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
7USE_GREP(NEWTOY(grep, "ZzEFHabhinorsvwclqe*f*m#x[!wx][!EFw]", TOYFLAG_BIN))
8USE_EGREP(OLDTOY(egrep, grep, TOYFLAG_BIN))
9USE_FGREP(OLDTOY(fgrep, grep, TOYFLAG_BIN))
10
11config GREP
12  bool "grep"
13  default y
14  help
15    usage: grep [-EFivwcloqsHbhn] [-m MAX] [-e REGEX]... [-f REGFILE] [FILE]...
16
17    Show lines matching regular expressions. If no -e, first argument is
18    regular expression to match. With no files (or "-" filename) read stdin.
19    Returns 0 if matched, 1 if no match found.
20
21    -e  Regex to match. (May be repeated.)
22    -f  File containing regular expressions to match.
23
24    match type:
25    -E  extended regex syntax    -F  fixed (match literal string)
26    -i  case insensitive         -m  stop after this many lines matched
27    -r  recursive (on dir)       -v  invert match
28    -w  whole word (implies -E)  -x  whole line
29    -z  input NUL terminated
30
31    display modes: (default: matched line)
32    -c  count of matching lines  -l  show matching filenames
33    -o  only matching part       -q  quiet (errors only)
34    -s  silent (no error msg)    -Z  output NUL terminated
35
36    output prefix (default: filename if checking more than 1 file)
37    -H  force filename           -b  byte offset of match
38    -h  hide filename            -n  line number of match
39
40config EGREP
41  bool
42  default y
43  depends on GREP
44
45config FGREP
46  bool
47  default y
48  depends on GREP
49*/
50
51#define FOR_grep
52#include "toys.h"
53#include <regex.h>
54
55GLOBALS(
56  long m;
57  struct arg_list *f;
58  struct arg_list *e;
59
60  struct arg_list *regex;
61)
62
63static void do_grep(int fd, char *name)
64{
65  FILE *file = fdopen(fd, "r");
66  long offset = 0;
67  int lcount = 0, mcount = 0, which = toys.optflags & FLAG_w ? 2 : 0;
68  char indelim = '\n' * !(toys.optflags&FLAG_z),
69       outdelim = '\n' * !(toys.optflags&FLAG_Z);
70
71  if (!fd) name = "(standard input)";
72
73  if (!file) {
74    perror_msg("%s", name);
75    return;
76  }
77
78  for (;;) {
79    char *line = 0, *start;
80    regmatch_t matches[3];
81    size_t unused;
82    long len;
83    int mmatch = 0;
84
85    lcount++;
86    if (0 > (len = getdelim(&line, &unused, indelim, file))) break;
87    if (line[len-1] == indelim) line[len-1] = 0;
88
89    start = line;
90
91    for (;;)
92    {
93      int rc = 0, skip = 0;
94
95      if (toys.optflags & FLAG_F) {
96        struct arg_list *seek, fseek;
97        char *s = 0;
98
99        for (seek = TT.e; seek; seek = seek->next) {
100          if (toys.optflags & FLAG_x) {
101            int i = (toys.optflags & FLAG_i);
102
103            if ((i ? strcasecmp : strcmp)(seek->arg, line)) s = line;
104          } else if (!*seek->arg) {
105            seek = &fseek;
106            fseek.arg = s = line;
107            break;
108          }
109          if (toys.optflags & FLAG_i) {
110            long ll = strlen(seek->arg);;
111
112            // Alas, posix hasn't got strcasestr()
113            for (s = line; *s; s++) if (!strncasecmp(s, seek->arg, ll)) break;
114            if (!*s) s = 0;
115          } else s = strstr(line, seek->arg);
116          if (s) break;
117        }
118
119        if (s) {
120          matches[which].rm_so = (s-line);
121          skip = matches[which].rm_eo = (s-line)+strlen(seek->arg);
122        } else rc = 1;
123      } else {
124        rc = regexec((regex_t *)toybuf, start, 3, matches,
125                     start==line ? 0 : REG_NOTBOL);
126        skip = matches[which].rm_eo;
127      }
128
129      if (toys.optflags & FLAG_x)
130        if (matches[which].rm_so || line[matches[which].rm_eo]) rc = 1;
131
132      if (toys.optflags & FLAG_v) {
133        if (toys.optflags & FLAG_o) {
134          if (rc) skip = matches[which].rm_eo = strlen(start);
135          else if (!matches[which].rm_so) {
136            start += skip;
137            continue;
138          } else matches[which].rm_eo = matches[which].rm_so;
139        } else {
140          if (!rc) break;
141          matches[which].rm_eo = strlen(start);
142        }
143        matches[which].rm_so = 0;
144      } else if (rc) break;
145
146      mmatch++;
147      toys.exitval = 0;
148      if (toys.optflags & FLAG_q) xexit();
149      if (toys.optflags & FLAG_l) {
150        printf("%s%c", name, outdelim);
151        free(line);
152        fclose(file);
153        return;
154      }
155      if (toys.optflags & FLAG_o)
156        if (matches[which].rm_eo == matches[which].rm_so)
157          break;
158
159      if (!(toys.optflags & FLAG_c)) {
160        if (toys.optflags & FLAG_H) printf("%s:", name);
161        if (toys.optflags & FLAG_n) printf("%d:", lcount);
162        if (toys.optflags & FLAG_b)
163          printf("%ld:", offset + (start-line) +
164              ((toys.optflags & FLAG_o) ? matches[which].rm_so : 0));
165        if (!(toys.optflags & FLAG_o)) xprintf("%s%c", line, outdelim);
166        else {
167          xprintf("%.*s%c", matches[which].rm_eo - matches[which].rm_so,
168                  start + matches[which].rm_so, outdelim);
169        }
170      }
171
172      start += skip;
173      if (!(toys.optflags & FLAG_o) || !*start) break;
174    }
175    offset += len;
176
177    free(line);
178
179    if (mmatch) mcount++;
180    if ((toys.optflags & FLAG_m) && mcount >= TT.m) break;
181  }
182
183  if (toys.optflags & FLAG_c) {
184    if (toys.optflags & FLAG_H) printf("%s:", name);
185    xprintf("%d%c", mcount, outdelim);
186  }
187
188  // loopfiles will also close the fd, but this frees an (opaque) struct.
