*: add most of the required setup_common_bufsiz() calls
[oweals/busybox.git] / findutils / grep.c
index ff6742a690d619e3c7ed345a09164dcdf9aa798c..b072cd44126f7f4abb16dcc0fa3edde752d47226 100644 (file)
@@ -18,9 +18,9 @@
  * (C) 2006 Jac Goudsmit added -o option
  */
 
-//applet:IF_GREP(APPLET(grep, _BB_DIR_BIN, _BB_SUID_DROP))
-//applet:IF_FEATURE_GREP_EGREP_ALIAS(APPLET_ODDNAME(egrep, grep, _BB_DIR_BIN, _BB_SUID_DROP, egrep))
-//applet:IF_FEATURE_GREP_FGREP_ALIAS(APPLET_ODDNAME(fgrep, grep, _BB_DIR_BIN, _BB_SUID_DROP, fgrep))
+//applet:IF_GREP(APPLET(grep, BB_DIR_BIN, BB_SUID_DROP))
+//applet:IF_FEATURE_GREP_EGREP_ALIAS(APPLET_ODDNAME(egrep, grep, BB_DIR_BIN, BB_SUID_DROP, egrep))
+//applet:IF_FEATURE_GREP_FGREP_ALIAS(APPLET_ODDNAME(fgrep, grep, BB_DIR_BIN, BB_SUID_DROP, fgrep))
 
 //kbuild:lib-$(CONFIG_GREP) += grep.o
 
@@ -58,6 +58,7 @@
 //config:        Print the specified number of context lines (-C).
 
 #include "libbb.h"
+#include "common_bufsiz.h"
 #include "xregex.h"
 
 
@@ -72,7 +73,6 @@
 //usage:       "PATTERN/-e PATTERN.../-f FILE [FILE]..."
 //usage:#define grep_full_usage "\n\n"
 //usage:       "Search for PATTERN in FILEs (or stdin)\n"
-//usage:     "\nOptions:"
 //usage:     "\n       -H      Add 'filename:' prefix"
 //usage:     "\n       -h      Do not add 'filename:' prefix"
 //usage:     "\n       -n      Add 'line_no:' prefix"
@@ -86,6 +86,7 @@
 //usage:     "\n       -r      Recurse"
 //usage:     "\n       -i      Ignore case"
 //usage:     "\n       -w      Match whole words only"
+//usage:     "\n       -x      Match whole lines only"
 //usage:     "\n       -F      PATTERN is a literal (not regexp)"
 //usage:       IF_FEATURE_GREP_EGREP_ALIAS(
 //usage:     "\n       -E      PATTERN is an extended regexp"
 //usage:#define fgrep_full_usage ""
 
