LXR toybox/toys/posix/grep.c

   1/* grep.c - print lines what match given regular expression
   2 *
   3 * Copyright 2013 CE Strake <strake888 at gmail.com>
   4 *
   5 * See http://pubs.opengroup.org/onlinepubs/9699919799/utilities/grep.html
   6 *
   7 * TODO: --color, "Binary file %s matches"
   8 *
   9 * Posix doesn't even specify -r, documenting deviations from it is silly.
  10
  11USE_GREP(NEWTOY(grep, "S(exclude)*M(include)*C#B#A#ZzEFHabhinorsvwclqe*f*m#x[!wx][!EFw]", TOYFLAG_BIN))
  12USE_EGREP(OLDTOY(egrep, grep, TOYFLAG_BIN))
  13USE_FGREP(OLDTOY(fgrep, grep, TOYFLAG_BIN))
  14
  15config GREP
  16  bool "grep"
  17  default y
  18  help
  19    usage: grep [-EFrivwcloqsHbhn] [-ABC NUM] [-m MAX] [-e REGEX]... [-MS PATTERN]... [-f REGFILE] [FILE]...
  20
  21    Show lines matching regular expressions. If no -e, first argument is
  22    regular expression to match. With no files (or "-" filename) read stdin.
  23    Returns 0 if matched, 1 if no match found.
  24
  25    -e  Regex to match. (May be repeated.)
  26    -f  File listing regular expressions to match.
  27
  28    file search:
  29    -r  Recurse into subdirectories (defaults FILE to ".")
  30    -M  Match filename pattern (--include)
  31    -S  Skip filename pattern (--exclude)
  32
  33    match type:
  34    -A  Show NUM lines after     -B  Show NUM lines before match
  35    -C  NUM lines context (A+B)  -E  extended regex syntax
  36    -F  fixed (literal match)    -i  case insensitive
  37    -m  match MAX many lines     -v  invert match
  38    -w  whole word (implies -E)  -x  whole line
  39    -z  input NUL terminated
  40
  41    display modes: (default: matched line)
  42    -c  count of matching lines  -l  show matching filenames
  43    -o  only matching part       -q  quiet (errors only)
  44    -s  silent (no error msg)    -Z  output NUL terminated
  45
  46    output prefix (default: filename if checking more than 1 file)
  47    -H  force filename           -b  byte offset of match
  48    -h  hide filename            -n  line number of match
  49
  50config EGREP
  51  bool
  52  default y
  53  depends on GREP
  54
  55config FGREP
  56  bool
  57  default y
  58  depends on GREP
  59*/
  60
  61#define FOR_grep
  62#include "toys.h"
  63#include <regex.h>
  64
  65GLOBALS(
  66  long m;
  67  struct arg_list *f;
  68  struct arg_list *e;
  69  long a;
  70  long b;
  71  long c;
  72  struct arg_list *M;
  73  struct arg_list *S;
  74
  75  char indelim, outdelim;
  76  int found;
  77)
  78
  79// Emit line with various potential prefixes and delimiter
  80static void outline(char *line, char dash, char *name, long lcount, long bcount,
  81  int trim)
  82{
  83  if (name && (toys.optflags&FLAG_H)) printf("%s%c", name, dash);
  84  if (!line || (lcount && (toys.optflags&FLAG_n)))
  85    printf("%ld%c", lcount, line ? dash : TT.outdelim);
  86  if (bcount && (toys.optflags&FLAG_b)) printf("%ld%c", bcount-1, dash);
  87  if (line) xprintf("%.*s%c", trim, line, TT.outdelim);
  88}
  89
  90// Show matches in one file
  91static void do_grep(int fd, char *name)
  92{
  93  struct double_list *dlb = 0;
  94  FILE *file = fdopen(fd, "r");
  95  long lcount = 0, mcount = 0, offset = 0, after = 0, before = 0;
  96  char *bars = 0;
  97
  98  if (!fd) name = "(standard input)";
  99
 100  if (!file) {
 101    perror_msg("%s", name);
 102
 103    return;
 104  }
 105
 106  // Loop through lines of input
 107  for (;;) {
 108    char *line = 0, *start;
 109    regmatch_t matches;
 110    size_t unused;
 111    long len;
 112    int mmatch = 0;
 113
 114    lcount++;
 115    errno = 0;
 116    len = getdelim(&line, &unused, TT.indelim, file);
 117    if (errno) perror_msg("%s", name);
 118    if (len<1) break;
 119    if (line[len-1] == TT.indelim) line[len-1] = 0;
 120
 121    start = line;
 122
 123    // Loop through matches in this line
 124    do {
 125      int rc = 0, skip = 0;
 126
 127      // Handle non-regex matches
 128      if (toys.optflags & FLAG_F) {
 129        struct arg_list *seek, fseek;
 130        char *s = 0;
 131
 132        for (seek = TT.e; seek; seek = seek->next) {
 133          if (toys.optflags & FLAG_x) {
 134            int i = (toys.optflags & FLAG_i);
 135
 136            if ((i ? strcasecmp : strcmp)(seek->arg, line)) s = line;
 137          } else if (!*seek->arg) {
 138            seek = &fseek;
