lib/libxml2
1
0
Fork 0
mirror of https://gitlab.gnome.org/GNOME/libxml2.git synced 2025-07-29 11:41:22 +03:00
Code Activity

html: HTML5 character data states

Browse Source
This commit is contained in:
Nick Wellnhofer
2024-09-06 17:49:04 +02:00
parent 5951179239
commit f9ed30e972
8 changed files with 463 additions and 381 deletions
Download Patch File Download Diff File Expand all files Collapse all files

765
HTMLparser.c
View File

File diff suppressed because it is too large Load Diff

2
include/libxml/HTMLparser.h
View File

@ -67,6 +67,8 @@ struct _htmlElemDesc {
const char** attrs_opt; /* Optional Attributes */ const char** attrs_opt; /* Optional Attributes */
const char** attrs_depr; /* Additional deprecated attributes */ const char** attrs_depr; /* Additional deprecated attributes */
const char** attrs_req; /* Required attributes */ const char** attrs_req; /* Required attributes */
int dataMode;
}; };
/* /*

3
result/HTML/doc2.htm
View File

@ -22,7 +22,8 @@
</frameset> </frameset>
<noframes> <noframes>
<body bgcolor="#FFFFFF" text="#000000" link="#000080" vlink="#000080" alink="#000080" topmargin="0" leftmargin="0" marginheight="0" marginwidth="0"> <body bgcolor="#FFFFFF" text="#000000" link="#000080" vlink="#000080" alink="#000080"
topmargin="0" leftmargin="0" marginheight="0" marginwidth="0">
<p>This page uses frames, but your browser doesn't support them.</p> <p>This page uses frames, but your browser doesn't support them.</p>
</body> </body>
</noframes> </noframes>

15
result/HTML/doc2.htm.sax
View File

@ -47,20 +47,9 @@ SAX.startElement(frame, name='rbottom', noresize, src='doc2_files/contents.htm',
SAX.endElement(frame) SAX.endElement(frame)
SAX.endElement(frameset) SAX.endElement(frameset)
SAX.startElement(noframes) SAX.startElement(noframes)
SAX.characters( SAX.cdata(
, 4) &lt;body bgcolor="#FFFFFF", 238)
SAX.startElement(body, bgcolor='#FFFFFF', text='#000000', link='#000080', vlink='#000080', alink='#000080', topmargin='0', leftmargin='0', marginheight='0', marginwidth='0')
SAX.characters(
, 3)
SAX.startElement(p)
SAX.characters(This page uses frames, but you, 61)
SAX.endElement(p)
SAX.characters(
, 3)
SAX.endElement(body)
SAX.characters(
, 3)
SAX.endElement(noframes) SAX.endElement(noframes)
SAX.endElement(frameset) SAX.endElement(frameset)
SAX.endElement(html) SAX.endElement(html)

11
result/HTML/doc3.htm
View File

@ -47,7 +47,7 @@ eval("page" + id + " = window.open(URL, '" + id + "', 'toolbars=0, scrollbars=0,
<td valign="top" width="31"><a href="http://bp6.gamesquad.net/"><img align="bottom" border="0" height="74" src="doc3_files/logo.gif" width="252"></a></td> <td valign="top" width="31"><a href="http://bp6.gamesquad.net/"><img align="bottom" border="0" height="74" src="doc3_files/logo.gif" width="252"></a></td>
<td align="left" bgcolor="#000000"> <td align="left" bgcolor="#000000">
<img height="15" src="doc3_files/spacer.gif" width="15"><!-- START GAMESQUAD.NET IFRAME RICH MEDIA CODE --> <!-- <20> 2000 GameSquad.net All Rights Reserved. --><iframe border="0" frameborder="no" height="60" marginheight="0" marginwidth="0" scrolling="no" src="doc3_files/adcycle.htm" width="468"> <img height="15" src="doc3_files/spacer.gif" width="15"><!-- START GAMESQUAD.NET IFRAME RICH MEDIA CODE --> <!-- <20> 2000 GameSquad.net All Rights Reserved. --><iframe border="0" frameborder="no" height="60" marginheight="0" marginwidth="0" scrolling="no" src="doc3_files/adcycle.htm" width="468">
<a href="http://ads.gamesquad.net/addclick.exe/adclick.cgi?REGION=game|tech|ent&amp;id=1" target="_top"><img src="http://ads.gamesquad.net/addclick.exe/adcycle.cgi?group=52&amp;media=1&amp;id=1" width="468" height="60" border="0" alt="GSN ROS Ad"></a> <a href="http://ads.gamesquad.net/addclick.exe/adclick.cgi?REGION=game|tech|ent&id=1" target="_top"><img src="http://ads.gamesquad.net/addclick.exe/adcycle.cgi?group=52&media=1&id=1" width=468 height=60 border=0 ALT="GSN ROS Ad"></a>
</iframe><!-- END GAMESQUAD.NET IFRAME RICH MEDIA CODE --><br><img height="15" src="doc3_files/spacer.gif" width="400"> </td> </iframe><!-- END GAMESQUAD.NET IFRAME RICH MEDIA CODE --><br><img height="15" src="doc3_files/spacer.gif" width="400"> </td>
</tr> </tr>
<tr> <tr>
@ -454,7 +454,7 @@ eval("page" + id + " = window.open(URL, '" + id + "', 'toolbars=0, scrollbars=0,
may send me some news to be posted if you find any ( we don't want may send me some news to be posted if you find any ( we don't want
to flood Tim ;-) ).<br><br>My e-mail address is <a href="mailto:killz@i82hq.com">killz@i82hq.com</a><br><br>Ciao for to flood Tim ;-) ).<br><br>My e-mail address is <a href="mailto:killz@i82hq.com">killz@i82hq.com</a><br><br>Ciao for
now.<br><br>The <20><>.</font><br><br> now.<br><br>The <20><>.</font><br><br>
</font><center><iframe frameborder="0" height="60" marginheight="0" marginwidth="0" noresize scrolling="no" src="doc3_files/ad_iframe.htm" width="468"><a href="http://ads.adflight.com/go_static.asp?asid=7708" target="_top"><img width="468" height="60" border="0" alt="Advertisement" src="http://ads.adflight.com/ad_static.asp?pid=2097&amp;sid=1881&amp;asid=7708"></a></iframe></center> </font><center><iframe frameborder="0" height="60" marginheight="0" marginwidth="0" noresize scrolling="no" src="doc3_files/ad_iframe.htm" width="468"><a href="http://ads.adflight.com/go_static.asp?asid=7708" target="_top"><img width=468 height=60 border=0 alt="Advertisement" src="http://ads.adflight.com/ad_static.asp?pid=2097&sid=1881&asid=7708"></a></iframe></center>
</td> </td>
</tr> </tr>
</tbody> </tbody>
@ -512,7 +512,7 @@ eval("page" + id + " = window.open(URL, '" + id + "', 'toolbars=0, scrollbars=0,
document.write("SRC=http://www.goto.com/d/ssn/dynconsole/?t"); document.write("SRC=http://www.goto.com/d/ssn/dynconsole/?t");
document.write("ype=html&size=100x90&url=http://www.goto.co"); document.write("ype=html&size=100x90&url=http://www.goto.co");
document.write("m/d/search/ssn/&target=_blank&Partner=SSN80"); document.write("m/d/search/ssn/&target=_blank&Partner=SSN80");
document.write("42DF8478957377>"); document.write("42DF8478957377></IFRAME>");
} else if ((parseInt(navigator.appVersion) > 3) } else if ((parseInt(navigator.appVersion) > 3)
&& (navigator.appName == "Netscape")) { && (navigator.appName == "Netscape")) {
document.write("<SCRIPT language=javascript type=text/javas"); document.write("<SCRIPT language=javascript type=text/javas");
@ -520,14 +520,15 @@ eval("page" + id + " = window.open(URL, '" + id + "', 'toolbars=0, scrollbars=0,
document.write("SRC=http://www.goto.com/d/ssn/dynconsole/?t"); document.write("SRC=http://www.goto.com/d/ssn/dynconsole/?t");
document.write("ype=js&size=100x90&url=http://www.goto.com/"); document.write("ype=js&size=100x90&url=http://www.goto.com/");
document.write("d/search/ssn/&target=_blank&Partner=SSN8042"); document.write("d/search/ssn/&target=_blank&Partner=SSN8042");
document.write("DF8478957377>"); document.write("DF8478957377></SC");
document.write("RIPT>");
} else { } else {
document.write("<A TARGET=_blank "); document.write("<A TARGET=_blank ");
document.write("HREF=http://www.goto.com/d/search/ssn/?from"); document.write("HREF=http://www.goto.com/d/search/ssn/?from");
document.write("GIF=true>"); document.write("GIF=true>");
document.write("<IMG ismap "); document.write("<IMG ismap ");
document.write("SRC=http://www.goto.com/d/ssn/dynconsole/?t"); document.write("SRC=http://www.goto.com/d/ssn/dynconsole/?t");
document.write("ype=gif&size=100x90>"); document.write("ype=gif&size=100x90></A>");
} }
// --> // -->
</script> </script>

9
result/HTML/doc3.htm.err
View File

@ -22,15 +22,6 @@ om/ad_static.asp?pid=2097&sid=1881&asid=7708"></a></IFRAME></CENTER></LI></FONT>
./test/HTML/doc3.htm:772: HTML parser error : Unexpected end tag : form ./test/HTML/doc3.htm:772: HTML parser error : Unexpected end tag : form
archive</A></FONT> </FORM></CENTER></TD></TR></TBODY></TABLE><!-- archive</A></FONT> </FORM></CENTER></TD></TR></TBODY></TABLE><!--
^ ^
./test/HTML/doc3.htm:795: HTML parser error : Unexpected end tag : iframe
document.write("42DF8478957377></IFRAME>");
^
./test/HTML/doc3.htm:804: HTML parser error : Unexpected end tag : sc");
document.write("RIPT>");
^
./test/HTML/doc3.htm:811: HTML parser error : Unexpected end tag : a
document.write("ype=gif&size=100x90></A>");
^
./test/HTML/doc3.htm:820: HTML parser error : Unexpected end tag : a ./test/HTML/doc3.htm:820: HTML parser error : Unexpected end tag : a
</A></A></B><B></NOSCRIPT></B><B><!-- END GoTo.com Search Box --></B </A></A></B><B></NOSCRIPT></B><B><!-- END GoTo.com Search Box --></B
^ ^

