline mode the PCRE_FIRSTLINE
1554: option is used for compiling, so that any match is constrained to be in the
1555: first line. */
1556:
1557: t = end_of_line(t, endptr, &endlinelength);
1558: linelength = t - ptr - endlinelength;
1559: length = multiline? (size_t)(endptr - ptr) : linelength;
1560:
1561: /* Check to see if the line we are looking at extends right to the very end
1562: of the buffer without a line terminator. This means the line is too long to
1563: handle. */
1564:
1565: if (endlinelength == 0 && t == main_buffer + bufsize)
1566: {
1567: fprintf(stderr, "pcregrep: line %d%s%s is too long for the internal buffer\n"
1568: "pcregrep: check the --buffer-size option\n",
1569: linenumber,
1570: (filename == NULL)? "" : " of file ",
1571: (filename == NULL)? "" : filename);
1572: return 2;
1573: }
1574:
1575: /* Extra processing for Jeffrey Friedl's debugging. */
1576:
1577: #ifdef JFRIEDL_DEBUG
1578: if (jfriedl_XT || jfriedl_XR)
1579: {
1.1.1.4 ! misho 1580: # include <sys/time.h>
! 1581: # include <time.h>
1.1 misho 1582: struct timeval start_time, end_time;
1583: struct timezone dummy;
1584: int i;
1585:
1586: if (jfriedl_XT)
1587: {
1588: unsigned long newlen = length * jfriedl_XT + strlen(jfriedl_prefix) + strlen(jfriedl_postfix);
1589: const char *orig = ptr;
1590: ptr = malloc(newlen + 1);
1591: if (!ptr) {
1592: printf("out of memory");
1593: pcregrep_exit(2);
1594: }
1595: endptr = ptr;
1596: strcpy(endptr, jfriedl_prefix); endptr += strlen(jfriedl_prefix);
1597: for (i = 0; i < jfriedl_XT; i++) {
1598: strncpy(endptr, orig, length);
1599: endptr += length;
1600: }
1601: strcpy(endptr, jfriedl_postfix); endptr += strlen(jfriedl_postfix);
1602: length = newlen;
1603: }
1604:
1605: if (gettimeofday(&start_time, &dummy) != 0)
1606: perror("bad gettimeofday");
1607:
1608:
1609: for (i = 0; i < jfriedl_XR; i++)
1.1.1.4 ! misho 1610: match = (pcre_exec(patterns->compiled, patterns->hint, ptr, length, 0,
1.1 misho 1611: PCRE_NOTEMPTY, offsets, OFFSET_SIZE) >= 0);
1612:
1613: if (gettimeofday(&end_time, &dummy) != 0)
1614: perror("bad gettimeofday");
1615:
1616: double delta = ((end_time.tv_sec + (end_time.tv_usec / 1000000.0))
1617: -
1618: (start_time.tv_sec + (start_time.tv_usec / 1000000.0)));
1619:
1620: printf("%s TIMER[%.4f]\n", match ? "MATCH" : "FAIL", delta);
1621: return 0;
1622: }
1623: #endif
1624:
1.1.1.4 ! misho 1625: /* We come back here after a match when show_only_matching is set, in order
! 1626: to find any further matches in the same line. This applies to
! 1627: --only-matching, --file-offsets, and --line-offsets. */
1.1 misho 1628:
1629: ONLY_MATCHING_RESTART:
1630:
1631: /* Run through all the patterns until one matches or there is an error other
1632: than NOMATCH. This code is in a subroutine so that it can be re-used for
1.1.1.4 ! misho 1633: finding subsequent matches when colouring matched lines. After finding one
! 1634: match, set PCRE_NOTEMPTY to disable any further matches of null strings in
! 1635: this line. */
1.1 misho 1636:
1.1.1.4 ! misho 1637: match = match_patterns(matchptr, length, options, startoffset, offsets, &mrc);
! 1638: options = PCRE_NOTEMPTY;
1.1 misho 1639:
1640: /* If it's a match or a not-match (as required), do what's wanted. */
1641:
1642: if (match != invert)
1643: {
1644: BOOL hyphenprinted = FALSE;
1645:
1646: /* We've failed if we want a file that doesn't have any matches. */
1647:
1648: if (filenames == FN_NOMATCH_ONLY) return 1;
1649:
1650: /* Just count if just counting is wanted. */
1651:
1652: if (count_only) count++;
1653:
1.1.1.3 misho 1654: /* When handling a binary file and binary-files==binary, the "binary"
1655: variable will be set true (it's false in all other cases). In this
1656: situation we just want to output the file name. No need to scan further. */
1657:
1658: else if (binary)
1659: {
1660: fprintf(stdout, "Binary file %s matches\n", filename);
1661: return 0;
1662: }
1663:
1.1 misho 1664: /* If all we want is a file name, there is no need to scan any more lines
1665: in the file. */
1666:
1667: else if (filenames == FN_MATCH_ONLY)
1668: {
1669: fprintf(stdout, "%s\n", printname);
1670: return 0;
1671: }
1672:
1673: /* Likewise, if all we want is a yes/no answer. */
1674:
1675: else if (quiet) return 0;
1676:
1.1.1.4 ! misho 1677: /* The --only-matching option prints just the substring that matched,
! 1678: and/or one or more captured portions of it, as long as these strings are
! 1679: not empty. The --file-offsets and --line-offsets options output offsets for
! 1680: the matching substring (all three set show_only_matching). None of these
! 1681: mutually exclusive options prints any context. Afterwards, adjust the start
! 1682: and then jump back to look for further matches in the same line. If we are
! 1683: in invert mode, however, nothing is printed and we do not restart - this
! 1684: could still be useful because the return code is set. */
1.1 misho 1685:
1.1.1.4 ! misho 1686: else if (show_only_matching)
1.1 misho 1687: {
1688: if (!invert)
1689: {
1690: if (printname != NULL) fprintf(stdout, "%s:", printname);
1691: if (number) fprintf(stdout, "%d:", linenumber);
1.1.1.4 ! misho 1692:
! 1693: /* Handle --line-offsets */
! 1694:
1.1 misho 1695: if (line_offsets)
1696: fprintf(stdout, "%d,%d\n", (int)(matchptr + offsets[0] - ptr),
1697: offsets[1] - offsets[0]);
1.1.1.4 ! misho 1698:
! 1699: /* Handle --file-offsets */
! 1700:
1.1 misho 1701: else if (file_offsets)
1702: fprintf(stdout, "%d,%d\n",
1703: (int)(filepos + matchptr + offsets[0] - ptr),
1704: offsets[1] - offsets[0]);
1.1.1.4 ! misho 1705:
! 1706: /* Handle --only-matching, which may occur many times */
! 1707:
! 1708: else
1.1 misho 1709: {
1.1.1.4 ! misho 1710: BOOL printed = FALSE;
! 1711: omstr *om;
! 1712:
! 1713: for (om = only_matching; om != NULL; om = om->next)
1.1 misho 1714: {
1.1.1.4 ! misho 1715: int n = om->groupnum;
! 1716: if (n < mrc)
! 1717: {
! 1718: int plen = offsets[2*n + 1] - offsets[2*n];
! 1719: if (plen > 0)
! 1720: {
! 1721: if (printed) fprintf(stdout, "%s", om_separator);
! 1722: if (do_colour) fprintf(stdout, "%c[%sm", 0x1b, colour_string);
! 1723: FWRITE(matchptr + offsets[n*2], 1, plen, stdout);
! 1724: if (do_colour) fprintf(stdout, "%c[00m", 0x1b);
! 1725: printed = TRUE;
! 1726: }
! 1727: }
1.1 misho 1728: }
1.1.1.4 ! misho 1729:
! 1730: if (printed || printname != NULL || number) fprintf(stdout, "\n");
1.1 misho 1731: }
1.1.1.4 ! misho 1732:
! 1733: /* Prepare to repeat to find the next match */
! 1734:
1.1 misho 1735: match = FALSE;
1736: if (line_buffered) fflush(stdout);
1737: rc = 0; /* Had some success */
1738: startoffset = offsets[1]; /* Restart after the match */
1739: goto ONLY_MATCHING_RESTART;
1740: }
1741: }
1742:
1743: /* This is the default case when none of the above options is set. We print
1744: the matching lines(s), possibly preceded and/or followed by other lines of
1745: context. */
1746:
1747: else
1748: {
1749: /* See if there is a requirement to print some "after" lines from a
1750: previous match. We never print any overlaps. */
1751:
1752: if (after_context > 0 && lastmatchnumber > 0)
1753: {
1754: int ellength;
1755: int linecount = 0;
1756: char *p = lastmatchrestart;
1757:
1758: while (p < ptr && linecount < after_context)
1759: {
1760: p = end_of_line(p, ptr, &ellength);
1761: linecount++;
1762: }
1763:
1764: /* It is important to advance lastmatchrestart during this printing so
1765: that it interacts correctly with any "before" printing below. Print
