New merge-emphasis & coerce-endtags options added.

Fixes #19.
htacg · Mar 17, 2012 · 1052c2b · 1052c2b
1 parent 3ed33a1
commit 1052c2b
Show file tree

Hide file tree

Showing 7 changed files with 80 additions and 5 deletions.
diff --git a/include/tidyenum.h b/include/tidyenum.h
@@ -112,6 +112,7 @@ typedef enum
   TidyQuiet,           /**< No 'Parsing X', guessed DTD or summary */
   TidyIndentContent,   /**< Indent content of appropriate tags */
                        /**< "auto" does text/block level content indentation */
+  TidyCoerceEndTags,   /**< Coerce end tags from start tags where probably intended */
   TidyHideEndTags,     /**< Suppress optional end tags */
   TidyXmlTags,         /**< Treat input as XML */
   TidyXmlOut,          /**< Create output as XML */
@@ -199,6 +200,7 @@ typedef enum
 #else
   TidyPunctWrapNotUsed,
 #endif
+  TidyMergeEmphasis,       /**< Merge nested B and I elements */
   TidyMergeDivs,       /**< Merge multiple DIVs */
   TidyDecorateInferredUL,  /**< Mark inferred UL elements with no indent CSS */
   TidyPreserveEntities,    /**< Preserve entities */

diff --git a/quickref.html b/quickref.html
@@ -8,7 +8,7 @@
   <body>
     <h1 id="top">Quick Reference</h1>
     <h2>HTML Tidy Configuration Options</h2>
-    <p>Version: <a href="https://github.com/w3c/tidy-html5/tree/5c4d2e5">https://github.com/w3c/tidy-html5/tree/5c4d2e5</a></p>
+    <p>Version: <a href="https://github.com/w3c/tidy-html5/tree/3a9a794">https://github.com/w3c/tidy-html5/tree/3a9a794</a></p>
     <p>
       <a class="h3" href="#MarkupHeader">HTML, XHTML, XML</a>
       <br />
@@ -88,6 +88,13 @@ <h2>HTML Tidy Configuration Options</h2>
         <td>Boolean</td>
         <td>no</td>
       </tr>
+      <tr>
+        <td>
+          <a href="#coerce-endtags">coerce-endtags</a>
+        </td>
+        <td>Boolean</td>
+        <td>yes</td>
+      </tr>
       <tr>
         <td>
           <a href="#css-prefix">css-prefix</a>
@@ -251,6 +258,13 @@ <h2>HTML Tidy Configuration Options</h2>
         <td>AutoBool</td>
         <td>auto</td>
       </tr>
+      <tr>
+        <td>
+          <a href="#merge-emphasis">merge-emphasis</a>
+        </td>
+        <td>Boolean</td>
+        <td>yes</td>
+      </tr>
       <tr>
         <td>
           <a href="#merge-spans">merge-spans</a>
@@ -866,6 +880,23 @@ <h2>HTML Tidy Configuration Options</h2>
       <tr>
         <td>&#160;</td>
       </tr>
+      <tr>
+        <td class="tabletitle" valign="top" id="coerce-endtags">coerce-endtags</td>
+        <td class="tabletitlelink" valign="top" align="right">
+          <a href="#top">Top</a>
+        </td>
+      </tr>
+      <tr>
+        <td valign="top">Type: <strong>Boolean</strong><br />
+            Default: <strong>yes</strong><br />Example: <strong>y/n, yes/no, t/f, true/false, 1/0</strong></td>
+        <td align="right" valign="top"></td>
+      </tr>
+      <tr>
+        <td colspan="2">This option specifies if Tidy should coerce a start tag into an end tag in cases where it looks like an end tag was probably intended; for example, given &lt;span&gt;foo &lt;b&gt;bar&lt;b&gt; baz&lt;/span&gt;, Tidy will output &lt;span&gt;foo &lt;b&gt;bar&lt;/b&gt; baz&lt;/span&gt;. </td>
+      </tr>
+      <tr>
+        <td>&#160;</td>
+      </tr>
       <tr>
         <td class="tabletitle" valign="top" id="css-prefix">css-prefix</td>
         <td class="tabletitlelink" valign="top" align="right">
@@ -1271,6 +1302,23 @@ <h2>HTML Tidy Configuration Options</h2>
       <tr>
         <td>&#160;</td>
       </tr>
+      <tr>
+        <td class="tabletitle" valign="top" id="merge-emphasis">merge-emphasis</td>
+        <td class="tabletitlelink" valign="top" align="right">
+          <a href="#top">Top</a>
+        </td>
+      </tr>
+      <tr>
+        <td valign="top">Type: <strong>Boolean</strong><br />
+            Default: <strong>yes</strong><br />Example: <strong>y/n, yes/no, t/f, true/false, 1/0</strong></td>
+        <td align="right" valign="top"></td>
+      </tr>
+      <tr>
+        <td colspan="2">This option specifies if Tidy should merge nested &lt;b&gt; and &lt;i&gt; elements; for example, for the case &lt;b class="rtop-2"&gt;foo &lt;b class="r2-2"&gt;bar&lt;/b&gt; baz&lt;/b&gt;, Tidy will output  &lt;b class="rtop-2"&gt;foo bar baz&lt;/b&gt;. Note that if you set this option to "no" Tidy will by default still attempt to correct some cases of nested &lt;b&gt; and &lt;i&gt; elements; for example, given &lt;span&gt;foo &lt;b&gt;bar &lt;b&gt;baz&lt;/b&gt;&lt;/b&gt; &lt;/span&gt;, Tidy will output &lt;span&gt;foo &lt;b&gt;bar baz&lt;/b&gt;&lt;/span&gt;. To suppress that behavior, set the coerce-endtags option to "no". </td>
+      </tr>
+      <tr>
+        <td>&#160;</td>
+      </tr>
       <tr>
         <td class="tabletitle" valign="top" id="merge-spans">merge-spans</td>
         <td class="tabletitlelink" valign="top" align="right">

