1 /* grep.c - print lines what match given regular expression
3 * Copyright 2013 CE Strake <strake888 at gmail.com>
5 * See http://pubs.opengroup.org/onlinepubs/9699919799/utilities/grep.html
9 USE_GREP(NEWTOY(grep, "C#B#A#ZzEFHabhinorsvwclqe*f*m#x[!wx][!EFw]", TOYFLAG_BIN))
10 USE_EGREP(OLDTOY(egrep, grep, TOYFLAG_BIN))
11 USE_FGREP(OLDTOY(fgrep, grep, TOYFLAG_BIN))
17 usage: grep [-EFivwcloqsHbhn] [-A NUM] [-m MAX] [-e REGEX]... [-f REGFILE] [FILE]...
19 Show lines matching regular expressions. If no -e, first argument is
20 regular expression to match. With no files (or "-" filename) read stdin.
21 Returns 0 if matched, 1 if no match found.
23 -e Regex to match. (May be repeated.)
24 -f File containing regular expressions to match.
27 -A Show NUM lines after -B Show NUM lines before match
28 -C NUM lines context (A+B) -E extended regex syntax
29 -F fixed (literal match) -i case insensitive
30 -m match MAX many lines -r recursive (on dir)
31 -v invert match -w whole word (implies -E)
32 -x whole line -z input NUL terminated
34 display modes: (default: matched line)
35 -c count of matching lines -l show matching filenames
36 -o only matching part -q quiet (errors only)
37 -s silent (no error msg) -Z output NUL terminated
39 output prefix (default: filename if checking more than 1 file)
40 -H force filename -b byte offset of match
41 -h hide filename -n line number of match
66 char indelim, outdelim;
69 // Emit line with various potential prefixes and delimiter
70 static void outline(char *line, char dash, char *name, long lcount, long bcount,
73 if (name && (toys.optflags&FLAG_H)) printf("%s%c", name, dash);
74 if (!line || (lcount && (toys.optflags&FLAG_n)))
75 printf("%ld%c", lcount, line ? dash : TT.outdelim);
76 if (bcount && (toys.optflags&FLAG_b)) printf("%ld%c", bcount-1, dash);
77 if (line) xprintf("%.*s%c", trim, line, TT.outdelim);
80 // Show matches in one file
81 static void do_grep(int fd, char *name)
83 struct double_list *dlb = 0;
84 FILE *file = fdopen(fd, "r");
85 long lcount = 0, mcount = 0, offset = 0, after = 0, before = 0;
88 if (!fd) name = "(standard input)";
91 perror_msg("%s", name);
96 // Loop through lines of input
98 char *line = 0, *start;
106 len = getdelim(&line, &unused, TT.indelim, file);
107 if (errno) perror_msg("%s", name);
109 if (line[len-1] == TT.indelim) line[len-1] = 0;
113 // Loop through matches in this line
115 int rc = 0, skip = 0;
117 // Handle non-regex matches
118 if (toys.optflags & FLAG_F) {
119 struct arg_list *seek, fseek;
122 for (seek = TT.e; seek; seek = seek->next) {
123 if (toys.optflags & FLAG_x) {
124 int i = (toys.optflags & FLAG_i);
126 if ((i ? strcasecmp : strcmp)(seek->arg, line)) s = line;
127 } else if (!*seek->arg) {
129 fseek.arg = s = line;
132 if (toys.optflags & FLAG_i) s = strnstr(line, seek->arg);
133 else s = strstr(line, seek->arg);
138 matches.rm_so = (s-line);
139 skip = matches.rm_eo = (s-line)+strlen(seek->arg);
142 rc = regexec((regex_t *)toybuf, start, 1, &matches,
143 start==line ? 0 : REG_NOTBOL);
144 skip = matches.rm_eo;
147 if (toys.optflags & FLAG_x)
148 if (matches.rm_so || line[matches.rm_eo]) rc = 1;
150 if (!rc && (toys.optflags & FLAG_w)) {
153 if ((start+matches.rm_so)!=line) {
154 c = start[matches.rm_so-1];
155 if (!isalnum(c) && c != '_') c = 0;
158 c = start[matches.rm_eo];
159 if (!isalnum(c) && c != '_') c = 0;
162 start += matches.rm_so+1;
168 if (toys.optflags & FLAG_v) {
169 if (toys.optflags & FLAG_o) {
170 if (rc) skip = matches.rm_eo = strlen(start);
171 else if (!matches.rm_so) {
174 } else matches.rm_eo = matches.rm_so;
177 matches.rm_eo = strlen(start);
180 } else if (rc) break;
182 // At least one line we didn't print since match while -ABC active
189 if (toys.optflags & FLAG_q) xexit();
190 if (toys.optflags & FLAG_l) {
191 xprintf("%s%c", name, TT.outdelim);
196 if (toys.optflags & FLAG_o)
197 if (matches.rm_eo == matches.rm_so)
200 if (!(toys.optflags & FLAG_c)) {
201 long bcount = 1 + offset + (start-line) +
202 ((toys.optflags & FLAG_o) ? matches.rm_so : 0);
204 if (!(toys.optflags & FLAG_o)) {
206 struct double_list *dl = dlist_pop(&dlb);
208 outline(dl->data, '-', name, lcount-before, 0, -1);
214 outline(line, ':', name, lcount, bcount, -1);
215 if (TT.a) after = TT.a+1;
216 } else outline(start+matches.rm_so, ':', name, lcount, bcount,
217 matches.rm_eo-matches.rm_so);
221 if (!(toys.optflags & FLAG_o)) break;
225 if (mmatch) mcount++;
227 int discard = (after || TT.b);
229 if (after && --after) {
230 outline(line, '-', name, lcount, 0, -1);
233 if (discard && TT.b) {
234 dlist_add(&dlb, line);
237 struct double_list *dl;
239 dl = dlist_pop(&dlb);
245 // If we discarded a line while displaying context, show bars before next
246 // line (but don't show them now in case that was last match in file)
247 if (discard && mcount) bars = "--";
251 if ((toys.optflags & FLAG_m) && mcount >= TT.m) break;
254 if (toys.optflags & FLAG_c) outline(0, ':', name, mcount, 0, -1);
256 // loopfiles will also close the fd, but this frees an (opaque) struct.