1766: each line's data using fwrite() in case there are binary zeroes. */
1767:
1768: while (lastmatchrestart < p)
1769: {
1770: char *pp = lastmatchrestart;
1771: if (printname != NULL) fprintf(stdout, "%s-", printname);
1772: if (number) fprintf(stdout, "%d-", lastmatchnumber++);
1773: pp = end_of_line(pp, endptr, &ellength);
1774: FWRITE(lastmatchrestart, 1, pp - lastmatchrestart, stdout);
1775: lastmatchrestart = pp;
1776: }
1777: if (lastmatchrestart != ptr) hyphenpending = TRUE;
1778: }
1779:
1780: /* If there were non-contiguous lines printed above, insert hyphens. */
1781:
1782: if (hyphenpending)
1783: {
1784: fprintf(stdout, "--\n");
1785: hyphenpending = FALSE;
1786: hyphenprinted = TRUE;
1787: }
1788:
1789: /* See if there is a requirement to print some "before" lines for this
1790: match. Again, don't print overlaps. */
1791:
1792: if (before_context > 0)
1793: {
1794: int linecount = 0;
1795: char *p = ptr;
1796:
1797: while (p > main_buffer && (lastmatchnumber == 0 || p > lastmatchrestart) &&
1798: linecount < before_context)
1799: {
1800: linecount++;
1801: p = previous_line(p, main_buffer);
1802: }
1803:
1804: if (lastmatchnumber > 0 && p > lastmatchrestart && !hyphenprinted)
1805: fprintf(stdout, "--\n");
1806:
1807: while (p < ptr)
1808: {
1809: int ellength;
1810: char *pp = p;
1811: if (printname != NULL) fprintf(stdout, "%s-", printname);
1812: if (number) fprintf(stdout, "%d-", linenumber - linecount--);
1813: pp = end_of_line(pp, endptr, &ellength);
1814: FWRITE(p, 1, pp - p, stdout);
1815: p = pp;
1816: }
1817: }
1818:
1819: /* Now print the matching line(s); ensure we set hyphenpending at the end
1820: of the file if any context lines are being output. */
1821:
1822: if (after_context > 0 || before_context > 0)
1823: endhyphenpending = TRUE;
1824:
1825: if (printname != NULL) fprintf(stdout, "%s:", printname);
1826: if (number) fprintf(stdout, "%d:", linenumber);
1827:
1828: /* In multiline mode, we want to print to the end of the line in which
1829: the end of the matched string is found, so we adjust linelength and the
1830: line number appropriately, but only when there actually was a match
1831: (invert not set). Because the PCRE_FIRSTLINE option is set, the start of
1832: the match will always be before the first newline sequence. */
1833:
1834: if (multiline & !invert)
1835: {
1836: char *endmatch = ptr + offsets[1];
1837: t = ptr;
1838: while (t < endmatch)
1839: {
1840: t = end_of_line(t, endptr, &endlinelength);
1841: if (t < endmatch) linenumber++; else break;
1842: }
1843: linelength = t - ptr - endlinelength;
1844: }
1845:
1846: /*** NOTE: Use only fwrite() to output the data line, so that binary
1847: zeroes are treated as just another data character. */
1848:
1849: /* This extra option, for Jeffrey Friedl's debugging requirements,
1850: replaces the matched string, or a specific captured string if it exists,
1851: with X. When this happens, colouring is ignored. */
1852:
1853: #ifdef JFRIEDL_DEBUG
1854: if (S_arg >= 0 && S_arg < mrc)
1855: {
1856: int first = S_arg * 2;
1857: int last = first + 1;
1858: FWRITE(ptr, 1, offsets[first], stdout);
1859: fprintf(stdout, "X");
1860: FWRITE(ptr + offsets[last], 1, linelength - offsets[last], stdout);
1861: }
1862: else
1863: #endif
1864:
1865: /* We have to split the line(s) up if colouring, and search for further
1866: matches, but not of course if the line is a non-match. */
1867:
1868: if (do_colour && !invert)
1869: {
1870: int plength;
1871: FWRITE(ptr, 1, offsets[0], stdout);
1872: fprintf(stdout, "%c[%sm", 0x1b, colour_string);
1873: FWRITE(ptr + offsets[0], 1, offsets[1] - offsets[0], stdout);
1874: fprintf(stdout, "%c[00m", 0x1b);
1875: for (;;)
1876: {
1877: startoffset = offsets[1];
1878: if (startoffset >= (int)linelength + endlinelength ||
1.1.1.4 ! misho 1879: !match_patterns(matchptr, length, options, startoffset, offsets,
! 1880: &mrc))
1.1 misho 1881: break;
1882: FWRITE(matchptr + startoffset, 1, offsets[0] - startoffset, stdout);
1883: fprintf(stdout, "%c[%sm", 0x1b, colour_string);
1884: FWRITE(matchptr + offsets[0], 1, offsets[1] - offsets[0], stdout);
1885: fprintf(stdout, "%c[00m", 0x1b);
1886: }
1887:
1888: /* In multiline mode, we may have already printed the complete line
1889: and its line-ending characters (if they matched the pattern), so there
1890: may be no more to print. */
1891:
1892: plength = (int)((linelength + endlinelength) - startoffset);
1893: if (plength > 0) FWRITE(ptr + startoffset, 1, plength, stdout);
1894: }
1895:
1896: /* Not colouring; no need to search for further matches */
1897:
1898: else FWRITE(ptr, 1, linelength + endlinelength, stdout);
1899: }
1900:
1901: /* End of doing what has to be done for a match. If --line-buffered was
1902: given, flush the output. */
1903:
1904: if (line_buffered) fflush(stdout);
1905: rc = 0; /* Had some success */
1906:
1907: /* Remember where the last match happened for after_context. We remember
1908: where we are about to restart, and that line's number. */
1909:
1910: lastmatchrestart = ptr + linelength + endlinelength;
1911: lastmatchnumber = linenumber + 1;
1912: }
1913:
1914: /* For a match in multiline inverted mode (which of course did not cause
1915: anything to be printed), we have to move on to the end of the match before
1916: proceeding. */
1917:
1918: if (multiline && invert && match)
1919: {
1920: int ellength;
1921: char *endmatch = ptr + offsets[1];
1922: t = ptr;
1923: while (t < endmatch)
1924: {
1925: t = end_of_line(t, endptr, &ellength);
1926: if (t <= endmatch) linenumber++; else break;
1927: }
1928: endmatch = end_of_line(endmatch, endptr, &ellength);
1929: linelength = endmatch - ptr - ellength;
1930: }
1931:
1932: /* Advance to after the newline and increment the line number. The file
1933: offset to the current line is maintained in filepos. */
1934:
1935: ptr += linelength + endlinelength;
1936: filepos += (int)(linelength + endlinelength);
1937: linenumber++;
1938:
1939: /* If input is line buffered, and the buffer is not yet full, read another
1940: line and add it into the buffer. */
1941:
1942: if (input_line_buffered && bufflength < (size_t)bufsize)
1943: {
1944: int add = read_one_line(ptr, bufsize - (int)(ptr - main_buffer), in);
1945: bufflength += add;
1946: endptr += add;
1947: }
1948:
1949: /* If we haven't yet reached the end of the file (the buffer is full), and
1950: the current point is in the top 1/3 of the buffer, slide the buffer down by
1951: 1/3 and refill it. Before we do this, if some unprinted "after" lines are
1952: about to be lost, print them. */
1953:
1954: if (bufflength >= (size_t)bufsize && ptr > main_buffer + 2*bufthird)
1955: {
1956: if (after_context > 0 &&
1957: lastmatchnumber > 0 &&
1958: lastmatchrestart < main_buffer + bufthird)
1959: {
1960: do_after_lines(lastmatchnumber, lastmatchrestart, endptr, printname);
1961: lastmatchnumber = 0;
1962: }
1963:
1964: /* Now do the shuffle */
1965:
1966: memmove(main_buffer, main_buffer + bufthird, 2*bufthird);
1967: ptr -= bufthird;
1968:
1969: #ifdef SUPPORT_LIBZ
1970: if (frtype == FR_LIBZ)
1971: bufflength = 2*bufthird +
1972: gzread (ingz, main_buffer + 2*bufthird, bufthird);
1973: else
1974: #endif
1975:
1976: #ifdef SUPPORT_LIBBZ2
1977: if (frtype == FR_LIBBZ2)
1978: bufflength = 2*bufthird +
1979: BZ2_bzread(inbz2, main_buffer + 2*bufthird, bufthird);
1980: else
1981: #endif
1982:
1983: bufflength = 2*bufthird +
1984: (input_line_buffered?