 #define OPTSTR_GREP \
-       "lnqvscFiHhe:f:Lorm:w" \
+       "lnqvscFiHhe:f:Lorm:wx" \
        IF_FEATURE_GREP_CONTEXT("A:B:C:") \
        IF_FEATURE_GREP_EGREP_ALIAS("E") \
        IF_EXTRA_COMPAT("z") \
@@ -139,6 +140,7 @@ enum {
        OPTBIT_r, /* recurse dirs */
        OPTBIT_m, /* -m MAX_MATCHES */
        OPTBIT_w, /* -w whole word match */
+       OPTBIT_x, /* -x whole line match */
        IF_FEATURE_GREP_CONTEXT(    OPTBIT_A ,) /* -A NUM: after-match context */
        IF_FEATURE_GREP_CONTEXT(    OPTBIT_B ,) /* -B NUM: before-match context */
        IF_FEATURE_GREP_CONTEXT(    OPTBIT_C ,) /* -C NUM: -A and -B combined */
@@ -161,6 +163,7 @@ enum {
        OPT_r = 1 << OPTBIT_r,
        OPT_m = 1 << OPTBIT_m,
        OPT_w = 1 << OPTBIT_w,
+       OPT_x = 1 << OPTBIT_x,
        OPT_A = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_A)) + 0,
        OPT_B = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_B)) + 0,
        OPT_C = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_C)) + 0,
@@ -199,11 +202,10 @@ struct globals {
        llist_t *pattern_head;   /* growable list of patterns to match */
        const char *cur_file;    /* the current file we are reading */
 } FIX_ALIASING;
-#define G (*(struct globals*)&bb_common_bufsiz1)
+#define G (*(struct globals*)bb_common_bufsiz1)
 #define INIT_G() do { \
-       struct G_sizecheck { \
-               char G_sizecheck[sizeof(G) > COMMON_BUFSIZE ? -1 : 1]; \
-       }; \
+       setup_common_bufsiz(); \
+       BUILD_BUG_ON(sizeof(G) > COMMON_BUFSIZE); \
 } while (0)
 #define max_matches       (G.max_matches         )
 #if !ENABLE_EXTRA_COMPAT
@@ -342,11 +344,42 @@ static int grep_file(FILE *file)
                while (pattern_ptr) {
                        gl = (grep_list_data_t *)pattern_ptr->data;
                        if (FGREP_FLAG) {
-                               found |= (((option_mask32 & OPT_i)
-                                       ? strcasestr(line, gl->pattern)
-                                       : strstr(line, gl->pattern)
-                                       ) != NULL);
+                               char *match;
+                               char *str = line;
+ opt_f_again:
+                               match = ((option_mask32 & OPT_i)
+                                       ? strcasestr(str, gl->pattern)
+                                       : strstr(str, gl->pattern)
+                                       );
+                               if (match) {
+                                       if (option_mask32 & OPT_x) {
+                                               if (match != str)
+                                                       goto opt_f_not_found;
+                                               if (str[strlen(gl->pattern)] != '\0')
+                                                       goto opt_f_not_found;
+                                       } else
+                                       if (option_mask32 & OPT_w) {
+                                               char c = (match != str) ? match[-1] : ' ';
+                                               if (!isalnum(c) && c != '_') {
+                                                       c = match[strlen(gl->pattern)];
+                                                       if (!c || (!isalnum(c) && c != '_'))
+                                                               goto opt_f_found;
+                                               }
+                                               str = match + 1;
+                                               goto opt_f_again;
+                                       }
+ opt_f_found:
+                                       found = 1;
+ opt_f_not_found: ;
+                               }
                        } else {
+#if ENABLE_EXTRA_COMPAT
+                               unsigned start_pos;
+#else
+                               int match_flg;
+#endif
+                               char *match_at;
+
                                if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
                                        gl->flg_mem_alocated_compiled |= COMPILED;
 #if !ENABLE_EXTRA_COMPAT
@@ -361,26 +394,58 @@ static int grep_file(FILE *file)
 #if !ENABLE_EXTRA_COMPAT
                                gl->matched_range.rm_so = 0;
                                gl->matched_range.rm_eo = 0;
+                               match_flg = 0;
+#else
+                               start_pos = 0;
 #endif
+                               match_at = line;
+ opt_w_again:
+//bb_error_msg("'%s' start_pos:%d line_len:%d", match_at, start_pos, line_len);
                                if (
 #if !ENABLE_EXTRA_COMPAT
-                                       regexec(&gl->compiled_regex, line, 1, &gl->matched_range, 0) == 0
+                                       regexec(&gl->compiled_regex, match_at, 1, &gl->matched_range, match_flg) == 0
 #else
-                                       re_search(&gl->compiled_regex, line, line_len,
-                                                       /*start:*/ 0, /*range:*/ line_len,
+                                       re_search(&gl->compiled_regex, match_at, line_len,
+                                                       start_pos, /*range:*/ line_len,
                                                        &gl->matched_range) >= 0
 #endif
                                ) {
-                                       if (!(option_mask32 & OPT_w))
+                                       if (option_mask32 & OPT_x) {
+                                               found = (gl->matched_range.rm_so == 0
+                                                        && match_at[gl->matched_range.rm_eo] == '\0');
+                                       } else
+                                       if (!(option_mask32 & OPT_w)) {
                                                found = 1;
-                                       else {
+                                       else {
                                                char c = ' ';
-                                               if (gl->matched_range.rm_so)
-                                                       c = line[gl->matched_range.rm_so - 1];
+                                               if (match_at > line || gl->matched_range.rm_so != 0) {
+                                                       c = match_at[gl->matched_range.rm_so - 1];
+                                               }
                                                if (!isalnum(c) && c != '_') {
-                                                       c = line[gl->matched_range.rm_eo];
-                                                       if (!c || (!isalnum(c) && c != '_'))
-                                                               found = 1;
+                                                       c = match_at[gl->matched_range.rm_eo];
+                                               }
+                                               if (!isalnum(c) && c != '_') {
+                                                       found = 1;
+                                               } else {
+                       /*
+                        * Why check gl->matched_range.rm_eo?
+                        * Zero-length match makes -w skip the line:
+                        * "echo foo | grep ^" prints "foo",
+                        * "echo foo | grep -w ^" prints nothing.
+                        * Without such check, we can loop forever.
+                        */
+#if !ENABLE_EXTRA_COMPAT
+                                                       if (gl->matched_range.rm_eo != 0) {
+                                                               match_at += gl->matched_range.rm_eo;
+                                                               match_flg |= REG_NOTBOL;
+                                                               goto opt_w_again;
+                                                       }
+#else
+                                                       if (gl->matched_range.rm_eo > start_pos) {
+                                                               start_pos = gl->matched_range.rm_eo;
+                                                               goto opt_w_again;
+                                                       }
+#endif
                                                }
                                        }
                                }
@@ -563,20 +628,20 @@ static char *add_grep_list_data(char *pattern)
 