diff --git a/src/config.c b/src/config.c
@@ -239,6 +239,7 @@ static const TidyOptionImpl option_defs[] =
   { TidyShowWarnings,            DG, "show-warnings",               BL, yes,             ParseBool,         boolPicks       },
   { TidyQuiet,                   MS, "quiet",                       BL, no,              ParseBool,         boolPicks       },
   { TidyIndentContent,           PP, "indent",                      IN, TidyNoState,     ParseAutoBool,     autoBoolPicks   },
+  { TidyCoerceEndTags,           MU, "coerce-endtags",              BL, yes,             ParseBool,         boolPicks       },
   { TidyHideEndTags,             MU, "hide-endtags",                BL, no,              ParseBool,         boolPicks       },
   { TidyXmlTags,                 MU, "input-xml",                   BL, no,              ParseBool,         boolPicks       },
   { TidyXmlOut,                  MU, "output-xml",                  BL, no,              ParseBool,         boolPicks       },
@@ -311,6 +312,7 @@ static const TidyOptionImpl option_defs[] =
 #if SUPPORT_ASIAN_ENCODINGS
   { TidyPunctWrap,               PP, "punctuation-wrap",            BL, no,              ParseBool,         boolPicks       },
 #endif
+  { TidyMergeEmphasis,           MU, "merge-emphasis",              BL, yes,             ParseBool,         boolPicks       },
   { TidyMergeDivs,               MU, "merge-divs",                  IN, TidyAutoState,   ParseAutoBool,     autoBoolPicks   },
   { TidyDecorateInferredUL,      MU, "decorate-inferred-ul",        BL, no,              ParseBool,         boolPicks       },
   { TidyPreserveEntities,        MU, "preserve-entities",           BL, no,              ParseBool,         boolPicks       },