1985: read_one_line(main_buffer + 2*bufthird, bufthird, in) :
1986: fread(main_buffer + 2*bufthird, 1, bufthird, in));
1987: endptr = main_buffer + bufflength;
1988:
1989: /* Adjust any last match point */
1990:
1991: if (lastmatchnumber > 0) lastmatchrestart -= bufthird;
1992: }
1993: } /* Loop through the whole file */
1994:
1995: /* End of file; print final "after" lines if wanted; do_after_lines sets
1996: hyphenpending if it prints something. */
1997:
1.1.1.4 ! misho 1998: if (!show_only_matching && !count_only)
1.1 misho 1999: {
2000: do_after_lines(lastmatchnumber, lastmatchrestart, endptr, printname);
2001: hyphenpending |= endhyphenpending;
2002: }
2003:
2004: /* Print the file name if we are looking for those without matches and there
2005: were none. If we found a match, we won't have got this far. */
2006:
2007: if (filenames == FN_NOMATCH_ONLY)
2008: {
2009: fprintf(stdout, "%s\n", printname);
2010: return 0;
2011: }
2012:
2013: /* Print the match count if wanted */
2014:
2015: if (count_only)
2016: {
2017: if (count > 0 || !omit_zero_count)
2018: {
2019: if (printname != NULL && filenames != FN_NONE)
2020: fprintf(stdout, "%s:", printname);
2021: fprintf(stdout, "%d\n", count);
2022: }
2023: }
2024:
2025: return rc;
2026: }
2027:
2028:
2029:
2030: /*************************************************
2031: * Grep a file or recurse into a directory *
2032: *************************************************/
2033:
2034: /* Given a path name, if it's a directory, scan all the files if we are
2035: recursing; if it's a file, grep it.
2036:
2037: Arguments:
2038: pathname the path to investigate
2039: dir_recurse TRUE if recursing is wanted (-r or -drecurse)
2040: only_one_at_top TRUE if the path is the only one at toplevel
2041:
1.1.1.4 ! misho 2042: Returns: -1 the file/directory was skipped
! 2043: 0 if there was at least one match
1.1 misho 2044: 1 if there were no matches
2045: 2 there was some kind of error
2046:
2047: However, file opening failures are suppressed if "silent" is set.
2048: */
2049:
2050: static int
2051: grep_or_recurse(char *pathname, BOOL dir_recurse, BOOL only_one_at_top)
2052: {
2053: int rc = 1;
2054: int frtype;
2055: void *handle;
1.1.1.4 ! misho 2056: char *lastcomp;
1.1 misho 2057: FILE *in = NULL; /* Ensure initialized */
2058:
2059: #ifdef SUPPORT_LIBZ
2060: gzFile ingz = NULL;
2061: #endif
2062:
2063: #ifdef SUPPORT_LIBBZ2
2064: BZFILE *inbz2 = NULL;
2065: #endif
2066:
1.1.1.3 misho 2067: #if defined SUPPORT_LIBZ || defined SUPPORT_LIBBZ2
1.1.1.2 misho 2068: int pathlen;
2069: #endif
2070:
1.1 misho 2071: /* If the file name is "-" we scan stdin */
2072:
2073: if (strcmp(pathname, "-") == 0)
2074: {
2075: return pcregrep(stdin, FR_PLAIN, stdin_name,
2076: (filenames > FN_DEFAULT || (filenames == FN_DEFAULT && !only_one_at_top))?
2077: stdin_name : NULL);
2078: }
2079:
1.1.1.4 ! misho 2080: /* Inclusion and exclusion: --include-dir and --exclude-dir apply only to
! 2081: directories, whereas --include and --exclude apply to everything else. The test
! 2082: is against the final component of the path. */
! 2083:
! 2084: lastcomp = strrchr(pathname, FILESEP);
! 2085: lastcomp = (lastcomp == NULL)? pathname : lastcomp + 1;
! 2086:
! 2087: /* If the file is a directory, skip if not recursing or if explicitly excluded.
! 2088: Otherwise, scan the directory and recurse for each path within it. The scanning
! 2089: code is localized so it can be made system-specific. */
! 2090:
! 2091: if (isdirectory(pathname))
! 2092: {
! 2093: if (dee_action == dee_SKIP ||
! 2094: !test_incexc(lastcomp, include_dir_patterns, exclude_dir_patterns))
! 2095: return -1;
1.1 misho 2096:
2097: if (dee_action == dee_RECURSE)
2098: {
2099: char buffer[1024];
2100: char *nextfile;
2101: directory_type *dir = opendirectory(pathname);
2102:
2103: if (dir == NULL)
2104: {
2105: if (!silent)
2106: fprintf(stderr, "pcregrep: Failed to open directory %s: %s\n", pathname,
2107: strerror(errno));
2108: return 2;
2109: }
2110:
2111: while ((nextfile = readdirectory(dir)) != NULL)
2112: {
1.1.1.4 ! misho 2113: int frc;
! 2114: sprintf(buffer, "%.512s%c%.128s", pathname, FILESEP, nextfile);
1.1 misho 2115: frc = grep_or_recurse(buffer, dir_recurse, FALSE);
2116: if (frc > 1) rc = frc;
2117: else if (frc == 0 && rc == 1) rc = 0;
2118: }
2119:
2120: closedirectory(dir);
2121: return rc;
2122: }
2123: }
2124:
2125: /* If the file is not a directory and not a regular file, skip it if that's
1.1.1.4 ! misho 2126: been requested. Otherwise, check for explicit include/exclude. */
1.1 misho 2127:
1.1.1.4 ! misho 2128: else if ((!isregfile(pathname) && DEE_action == DEE_SKIP) ||
! 2129: !test_incexc(lastcomp, include_patterns, exclude_patterns))
! 2130: return -1;
1.1 misho 2131:
2132: /* Control reaches here if we have a regular file, or if we have a directory
2133: and recursion or skipping was not requested, or if we have anything else and
2134: skipping was not requested. The scan proceeds. If this is the first and only
2135: argument at top level, we don't show the file name, unless we are only showing
2136: the file name, or the filename was forced (-H). */
2137:
1.1.1.3 misho 2138: #if defined SUPPORT_LIBZ || defined SUPPORT_LIBBZ2
1.1 misho 2139: pathlen = (int)(strlen(pathname));
1.1.1.2 misho 2140: #endif
1.1 misho 2141:
2142: /* Open using zlib if it is supported and the file name ends with .gz. */
2143:
2144: #ifdef SUPPORT_LIBZ
2145: if (pathlen > 3 && strcmp(pathname + pathlen - 3, ".gz") == 0)
2146: {
2147: ingz = gzopen(pathname, "rb");
2148: if (ingz == NULL)
2149: {
2150: if (!silent)
2151: fprintf(stderr, "pcregrep: Failed to open %s: %s\n", pathname,
2152: strerror(errno));
2153: return 2;
2154: }
2155: handle = (void *)ingz;
2156: frtype = FR_LIBZ;
2157: }
2158: else
2159: #endif
2160:
2161: /* Otherwise open with bz2lib if it is supported and the name ends with .bz2. */
2162:
2163: #ifdef SUPPORT_LIBBZ2
2164: if (pathlen > 4 && strcmp(pathname + pathlen - 4, ".bz2") == 0)
2165: {
2166: inbz2 = BZ2_bzopen(pathname, "rb");
2167: handle = (void *)inbz2;
2168: frtype = FR_LIBBZ2;
2169: }
2170: else
2171: #endif
2172:
2173: /* Otherwise use plain fopen(). The label is so that we can come back here if
2174: an attempt to read a .bz2 file indicates that it really is a plain file. */
2175:
2176: #ifdef SUPPORT_LIBBZ2
2177: PLAIN_FILE:
2178: #endif
2179: {
2180: in = fopen(pathname, "rb");
2181: handle = (void *)in;
2182: frtype = FR_PLAIN;
2183: }
2184:
2185: /* All the opening methods return errno when they fail. */
2186:
2187: if (handle == NULL)
2188: {
2189: if (!silent)
2190: fprintf(stderr, "pcregrep: Failed to open %s: %s\n", pathname,
2191: strerror(errno));
2192: return 2;
2193: }
2194:
2195: /* Now grep the file */
2196:
2197: rc = pcregrep(handle, frtype, pathname, (filenames > FN_DEFAULT ||
2198: (filenames == FN_DEFAULT && !only_one_at_top))? pathname : NULL);
2199:
2200: /* Close in an appropriate manner. */
2201:
2202: #ifdef SUPPORT_LIBZ
2203: if (frtype == FR_LIBZ)
2204: gzclose(ingz);
2205: else
2206: #endif
2207:
2208: /* If it is a .bz2 file and the result is 3, it means that the first attempt to
2209: read failed. If the error indicates that the file isn't in fact bzipped, try
2210: again as a normal file. */
2211:
2212: #ifdef SUPPORT_LIBBZ2
2213: if (frtype == FR_LIBBZ2)
2214: {
2215: if (rc == 3)
2216: {
2217: int errnum;
2218: const char *err = BZ2_bzerror(inbz2, &errnum);
2219: if (errnum == BZ_DATA_ERROR_MAGIC)
2220: {
2221: BZ2_bzclose(inbz2);
2222: goto PLAIN_FILE;
2223: }
2224: else if (!silent)
2225: fprintf(stderr, "pcregrep: Failed to read %s using bzlib: %s\n",
2226: pathname, err);
2227: rc = 2; /* The normal "something went wrong" code */
2228: }
2229: BZ2_bzclose(inbz2);
2230: }
2231: else
2232: #endif
2233:
2234: /* Normal file close */
2235:
2236: fclose(in);
2237:
2238: /* Pass back the yield from pcregrep(). */
2239:
2240: return rc;
2241: }
2242:
2243:
2244:
2245: /*************************************************
2246: * Handle a single-letter, no data option *
2247: *************************************************/
2248:
2249: static int
2250: handle_option(int letter, int options)
2251: {
2252: switch(letter)
2253: {
2254: case N_FOFFSETS: file_offsets = TRUE; break;
2255: case N_HELP: help(); pcregrep_exit(0);
2256: case N_LBUFFER: line_buffered = TRUE; break;
2257: case N_LOFFSETS: line_offsets = number = TRUE; break;
2258: case N_NOJIT: study_options &= ~PCRE_STUDY_JIT_COMPILE; break;
1.1.1.3 misho 2259: case 'a': binary_files = BIN_TEXT; break;
1.1 misho 2260: case 'c': count_only = TRUE; break;
2261: case 'F': process_options |= PO_FIXED_STRINGS; break;
2262: case 'H': filenames = FN_FORCE; break;
1.1.1.3 misho 2263: case 'I': binary_files = BIN_NOMATCH; break;
1.1 misho 2264: case 'h': filenames = FN_NONE; break;
2265: case 'i': options |= PCRE_CASELESS; break;
2266: case 'l': omit_zero_count = TRUE; filenames = FN_MATCH_ONLY; break;
2267: case 'L': filenames = FN_NOMATCH_ONLY; break;
2268: case 'M': multiline = TRUE; options |= PCRE_MULTILINE|PCRE_FIRSTLINE; break;
2269: case 'n': number = TRUE; break;
1.1.1.4 ! misho 2270:
! 2271: case 'o':
! 2272: only_matching_last = add_number(0, only_matching_last);
! 2273: if (only_matching == NULL) only_matching = only_matching_last;
! 2274: break;
! 2275:
1.1 misho 2276: case 'q': quiet = TRUE; break;
2277: case 'r': dee_action = dee_RECURSE; break;
2278: case 's': silent = TRUE; break;
2279: case 'u': options |= PCRE_UTF8; utf8 = TRUE; break;
2280: case 'v': invert = TRUE; break;
2281: case 'w': process_options |= PO_WORD_MATCH; break;
2282: case 'x': process_options |= PO_LINE_MATCH; break;
2283:
2284: case 'V':
1.1.1.4 ! misho 2285: fprintf(stdout, "pcregrep version %s\n", pcre_version());
1.1 misho 2286: pcregrep_exit(0);
2287: break;
2288:
2289: default:
2290: fprintf(stderr, "pcregrep: Unknown option -%c\n", letter);
2291: pcregrep_exit(usage(2));
2292: }
2293:
2294: return options;
2295: }
2296:
2297:
2298:
2299:
2300: /*************************************************
2301: * Construct printed ordinal *
2302: *************************************************/
2303:
2304: /* This turns a number into "1st", "3rd", etc. */
2305:
2306: static char *
2307: ordin(int n)
2308: {
2309: static char buffer[8];
2310: char *p = buffer;
2311: sprintf(p, "%d", n);
2312: while (*p != 0) p++;
2313: switch (n%10)
2314: {
2315: case 1: strcpy(p, "st"); break;
2316: case 2: strcpy(p, "nd"); break;
2317: case 3: strcpy(p, "rd"); break;
2318: default: strcpy(p, "th"); break;
2319: }
2320: return buffer;
2321: }
2322:
2323:
2324:
2325: /*************************************************
2326: * Compile a single pattern *
2327: *************************************************/
2328:
1.1.1.4 ! misho 2329: /* Do nothing if the pattern has already been compiled. This is the case for
! 2330: include/exclude patterns read from a file.
! 2331:
! 2332: When the -F option has been used, each "pattern" may be a list of strings,
! 2333: separated by line breaks. They will be matched literally. We split such a
! 2334: string and compile the first substring, inserting an additional block into the
! 2335: pattern chain.
1.1 misho 2336:
2337: Arguments:
1.1.1.4 ! misho 2338: p points to the pattern block
1.1 misho 2339: options the PCRE options
1.1.1.4 ! misho 2340: popts the processing options
! 2341: fromfile TRUE if the pattern was read from a file
! 2342: fromtext file name or identifying text (e.g. "include")
1.1 misho 2343: count 0 if this is the only command line pattern, or
2344: number of the command line pattern, or
2345: linenumber for a pattern from a file
2346:
2347: Returns: TRUE on success, FALSE after an error
2348: */
2349:
2350: static BOOL
1.1.1.4 ! misho 2351: compile_pattern(patstr *p, int options, int popts, int fromfile,
! 2352: const char *fromtext, int count)
1.1 misho 2353: {
2354: char buffer[PATBUFSIZE];
2355: const char *error;
1.1.1.4 ! misho 2356: char *ps = p->string;
! 2357: int patlen = strlen(ps);
1.1 misho 2358: int errptr;
2359:
1.1.1.4 ! misho 2360: if (p->compiled != NULL) return TRUE;
1.1 misho 2361:
1.1.1.4 ! misho 2362: if ((popts & PO_FIXED_STRINGS) != 0)
1.1 misho 2363: {
1.1.1.4 ! misho 2364: int ellength;
! 2365: char *eop = ps + patlen;
! 2366: char *pe = end_of_line(ps, eop, &ellength);
! 2367:
! 2368: if (ellength != 0)
! 2369: {
! 2370: if (add_pattern(pe, p) == NULL) return FALSE;
! 2371: patlen = (int)(pe - ps - ellength);
! 2372: }
1.1 misho 2373: }
2374:
1.1.1.4 ! misho 2375: sprintf(buffer, "%s%.*s%s", prefix[popts], patlen, ps, suffix[popts]);
! 2376: p->compiled = pcre_compile(buffer, options, &error, &errptr, pcretables);
! 2377: if (p->compiled != NULL) return TRUE;
! 2378:
1.1 misho 2379: /* Handle compile errors */
2380:
1.1.1.4 ! misho 2381: errptr -= (int)strlen(prefix[popts]);
! 2382: if (errptr > patlen) errptr = patlen;
1.1 misho 2383:
1.1.1.4 ! misho 2384: if (fromfile)
1.1 misho 2385: {
1.1.1.4 ! misho 2386: fprintf(stderr, "pcregrep: Error in regex in line %d of %s "
! 2387: "at offset %d: %s\n", count, fromtext, errptr, error);
1.1 misho 2388: }
2389: else
2390: {
1.1.1.4 ! misho 2391: if (count == 0)
! 2392: fprintf(stderr, "pcregrep: Error in %s regex at offset %d: %s\n",
! 2393: fromtext, errptr, error);
! 2394: else
! 2395: fprintf(stderr, "pcregrep: Error in %s %s regex at offset %d: %s\n",
! 2396: ordin(count), fromtext, errptr, error);
1.1 misho 2397: }
2398:
2399: return FALSE;
2400: }
2401:
2402:
2403:
2404: /*************************************************
1.1.1.4 ! misho 2405: * Read and compile a file of patterns *
1.1 misho 2406: *************************************************/
2407:
1.1.1.4 ! misho 2408: /* This is used for --filelist, --include-from, and --exclude-from.
1.1 misho 2409:
2410: Arguments:
1.1.1.4 ! misho 2411: name the name of the file; "-" is stdin
! 2412: patptr pointer to the pattern chain anchor
! 2413: patlastptr pointer to the last pattern pointer
! 2414: popts the process options to pass to pattern_compile()
1.1 misho 2415:
1.1.1.4 ! misho 2416: Returns: TRUE if all went well
1.1 misho 2417: */
2418:
2419: static BOOL
1.1.1.4 ! misho 2420: read_pattern_file(char *name, patstr **patptr, patstr **patlastptr, int popts)
1.1 misho 2421: {
1.1.1.4 ! misho 2422: int linenumber = 0;
! 2423: FILE *f;
! 2424: char *filename;
! 2425: char buffer[PATBUFSIZE];
! 2426:
! 2427: if (strcmp(name, "-") == 0)
1.1 misho 2428: {
1.1.1.4 ! misho 2429: f = stdin;
! 2430: filename = stdin_name;
! 2431: }
! 2432: else
! 2433: {
! 2434: f = fopen(name, "r");
! 2435: if (f == NULL)
! 2436: {
! 2437: fprintf(stderr, "pcregrep: Failed to open %s: %s\n", name, strerror(errno));
! 2438: return FALSE;
! 2439: }
! 2440: filename = name;
! 2441: }
! 2442:
! 2443: while (fgets(buffer, PATBUFSIZE, f) != NULL)
! 2444: {
! 2445: char *s = buffer + (int)strlen(buffer);
! 2446: while (s > buffer && isspace((unsigned char)(s[-1]))) s--;
! 2447: *s = 0;
! 2448: linenumber++;
! 2449: if (buffer[0] == 0) continue; /* Skip blank lines */
! 2450:
! 2451: /* Note: this call to add_pattern() puts a pointer to the local variable
! 2452: "buffer" into the pattern chain. However, that pointer is used only when
! 2453: compiling the pattern, which happens immediately below, so we flatten it
! 2454: afterwards, as a precaution against any later code trying to use it. */
! 2455:
! 2456: *patlastptr = add_pattern(buffer, *patlastptr);
! 2457: if (*patlastptr == NULL) return FALSE;
! 2458: if (*patptr == NULL) *patptr = *patlastptr;
! 2459:
! 2460: /* This loop is needed because compiling a "pattern" when -F is set may add
! 2461: on additional literal patterns if the original contains a newline. In the
! 2462: common case, it never will, because fgets() stops at a newline. However,
! 2463: the -N option can be used to give pcregrep a different newline setting. */
! 2464:
1.1 misho 2465: for(;;)
2466: {
1.1.1.4 ! misho 2467: if (!compile_pattern(*patlastptr, pcre_options, popts, TRUE, filename,
! 2468: linenumber))
1.1 misho 2469: return FALSE;
1.1.1.4 ! misho 2470: (*patlastptr)->string = NULL; /* Insurance */
! 2471: if ((*patlastptr)->next == NULL) break;
! 2472: *patlastptr = (*patlastptr)->next;
1.1 misho 2473: }
2474: }
1.1.1.4 ! misho 2475:
! 2476: if (f != stdin) fclose(f);
! 2477: return TRUE;
1.1 misho 2478: }
2479:
2480:
2481:
2482: /*************************************************
2483: * Main program *
2484: *************************************************/
2485:
2486: /* Returns 0 if something matched, 1 if nothing matched, 2 after an error. */
2487:
2488: int
2489: main(int argc, char **argv)
2490: {
2491: int i, j;
2492: int rc = 1;
2493: BOOL only_one_at_top;
1.1.1.4 ! misho 2494: patstr *cp;
! 2495: fnstr *fn;
1.1 misho 2496: const char *locale_from = "--locale";
2497: const char *error;
2498:
2499: #ifdef SUPPORT_PCREGREP_JIT
2500: pcre_jit_stack *jit_stack = NULL;
2501: #endif
2502:
2503: /* Set the default line ending value from the default in the PCRE library;
2504: "lf", "cr", "crlf", and "any" are supported. Anything else is treated as "lf".
2505: Note that the return values from pcre_config(), though derived from the ASCII
2506: codes, are the same in EBCDIC environments, so we must use the actual values
2507: rather than escapes such as as '\r'. */
2508:
2509: (void)pcre_config(PCRE_CONFIG_NEWLINE, &i);
2510: switch(i)
2511: {
2512: default: newline = (char *)"lf"; break;
2513: case 13: newline = (char *)"cr"; break;
2514: case (13 << 8) | 10: newline = (char *)"crlf"; break;
2515: case -1: newline = (char *)"any"; break;
2516: case -2: newline = (char *)"anycrlf"; break;
2517: }
2518:
2519: /* Process the options */
2520:
2521: for (i = 1; i < argc; i++)
2522: {
2523: option_item *op = NULL;
2524: char *option_data = (char *)""; /* default to keep compiler happy */
2525: BOOL longop;
2526: BOOL longopwasequals = FALSE;
2527:
2528: if (argv[i][0] != '-') break;
2529:
2530: /* If we hit an argument that is just "-", it may be a reference to STDIN,
2531: but only if we have previously had -e or -f to define the patterns. */
2532:
2533: if (argv[i][1] == 0)
2534: {
1.1.1.4 ! misho 2535: if (pattern_files != NULL || patterns != NULL) break;
1.1 misho 2536: else pcregrep_exit(usage(2));
2537: }
2538:
2539: /* Handle a long name option, or -- to terminate the options */
2540:
2541: if (argv[i][1] == '-')
2542: {
2543: char *arg = argv[i] + 2;
2544: char *argequals = strchr(arg, '=');
2545:
2546: if (*arg == 0) /* -- terminates options */
2547: {
2548: i++;
2549: break; /* out of the options-handling loop */
2550: }
2551:
2552: longop = TRUE;
2553:
2554: /* Some long options have data that follows after =, for example file=name.
2555: Some options have variations in the long name spelling: specifically, we
2556: allow "regexp" because GNU grep allows it, though I personally go along
2557: with Jeffrey Friedl and Larry Wall in preferring "regex" without the "p".
2558: These options are entered in the table as "regex(p)". Options can be in
2559: both these categories. */
2560:
2561: for (op = optionlist; op->one_char != 0; op++)
2562: {
2563: char *opbra = strchr(op->long_name, '(');
2564: char *equals = strchr(op->long_name, '=');
2565:
2566: /* Handle options with only one spelling of the name */
2567:
2568: if (opbra == NULL) /* Does not contain '(' */
2569: {
2570: if (equals == NULL) /* Not thing=data case */
2571: {
2572: if (strcmp(arg, op->long_name) == 0) break;
2573: }
2574: else /* Special case xxx=data */
2575: {
2576: int oplen = (int)(equals - op->long_name);
2577: int arglen = (argequals == NULL)?
2578: (int)strlen(arg) : (int)(argequals - arg);
2579: if (oplen == arglen && strncmp(arg, op->long_name, oplen) == 0)
2580: {
2581: option_data = arg + arglen;
2582: if (*option_data == '=')
2583: {
2584: option_data++;
2585: longopwasequals = TRUE;
2586: }
2587: break;
2588: }
2589: }
2590: }
2591:
2592: /* Handle options with an alternate spelling of the name */
2593:
2594: else
2595: {
2596: char buff1[24];
2597: char buff2[24];
2598:
2599: int baselen = (int)(opbra - op->long_name);
2600: int fulllen = (int)(strchr(op->long_name, ')') - op->long_name + 1);
2601: int arglen = (argequals == NULL || equals == NULL)?