189  fclose(file);
190}
191
192static void parse_regex(void)
193{
194  struct arg_list *al, *new, *list = NULL;
195  long len = 0;
196  char *s, *ss;
197
198  // Add all -f lines to -e list. (Yes, this is leaking allocation context for
199  // exit to free. Not supporting nofork for this command any time soon.)
200  al = TT.f ? TT.f : TT.e;
201  while (al) {
202    if (TT.f) s = ss = xreadfile(al->arg, 0, 0);
203    else s = ss = al->arg;
204
205    do {
206      ss = strchr(s, '\n');
207      if (ss) *(ss++) = 0;
208      new = xmalloc(sizeof(struct arg_list));
209      new->next = list;
210      new->arg = s;
211      list = new;
212      s = ss;
213    } while (ss && *s);
214    al = al->next;
215    if (!al && TT.f) {
216      TT.f = 0;
217      al = TT.e;
218    }
219  }
220  TT.e = list;
221
222  if (!(toys.optflags & FLAG_F)) {
223    int w = toys.optflags & FLAG_w;
224    char *regstr;
225
226    // Convert strings to one big regex
227    if (w) len = 36;
228    for (al = TT.e; al; al = al->next)
229      len += strlen(al->arg)+1+!(toys.optflags & FLAG_E);
230
231    regstr = s = xmalloc(len);
232    if (w) s = stpcpy(s, "(^|[^_[:alnum:]])(");
233    for (al = TT.e; al; al = al->next) {
234      s = stpcpy(s, al->arg);
235      if (!(toys.optflags & FLAG_E)) *(s++) = '\\';
236      *(s++) = '|';
237    }
238    *(s-=(1+!(toys.optflags & FLAG_E))) = 0;
239    if (w) strcpy(s, ")($|[^_[:alnum:]])");
240
241    w = regcomp((regex_t *)toybuf, regstr,
242                ((toys.optflags & FLAG_E) ? REG_EXTENDED : 0) |
243                ((toys.optflags & FLAG_i) ? REG_ICASE    : 0));
244
245    if (w) {
246      regerror(w, (regex_t *)toybuf, toybuf+sizeof(regex_t),
247               sizeof(toybuf)-sizeof(regex_t));
248      error_exit("bad REGEX: %s", toybuf);
249    }
250  }
251}
252
253static int do_grep_r(struct dirtree *new)
254{
255  char *name;
256
257  if (new->parent && !dirtree_notdotdot(new)) return 0;
258  if (S_ISDIR(new->st.st_mode)) return DIRTREE_RECURSE;
259
260  // "grep -r onefile" doesn't show filenames, but "grep -r onedir" should.
261  if (new->parent && !(toys.optflags & FLAG_h)) toys.optflags |= FLAG_H;
262
263  name = dirtree_path(new, 0);
264  do_grep(openat(dirtree_parentfd(new), new->name, 0), name);
265  free(name);
266
267  return 0;
268}
269
270void grep_main(void)
271{
272  char **ss;
273
274  // Handle egrep and fgrep
275  if (*toys.which->name == 'e' || (toys.optflags & FLAG_w))
276    toys.optflags |= FLAG_E;
277  if (*toys.which->name == 'f') toys.optflags |= FLAG_F;
278
279  if (!TT.e && !TT.f) {
280    if (!*toys.optargs) error_exit("no REGEX");
281    TT.e = xzalloc(sizeof(struct arg_list));
282    TT.e->arg = *(toys.optargs++);
283    toys.optc--;
284  }
285
286  parse_regex();
287
288  if (!(toys.optflags & FLAG_h) && toys.optc>1) toys.optflags |= FLAG_H;
289
290  toys.exitval = 1;
291  if (toys.optflags & FLAG_s) {
292    close(2);
293    xopen("/dev/null", O_RDWR);
294  }
295
296  if (toys.optflags & FLAG_r) {
297    for (ss=toys.optargs; *ss; ss++) {
298      if (!strcmp(*ss, "-")) do_grep(0, *ss);
299      else dirtree_read(*ss, do_grep_r);
300    }
301  } else loopfiles_rw(toys.optargs, O_RDONLY, 0, 1, do_grep);
302}
303