Please describe the problem:
Parsing the attached (and valid) HTML file with libxml2's HTMLparser leads to a truncated document (try "xmllint --html utf8.html"). The same bug occurs on the corresponding iso-8859-* encoded document, too, so it isn't limited to a particular encoding. Having a look at HTMLparser.c and friends didn't really help, however I tried to narrow it down a bit.

The problem seems to be that on the first non-ASCII character read

      /*
       * Humm this is bad, do an automatic flow conversion
       */
      xmlSwitchEncoding(ctxt, XML_CHAR_ENCODING_8859_1);
      ctxt->charset = XML_CHAR_ENCODING_UTF8;
      return(xmlCurrentChar(ctxt, len));

(HTMLparser.c:386-391) is executed _before_ the <meta>-tag is parsed and

      if ((http) && (content != NULL))
          htmlCheckEncoding(ctxt, content);

(HTMLparser.c:3404-3405) is executed, which happens some characters later.

Accordingly each of the following makes the problem vanish:
* commenting out lines 3404-3405 of HTMLparser.c
* commenting out lines 389-390 of HTMLparser.c
* swapping lines 4 <-> 5 of utf8.html (which is effectively the same as the above)

Don't know if this is the right trace, any hints or ideas?

Steps to reproduce:
xmllint --html utf8.html


Actual results:
xmllint prints HTML code with truncated text between <p> and </p> tags.

Expected results:
xmllint should print the document with complete text between <p> and </p> tags.

Does this happen every time?
yes

Other information: