Changeset 2178

Timestamp:

09/26/08 23:59:57 (2 months ago)

Author:

karpet

Message:

some versions of html parser were passing through extra whitespace.
seems to be a specific libxml2 issue. in any case, added a new
whitespace check in both add to buf methods and perl bindings
(the latter where t/20-metanames.t was failing due to extra whitespace)

Files:

• libswish3/trunk/bindings/perl/XS/Property.xs (modified) (1 diff)
• libswish3/trunk/bindings/perl/t/20metanames.t (modified) (1 diff)
• libswish3/trunk/bindings/perl/xs_helpers.c (modified) (1 diff)
• libswish3/trunk/src/libswish3/config.c (modified) (3 diffs)
• libswish3/trunk/src/libswish3/libswish3.h (modified) (1 diff)
• libswish3/trunk/src/libswish3/namedbuffer.c (modified) (2 diffs)
• libswish3/trunk/src/libswish3/string.c (modified) (2 diffs)

libswish3/trunk/bindings/perl/XS/Property.xs

@@ -95 +95 @@
-        if (SWISH_DEBUG & SWISH_DEBUG_MEMORY) {
-            warn("DESTROYing swish_Property object %s  [0x%x] [ref_cnt = %d]", 
-
+long 
-        }
-        

libswish3/trunk/bindings/perl/t/20metanames.t

@@ -32 +32 @@
-    #diag(dump($data->metanames));
-
-wordlist
+tokens
-
-}

libswish3/trunk/bindings/perl/xs_helpers.c

@@ -565 +565 @@
-    int bump            = strlen(SWISH_TOKENPOS_BUMPER);
-    int len;
+
+    //warn("%s nb_content: '%s'\n", key, str);
-        
-    /* analogous to @strings = split(/SWISH_TOKENPOS_BUMPER/, str) */
-    while((tmp = xmlStrstr(str, (xmlChar*)SWISH_TOKENPOS_BUMPER)) != NULL)
-    {
+        //warn("%s split: '%s'\n", key, str);
-        len = tmp - str;
-)
+ && !swish_str_all_ws_len((xmlChar*)str, len)) {
-            av_push(strings, newSVpvn((char*)str, len));
-    
+}
-        str = tmp + bump;  /* move the pointer up */
-    }
-    
-    /* no match and/or last match */
-
+
+
+
+
-        av_push(strings, newSVpvn((char*)str, strlen((char*)str)));
-    }

libswish3/trunk/src/libswish3/config.c

@@ -155 +155 @@
-    if (SWISH_DEBUG & SWISH_DEBUG_MEMORY) {
-        SWISH_DEBUG_MSG("freeing config");
-int)config, (
+long int)config, (long 
-        swish_mem_debug();
-    }
@@ -234 +234 @@
-
-    if (SWISH_DEBUG & SWISH_DEBUG_MEMORY) {
-
+long 
-    }
-
@@ -412 +412 @@
-{
-    SWISH_DEBUG_MSG("config->ref_cnt = %d", config->ref_cnt);
-
-
-int)config, (
+long 
+long 
+long int)config, (long 
-
-    xmlHashScan(config->misc, (xmlHashScanner)config_printer, "misc conf");

libswish3/trunk/src/libswish3/libswish3.h

@@ -468 +468 @@
-xmlChar *           swish_str_skip_ws(xmlChar *s);
-void                swish_str_trim_ws(xmlChar *string);
-
+
+
-void                swish_debug_wchars( const wchar_t * widechars );
-int                 swish_wchar_t_comp(const void *s1, const void *s2);

libswish3/trunk/src/libswish3/namedbuffer.c

@@ -171 +171 @@
-    xmlBufferPtr buf = swish_hash_fetch(nb->hash, name);
-
+/* if the str is nothing but whitespace, skip it */
+    if (swish_str_all_ws(str)) {
+        if (SWISH_DEBUG & SWISH_DEBUG_NAMEDBUFFER)
+            SWISH_DEBUG_MSG("skipping all whitespace string '%s'", str);
+
+        return;
+    }
+
-    if (!buf) {
-        if (autovivify) {
@@ -196 +204 @@
-    }
-    else {
-
+
+
-        swish_append_buffer(buf, str, len);
-    }

libswish3/trunk/src/libswish3/string.c

@@ -541 +541 @@
-}
-
-
+
-swish_str_all_ws(
-    xmlChar *s
-)
-{
-
-
+
+
+
+
+
+
+
+
+
+
-    for (i = 0; i < len; i++) {
-        if (!isspace((int)s[i])) {
@@ -555 +563 @@
-    return 1;
-}
+
-
-void