260 static void parse_regex(void)
262 struct arg_list *al, *new, *list = NULL;
266 // Add all -f lines to -e list. (Yes, this is leaking allocation context for
267 // exit to free. Not supporting nofork for this command any time soon.)
268 al = TT.f ? TT.f : TT.e;
270 if (TT.f) s = ss = xreadfile(al->arg, 0, 0);
271 else s = ss = al->arg;
273 // Split lines at \n, add individual lines to new list.
275 ss = strchr(s, '\n');
277 new = xmalloc(sizeof(struct arg_list));
284 // Advance, when we run out of -f switch to -e.
293 if (!(toys.optflags & FLAG_F)) {
297 // Convert strings to one big regex
298 for (al = TT.e; al; al = al->next)
299 len += strlen(al->arg)+1+!(toys.optflags & FLAG_E);
301 regstr = s = xmalloc(len);
302 for (al = TT.e; al; al = al->next) {
303 s = stpcpy(s, al->arg);
304 if (!(toys.optflags & FLAG_E)) *(s++) = '\\';
307 *(s-=(1+!(toys.optflags & FLAG_E))) = 0;
309 i = regcomp((regex_t *)toybuf, regstr,
310 ((toys.optflags & FLAG_E) ? REG_EXTENDED : 0) |
311 ((toys.optflags & FLAG_i) ? REG_ICASE : 0));
314 regerror(i, (regex_t *)toybuf, toybuf+sizeof(regex_t),
315 sizeof(toybuf)-sizeof(regex_t));
316 error_exit("bad REGEX: %s", toybuf);
321 static int do_grep_r(struct dirtree *new)
325 if (!dirtree_notdotdot(new)) return 0;
326 if (S_ISDIR(new->st.st_mode)) return DIRTREE_RECURSE;
328 // "grep -r onefile" doesn't show filenames, but "grep -r onedir" should.
329 if (new->parent && !(toys.optflags & FLAG_h)) toys.optflags |= FLAG_H;
331 name = dirtree_path(new, 0);
332 do_grep(openat(dirtree_parentfd(new), new->name, 0), name);
340 char **ss = toys.optargs;
342 if (!TT.a) TT.a = TT.c;
343 if (!TT.b) TT.b = TT.c;
345 TT.indelim = '\n' * !(toys.optflags&FLAG_z);
346 TT.outdelim = '\n' * !(toys.optflags&FLAG_Z);
348 // Handle egrep and fgrep
349 if (*toys.which->name == 'e') toys.optflags |= FLAG_E;
350 if (*toys.which->name == 'f') toys.optflags |= FLAG_F;
352 if (!TT.e && !TT.f) {
353 if (!*ss) error_exit("no REGEX");
354 TT.e = xzalloc(sizeof(struct arg_list));
361 if (!(toys.optflags & FLAG_h) && toys.optc>1) toys.optflags |= FLAG_H;
364 if (toys.optflags & FLAG_s) {
366 xopen_stdio("/dev/null", O_RDWR);
369 if (toys.optflags & FLAG_r) {
370 // Iterate through -r arguments. Use "." as default if none provided.
371 for (ss = *ss ? ss : (char *[]){".", 0}; *ss; ss++) {
372 if (!strcmp(*ss, "-")) do_grep(0, *ss);
373 else dirtree_read(*ss, do_grep_r);
375 } else loopfiles_rw(ss, O_RDONLY|WARN_ONLY, 0, do_grep);