- HTMLparser.c: Patch from Jonas Borgström
(htmlGetEndPriority): New function, returns
the priority of a certain element.
(htmlAutoCloseOnClose): Only close inline elements if they
all have lower or equal priority.
- result/HTML: this of course changed a number of tests results.
Daniel
diff --git a/ChangeLog b/ChangeLog
index 2a45a17..7a5516b 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,12 @@
+Fri May 11 16:07:13 CEST 2001 Daniel Veillard <Daniel.Veillard@imag.fr>
+
+ * HTMLparser.c: Patch from Jonas Borgström
+ (htmlGetEndPriority): New function, returns
+ the priority of a certain element.
+ (htmlAutoCloseOnClose): Only close inline elements if they
+ all have lower or equal priority.
+ * result/HTML: this of course changed a number of tests results.
+
Thu May 10 17:30:22 CEST 2001 Daniel Veillard <Daniel.Veillard@imag.fr>
* xmlIO.c catalog.c: plugged in the default catalog resolution
diff --git a/HTMLparser.c b/HTMLparser.c
index 870201d..5e79464 100644
--- a/HTMLparser.c
+++ b/HTMLparser.c
@@ -565,15 +565,32 @@
};
/*
- * end tags that imply the end of the inside elements
- */
-const char *htmlEndClose[] = {
-"head",
-"body",
-"html",
-NULL
-};
+ * This table is used by the htmlparser to know what to do with
+ * broken html pages. By assigning different priorities to different
+ * elements the parser can decide how to handle extra endtags.
+ * Endtags are only allowed to close elements with lower or equal
+ * priority.
+ */
+typedef struct {
+ const char *name;
+ int priority;
+} elementPriority;
+
+const elementPriority htmlEndPriority[] = {
+ {"div", 150},
+ {"td", 160},
+ {"th", 160},
+ {"tr", 170},
+ {"thead", 180},
+ {"tbody", 180},
+ {"tfoot", 180},
+ {"table", 190},
+ {"head", 200},
+ {"body", 200},
+ {"html", 220},
+ {NULL, 100} /* Default priority */
+};
static const char** htmlStartCloseIndex[100];
static int htmlStartCloseIndexinitialized = 0;
@@ -628,6 +645,23 @@
}
/**
+ * htmlGetEndPriority:
+ * @name: The name of the element to look up the priority for.
+ *
+ * Return value: The "endtag" priority.
+ **/
+static int
+htmlGetEndPriority (const xmlChar *name) {
+ int i = 0;
+
+ while ((htmlEndPriority[i].name != NULL) &&
+ (!xmlStrEqual((const xmlChar *)htmlEndPriority[i].name, name)))
+ i++;
+
+ return(htmlEndPriority[i].priority);
+}
+
+/**
* htmlCheckAutoClose:
* @newtag: The new tag name
* @oldtag: The old tag name
@@ -674,7 +708,7 @@
htmlAutoCloseOnClose(htmlParserCtxtPtr ctxt, const xmlChar *newtag) {
htmlElemDescPtr info;
xmlChar *oldname;
- int i, endCloses = 0;
+ int i, priority;
#ifdef DEBUG
xmlGenericError(xmlGenericErrorContext,"Close of %s stack: %d elements\n", newtag, ctxt->nameNr);
@@ -682,15 +716,20 @@
xmlGenericError(xmlGenericErrorContext,"%d : %s\n", i, ctxt->nameTab[i]);
#endif
+ priority = htmlGetEndPriority (newtag);
+
for (i = (ctxt->nameNr - 1);i >= 0;i--) {
+
if (xmlStrEqual(newtag, ctxt->nameTab[i])) break;
+ /*
+ * A missplaced endtagad can only close elements with lower
+ * or equal priority, so if we find an element with higher
+ * priority before we find an element with
+ * matching name, we just ignore this endtag
+ */
+ if (htmlGetEndPriority (ctxt->nameTab[i]) > priority) return;
}
if (i < 0) return;
- for (i = 0; (htmlEndClose[i] != NULL);i++)