2602: (int)strlen(arg) : (int)(argequals - arg);
2603:
2604: sprintf(buff1, "%.*s", baselen, op->long_name);
2605: sprintf(buff2, "%s%.*s", buff1, fulllen - baselen - 2, opbra + 1);
2606:
2607: if (strncmp(arg, buff1, arglen) == 0 ||
2608: strncmp(arg, buff2, arglen) == 0)
2609: {
2610: if (equals != NULL && argequals != NULL)
2611: {
2612: option_data = argequals;
2613: if (*option_data == '=')
2614: {
2615: option_data++;
2616: longopwasequals = TRUE;
2617: }
2618: }
2619: break;
2620: }
2621: }
2622: }
2623:
2624: if (op->one_char == 0)
2625: {
2626: fprintf(stderr, "pcregrep: Unknown option %s\n", argv[i]);
2627: pcregrep_exit(usage(2));
2628: }
2629: }
2630:
2631: /* Jeffrey Friedl's debugging harness uses these additional options which
2632: are not in the right form for putting in the option table because they use
2633: only one hyphen, yet are more than one character long. By putting them
2634: separately here, they will not get displayed as part of the help() output,
2635: but I don't think Jeffrey will care about that. */
2636:
2637: #ifdef JFRIEDL_DEBUG
2638: else if (strcmp(argv[i], "-pre") == 0) {
2639: jfriedl_prefix = argv[++i];
2640: continue;
2641: } else if (strcmp(argv[i], "-post") == 0) {
2642: jfriedl_postfix = argv[++i];
2643: continue;
2644: } else if (strcmp(argv[i], "-XT") == 0) {
2645: sscanf(argv[++i], "%d", &jfriedl_XT);
2646: continue;
2647: } else if (strcmp(argv[i], "-XR") == 0) {
2648: sscanf(argv[++i], "%d", &jfriedl_XR);
2649: continue;
2650: }
2651: #endif
2652:
2653:
2654: /* One-char options; many that have no data may be in a single argument; we
2655: continue till we hit the last one or one that needs data. */
2656:
2657: else
2658: {
2659: char *s = argv[i] + 1;
2660: longop = FALSE;
1.1.1.4 ! misho 2661:
1.1 misho 2662: while (*s != 0)
2663: {
2664: for (op = optionlist; op->one_char != 0; op++)
2665: {
2666: if (*s == op->one_char) break;
2667: }
2668: if (op->one_char == 0)
2669: {
2670: fprintf(stderr, "pcregrep: Unknown option letter '%c' in \"%s\"\n",
2671: *s, argv[i]);
2672: pcregrep_exit(usage(2));
2673: }
2674:
1.1.1.4 ! misho 2675: option_data = s+1;
! 2676:
! 2677: /* Break out if this is the last character in the string; it's handled
! 2678: below like a single multi-char option. */
! 2679:
! 2680: if (*option_data == 0) break;
! 2681:
! 2682: /* Check for a single-character option that has data: OP_OP_NUMBER(S)
! 2683: are used for ones that either have a numerical number or defaults, i.e.
! 2684: the data is optional. If a digit follows, there is data; if not, carry on
1.1 misho 2685: with other single-character options in the same string. */
2686:
1.1.1.4 ! misho 2687: if (op->type == OP_OP_NUMBER || op->type == OP_OP_NUMBERS)
1.1 misho 2688: {
2689: if (isdigit((unsigned char)s[1])) break;
2690: }
1.1.1.4 ! misho 2691: else /* Check for an option with data */
1.1 misho 2692: {
1.1.1.4 ! misho 2693: if (op->type != OP_NODATA) break;
1.1 misho 2694: }
2695:
2696: /* Handle a single-character option with no data, then loop for the
2697: next character in the string. */
2698:
2699: pcre_options = handle_option(*s++, pcre_options);
2700: }
2701: }
2702:
2703: /* At this point we should have op pointing to a matched option. If the type
2704: is NO_DATA, it means that there is no data, and the option might set
2705: something in the PCRE options. */
2706:
2707: if (op->type == OP_NODATA)
2708: {
2709: pcre_options = handle_option(op->one_char, pcre_options);
2710: continue;
2711: }
2712:
1.1.1.4 ! misho 2713: /* If the option type is OP_OP_STRING or OP_OP_NUMBER(S), it's an option that
1.1 misho 2714: either has a value or defaults to something. It cannot have data in a
2715: separate item. At the moment, the only such options are "colo(u)r",
2716: "only-matching", and Jeffrey Friedl's special -S debugging option. */
2717:
2718: if (*option_data == 0 &&
1.1.1.4 ! misho 2719: (op->type == OP_OP_STRING || op->type == OP_OP_NUMBER ||
! 2720: op->type == OP_OP_NUMBERS))
1.1 misho 2721: {
2722: switch (op->one_char)
2723: {
2724: case N_COLOUR:
2725: colour_option = (char *)"auto";
2726: break;
2727:
2728: case 'o':
1.1.1.4 ! misho 2729: only_matching_last = add_number(0, only_matching_last);
! 2730: if (only_matching == NULL) only_matching = only_matching_last;
1.1 misho 2731: break;
2732:
2733: #ifdef JFRIEDL_DEBUG
2734: case 'S':
2735: S_arg = 0;
2736: break;
2737: #endif
2738: }
2739: continue;
2740: }
2741:
2742: /* Otherwise, find the data string for the option. */
2743:
2744: if (*option_data == 0)
2745: {
2746: if (i >= argc - 1 || longopwasequals)
2747: {
2748: fprintf(stderr, "pcregrep: Data missing after %s\n", argv[i]);
2749: pcregrep_exit(usage(2));
2750: }
2751: option_data = argv[++i];
2752: }
2753:
1.1.1.4 ! misho 2754: /* If the option type is OP_OP_NUMBERS, the value is a number that is to be
! 2755: added to a chain of numbers. */
! 2756:
! 2757: if (op->type == OP_OP_NUMBERS)
! 2758: {
! 2759: unsigned long int n = decode_number(option_data, op, longop);
! 2760: omdatastr *omd = (omdatastr *)op->dataptr;
! 2761: *(omd->lastptr) = add_number((int)n, *(omd->lastptr));
! 2762: if (*(omd->anchor) == NULL) *(omd->anchor) = *(omd->lastptr);
! 2763: }
! 2764:
! 2765: /* If the option type is OP_PATLIST, it's the -e option, or one of the
! 2766: include/exclude options, which can be called multiple times to create lists
! 2767: of patterns. */
1.1 misho 2768:
1.1.1.4 ! misho 2769: else if (op->type == OP_PATLIST)
1.1 misho 2770: {
1.1.1.4 ! misho 2771: patdatastr *pd = (patdatastr *)op->dataptr;
! 2772: *(pd->lastptr) = add_pattern(option_data, *(pd->lastptr));
! 2773: if (*(pd->lastptr) == NULL) goto EXIT2;
! 2774: if (*(pd->anchor) == NULL) *(pd->anchor) = *(pd->lastptr);
! 2775: }
! 2776:
! 2777: /* If the option type is OP_FILELIST, it's one of the options that names a
! 2778: file. */
! 2779:
! 2780: else if (op->type == OP_FILELIST)
! 2781: {
! 2782: fndatastr *fd = (fndatastr *)op->dataptr;
! 2783: fn = (fnstr *)malloc(sizeof(fnstr));
! 2784: if (fn == NULL)
1.1 misho 2785: {
1.1.1.4 ! misho 2786: fprintf(stderr, "pcregrep: malloc failed\n");
! 2787: goto EXIT2;
1.1 misho 2788: }
1.1.1.4 ! misho 2789: fn->next = NULL;
! 2790: fn->name = option_data;
! 2791: if (*(fd->anchor) == NULL)
! 2792: *(fd->anchor) = fn;
! 2793: else
! 2794: (*(fd->lastptr))->next = fn;
! 2795: *(fd->lastptr) = fn;
1.1 misho 2796: }
2797:
1.1.1.3 misho 2798: /* Handle OP_BINARY_FILES */
2799:
2800: else if (op->type == OP_BINFILES)
2801: {
2802: if (strcmp(option_data, "binary") == 0)
2803: binary_files = BIN_BINARY;
2804: else if (strcmp(option_data, "without-match") == 0)
2805: binary_files = BIN_NOMATCH;
2806: else if (strcmp(option_data, "text") == 0)
2807: binary_files = BIN_TEXT;
2808: else
2809: {
2810: fprintf(stderr, "pcregrep: unknown value \"%s\" for binary-files\n",
2811: option_data);
2812: pcregrep_exit(usage(2));
2813: }
2814: }
2815:
1.1.1.4 ! misho 2816: /* Otherwise, deal with a single string or numeric data value. */
1.1 misho 2817:
2818: else if (op->type != OP_NUMBER && op->type != OP_LONGNUMBER &&
2819: op->type != OP_OP_NUMBER)
2820: {
2821: *((char **)op->dataptr) = option_data;
2822: }
2823: else
2824: {
1.1.1.4 ! misho 2825: unsigned long int n = decode_number(option_data, op, longop);
! 2826: if (op->type == OP_LONGNUMBER) *((unsigned long int *)op->dataptr) = n;
! 2827: else *((int *)op->dataptr) = n;
1.1 misho 2828: }
2829: }
2830:
2831: /* Options have been decoded. If -C was used, its value is used as a default
2832: for -A and -B. */
2833:
2834: if (both_context > 0)
2835: {
2836: if (after_context == 0) after_context = both_context;
2837: if (before_context == 0) before_context = both_context;
2838: }
2839:
2840: /* Only one of --only-matching, --file-offsets, or --line-offsets is permitted.
