Use a hash table for anchors #64

src/attrs.c

@@ -941,20 +941,33 @@ static void FreeAnchor(TidyDocImpl* doc, Anchor *a)
     TidyDocFree( doc, a );
 }
 
+static uint anchorNameHash(ctmbstr s)
+{
+    uint hashval;
+
+    for (hashval = 0; *s != '\0'; s++) {
+        tmbchar c = TY_(ToLower)( *s );
+        hashval = c + 31*hashval;
+    }
+
+    return hashval % ANCHOR_HASH_SIZE;
+}
+
 /* removes anchor for specific node */
-void TY_(RemoveAnchorByNode)( TidyDocImpl* doc, Node *node )
+void TY_(RemoveAnchorByNode)( TidyDocImpl* doc, ctmbstr name, Node *node )
 {
     TidyAttribImpl* attribs = &doc->attribs;
     Anchor *delme = NULL, *curr, *prev = NULL;
+    uint h = anchorNameHash(name);
 
-    for ( curr=attribs->anchor_list; curr!=NULL; curr=curr->next )
+    for ( curr=attribs->anchor_hash[h]; curr!=NULL; curr=curr->next )
     {
         if ( curr->node == node )
         {
             if ( prev )
                 prev->next = curr->next;
             else
-                attribs->anchor_list = curr->next;
+                attribs->anchor_hash[h] = curr->next;
             delme = curr;
             break;
         }
@@ -981,18 +994,19 @@ static Anchor* AddAnchor( TidyDocImpl* doc, ctmbstr name, Node *node )
 {
     TidyAttribImpl* attribs = &doc->attribs;
     Anchor *a = NewAnchor( doc, name, node );
+    uint h = anchorNameHash(name);
 
-    if ( attribs->anchor_list == NULL)
+    if ( attribs->anchor_hash[h] == NULL)
-         attribs->anchor_list = a;
+         attribs->anchor_hash[h] = a;
     else
     {
-        Anchor *here =  attribs->anchor_list;
+        Anchor *here =  attribs->anchor_hash[h];
         while (here->next)
             here = here->next;
         here->next = a;
     }
 
-    return attribs->anchor_list;
+    return attribs->anchor_hash[h];
 }
 
 /* return node associated with anchor */
@@ -1000,10 +1014,11 @@ static Node* GetNodeByAnchor( TidyDocImpl* doc, ctmbstr name )
 {
     TidyAttribImpl* attribs = &doc->attribs;
     Anchor *found;
+    uint h = anchorNameHash(name);
     tmbstr lname = TY_(tmbstrdup)(doc->allocator, name);
     lname = TY_(tmbstrtolower)(lname);
 
-    for ( found = attribs->anchor_list; found != NULL; found = found->next )
+    for ( found = attribs->anchor_hash[h]; found != NULL; found = found->next )
     {
         if ( TY_(tmbstrcmp)(found->name, lname) == 0 )
             break;
@@ -1020,10 +1035,13 @@ void TY_(FreeAnchors)( TidyDocImpl* doc )
 {
     TidyAttribImpl* attribs = &doc->attribs;
     Anchor* a;
-    while (NULL != (a = attribs->anchor_list) )
+    uint h;
-    {
+    for (h = 0; h < ANCHOR_HASH_SIZE; h++) {
-        attribs->anchor_list = a->next;
+        while (NULL != (a = attribs->anchor_hash[h]) )
-        FreeAnchor(doc, a);
+        {
+            attribs->anchor_hash[h] = a->next;
+            FreeAnchor(doc, a);
+        }
     }
 }
 

src/attrs.h

@@ -55,10 +55,15 @@ struct _AttrHash
 typedef struct _AttrHash AttrHash;
 #endif
 
+enum
+{
+    ANCHOR_HASH_SIZE=1021u
+};
+
 struct _TidyAttribImpl
 {
     /* anchor/node lookup */
-    Anchor*    anchor_list;
+    Anchor*    anchor_hash[ANCHOR_HASH_SIZE];
 
     /* Declared literal attributes */
     Attribute* declared_attr_list;
@@ -117,7 +122,7 @@ Bool TY_(IsValidHTMLID)(ctmbstr id);
 Bool TY_(IsValidXMLID)(ctmbstr id);
 
 /* removes anchor for specific node */
-void TY_(RemoveAnchorByNode)( TidyDocImpl* doc, Node *node );
+void TY_(RemoveAnchorByNode)( TidyDocImpl* doc, ctmbstr name, Node *node );
 
 /* free all anchors */
 void TY_(FreeAnchors)( TidyDocImpl* doc );

src/clean.c

@@ -2638,17 +2638,19 @@ void TY_(FixAnchors)(TidyDocImpl* doc, Node *node, Bool wantName, Bool wantId)
 
             if (id && !wantId
                 /* make sure that Name has been emitted if requested */
-                && (hadName || !wantName || NameEmitted) )
+                && (hadName || !wantName || NameEmitted) ) {
+                if (!wantId && !wantName)
+                    TY_(RemoveAnchorByNode)(doc, id->value, node);
                 TY_(RemoveAttribute)(doc, node, id);
+            }
 
             if (name && !wantName
                 /* make sure that Id has been emitted if requested */
-                && (hadId || !wantId || IdEmitted) )
+                && (hadId || !wantId || IdEmitted) ) {
+                if (!wantId && !wantName)
+                    TY_(RemoveAnchorByNode)(doc, name->value, node);
                 TY_(RemoveAttribute)(doc, node, name);
-
+            }
-            if (TY_(AttrGetById)(node, TidyAttr_NAME) == NULL &&
-                TY_(AttrGetById)(node, TidyAttr_ID) == NULL)
-                TY_(RemoveAnchorByNode)(doc, node);
         }
 
         if (node->content)

src/lexer.c

@@ -1128,7 +1128,7 @@ void TY_(FreeAttrs)( TidyDocImpl* doc, Node *node )
             if ( (attrIsID(av) || attrIsNAME(av)) &&
                  TY_(IsAnchorElement)(doc, node) )
             {
-                TY_(RemoveAnchorByNode)( doc, node );
+                TY_(RemoveAnchorByNode)( doc, av->value, node );
             }
         }