Issue #70 - Initial implmentation of SVG support.

An immense thanks to Ger Hobbelt who had already done this
in his github.com/GerHobbelt/htmltidy fork.

The two sources have diverges so was not a simple cut
an paste. But again thanks Ger for this.

include/tidyenum.h

@@ -423,6 +423,7 @@ typedef enum
   TidyTag_STYLE,    /**< STYLE */
   TidyTag_SUB,      /**< SUB */
   TidyTag_SUP,      /**< SUP */
+  TidyTag_SVG,      /**< SVG  (HTML5) */
   TidyTag_TABLE,    /**< TABLE */
   TidyTag_TBODY,    /**< TBODY */
   TidyTag_TD,       /**< TD */
@@ -783,8 +784,15 @@ typedef enum
   TidyAttr_ARIA_VALUENOW,
   TidyAttr_ARIA_VALUETEXT,
 
-
+  /* SVG attributes (SVG 1.1) */
-
+  TidyAttr_X,					/**< X= */
+  TidyAttr_Y,                   /**< Y= */
+  TidyAttr_VIEWBOX,             /**< VIEWBOX= */
+  TidyAttr_PRESERVEASPECTRATIO, /**< PRESERVEASPECTRATIO= */
+  TidyAttr_ZOOMANDPAN,          /**< ZOOMANDPAN= */
+  TidyAttr_BASEPROFILE,         /**< BASEPROFILE= */
+  TidyAttr_CONTENTSCRIPTTYPE,   /**< CONTENTSCRIPTTYPE= */
+  TidyAttr_CONTENTSTYLETYPE,    /**< CONTENTSTYLETYPE= */
 
   N_TIDY_ATTRIBS              /**< Must be last */
 } TidyAttrId;

src/attrdict.c

@@ -11693,6 +11693,41 @@ const AttrVersion TY_(W3CAttrsFor_SUP)[] =
   { TidyAttr_UNKNOWN,               0                                                                          },
 };
 
+const AttrVersion TY_(W3CAttrsFor_SVG)[] =
+{
+  { TidyAttr_ALIGN,                 xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_CLASS,                 xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_DIR,                   xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_ID,                    xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_HEIGHT,                xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10|HT50|XH50 },
+  { TidyAttr_LANG,                  xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|xxxx|xxxx|HT50|XH50 },
+  { TidyAttr_OnCLICK,               xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_OnDBLCLICK,            xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_OnKEYDOWN,             xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_OnKEYPRESS,            xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_OnKEYUP,               xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_OnMOUSEDOWN,           xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_OnMOUSEMOVE,           xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_OnMOUSEOUT,            xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_OnMOUSEOVER,           xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_OnMOUSEUP,             xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_STYLE,                 xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_TITLE,                 xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_WIDTH,                 xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10|HT50|XH50 },
+  { TidyAttr_XML_LANG,              xxxx|xxxx|xxxx|xxxx|X10T|xxxx|xxxx|X10F|xxxx|xxxx|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_XMLNS,                 xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_X,                     xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_Y,                     xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_VIEWBOX,               xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_PRESERVEASPECTRATIO,   xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_ZOOMANDPAN,            xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_VERSION,               xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_BASEPROFILE,           xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_CONTENTSCRIPTTYPE,     xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_CONTENTSTYLETYPE,      xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|xxxx|H41S|X10S|XH11|xxxx|HT50|XH50 },
+  { TidyAttr_UNKNOWN,               0                                                                          },
+};
+
 const AttrVersion TY_(W3CAttrsFor_TABLE)[] = 
 {
   { TidyAttr_ACCESSKEY,             xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|xxxx|HT50|XH50 },

src/attrdict.h

@@ -103,6 +103,7 @@ extern const AttrVersion TY_(W3CAttrsFor_STRONG)[];
 extern const AttrVersion TY_(W3CAttrsFor_STYLE)[];
 extern const AttrVersion TY_(W3CAttrsFor_SUB)[];
 extern const AttrVersion TY_(W3CAttrsFor_SUP)[];
+extern const AttrVersion TY_(W3CAttrsFor_SVG)[];
 extern const AttrVersion TY_(W3CAttrsFor_TABLE)[];
 extern const AttrVersion TY_(W3CAttrsFor_TBODY)[];
 extern const AttrVersion TY_(W3CAttrsFor_TD)[];

src/attrs.c

@@ -397,6 +397,24 @@ static const Attribute attribute_defs [] =
   { TidyAttr_ARIA_VALUENOW,           "aria-valuenow",           CH_PCDATA   },
   { TidyAttr_ARIA_VALUETEXT,          "aria-valuetext",          CH_PCDATA   },
 