- if (xmlStrEqual(newtag, (const xmlChar *) htmlEndClose[i])) {
- endCloses = 1;
- break;
- }
while (!xmlStrEqual(newtag, ctxt->name)) {
info = htmlTagLookup(ctxt->name);
@@ -707,8 +746,6 @@
"Opening and ending tag mismatch: %s and %s\n",
newtag, ctxt->name);
ctxt->wellFormed = 0;
- } else if (endCloses == 0) {
- return;
}
if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL))
ctxt->sax->endElement(ctxt->userData, ctxt->name);
diff --git a/result/HTML/doc3.htm b/result/HTML/doc3.htm
index fd9f514..3ed7223 100644
--- a/result/HTML/doc3.htm
+++ b/result/HTML/doc3.htm
@@ -97,8 +97,7 @@
</td></tr></tbody></table>
</center></td></tr></tbody></table></td></tr>
</tbody></table>
-<table bgcolor="#003399" border="0" cellspacing="6" width="80%"><tbody>
-<tr>
+<table bgcolor="#003399" border="0" cellspacing="6" width="80%"><tbody><tr>
<td bgcolor="black" valign="top" width="10%"><table border="0" cellpadding="3" cellspacing="0" width="100%"><tbody><tr><td width="100%">
<img height="1" src="doc3_files/spacer.gif" width="111">
<br>
@@ -784,8 +783,7 @@
</td></tr></tbody></table>
</center>
</td>
-<td bgcolor="silver" valign="top" width="10%">
-<center>
+<td bgcolor="silver" valign="top" width="10%"><center>
<p>
<table bgcolor="silver" border="0" cellpadding="0" cellspacing="0" width="100%"><tbody><tr><td colstart="1"><center>
<!-- <FORM ACTION="/cgi-bin/subscribe.pl" METHOD="POST" ENCTYPE="application/x-www-form-urlencoded">
@@ -853,11 +851,10 @@
}
// -->
</script>
-<b><noscript>
+<b><noscript></noscript></b>
<a href="http://www.goto.com/d/search/ssn/?fromGIF=true" target="_blank"><img align="bottom" border="0" height="90" ismap src="doc3_files/100x90.gif" width="100"></a>
<b><a href="http://www.goto.com/d/search/ssn/?fromGIF=true" target="_blank"></a></b>
<b></b>
-</noscript></b>
<b><!-- END GoTo.com Search Box --></b>
<!-- Pricewatch Search Box --><form action="http://www.pricewatch.com/search/search.asp" method="get" target="_Blank"><center><p>
<b><font color="white" face="ARIAL, HELVETICA" size="1">PC Price
@@ -880,10 +877,9 @@
</a>
</td></tr></tbody></table>
<table bgcolor="silver" border="0" cellpadding="0" cellspacing="0" height="100%" width="100%"><tbody><tr><td width="100%"> </td></tr></tbody></table>
-</center>
+</center></td>
+</tr></tbody></table>
<!-- </TABLE>--><center></center>
-</td>
-</tr>
<tr><td colspan="3" valign="TOP" height="70"> </td></tr>
<table border="0" width="780"><tbody>
<tr><td width="780"><p align="center">
@@ -893,8 +889,7 @@
</p></td></tr>
<!-- <TR> <TD WIDTH="780"> <P ALIGN="CENTER"><FONT SIZE="1" COLOR="#999999" FACE="Verdana,arial">Site design by Tim Brinkley</FONT> </TD> </TR> -->
</tbody></table>
-<script> window.open=NS_ActualOpen; </script>
-</tbody></table>
</div>
+<script> window.open=NS_ActualOpen; </script>
</body>
</html>
diff --git a/result/HTML/doc3.htm.err b/result/HTML/doc3.htm.err
index 86783ea..949fefb 100644
--- a/result/HTML/doc3.htm.err
+++ b/result/HTML/doc3.htm.err
@@ -37,12 +37,9 @@
./test/HTML/doc3.htm:742: error: Unexpected end tag : p
=7708"></a></IFRAME></CENTER></LI></FONT></TD></TR></TBODY></TABLE></CENTER></P
^
-./test/HTML/doc3.htm:767: error: Opening and ending tag mismatch: font and form
+./test/HTML/doc3.htm:767: error: Unexpected end tag : form
archive</A></FONT> </FORM></CENTER></TD></TR></TBODY></TABLE><!--
- ^
-./test/HTML/doc3.htm:767: error: Opening and ending tag mismatch: center and font
- archive</A></FONT> </FORM></CENTER></TD></TR></TBODY></TABLE><!--
- ^
+ ^
./test/HTML/doc3.htm:790: error: Unexpected end tag : iframe
document.write("42DF8478957377></IFRAME>");
^
@@ -55,13 +52,10 @@
./test/HTML/doc3.htm:806: error: Unexpected end tag : a
document.write("ype=gif&size=100x90></A>");
^
-./test/HTML/doc3.htm:810: error: Opening and ending tag mismatch: b and noscript
- <B><NOSCRIPT></B><A
- ^
./test/HTML/doc3.htm:815: error: Unexpected end tag : a
</A></A></B><B></NOSCRIPT></B><B><!-- END GoTo.com Search Box --></
^
-./test/HTML/doc3.htm:815: error: Opening and ending tag mismatch: noscript and b
+./test/HTML/doc3.htm:815: error: Unexpected end tag : noscript
</A></A></B><B></NOSCRIPT></B><B><!-- END GoTo.com Search Box --></
^
./test/HTML/doc3.htm:821: error: Opening and ending tag mismatch: form and center
@@ -76,18 +70,12 @@
./test/HTML/doc3.htm:834: error: Unexpected end tag : p
width="100%"> </TD></TR></TBODY></TABLE></P></CENTER></TR></TBODY></TABLE>
^
-./test/HTML/doc3.htm:834: error: Opening and ending tag mismatch: tr and td
-width="100%"> </TD></TR></TBODY></TABLE></P></CENTER></TR></TBODY></TABLE>
- ^
-./test/HTML/doc3.htm:834: error: Opening and ending tag mismatch: tbody and td
-width="100%"> </TD></TR></TBODY></TABLE></P></CENTER></TR></TBODY></TABLE>
- ^
-./test/HTML/doc3.htm:834: error: Opening and ending tag mismatch: table and td
-width="100%"> </TD></TR></TBODY></TABLE></P></CENTER></TR></TBODY></TABLE>
- ^
-./test/HTML/doc3.htm:836: error: Opening and ending tag mismatch: table and tbody
+./test/HTML/doc3.htm:835: error: Unexpected end tag : td
+<CENTER></CENTER></TD></TR><TR><TD COLSPAN="3" VALIGN="TOP"
+ ^
+./test/HTML/doc3.htm:835: error: Unexpected end tag : tr
+<CENTER></CENTER></TD></TR><TR><TD COLSPAN="3" VALIGN="TOP"
+ ^
+./test/HTML/doc3.htm:836: error: Unexpected end tag : table
HEIGHT="70"> </TD> </TR></TABLE>
^
-./test/HTML/doc3.htm:844: error: Opening and ending tag mismatch: div and tbody
-,arial">Site design by Tim Brinkley</FONT> </TD> </TR> --></TBODY></TABLE></DI
- ^
diff --git a/result/HTML/doc3.htm.sax b/result/HTML/doc3.htm.sax
index 7c35661..28d6136 100644
--- a/result/HTML/doc3.htm.sax
+++ b/result/HTML/doc3.htm.sax
Binary files differ
diff --git a/result/HTML/wired.html b/result/HTML/wired.html
index e62d5c5..6653097 100644
--- a/result/HTML/wired.html
+++ b/result/HTML/wired.html
@@ -127,17 +127,17 @@
<tr><td bgcolor="#CCFFCC"><font size="1" face="Verdana, Arial, Helvetica, sans-serif" color="#000000"><a href="/news/news/reuters/sports/">Sports</a></font></td></tr>
<tr><td bgcolor="#99FF99"><font size="1" face="Verdana, Arial, Helvetica, sans-serif" color="#000000"><a href="/news/news/reuters/business/">Finance</a></font></td></tr>
<!-- End upper left nav --><!-- Begin lower Left Nav --><tr><td bgcolor="#FF0000"><font face="Verdana, Arial, Helvetica, sans-serif" color="#FFFFFF"><b><font size="1">FREE DELIVERY</font></b></font></td></tr>
-<tr><td bgcolor="#99FF99"><table cellspacing="0" cellpadding="0" border="0">
