Changeset 933

Timestamp:

08/13/02 18:45:11 (6 years ago)

Author:

whmoseley

Message:

A collection of updates. See CHANGES below.

Mostly fixed a bug in the parsing of meta tags when IgnoreTags? is used.

Files:

• trunk/swish-e/MANIFEST (modified) (1 diff)
• trunk/swish-e/conf/example4.config (modified) (2 diffs)
• trunk/swish-e/example/swish.cgi (modified) (1 diff)
• trunk/swish-e/pod/CHANGES.pod (modified) (1 diff)
• trunk/swish-e/pod/SWISH-CONFIG.pod (modified) (2 diffs)
• trunk/swish-e/prog-bin/spider.pl (modified) (1 diff)
• trunk/swish-e/src/parse_conffile.c (modified) (1 diff)
• trunk/swish-e/src/parser.c (modified) (8 diffs)
• trunk/swish-e/src/swish.h (modified) (1 diff)
• trunk/swish-e/src/swish2.c (modified) (1 diff)

trunk/swish-e/MANIFEST

@@ -231 +231 @@
-src/txt.c
-src/txt.h
+src/vsnprintf.c
-src/vms/acconfig.h_vms
-src/vms/build_swish-e.com

trunk/swish-e/conf/example4.config

@@ -63 +63 @@
-
-
-# Put yes to ignore the total number of words in the file
-# when calculating ranking. Often better with merges and
-# small files. Default is no.
-
-IgnoreTotalWordCountWhenRanking yes
-
-
-# Set the stopwords (words to ignore when searching and when indexing)
-# Carefully think about this feature before using a list of stopwords
@@ -83 +76 @@
-# swish will determine what stopwords to use.  But please
-# read the documentation before using the IgnoreLimit directive.
-
-
-
-
-
-
-
+
-
-

trunk/swish-e/example/swish.cgi

@@ -1370 +1370 @@
-    @properties = ( 'swishreccount', @properties );
-
-
+
+
+
+
+
+
-    $self->swish_command( -H => 9 );
-

trunk/swish-e/pod/CHANGES.pod

@@ -362 +362 @@
-with libxml2 parser.
-
+=item * New directive: IndexAltTagMetaName
+
+Allows indexing of image ALT tags.  Only available when using the libxml2 parser.
+
-=item * New directive: AbsoluteLinks
-

trunk/swish-e/pod/SWISH-CONFIG.pod

@@ -245 +245 @@
-
-L<IndexAdmin|/"item_IndexAdmin"> *text*
+
+=item *
+
+L<IndexAltTagMetaName/"item_IndexAltTagMetaName"> *tagname*|as-text
-
-=item *
@@ -1022 +1026 @@
-    ImageLinksMetaName swishdefault
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-=item AbsoluteLinks [yes|NO]

trunk/swish-e/prog-bin/spider.pl

@@ -256 +256 @@
-    eval { spider( $server, $uri ) };
-    print STDERR $@ if $@;
+
+
+    # provide a way to call a function in the config file when all done
+    check_user_function( 'spider_done', undef, $server );
+    
-
-    return if $server->{quiet};

trunk/swish-e/src/parse_conffile.c

@@ -596 +596 @@
-        }
-
+
+        /* What to do with IMG ATL tags? */
+        if (strcasecmp(w0, "IndexAltTagMetaName") == 0)
+        {
+            if (sl->n <= 1)
+                progerr("%s: requires one value", w0);
+
+            if ( strcasecmp( sl->word[1], "as-text" ) == 0)
+            {
+                sw->IndexAltTag = 1;
+                if ( sw->IndexAltTagMeta )
+                {
+                    efree( sw->IndexAltTagMeta );
+                    sw->IndexAltTagMeta = NULL;
+                }
+            }
+            else
+            {
+                sw->IndexAltTag = 1;
+                if ( sw->IndexAltTagMeta )
+                {
+                    efree( sw->IndexAltTagMeta );
+                    sw->IndexAltTagMeta = NULL;
+                }
+                sw->IndexAltTagMeta = estrdup( sl->word[1] );
+            }
+            continue;
+        }
+
+                
+
-        
-        /* Meta name to extract out <img src> links */