diff --git a/src/localize.c b/src/localize.c
@@ -470,6 +470,12 @@ static const TidyOptionDoc option_docs[] =
   {TidyHideComments,
    "This option specifies if Tidy should print out comments. "
   },
+  {TidyCoerceEndTags,
+   "This option specifies if Tidy should coerce a start tag into an end tag "
+   "in cases where it looks like an end tag was probably intended; "
+   "for example, given &lt;span&gt;foo &lt;b&gt;bar&lt;b&gt; baz&lt;/span&gt;, "
+   "Tidy will output &lt;span&gt;foo &lt;b&gt;bar&lt;/b&gt; baz&lt;/span&gt;. "
+  },
   {TidyHideEndTags,
    "This option specifies if Tidy should omit optional end-tags when "
    "generating the pretty printed markup. This option is ignored if you are "
@@ -504,6 +510,18 @@ static const TidyOptionDoc option_docs[] =
    "that takes a list of predefined values to lower case. This is required "
    "for XHTML documents. "
   },
+  {TidyMergeEmphasis,
+   "This option specifies if Tidy should merge nested &lt;b&gt; and &lt;i&gt; "
+   "elements; for example, for the case "
+   "&lt;b class=\"rtop-2\"&gt;foo &lt;b class=\"r2-2\"&gt;bar&lt;/b&gt; baz&lt;/b&gt;, "
+   "Tidy will output  &lt;b class=\"rtop-2\"&gt;foo bar baz&lt;/b&gt;. "
+   "Note that if you set this option to \"no\" Tidy will by default still "
+   "attempt to correct some cases of nested &lt;b&gt; and &lt;i&gt; elements; "
+   "for example, given "
+   "&lt;span&gt;foo &lt;b&gt;bar &lt;b&gt;baz&lt;/b&gt;&lt;/b&gt; &lt;/span&gt;, "
+   "Tidy will output &lt;span&gt;foo &lt;b&gt;bar baz&lt;/b&gt;&lt;/span&gt;. "
+   "To suppress that behavior, set the coerce-endtags option to \"no\". "
+  },
   {TidyMergeDivs,
    "Can be used to modify behavior of -c (--clean yes) option. "
    "This option specifies if Tidy should merge nested &lt;div&gt; such as "

diff --git a/src/parser.c b/src/parser.c
@@ -1400,6 +1400,7 @@ void TY_(ParseInline)( TidyDocImpl* doc, Node *element, GetTokenMode mode )
              && !nodeIsSUP(node)
              && !nodeIsQ(node)
              && !nodeIsSPAN(node)
+             && cfgBool(doc, TidyCoerceEndTags)
            )
         {
             /* proceeds only if "node" does not have any attribute and
@@ -1617,7 +1618,8 @@ void TY_(ParseInline)( TidyDocImpl* doc, Node *element, GetTokenMode mode )
             /* #427827 - fix by Randy Waki and Bjoern Hoehrmann 23 Aug 00 */
             /* other fixes by Dave Raggett */
             /* if (node->attributes == NULL) */
-            if (node->type != EndTag && node->attributes == NULL)
+            if (node->type != EndTag && node->attributes == NULL
+                && cfgBool(doc, TidyCoerceEndTags) )
             {
                 node->type = EndTag;
                 TY_(ReportError)(doc, element, node, COERCE_TO_ENDTAG);
@@ -3022,7 +3024,8 @@ void TY_(ParseTitle)(TidyDocImpl* doc, Node *title, GetTokenMode ARG_UNUSED(mode
     Node *node;
     while ((node = TY_(GetToken)(doc, MixedContent)) != NULL)
     {
-        if (node->tag == title->tag && node->type == StartTag)
+        if (node->tag == title->tag && node->type == StartTag
+            && cfgBool(doc, TidyCoerceEndTags) )
         {
             TY_(ReportError)(doc, title, node, COERCE_TO_ENDTAG);
             node->type = EndTag;

diff --git a/src/tidylib.c b/src/tidylib.c
@@ -1246,13 +1246,15 @@ int         tidyDocCleanAndRepair( TidyDocImpl* doc )
     Bool tidyMark = cfgBool( doc, TidyMark );
     Bool tidyXmlTags = cfgBool( doc, TidyXmlTags );
     Bool wantNameAttr = cfgBool( doc, TidyAnchorAsName );
+    Bool mergeEmphasis = cfgBool( doc, TidyMergeEmphasis );
     Node* node;
 
     if (tidyXmlTags)
        return tidyDocStatus( doc );
 
     /* simplifies <b><b> ... </b> ...</b> etc. */
-    TY_(NestedEmphasis)( doc, &doc->root );
+    if ( mergeEmphasis )
+        TY_(NestedEmphasis)( doc, &doc->root );
 
     /* cleans up <dir>indented text</dir> etc. */
     TY_(List2BQ)( doc, &doc->root );

diff --git a/src/version.h b/src/version.h
@@ -1 +1 @@
-static const char TY_(release_date)[] = "https://github.com/w3c/tidy-html5/tree/3a9a794";
+static const char TY_(release_date)[] = "https://github.com/w3c/tidy-html5/tree/3ed33a1";