+  { TidyAttr_X,                        "x",                      CH_PCDATA    }, /* for <svg> */
+  { TidyAttr_Y,                        "y",                      CH_PCDATA    }, /* for <svg> */
+#if 0   /* with uppercase chars taken directly from W3C; are these case-insensitive everywhere? */
+  { TidyAttr_VIEWBOX,           "viewBox",           VERS_INLINE_SVG,   CH_PCDATA    }, /* for <svg> */
+  { TidyAttr_PRESERVEASPECTRATIO, "preserveAspectRatio", VERS_INLINE_SVG, CH_PCDATA  }, /* for <svg> */
+  { TidyAttr_ZOOMANDPAN,        "zoomAndPan",        VERS_INLINE_SVG,   CH_PCDATA    }, /* for <svg> */
+  { TidyAttr_BASEPROFILE,       "baseProfile",       VERS_INLINE_SVG,   CH_PCDATA    }, /* for <svg> */
+  { TidyAttr_CONTENTSCRIPTTYPE, "contentScriptType", VERS_INLINE_SVG,   CH_PCDATA    }, /* for <svg> */
+  { TidyAttr_CONTENTSTYLETYPE,  "contentStyleType",  VERS_INLINE_SVG,   CH_PCDATA    }, /* for <svg> */
+#else
+  { TidyAttr_VIEWBOX,                  "viewbox",                CH_PCDATA    }, /* for <svg> */
+  { TidyAttr_PRESERVEASPECTRATIO,      "preserveaspectratio",    CH_PCDATA    }, /* for <svg> */
+  { TidyAttr_ZOOMANDPAN,               "zoomandpan",             CH_PCDATA    }, /* for <svg> */
+  { TidyAttr_BASEPROFILE,              "baseprofile",            CH_PCDATA    }, /* for <svg> */
+  { TidyAttr_CONTENTSCRIPTTYPE,        "contentscripttype",      CH_PCDATA    }, /* for <svg> */
+  { TidyAttr_CONTENTSTYLETYPE,         "contentstyletype",       CH_PCDATA    }, /* for <svg> */
+#endif
+
   /* this must be the final entry */
   { N_TIDY_ATTRIBS,             NULL,                    NULL         }
 };

src/lexer.h

@@ -525,6 +525,7 @@ typedef enum
   MixedContent,
   Preformatted,
   IgnoreMarkup,
+  OtherNamespace,
   CdataContent
 } GetTokenMode;
 

src/parser.c

@@ -1359,6 +1359,179 @@ void TY_(ParseBlock)( TidyDocImpl* doc, Node *element, GetTokenMode mode)
 #endif
 }
 