30
result/HTML/doc3.htm.sax
View File

@ -83,14 +83,8 @@ SAX.comment( START GAMESQUAD.NET IFRAME RICH MEDIA CODE )
SAX.characters( , 1) SAX.characters( , 1)
SAX.comment( © 2000 GameSquad.net All Rights Reserved. ) SAX.comment( © 2000 GameSquad.net All Rights Reserved. )
SAX.startElement(iframe, border='0', frameborder='no', height='60', marginheight='0', marginwidth='0', scrolling='no', src='doc3_files/adcycle.htm', width='468') SAX.startElement(iframe, border='0', frameborder='no', height='60', marginheight='0', marginwidth='0', scrolling='no', src='doc3_files/adcycle.htm', width='468')
SAX.characters( SAX.cdata(
, 1) &lt;a href="http://ads.gamesq, 235)
SAX.startElement(a, href='http://ads.gamesquad.net/addclick.exe/adclick.cgi?REGION=game|tech|ent&amp;id=1', target='_top')
SAX.startElement(img, src='http://ads.gamesquad.net/addclick.exe/adcycle.cgi?group=52&amp;media=1&amp;id=1', width='468', height='60', border='0', alt='GSN ROS Ad')
SAX.endElement(img)
SAX.endElement(a)
SAX.characters(
, 1)
SAX.endElement(iframe) SAX.endElement(iframe)
SAX.comment( END GAMESQUAD.NET IFRAME RICH MEDIA CODE ) SAX.comment( END GAMESQUAD.NET IFRAME RICH MEDIA CODE )
SAX.startElement(br) SAX.startElement(br)
@ -2563,10 +2557,7 @@ SAX.characters(
SAX.characters( SAX.characters(
, 14) , 14)
SAX.endElement(font) SAX.endElement(font)
SAX.startElement(center) SAX.startElement(center)
SAX.startElement(iframe, frameborder='0', height='60', marginheight='0', marginwidth='0', noresize, scrolling='no', src='doc3_files/ad_iframe.htm', width='468')
SAX.startElement(a, href='http://ads.adflight.com/go_static.asp?asid=7708', target='_top')
SAX.startElement(img, width='468', height='60', border='0', alt='Advertisement', src='http://ads.adflight.com/ad_static.asp?pid=2097&amp;sid=1881&amp;asid=7708')
SAX.startElement(iframe, frameborder='0', height='60', marginheight='0', marginwidth='0', noresize, scrolling='no', src='doc3_files/ad_iframe.htm', width='468') SAX.startElement(iframe, frameborder='0', height='60', marginheight='0', marginwidth='0', noresize, scrolling='no', src='doc3_files/ad_iframe.htm', width='468')
SAX.cdata(&lt;a href="http://ads.adfligh, 202) SAX.cdata(&lt;a href="http://ads.adfligh, 202)
SAX.endElement(iframe) SAX.endElement(iframe)
@ -2687,18 +2678,9 @@ SAX.characters(
SAX.characters( SAX.characters(
, 14) , 14)
SAX.startElement(script, language='javascript', type='text/javascript') SAX.startElement(script, language='javascript', type='text/javascript')
SAX.cdata( SAX.cdata(
&lt;!-- &lt;!--
if ((, 532) if ((, 1000)
SAX.error: Unexpected end tag : iframe
SAX.cdata(");
} else if ((parseI, 463)
SAX.error: Unexpected end tag : sc");
SAX.cdata(");
} else {
d, 328)
SAX.error: Unexpected end tag : a
SAX.cdata(");
SAX.cdata(377&gt;&lt;/SC"); SAX.cdata(377&gt;&lt;/SC");
docu, 410) docu, 410)
SAX.endElement(script) SAX.endElement(script)

9
runtest.c
View File

@ -2024,8 +2024,13 @@ pushBoundaryTest(const char *filename, const char *result,
} else if (isText) { } else if (isText) {
int c = *ctxt->input->cur; int c = *ctxt->input->cur;
/* 3 bytes for partial UTF-8 */ if ((options & XML_PARSE_HTML) &&
max = ((c == '<') || (c == '&')) ? 1 : 3; (ctxt->endCheckState)) {
max = strlen((const char *) ctxt->name) + 2;
} else {
/* 3 bytes for partial UTF-8 */
max = ((c == '<') || (c == '&')) ? 1 : 3;
}
} else if (ctxt->instate == XML_PARSER_CDATA_SECTION) { } else if (ctxt->instate == XML_PARSER_CDATA_SECTION) {
/* 2 bytes for terminator, 3 bytes for UTF-8 */ /* 2 bytes for terminator, 3 bytes for UTF-8 */
max = 5; max = 5;