-<tr><td bgcolor="#99FF99"><form action="http://r.hotwired.com/r/hw_wm_r_nav_nwsltr/http://perl.hotwired.com/massmail/cgiParser.cgi" method="get" target="_top">
+<tr><td bgcolor="#99FF99"><table cellspacing="0" cellpadding="0" border="0"><tr>
+<td bgcolor="#99FF99"><form action="http://r.hotwired.com/r/hw_wm_r_nav_nwsltr/http://perl.hotwired.com/massmail/cgiParser.cgi" method="get" target="_top">
<input type="hidden" name="success_page" value="http://www.hotwired.com/email/signup/wirednews-ascii.html">
<input type="hidden" name="failure_page" value="http://www.hotwired.com/email/signup/wirednews-ascii.html">
<input type="hidden" name="LIST" value="wn_ascii">
<input type="hidden" name="SOURCE" value="other">
<input type="hidden" name="ACTION" value="subscribe">
<input type="TEXT" name="from" size="10" value="enter email">
-
- <td valign="top" bgcolor="#99FF99"><input type="SUBMIT" name="SUBMIT" value="GO"></td>
-</form></td></tr>
+</form></td>
+<td valign="top" bgcolor="#99FF99"><input type="SUBMIT" name="SUBMIT" value="GO"></td>
+</tr></table></td></tr>
<tr><td bgcolor="#FF0000"><font face="Verdana, Arial, Helvetica, sans-serif" color="#FFFFFF"><b><font size="1">STOCKS</font></b></font></td></tr>
<tr><td bgcolor="#99FF99"><font face="Verdana, Arial, Helvetica, sans-serif" size="1">Get Quote:</font></td></tr>
<tr><td bgcolor="#99FF99" marginwidth="0" marginheight="0"><form method="get" action="http://r.wired.com/r/10020/http://stocks.wired.com/stocks_quotes.asp">
@@ -164,8 +164,7 @@
</font></td></tr>
<tr><td bgcolor="#99FF99"><font size="1" face="Verdana, Arial, Helvetica, sans-serif" color="#000000"><a href="http://redirect.wired.com/redir/53/http://stocks.wired.com/stocks_portfolios.asp">Portfolios</a></font></td></tr>
<!-- BEGIN B&N spot --><tr><td bgcolor="#FF0000"><font size="1" face="Verdana, Arial, Helvetica, sans-serif" color="#FFFFFF"><b>FIND A BOOK</b></font></td></tr>
-<tr><td bgcolor="#CCFFCC">
-<table cellspacing="0" cellpadding="0" border="0" width="145">
+<tr><td bgcolor="#CCFFCC"><table cellspacing="0" cellpadding="0" border="0" width="145">
<tr><td bgcolor="#CCFFCC"><form action="http://r.wired.com/r/wn_nav_c_bn/http://barnesandnoble.bfast.com/booklink/click">
<input type="hidden" name="sourceid" value="383471">
<input type="hidden" name="categoryid" value="categorydropdown">
@@ -207,6 +206,7 @@
</option>
</select></font>
+</form></td></tr>
<tr align="left" valign="top"><td valign="top" bgcolor="#CCFFCC">
<input type="submit" value="GO">
<img src="http://barnesandnoble.bfast.com/booklink/serve?sourceid=383471&is_search=Y" border="0" align="top">
@@ -219,7 +219,7 @@
</font>
<br clear="all">
</p></td></tr>
-</form></td></tr>
+</table></td></tr>
<!-- END B&N spot --><!-- BEGIN MAGAZINE SPOT --><tr><td bgcolor="#000000"><font color="#FFFFFF" face="Verdana, Arial, Helvetica, sans-serif" size="1"><b>WIRED
MAGAZINE </b></font></td></tr>
<tr><td bgcolor="#FFFF99" align="CENTER"><font face="verdana, arial, helvetica, sans-serif" size="1">
@@ -629,8 +629,5 @@
</tr>
</table>
<br>
-</td></tr>
-</table></td></tr>
-</table>
</body>
</html>
diff --git a/result/HTML/wired.html.err b/result/HTML/wired.html.err
index ecf58e8..82415e0 100644
--- a/result/HTML/wired.html.err
+++ b/result/HTML/wired.html.err
@@ -181,27 +181,15 @@