1.1.1.4 ! misho 2841: However, all three set show_only_matching because they display, each in their
! 2842: own way, only the data that has matched. */
1.1 misho 2843:
1.1.1.4 ! misho 2844: if ((only_matching != NULL && (file_offsets || line_offsets)) ||
1.1 misho 2845: (file_offsets && line_offsets))
2846: {
2847: fprintf(stderr, "pcregrep: Cannot mix --only-matching, --file-offsets "
2848: "and/or --line-offsets\n");
2849: pcregrep_exit(usage(2));
2850: }
2851:
1.1.1.4 ! misho 2852: if (only_matching != NULL || file_offsets || line_offsets)
! 2853: show_only_matching = TRUE;
1.1 misho 2854:
2855: /* If a locale has not been provided as an option, see if the LC_CTYPE or
2856: LC_ALL environment variable is set, and if so, use it. */
2857:
2858: if (locale == NULL)
2859: {
2860: locale = getenv("LC_ALL");
2861: locale_from = "LCC_ALL";
2862: }
2863:
2864: if (locale == NULL)
2865: {
2866: locale = getenv("LC_CTYPE");
2867: locale_from = "LC_CTYPE";
2868: }
2869:
2870: /* If a locale has been provided, set it, and generate the tables the PCRE
2871: needs. Otherwise, pcretables==NULL, which causes the use of default tables. */
2872:
2873: if (locale != NULL)
2874: {
2875: if (setlocale(LC_CTYPE, locale) == NULL)
2876: {
2877: fprintf(stderr, "pcregrep: Failed to set locale %s (obtained from %s)\n",
2878: locale, locale_from);
2879: return 2;
2880: }
2881: pcretables = pcre_maketables();
2882: }
2883:
2884: /* Sort out colouring */
2885:
2886: if (colour_option != NULL && strcmp(colour_option, "never") != 0)
2887: {
2888: if (strcmp(colour_option, "always") == 0) do_colour = TRUE;
2889: else if (strcmp(colour_option, "auto") == 0) do_colour = is_stdout_tty();
2890: else
2891: {
2892: fprintf(stderr, "pcregrep: Unknown colour setting \"%s\"\n",
2893: colour_option);
2894: return 2;
2895: }
2896: if (do_colour)
2897: {
2898: char *cs = getenv("PCREGREP_COLOUR");
2899: if (cs == NULL) cs = getenv("PCREGREP_COLOR");
2900: if (cs != NULL) colour_string = cs;
2901: }
2902: }
2903:
2904: /* Interpret the newline type; the default settings are Unix-like. */
2905:
2906: if (strcmp(newline, "cr") == 0 || strcmp(newline, "CR") == 0)
2907: {
2908: pcre_options |= PCRE_NEWLINE_CR;
2909: endlinetype = EL_CR;
2910: }
2911: else if (strcmp(newline, "lf") == 0 || strcmp(newline, "LF") == 0)
2912: {
2913: pcre_options |= PCRE_NEWLINE_LF;
2914: endlinetype = EL_LF;
2915: }
2916: else if (strcmp(newline, "crlf") == 0 || strcmp(newline, "CRLF") == 0)
2917: {
2918: pcre_options |= PCRE_NEWLINE_CRLF;
2919: endlinetype = EL_CRLF;
2920: }
2921: else if (strcmp(newline, "any") == 0 || strcmp(newline, "ANY") == 0)
2922: {
2923: pcre_options |= PCRE_NEWLINE_ANY;
2924: endlinetype = EL_ANY;
2925: }
2926: else if (strcmp(newline, "anycrlf") == 0 || strcmp(newline, "ANYCRLF") == 0)
2927: {
2928: pcre_options |= PCRE_NEWLINE_ANYCRLF;
2929: endlinetype = EL_ANYCRLF;
2930: }
2931: else
2932: {
2933: fprintf(stderr, "pcregrep: Invalid newline specifier \"%s\"\n", newline);
2934: return 2;
2935: }
2936:
2937: /* Interpret the text values for -d and -D */
2938:
2939: if (dee_option != NULL)
2940: {
2941: if (strcmp(dee_option, "read") == 0) dee_action = dee_READ;
2942: else if (strcmp(dee_option, "recurse") == 0) dee_action = dee_RECURSE;
2943: else if (strcmp(dee_option, "skip") == 0) dee_action = dee_SKIP;
2944: else
2945: {
2946: fprintf(stderr, "pcregrep: Invalid value \"%s\" for -d\n", dee_option);
2947: return 2;
2948: }
2949: }
2950:
2951: if (DEE_option != NULL)
2952: {
2953: if (strcmp(DEE_option, "read") == 0) DEE_action = DEE_READ;
2954: else if (strcmp(DEE_option, "skip") == 0) DEE_action = DEE_SKIP;
2955: else
2956: {
2957: fprintf(stderr, "pcregrep: Invalid value \"%s\" for -D\n", DEE_option);
2958: return 2;
2959: }
2960: }
2961:
2962: /* Check the values for Jeffrey Friedl's debugging options. */
2963:
2964: #ifdef JFRIEDL_DEBUG
2965: if (S_arg > 9)
2966: {
2967: fprintf(stderr, "pcregrep: bad value for -S option\n");
2968: return 2;
2969: }
2970: if (jfriedl_XT != 0 || jfriedl_XR != 0)
2971: {
2972: if (jfriedl_XT == 0) jfriedl_XT = 1;
2973: if (jfriedl_XR == 0) jfriedl_XR = 1;
2974: }
2975: #endif
2976:
1.1.1.4 ! misho 2977: /* Get memory for the main buffer. */
1.1 misho 2978:
2979: bufsize = 3*bufthird;
2980: main_buffer = (char *)malloc(bufsize);
2981:
1.1.1.4 ! misho 2982: if (main_buffer == NULL)
1.1 misho 2983: {
2984: fprintf(stderr, "pcregrep: malloc failed\n");
2985: goto EXIT2;
2986: }
2987:
1.1.1.4 ! misho 2988: /* If no patterns were provided by -e, and there are no files provided by -f,
1.1 misho 2989: the first argument is the one and only pattern, and it must exist. */
2990:
1.1.1.4 ! misho 2991: if (patterns == NULL && pattern_files == NULL)
1.1 misho 2992: {
2993: if (i >= argc) return usage(2);
1.1.1.4 ! misho 2994: patterns = patterns_last = add_pattern(argv[i++], NULL);
! 2995: if (patterns == NULL) goto EXIT2;
1.1 misho 2996: }
2997:
2998: /* Compile the patterns that were provided on the command line, either by
1.1.1.4 ! misho 2999: multiple uses of -e or as a single unkeyed pattern. We cannot do this until
! 3000: after all the command-line options are read so that we know which PCRE options
! 3001: to use. When -F is used, compile_pattern() may add another block into the
! 3002: chain, so we must not access the next pointer till after the compile. */
1.1 misho 3003:
1.1.1.4 ! misho 3004: for (j = 1, cp = patterns; cp != NULL; j++, cp = cp->next)
1.1 misho 3005: {
1.1.1.4 ! misho 3006: if (!compile_pattern(cp, pcre_options, process_options, FALSE, "command-line",
! 3007: (j == 1 && patterns->next == NULL)? 0 : j))
1.1 misho 3008: goto EXIT2;
3009: }
3010:
1.1.1.4 ! misho 3011: /* Read and compile the regular expressions that are provided in files. */
1.1 misho 3012:
1.1.1.4 ! misho 3013: for (fn = pattern_files; fn != NULL; fn = fn->next)
1.1 misho 3014: {
1.1.1.4 ! misho 3015: if (!read_pattern_file(fn->name, &patterns, &patterns_last, process_options))
! 3016: goto EXIT2;
! 3017: }
1.1 misho 3018:
1.1.1.4 ! misho 3019: /* Study the regular expressions, as we will be running them many times. If an
! 3020: extra block is needed for a limit, set PCRE_STUDY_EXTRA_NEEDED so that one is