+/* [i_a] svg / math */
+
+struct MatchingDescendantData
+{
+	Node *found_node;
+	Bool *passed_marker_node;
+
+	/* input: */
+	TidyTagId matching_tagId;
+	Node *node_to_find;
+	Node *marker_node;
+};
+
+static NodeTraversalSignal FindDescendant_cb(TidyDocImpl* ARG_UNUSED(doc), Node* node, void *propagate)
+{
+	struct MatchingDescendantData *cb_data = (struct MatchingDescendantData *)propagate;
+
+	if (TagId(node) == cb_data->matching_tagId)
+	{
+		/* make sure we match up 'unknown' tags exactly! */
+		if (cb_data->matching_tagId != TidyTag_UNKNOWN ||
+			(node->element != NULL &&
+			cb_data->node_to_find != NULL &&
+			cb_data->node_to_find->element != NULL &&
+			0 == TY_(tmbstrcmp)(cb_data->node_to_find->element, node->element)))
+		{
+			cb_data->found_node = node;
+			return ExitTraversal;
+		}
+	}
+
+	if (cb_data->passed_marker_node && node == cb_data->marker_node)
+		*cb_data->passed_marker_node = yes;
+
+	return VisitParent;
+}
+
+/*
+Search the parent chain (from 'parent' upwards up to the root) for a node matching the
+given 'node'.
+
+When the search passes beyond the 'marker_node' (which is assumed to sit in the
+parent chain), this will be flagged by setting the boolean referenced by
+'is_parent_of_marker' to yes.
+
+'is_parent_of_marker' and 'marker_node' are optional parameters and may be NULL.
+*/
+static Node *FindMatchingDescendant( Node *parent, Node *node, Node *marker_node, Bool *is_parent_of_marker )
+{
+	struct MatchingDescendantData cb_data = { 0 };
+	cb_data.matching_tagId = TagId(node);
+	cb_data.node_to_find = node;
+	cb_data.marker_node = marker_node;
+
+	assert(node);
+
+	if (is_parent_of_marker)
+		*is_parent_of_marker = no;
+
+	TY_(TraverseNodeTree)(NULL, parent, FindDescendant_cb, &cb_data);
+	return cb_data.found_node;
+}
+
+/*
+   Act as a generic XML (sub)tree parser: collect each node and add it to the DOM, without any further validation.
+   TODO : add schema- or other-hierarchy-definition-based validation of the subtree here...
+*/
+void TY_(ParseNamespace)(TidyDocImpl* doc, Node *basenode, GetTokenMode mode)
+{
+    Lexer* lexer = doc->lexer;
+	Node *node;
+	Node *parent = basenode;
+	uint istackbase;
+
+	/* a la <table>: defer popping elements off the inline stack */
+	TY_(DeferDup)( doc );
+	istackbase = lexer->istackbase;
+	lexer->istackbase = lexer->istacksize;
+
+	mode = OtherNamespace; /* Preformatted; IgnoreWhitespace; */
+
+	while ((node = TY_(GetToken)(doc, mode)) != NULL)
+	{
+		/*
+		fix check to skip action in InsertMisc for regular/empty
+		nodes, which we don't want here...
+
+		The way we do it here is by checking and processing everything
+		and only what remains goes into InsertMisc()
+		*/
+
+		/* is this a close tag? And does it match the current parent node? */
+		if (node->type == EndTag)
+		{
+			/*
+			to prevent end tags flowing from one 'alternate namespace' we
+			check this in two phases: first we check if the tag is a
+			descendant of the current node, and when it is, we check whether
+			it is the end tag for a node /within/ or /outside/ the basenode.
+			*/
+			Bool outside;
+			Node *mp = FindMatchingDescendant(parent, node, basenode, &outside);
+
+			if (mp != NULL)
+			{
+				/*
+				when mp != parent as we might expect,
+				infer end tags until we 'hit' the matched
+				parent or the basenode
+				*/
+				Node *n;
+
+				for (n = parent;
+					 n != NULL && n != basenode->parent && n != mp;
+					 n = n->parent)
+				{
+					/* n->implicit = yes; */
+					n->closed = yes;
+					TY_(ReportError)(doc, n->parent, n, MISSING_ENDTAG_BEFORE);
+				}
+				assert(outside == no ? n == mp : 1);
+				assert(outside == yes ? n == basenode->parent : 1);
+
+				if (outside == no)
+				{
+					/* EndTag for a node within the basenode subtree. Roll on... */
+					n->closed = yes;
+					TY_(FreeNode)(doc, node);
+
+					node = n;
+					parent = node->parent;
+				}
+				else
+				{
+					/* EndTag for a node outside the basenode subtree: let the caller handle that. */
+					TY_(UngetToken)( doc );
+					node = basenode;
+					parent = node->parent;
+				}
+
+				/* when we've arrived at the end-node for the base node, it's quitting time */
+				if (node == basenode)
+				{
+					lexer->istackbase = istackbase;
+					assert(basenode->closed == yes);
+					return;
+				}
+			}
+			else
+			{
+				/* unmatched close tag: report an error and discard */
+				TY_(ReportError)(doc, parent, node, NON_MATCHING_ENDTAG);
+				TY_(ReportError)(doc, parent, node, DISCARDING_UNEXPECTED);
+				assert(parent);
+				assert(parent->tag != node->tag);
+			}
+		}
+		else if (node->type == StartTag)
+		{
+			/* add another child to the current parent */
+			TY_(InsertNodeAtEnd)(parent, node);
+			parent = node;
+		}
+		else
+		{
+			TY_(InsertNodeAtEnd)(parent, node);
+		}
+	}
+
+	TY_(ReportError)(doc, basenode->parent, basenode, MISSING_ENDTAG_FOR);
+}
+
+
 void TY_(ParseInline)( TidyDocImpl* doc, Node *element, GetTokenMode mode )
 {
 #if !defined(NDEBUG) && defined(_MSC_VER)