./test/HTML/wired.html:97: error: htmlParseEntityRef: expecting ';'
lue="http://www.hotbot.com/?SM=MC&DV=0&LG=any&RD=RG&DC=10&DE=2&_v=2&OPs=MDRTP&M
^
-./test/HTML/wired.html:165: error: Opening and ending tag mismatch: td and form
-</td>
- ^
-./test/HTML/wired.html:170: error: Opening and ending tag mismatch: tr and form
+./test/HTML/wired.html:170: error: Unexpected end tag : form
</tr> </form>
- ^
-./test/HTML/wired.html:171: error: Opening and ending tag mismatch: table and td
-</table></td>
- ^
-./test/HTML/wired.html:244: error: Opening and ending tag mismatch: td and form
- </select></font></td></tr>
- ^
-./test/HTML/wired.html:244: error: Opening and ending tag mismatch: tr and form
- </select></font></td></tr>
- ^
+ ^
./test/HTML/wired.html:248: error: htmlParseEntityRef: expecting ';'
MG SRC="http://barnesandnoble.bfast.com/booklink/serve?sourceid=383471&is_searc
^
-./test/HTML/wired.html:266: error: Opening and ending tag mismatch: table and td
- </table>
- ^
+./test/HTML/wired.html:265: error: Unexpected end tag : form
+ </tr> </form>
+ ^
./test/HTML/wired.html:346: error: Opening and ending tag mismatch: td and font
</td>
^
diff --git a/result/HTML/wired.html.sax b/result/HTML/wired.html.sax
index b90ae2d..0edd57e 100644
--- a/result/HTML/wired.html.sax
+++ b/result/HTML/wired.html.sax
@@ -778,7 +778,8 @@
SAX.characters( , 2)
SAX.characters(
, 1)
-SAX.error: Opening and ending tag mismatch: td and form
+SAX.endElement(form)
+SAX.endElement(td)
SAX.characters(
, 4)
SAX.startElement(td, valign='top', bgcolor='#99FF99')
@@ -792,12 +793,12 @@
SAX.endElement(td)
SAX.characters(
, 2)
-SAX.error: Opening and ending tag mismatch: tr and form
+SAX.endElement(tr)
SAX.characters( , 4)
-SAX.endElement(form)
+SAX.error: Unexpected end tag : form
SAX.characters(
, 1)
-SAX.error: Opening and ending tag mismatch: table and td
+SAX.endElement(table)
SAX.endElement(td)
SAX.characters(
, 3)
@@ -1074,8 +1075,9 @@
SAX.endElement(option)
SAX.endElement(select)
SAX.endElement(font)
-SAX.error: Opening and ending tag mismatch: td and form
-SAX.error: Opening and ending tag mismatch: tr and form
+SAX.endElement(form)
+SAX.endElement(td)
+SAX.endElement(tr)
SAX.characters(
, 2)
SAX.startElement(tr, align='left', valign='top')
@@ -1137,10 +1139,10 @@
, 9)
SAX.endElement(tr)
SAX.characters( , 2)
-SAX.endElement(form)
+SAX.error: Unexpected end tag : form
SAX.characters(
, 9)
-SAX.error: Opening and ending tag mismatch: table and td
+SAX.endElement(table)
SAX.characters(
, 2)
@@ -1461,14 +1463,14 @@
, 2)
SAX.endElement(table)
-SAX.characters(
+SAX.ignorableWhitespace(
, 2)
SAX.comment( end lower left side Navigation )
-SAX.characters(
+SAX.ignorableWhitespace(
, 1)
SAX.comment( CONTENT TABLE )
-SAX.characters(
+SAX.ignorableWhitespace(
, 2)
SAX.startElement(table, border='0', width='447', cellspacing='0', cellpadding='0', bordercolor='#66FF00')
@@ -2828,20 +2830,14 @@
SAX.characters(
, 1)
SAX.endElement(table)
-SAX.characters(
+SAX.ignorableWhitespace(
, 3)
SAX.startElement(br)
SAX.endElement(br)
-SAX.characters(
+SAX.ignorableWhitespace(
, 1)
-SAX.endElement(td)
-SAX.endElement(tr)
-SAX.endElement(table)
-SAX.endElement(td)
-SAX.endElement(tr)
-SAX.endElement(table)
SAX.endElement(body)
SAX.ignorableWhitespace(
, 1)