! 3021: returned, even if studying produces no data. */
1.1 misho 3022:
1.1.1.4 ! misho 3023: if (match_limit > 0 || match_limit_recursion > 0)
! 3024: study_options |= PCRE_STUDY_EXTRA_NEEDED;
1.1 misho 3025:
1.1.1.4 ! misho 3026: /* Unless JIT has been explicitly disabled, arrange a stack for it to use. */
1.1 misho 3027:
3028: #ifdef SUPPORT_PCREGREP_JIT
3029: if ((study_options & PCRE_STUDY_JIT_COMPILE) != 0)
3030: jit_stack = pcre_jit_stack_alloc(32*1024, 1024*1024);
3031: #endif
3032:
1.1.1.4 ! misho 3033: for (j = 1, cp = patterns; cp != NULL; j++, cp = cp->next)
1.1 misho 3034: {
1.1.1.4 ! misho 3035: cp->hint = pcre_study(cp->compiled, study_options, &error);
1.1 misho 3036: if (error != NULL)
3037: {
3038: char s[16];
1.1.1.4 ! misho 3039: if (patterns->next == NULL) s[0] = 0; else sprintf(s, " number %d", j);
1.1 misho 3040: fprintf(stderr, "pcregrep: Error while studying regex%s: %s\n", s, error);
3041: goto EXIT2;
3042: }
3043: #ifdef SUPPORT_PCREGREP_JIT
1.1.1.4 ! misho 3044: if (jit_stack != NULL && cp->hint != NULL)
! 3045: pcre_assign_jit_stack(cp->hint, NULL, jit_stack);
1.1 misho 3046: #endif
3047: }
3048:
3049: /* If --match-limit or --recursion-limit was set, put the value(s) into the
1.1.1.4 ! misho 3050: pcre_extra block for each pattern. There will always be an extra block because
! 3051: of the use of PCRE_STUDY_EXTRA_NEEDED above. */
1.1 misho 3052:
1.1.1.4 ! misho 3053: for (cp = patterns; cp != NULL; cp = cp->next)
1.1 misho 3054: {
1.1.1.4 ! misho 3055: if (match_limit > 0)
1.1 misho 3056: {
1.1.1.4 ! misho 3057: cp->hint->flags |= PCRE_EXTRA_MATCH_LIMIT;
! 3058: cp->hint->match_limit = match_limit;
! 3059: }
! 3060:
! 3061: if (match_limit_recursion > 0)
! 3062: {
! 3063: cp->hint->flags |= PCRE_EXTRA_MATCH_LIMIT_RECURSION;
! 3064: cp->hint->match_limit_recursion = match_limit_recursion;
1.1 misho 3065: }
3066: }
3067:
1.1.1.4 ! misho 3068: /* If there are include or exclude patterns read from the command line, compile
! 3069: them. -F, -w, and -x do not apply, so the third argument of compile_pattern is
! 3070: 0. */
1.1 misho 3071:
1.1.1.4 ! misho 3072: for (j = 0; j < 4; j++)
1.1 misho 3073: {
1.1.1.4 ! misho 3074: int k;
! 3075: for (k = 1, cp = *(incexlist[j]); cp != NULL; k++, cp = cp->next)
1.1 misho 3076: {
1.1.1.4 ! misho 3077: if (!compile_pattern(cp, pcre_options, 0, FALSE, incexname[j],
! 3078: (k == 1 && cp->next == NULL)? 0 : k))
! 3079: goto EXIT2;
1.1 misho 3080: }
3081: }
3082:
1.1.1.4 ! misho 3083: /* Read and compile include/exclude patterns from files. */
! 3084:
! 3085: for (fn = include_from; fn != NULL; fn = fn->next)
1.1 misho 3086: {
1.1.1.4 ! misho 3087: if (!read_pattern_file(fn->name, &include_patterns, &include_patterns_last, 0))
1.1 misho 3088: goto EXIT2;
3089: }
3090:
1.1.1.4 ! misho 3091: for (fn = exclude_from; fn != NULL; fn = fn->next)
1.1 misho 3092: {
1.1.1.4 ! misho 3093: if (!read_pattern_file(fn->name, &exclude_patterns, &exclude_patterns_last, 0))
1.1 misho 3094: goto EXIT2;
3095: }
3096:
1.1.1.4 ! misho 3097: /* If there are no files that contain lists of files to search, and there are
! 3098: no file arguments, search stdin, and then exit. */
! 3099:
! 3100: if (file_lists == NULL && i >= argc)
1.1 misho 3101: {
1.1.1.4 ! misho 3102: rc = pcregrep(stdin, FR_PLAIN, stdin_name,
! 3103: (filenames > FN_DEFAULT)? stdin_name : NULL);
! 3104: goto EXIT;
1.1 misho 3105: }
3106:
1.1.1.4 ! misho 3107: /* If any files that contains a list of files to search have been specified,
! 3108: read them line by line and search the given files. */
1.1.1.3 misho 3109:
1.1.1.4 ! misho 3110: for (fn = file_lists; fn != NULL; fn = fn->next)
1.1.1.3 misho 3111: {
3112: char buffer[PATBUFSIZE];
3113: FILE *fl;
1.1.1.4 ! misho 3114: if (strcmp(fn->name, "-") == 0) fl = stdin; else
1.1.1.3 misho 3115: {
1.1.1.4 ! misho 3116: fl = fopen(fn->name, "rb");
1.1.1.3 misho 3117: if (fl == NULL)
3118: {
1.1.1.4 ! misho 3119: fprintf(stderr, "pcregrep: Failed to open %s: %s\n", fn->name,
1.1.1.3 misho 3120: strerror(errno));
3121: goto EXIT2;
3122: }
3123: }
3124: while (fgets(buffer, PATBUFSIZE, fl) != NULL)
3125: {
3126: int frc;
3127: char *end = buffer + (int)strlen(buffer);
3128: while (end > buffer && isspace(end[-1])) end--;
3129: *end = 0;
3130: if (*buffer != 0)
3131: {
3132: frc = grep_or_recurse(buffer, dee_action == dee_RECURSE, FALSE);
3133: if (frc > 1) rc = frc;
3134: else if (frc == 0 && rc == 1) rc = 0;
3135: }
3136: }
1.1.1.4 ! misho 3137: if (fl != stdin) fclose(fl);
1.1.1.3 misho 3138: }
3139:
1.1.1.4 ! misho 3140: /* After handling file-list, work through remaining arguments. Pass in the fact
! 3141: that there is only one argument at top level - this suppresses the file name if
! 3142: the argument is not a directory and filenames are not otherwise forced. */
1.1 misho 3143:
1.1.1.4 ! misho 3144: only_one_at_top = i == argc - 1 && file_lists == NULL;
1.1 misho 3145:
3146: for (; i < argc; i++)
3147: {
3148: int frc = grep_or_recurse(argv[i], dee_action == dee_RECURSE,
3149: only_one_at_top);
3150: if (frc > 1) rc = frc;
3151: else if (frc == 0 && rc == 1) rc = 0;
3152: }
3153:
3154: EXIT:
3155: #ifdef SUPPORT_PCREGREP_JIT
3156: if (jit_stack != NULL) pcre_jit_stack_free(jit_stack);
3157: #endif
1.1.1.4 ! misho 3158:
1.1 misho 3159: if (main_buffer != NULL) free(main_buffer);
1.1.1.4 ! misho 3160:
! 3161: free_pattern_chain(patterns);
! 3162: free_pattern_chain(include_patterns);
! 3163: free_pattern_chain(include_dir_patterns);
! 3164: free_pattern_chain(exclude_patterns);
! 3165: free_pattern_chain(exclude_dir_patterns);
! 3166:
! 3167: free_file_chain(exclude_from);
! 3168: free_file_chain(include_from);
! 3169: free_file_chain(pattern_files);
! 3170: free_file_chain(file_lists);
! 3171:
! 3172: while (only_matching != NULL)
! 3173: {
! 3174: omstr *this = only_matching;
! 3175: only_matching = this->next;
! 3176: free(this);
1.1 misho 3177: }
1.1.1.4 ! misho 3178:
1.1 misho 3179: pcregrep_exit(rc);
3180:
3181: EXIT2:
3182: rc = 2;
3183: goto EXIT;
3184: }
3185:
3186: /* End of pcregrep */
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>