skip archive binaries; buffer size problem remains

styxyang · Sep 1, 2013 · ca1e32a · ca1e32a
1 parent 06d5710
commit ca1e32a
Show file tree

Hide file tree

Showing 6 changed files with 16 additions and 8 deletions.
diff --git a/magic.c b/magic.c
@@ -10,6 +10,7 @@
 #define M_VAX_COFF_EX      0x0178
 #define M_TAR_ZIP_LZW      0x1F9D
 #define M_TAR_ZIP_LZH      0x1FA0
+#define M_ARCH_BIN         0x213C617263683E0A
 #define M_PDF              0x25504446
 #define M_PSD              0x38425053
 #define M_BMP              0x424D
@@ -45,6 +46,7 @@ static uint8_t magic_index[][MENTRY_LEN + 1] = {
     {0x01, 0x78, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 2},
     {0x1F, 0x9D, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 2},
     {0x1F, 0xA0, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 2},
+    {0x21, 0x3C, 0x61, 0x72, 0x63, 0x68, 0x3E, 0x0A, 8},
     {0x25, 0x50, 0x44, 0x46, 0x00, 0x00, 0x00, 0x00, 4},
     {0x38, 0x42, 0x50, 0x53, 0x00, 0x00, 0x00, 0x00, 4},
     {0x42, 0x4D, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 2},

diff --git a/magic_defs b/magic_defs
@@ -34,3 +34,4 @@ CAFEBABE         MACH_O_FAT_CLASS
 0000002066747970 3GPP2_FTYP
 0000001866747970 MP4_FTYP
 0178             VAX_COFF_EX
+213C617263683E0A ARCH_BIN
diff --git a/options.c b/options.c
@@ -30,7 +30,6 @@ void parse_options(int argc, char *argv[]) {
     struct option longopts[] = {
         { "kmp",  no_argument,       &opt.str_matching_algo, 0 },
         { "file", required_argument, NULL,                   0 },
-        { "dir",  required_argument, NULL,                   0 }
     };
 
     /* since the arguments are not so many
@@ -44,8 +43,6 @@ void parse_options(int argc, char *argv[]) {
                 dprintf(INFO, "%d, %d\n", optopt, optind);
                 if (strcmp(longopts[opt_idx].name, "file") == 0) {
                     opt.input_file = optarg;
-                } else if (strcmp(longopts[opt_idx].name, "dir") == 0) {
-                    opt.input_dir = optarg;
                 } else if (opt.str_matching_algo == SM_ALGO_KMP) {
                     printf("using KMP algorithm\n");
                 }
@@ -61,14 +58,18 @@ void parse_options(int argc, char *argv[]) {
     /* argv += optind; */
 
     if (optind >= argc) {
-        dprintf(ERROR, "parse_options: missing pattern string");
         print_usage();
     }
-    opt.search_pattern = argv[optind];
+    opt.search_pattern = argv[optind++];
     opt.search_patlen = strlen(opt.search_pattern);
+
+    if (optind < argc) {
+        dprintf(ERROR, "%d %d", optind, argc);
+        opt.input_dirs = argv + optind;
+    }
     printf("search for \"%s\"-%zd\n", opt.search_pattern, opt.search_patlen);
 
-    if (!opt.input_file && !opt.input_dir) {
+    if (!opt.input_file && !opt.input_dirs) {
         dprintf(ERROR, "parse_options: missing search scope");
         printf("default in current directory\n");
         if (!opt.search_pattern) {

diff --git a/options.h b/options.h
@@ -10,7 +10,7 @@ typedef struct {
     char *search_pattern;
     int   search_patlen;
     char *input_file;
-    char *input_dir;
+    char **input_dirs;
 } cli_options;
 
 extern cli_options opt;

diff --git a/thread.c b/thread.c
@@ -43,6 +43,7 @@ static uint32_t begin_of_line(char *fb, uint32_t mid)
     return mid;
 }
 
+/* write matched data to buffer with line number `linum' and `lastlinum' */
 static void worker_writebuffer(char *fb, uint32_t pos,
                                uint32_t linum, uint32_t lastlinum)
 {
@@ -117,7 +118,8 @@ void *worker_thread(void *arg)
                 break;
             }
 
-            dprintf(INFO, "T%d <%s> match at %d", tid, fi.filename, matchpos + startpos);
+            dprintf(INFO, "T%d <%s> match at %d", tid, fi.filename,
+                    matchpos + startpos);
 
             memset(buf, 0, sizeof(buf));
             worker_writebuffer(fb, matchpos + startpos, linum, lastlinum);
@@ -128,6 +130,7 @@ void *worker_thread(void *arg)
             startpos += matchpos + opt.search_patlen;
         }
 
+        /* if offset is not zero, buffer contains something to write */
         if (off) {
             /* FIXME abstract as `write_filename' maybe */
             pthread_mutex_lock(&outmtx);

diff --git a/trie.c b/trie.c
@@ -64,6 +64,7 @@ struct trie *trie_lookup(struct trie *root, uint8_t *target, int size)
     return trie_walk(root, target, size, false, false);
 }
 
+/* will return the node regardless of whether it is a leaf */
 struct trie *trie_scan(struct trie *root, uint8_t *target, int size)
 {
     assert(root);