src/tags.c

@@ -108,6 +108,7 @@ static CheckAttribs CheckHTML;
 #define VERS_ELEM_STYLE      (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50)
 #define VERS_ELEM_SUB        (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50)
 #define VERS_ELEM_SUP        (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50)
+#define VERS_ELEM_SVG        (xxxx|xxxx|xxxx|H41T|X10T|xxxx|H41F|X10F|xxxx|H41S|X10S|XH11|xxxx|HT50|XH50)
 #define VERS_ELEM_TABLE      (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10|HT50|XH50)
 #define VERS_ELEM_TBODY      (xxxx|xxxx|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|xxxx|HT50|XH50)
 #define VERS_ELEM_TD         (xxxx|HT32|H40T|H41T|X10T|H40F|H41F|X10F|H40S|H41S|X10S|XH11|XB10|HT50|XH50)
@@ -247,6 +248,7 @@ static const Dict tag_defs[] =
   { TidyTag_STYLE,      "style",      VERS_ELEM_STYLE,      &TY_(W3CAttrsFor_STYLE)[0],      (CM_HEAD|CM_BLOCK),                            TY_(ParseScript),   NULL           },
   { TidyTag_SUB,        "sub",        VERS_ELEM_SUB,        &TY_(W3CAttrsFor_SUB)[0],        (CM_INLINE),                                   TY_(ParseInline),   NULL           },
   { TidyTag_SUP,        "sup",        VERS_ELEM_SUP,        &TY_(W3CAttrsFor_SUP)[0],        (CM_INLINE),                                   TY_(ParseInline),   NULL           },
+  { TidyTag_SVG,        "svg",        VERS_ELEM_SVG,        &TY_(W3CAttrsFor_SVG)[0],        (CM_INLINE|CM_BLOCK|CM_MIXED),                 TY_(ParseNamespace),NULL           },
   { TidyTag_TABLE,      "table",      VERS_ELEM_TABLE,      &TY_(W3CAttrsFor_TABLE)[0],      (CM_BLOCK),                                    TY_(ParseTableTag), CheckTABLE     },
   { TidyTag_TBODY,      "tbody",      VERS_ELEM_TBODY,      &TY_(W3CAttrsFor_TBODY)[0],      (CM_TABLE|CM_ROWGRP|CM_OPT),                   TY_(ParseRowGroup), NULL           },
   { TidyTag_TD,         "td",         VERS_ELEM_TD,         &TY_(W3CAttrsFor_TD)[0],         (CM_ROW|CM_OPT|CM_NO_INDENT),                  TY_(ParseBlock),    NULL           },
@@ -949,6 +951,41 @@ uint TY_(nodeHeaderLevel)( Node* node )
     return 0;
 }
 
+/* [i_a] generic node tree traversal; see also <tidy-int.h> */
+NodeTraversalSignal TY_(TraverseNodeTree)(TidyDocImpl* doc, Node* node, NodeTraversalCallBack *cb, void *propagate )
+{
+    while (node)
+    {
+        NodeTraversalSignal s = (*cb)(doc, node, propagate);
+
+        if (node->content && (s == ContinueTraversal || s == SkipSiblings))
+        {
+            s = TY_(TraverseNodeTree)(doc, node->content, cb, propagate);
+        }
+
+        switch (s)
+        {
+        case ExitTraversal:
+            return ExitTraversal;
+
+        case VisitParent:
+            node = node->parent;
+            continue;
+
+        case SkipSiblings:
+        case SkipChildrenAndSiblings:
+            return ContinueTraversal;
+
+        default:
+            node = node->next;
+            break;
+        }
+    }
+    return ContinueTraversal;
+}
+
+
+
 /*
  * local variables:
  * mode: c

src/tags.h

@@ -111,6 +111,7 @@ Parser TY_(ParseSelect);
 Parser TY_(ParseOptGroup);
 Parser TY_(ParseText);
 Parser TY_(ParseDatalist);
+Parser TY_(ParseNamespace);
 
 CheckAttribs TY_(CheckAttributes);
 

src/tidy-int.h

@@ -121,4 +121,26 @@ TidyOption   tidyImplToOption( const TidyOptionImpl* option );
 
 int          TY_(DocParseStream)( TidyDocImpl* impl, StreamIn* in );
 
+/*
+   [i_a] generic node tree traversal code; used in several spots.
+
+   Define your own callback, which returns one of the NodeTraversalSignal values
+   to instruct the tree traversal routine TraverseNodeTree() what to do.
+
+   Pass custom data to/from the callback using the 'propagate' reference.
+ */
+typedef enum
+{
+	ContinueTraversal,	     /* visit siblings and children */
+	SkipChildren,			 /* visit siblings of this node; ignore its children */
+	SkipSiblings,			 /* ignore subsequent siblings of this node; ignore their children; traverse  */
+	SkipChildrenAndSiblings, /* visit siblings of this node; ignore its children */
+	VisitParent,			 /* REVERSE traversal: visit the parent of the current node */
+	ExitTraversal			 /* terminate traversal on the spot */
+} NodeTraversalSignal;
+
+typedef NodeTraversalSignal NodeTraversalCallBack(TidyDocImpl* doc, Node* node, void *propagate);
+
+NodeTraversalSignal TY_(TraverseNodeTree)(TidyDocImpl* doc, Node* node, NodeTraversalCallBack *cb, void *propagate);
+
 #endif /* __TIDY_INT_H__ */