Ticket #587: omindex.patch

omindex.cc

@@ -80 +80 @@
 static bool follow_symlinks = false;
 static bool ignore_exclusions = false;
 static bool spelling = false;
+static size_t  max_size = 0;
 static bool verbose = false;
 static enum {
     EMPTY_BODY_WARN, EMPTY_BODY_INDEX, EMPTY_BODY_SKIP
@@ -382 +383 @@
         return;
     }
 
+    if (max_size>0) {
+        if (d.get_size() > max_size) {
+            skip(file, "Larger than size limit", SKIP_VERBOSE_ONLY);
+            return;
+        }
+    }
+
     string urlterm("U");
     urlterm += url;
 
@@ -1030 +1038 @@
         { "spelling",   no_argument,            NULL, 'S' },
         { "verbose",    no_argument,            NULL, 'v' },
         { "empty-docs", required_argument,      NULL, 'e' },
+        { "max-size",   required_argument,      NULL, 'm' },
         { 0, 0, NULL, 0 }
     };
 
@@ -1192 +1201 @@
 
     string dbpath;
     int getopt_ret;
-    while ((getopt_ret = gnu_getopt_long(argc, argv, "hvd:D:U:M:F:l:s:pfSVe:i",
+    while ((getopt_ret = gnu_getopt_long(argc, argv, "hvd:D:U:M:F:l:s:pfSVe:im:",
                                          longopts, NULL)) != -1) {
         switch (getopt_ret) {
         case 'h': {
@@ -1222 +1231 @@
 "  -f, --follow             follow symbolic links\n"
 "  -i, --ignore-exclusions  ignore meta robots tags and similar exclusions\n"
 "  -S, --spelling           index data for spelling correction\n"
+"  -m, --max-size           maximum size of file to index (in bytes)\n"
 "  -v, --verbose            show more information about what is happening\n"
 "      --overwrite          create the database anew (the default is to update\n"
 "                           if the database already exists)" << endl;
@@ -1326 +1336 @@
         case 'v':
             verbose = true;
             break;
+        case 'm': {
+            int arg = atoi(optarg);
+            if (arg < 0) arg = 0;
+            max_size = size_t(arg);
+            break;
+        }
         case ':': // missing param
             return 1;
         case '?': // unknown option: FIXME -> char