 static void load_regexes_from_file(llist_t *fopt)
 {
-       char *line;
-       FILE *f;
-
        while (fopt) {
+               char *line;
+               FILE *fp;
                llist_t *cur = fopt;
                char *ffile = cur->data;
 
                fopt = cur->link;
                free(cur);
-               f = xfopen_stdin(ffile);
-               while ((line = xmalloc_fgetline(f)) != NULL) {
+               fp = xfopen_stdin(ffile);
+               while ((line = xmalloc_fgetline(fp)) != NULL) {
                        llist_add_to(&pattern_head,
                                new_grep_list_data(line, ALLOCATED));
                }
+               fclose_if_not_stdin(fp);
        }
 }
 
@@ -603,7 +668,7 @@ static int grep_dir(const char *dir)
        int matched = 0;
        recursive_action(dir,
                /* recurse=yes */ ACTION_RECURSE |
-               /* followLinks=no */
+               /* followLinks=command line only */ ACTION_FOLLOWLINKS_L0 |
                /* depthFirst=yes */ ACTION_DEPTHFIRST,
                /* fileAction= */ file_action_grep,
                /* dirAction= */ NULL,
@@ -618,11 +683,15 @@ int grep_main(int argc UNUSED_PARAM, char **argv)
        FILE *file;
        int matched;
        llist_t *fopt = NULL;
-
-       /* do normal option parsing */
 #if ENABLE_FEATURE_GREP_CONTEXT
        int Copt, opts;
+#endif
+
+       /* For grep, exitcode of 1 is "not found". Other errors are 2: */
+       xfunc_error_retval = 2;
 
+       /* do normal option parsing */
+#if ENABLE_FEATURE_GREP_CONTEXT
        /* -H unsets -h; -C unsets -A,-B; -e,-f are lists;
         * -m,-A,-B,-C have numeric param */
        opt_complementary = "H-h:C-AB:e::f::m+:A+:B+:C+";
@@ -633,7 +702,7 @@ int grep_main(int argc UNUSED_PARAM, char **argv)
 
        if (opts & OPT_C) {
                /* -C unsets prev -A and -B, but following -A or -B
-                  may override it */
+                * may override it */
                if (!(opts & OPT_A)) /* not overridden */
                        lines_after = Copt;
                if (!(opts & OPT_B)) /* not overridden */
@@ -660,21 +729,25 @@ int grep_main(int argc UNUSED_PARAM, char **argv)
 #endif
        invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
 
-       if (pattern_head != NULL) {
-               /* convert char **argv to grep_list_data_t */
+       {       /* convert char **argv to grep_list_data_t */
                llist_t *cur;
-
                for (cur = pattern_head; cur; cur = cur->link)
                        cur->data = new_grep_list_data(cur->data, 0);
        }
-       if (option_mask32 & OPT_f)
+       if (option_mask32 & OPT_f) {
                load_regexes_from_file(fopt);
+               if (!pattern_head) { /* -f EMPTY_FILE? */
+                       /* GNU grep treats it as "nothing matches" */
+                       llist_add_to(&pattern_head, new_grep_list_data((char*) "", 0));
+                       invert_search ^= 1;
+               }
+       }
 
        if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
                option_mask32 |= OPT_F;
 
 #if !ENABLE_EXTRA_COMPAT
-       if (!(option_mask32 & (OPT_o | OPT_w)))
+       if (!(option_mask32 & (OPT_o | OPT_w | OPT_x)))
                reflags = REG_NOSUB;
 #endif