trunk/swish-e/src/parser.c

@@ -159 +159 @@
-    int                 abort;              // flag to stop parsing
-    char               *baseURL;            // for fixing up relative links
-
+ -- for hiding blocks with comments
-} PARSE_DATA;
-
@@ -183 +183 @@
-static void Convert_to_latin1( PARSE_DATA *parse_data, char *txt, int txtlen );
-static int parse_chunks( PARSE_DATA *parse_data );
+
+static void index_alt_tab( PARSE_DATA *parse_data, const char **attr );
-static char *extract_html_links( PARSE_DATA *parse_data, const char **attr, struct metaEntry *meta_entry, char *tag );
-static int read_next_chunk( FileProp *fprop, char *buf, int buf_size, int max_size );
@@ -631 +633 @@
-            /* Extract out links from images */
-            else if ( strcmp( tag, "img") == 0 )
+            {
+                if (parse_data->sw->IndexAltTag)
+                    index_alt_tab( parse_data, attr );
+                    
-                extract_html_links( parse_data, attr, parse_data->sw->images_meta, "src" );
+            }
-
-
@@ -900 +907 @@
-*       parse_data
-*       tag         = tag to look for as a metaname/property
-
+ (since might be different from start tag)
-*       meta_append = if zero, tells push that this is a new meta
-*       prop_append   otherwise, says it's a sibling of a previous call
@@ -929 +936 @@
-    {
-        /* shouldn't need to flush buffer since it's just blocking out a section and should be balanced */
+        /* but need to due to the weird way the char buffer is used (and shared with props) and how metatags are assigned to the buffer */
+        /* basically, since flush_buffer looks at the ignore flag and always clears the buffer, need to do it now */
+        /* flush_buffer really should not be in the business of checking the ignore flag, and rather we need to keep two buffers -- or maybe just always flush with any change */
+        
+        flush_buffer( parse_data, 1 );
+
-        push_stack( &parse_data->meta_stack, endtag, NULL, meta_append, 1 );
-        push_stack( &parse_data->prop_stack, endtag, NULL, prop_append, 1 );
@@ -1473 +1486 @@
-    char        *c;
-
-
-    /* anything to do? */
-    if ( !buf->cur )
@@ -1511 +1523 @@
-    {
-        /* Index the text */
-
+  // this really is wrong -- should not check ignore here.  Fix should be to use two buffers
-            parse_data->total_words +=
-                indexstring( sw, c, parse_data->filenum, structure, 0, NULL, &(parse_data->word_pos) );
@@ -1651 +1663 @@
-
-/*********************************************************************
+*   Index ALT tabs
+*
+*
+*********************************************************************/
+static void index_alt_tab( PARSE_DATA *parse_data, const char **attr )
+{
+    int  meta_append = 0;
+    int  prop_append = 0;
+    char *tagbuf     = parse_data->sw->IndexAltTagMeta;
+    char *alt_text   = extract_html_links( parse_data, attr, NULL, "alt");
+
+
+    if ( !alt_text )
+        return;
+
+    /* Index as regular text? */
+    if ( !parse_data->sw->IndexAltTagMeta )
+    {
+        char_hndl( parse_data, alt_text, strlen( alt_text ) );
+        return;
+    }
+    
+    flush_buffer( parse_data, 1 );
+    start_metaTag( parse_data, tagbuf, tagbuf, &meta_append, &prop_append, 0 );
+    char_hndl( parse_data, alt_text, strlen( alt_text ) );
+    end_metaTag( parse_data, tagbuf, 0 );
+}
+
+
+
+
+/*********************************************************************
-*   Extract out links for indexing
-*

trunk/swish-e/src/swish.h

@@ -879 +879 @@
-    struct metaEntry *images_meta;
-
+
+    /* if allocated the meta name to store alt tags as */
+    int               IndexAltTag;     
+    char             *IndexAltTagMeta;   // use this meta-tag, if set
+
-    /* for converting relative links in href's and img src tags absoulte */
-    int               AbsoluteLinks;

trunk/swish-e/src/swish2.c

@@ -174 +174 @@
-        /* FileRules?? */
-
+        /* meta name for ALT tags */
+        if ( sw->IndexAltTagMeta )
+        {
+            efree( sw->IndexAltTagMeta );
+            sw->IndexAltTagMeta = NULL;
+        }
+
+
-
-        while (tmpindexlist) {