[LUCENE-5191] SimpleHTMLEncoder in Highlighter module breaks Unicode outside BMP

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 4.5, 6.0
• Component/s: modules/highlighter
• Lucene Fields: New
• Assignee: Uwe Schindler
• Reporter: Uwe Schindler
• Created: 28/Aug/13 23:04
• Updated: 28/Aug/22 13:52
• Resolved: 29/Aug/13 22:10

Description

The highlighter provides a function to escape HTML, which does to much. To create valid HTML only ", <, >, & must be escaped, everything else can kept unescaped. The escaper unfortunately does also additionally escape everything > 127, which is unneeded if your web site has the correct encoding. It also produces huge amounts of HTML entities if used with eastern languages.

This would not be a bugf if the escaping would be correct, but it isn't, it escapes like that:

result.append("\&#").append((int)ch).append(";");

So it escapes not (as HTML needs) the unicode codepoint, instead it escapes the UTF-16 char, which is incorrect, e.g. for our all-time favourite Deseret:

U+10400 (deseret capital letter long i) would be escaped as &#55297;&#56320; and not as &#66560;.

So we should remove the stupid encoding of chars > 127 which is simply useless

See also: https://github.com/elasticsearch/elasticsearch/issues/3587