 139            fseek.arg = s = line;
 140            break;
 141          }
 142          if (toys.optflags & FLAG_i) s = strnstr(line, seek->arg);
 143          else s = strstr(line, seek->arg);
 144          if (s) break;
 145        }
 146
 147        if (s) {
 148          matches.rm_so = (s-line);
 149          skip = matches.rm_eo = (s-line)+strlen(seek->arg);
 150        } else rc = 1;
 151      } else {
 152        rc = regexec((regex_t *)toybuf, start, 1, &matches,
 153                     start==line ? 0 : REG_NOTBOL);
 154        skip = matches.rm_eo;
 155      }
 156
 157      if (toys.optflags & FLAG_x)
 158        if (matches.rm_so || line[matches.rm_eo]) rc = 1;
 159
 160      if (!rc && (toys.optflags & FLAG_w)) {
 161        char c = 0;
 162
 163        if ((start+matches.rm_so)!=line) {
 164          c = start[matches.rm_so-1];
 165          if (!isalnum(c) && c != '_') c = 0;
 166        }
 167        if (!c) {
 168          c = start[matches.rm_eo];
 169          if (!isalnum(c) && c != '_') c = 0;
 170        }
 171        if (c) {
 172          start += matches.rm_so+1;
 173
 174          continue;
 175        }
 176      }
 177
 178      if (toys.optflags & FLAG_v) {
 179        if (toys.optflags & FLAG_o) {
 180          if (rc) skip = matches.rm_eo = strlen(start);
 181          else if (!matches.rm_so) {
 182            start += skip;
 183            continue;
 184          } else matches.rm_eo = matches.rm_so;
 185        } else {
 186          if (!rc) break;
 187          matches.rm_eo = strlen(start);
 188        }
 189        matches.rm_so = 0;
 190      } else if (rc) break;
 191
 192      // At least one line we didn't print since match while -ABC active
 193      if (bars) {
 194        xputs(bars);
 195        bars = 0;
 196      }
 197      mmatch++;
 198      TT.found = 1;
 199      if (toys.optflags & FLAG_q) {
 200        toys.exitval = 0;
 201        xexit();
 202      }
 203      if (toys.optflags & FLAG_l) {
 204        xprintf("%s%c", name, TT.outdelim);
 205        free(line);
 206        fclose(file);
 207        return;
 208      }
 209      if (toys.optflags & FLAG_o)
 210        if (matches.rm_eo == matches.rm_so)
 211          break;
 212
 213      if (!(toys.optflags & FLAG_c)) {
 214        long bcount = 1 + offset + (start-line) +
 215          ((toys.optflags & FLAG_o) ? matches.rm_so : 0);
 216 
 217        if (!(toys.optflags & FLAG_o)) {
 218          while (dlb) {
 219            struct double_list *dl = dlist_pop(&dlb);
 220
 221            outline(dl->data, '-', name, lcount-before, 0, -1);
 222            free(dl->data);
 223            free(dl);
 224            before--;
 225          }
 226
 227          outline(line, ':', name, lcount, bcount, -1);
 228          if (TT.a) after = TT.a+1;
 229        } else outline(start+matches.rm_so, ':', name, lcount, bcount,
 230                       matches.rm_eo-matches.rm_so);
 231      }
 232
 233      start += skip;
 234      if (!(toys.optflags & FLAG_o)) break;
 235    } while (*start);
 236    offset += len;
 237
 238    if (mmatch) mcount++;
 239    else {
 240      int discard = (after || TT.b);
 241
 242      if (after && --after) {
 243        outline(line, '-', name, lcount, 0, -1);
 244        discard = 0;
 245      }
 246      if (discard && TT.b) {
 247        dlist_add(&dlb, line);
 248        line = 0;
 249        if (++before>TT.b) {
 250          struct double_list *dl;
 251
 252          dl = dlist_pop(&dlb);
 253          free(dl->data);
 254          free(dl);
 255          before--;
 256        } else discard = 0;
 257      }
 258      // If we discarded a line while displaying context, show bars before next
 259      // line (but don't show them now in case that was last match in file)
 260      if (discard && mcount) bars = "--";
 261    }
 262    free(line);
 263
 264    if ((toys.optflags & FLAG_m) && mcount >= TT.m) break;
 265  }
 266
 267  if (toys.optflags & FLAG_c) outline(0, ':', name, mcount, 0, -1);
 268
 269  // loopfiles will also close the fd, but this frees an (opaque) struct.
 270  fclose(file);
 271}
 272
 273static void parse_regex(void)
 274{
 275  struct arg_list *al, *new, *list = NULL;
 276  long len = 0;
 277  char *s, *ss;
 278
 279  // Add all -f lines to -e list. (Yes, this is leaking allocation context for
 280  // exit to free. Not supporting nofork for this command any time soon.)
 281  al = TT.f ? TT.f : TT.e;
 282  while (al) {
 283    if (TT.f) s = ss = xreadfile(al->arg, 0, 0);
 284    else s = ss = al->arg;
 285
 286    // Split lines at \n, add individual lines to new list.
 287    do {
 288      ss = strchr(s, '\n');
 289      if (ss) *(ss++) = 0;
 290      new = xmalloc(sizeof(struct arg_list));
 291      new->next = list;
 292      new->arg = s;
 293      list = new;
 294      s = ss;
 295    } while (ss && *s);
 296
 297    // Advance, when we run out of -f switch to -e.
 298    al = al->next;
 299    if (!al && TT.f) {
 300      TT.f = 0;
 301      al = TT.e;
 302    }
 303  }
 304  TT.e = list;
 305
 306  if (!(toys.optflags & FLAG_F)) {
 307    char *regstr;
 308    int i;
 309
 310    // Convert strings to one big regex
 311    for (al = TT.e; al; al = al->next)
 312      len += strlen(al->arg)+1+!(toys.optflags & FLAG_E);
 313
 314    regstr = s = xmalloc(len);
 315    for (al = TT.e; al; al = al->next) {
 316      s = stpcpy(s, al->arg);
 317      if (!(toys.optflags & FLAG_E)) *(s++) = '\\';
 318      *(s++) = '|';
 319    }
 320    *(s-=(1+!(toys.optflags & FLAG_E))) = 0;
 321
 322    i = regcomp((regex_t *)toybuf, regstr,
 323                ((toys.optflags & FLAG_E) ? REG_EXTENDED : 0) |
 324                ((toys.optflags & FLAG_i) ? REG_ICASE    : 0));
 325
 326    if (i) {
 327      regerror(i, (regex_t *)toybuf, toybuf+sizeof(regex_t),
 328               sizeof(toybuf)-sizeof(regex_t));
 329      error_exit("bad REGEX: %s", toybuf);
 330    }
 331  }
 332}
 333
 334static int do_grep_r(struct dirtree *new)
 335{
 336  char *name;
 337
 338  if (!dirtree_notdotdot(new)) return 0;
 339  if (S_ISDIR(new->st.st_mode)) return DIRTREE_RECURSE;
 340  if (TT.S || TT.M) {
 341    struct arg_list *al;
 342
 343    for (al = TT.S; al; al = al->next)
 344      if (!fnmatch(al->arg, new->name, 0)) return 0;
 345
 346    if (TT.M) {
 347      for (al = TT.M; al; al = al->next)
 348        if (!fnmatch(al->arg, new->name, 0)) break;
 349
 350      if (!al) return 0;
 351    }
 352  }
 353
 354  // "grep -r onefile" doesn't show filenames, but "grep -r onedir" should.
 355  if (new->parent && !(toys.optflags & FLAG_h)) toys.optflags |= FLAG_H;
 356
 357  name = dirtree_path(new, 0);
 358  do_grep(openat(dirtree_parentfd(new), new->name, 0), name);
 359  free(name);
 360
 361  return 0;
 362}
 363
 364void grep_main(void)
 365{
 366  char **ss = toys.optargs;
 367
 368  // Grep exits with 2 for errors
 369  toys.exitval = 2;
 370
 371  if (!TT.a) TT.a = TT.c;
 372  if (!TT.b) TT.b = TT.c;
 373
 374  TT.indelim = '\n' * !(toys.optflags&FLAG_z);
 375  TT.outdelim = '\n' * !(toys.optflags&FLAG_Z);
 376
 377  // Handle egrep and fgrep
 378  if (*toys.which->name == 'e') toys.optflags |= FLAG_E;
 379  if (*toys.which->name == 'f') toys.optflags |= FLAG_F;
 380
 381  if (!TT.e && !TT.f) {
 382    if (!*ss) error_exit("no REGEX");
 383    TT.e = xzalloc(sizeof(struct arg_list));
 384    TT.e->arg = *(ss++);
 385    toys.optc--;
 386  }
 387
 388  parse_regex();
 389
 390  if (!(toys.optflags & FLAG_h) && toys.optc>1) toys.optflags |= FLAG_H;
 391
 392  if (toys.optflags & FLAG_s) {
 393    close(2);
 394    xopen_stdio("/dev/null", O_RDWR);
 395  }
 396
 397  if (toys.optflags & FLAG_r) {
 398    // Iterate through -r arguments. Use "." as default if none provided.
 399    for (ss = *ss ? ss : (char *[]){".", 0}; *ss; ss++) {
 400      if (!strcmp(*ss, "-")) do_grep(0, *ss);
 401      else dirtree_read(*ss, do_grep_r);
 402    }
 403  } else loopfiles_rw(ss, O_RDONLY|WARN_ONLY, 0, do_grep);
 404  toys.exitval = !TT.found;
 405}
 406