Issue1335

classification
Title: html5lib trunk won't compile due to Jython Unicode pickiness
Type: Severity: normal
Components: Core Versions: 25rc4
process
Status: closed Resolution: wont fix
Dependencies: Superseder:
Assigned To: zyasoft Nosy List: dmbaggett, zyasoft
Priority: Keywords:

Created on 2009-05-01.13:12:39 by dmbaggett, last changed 2009-05-30.00:45:58 by zyasoft.

Messages
msg4625 (view) Author: Dave Baggett (dmbaggett) Date: 2009-05-01.13:12:36
This line of code from the html5lib trunk file inputstream.py:

invalid_unicode_re =
re.compile(u"[\u0001-\u0008\u000B\u000E-\u001F\u007F-\u009F\uD800-\uDFFF\uFDD0-\uFDEF\uFFFE\uFFFF\U0001FFFE\U0001FFFF\U0002FFFE\U0002FFFF\U0003FFFE\U0003FFFF\U0004FFFE\U0004FFFF\U0\
005FFFE\U0005FFFF\U0006FFFE\U0006FFFF\U0007FFFE\U0007FFFF\U0008FFFE\U0008FFFF\U0009FFFE\U0009FFFF\U000AFFFE\U000AFFFF\U000BFFFE\U000BFFFF\U000CFFFE\U000CFFFF\U000DFFFE\U000DFFFF\U000EFFFE\U000EFFFF\U00\
0FFFFE\U000FFFFF\U0010FFFE\U0010FFFF]")  

won't compile under Jython 2.5b3:

Sorry: UnicodeDecodeError: ('unicodeescape',
'u"[\\u0001-\\u0008\\u000B\\u000E-\\u001F\\u007F-\\u009F\\uD800-\\uDFFF\\uFDD0-\\uFDEF\\uFFFE\\uFFFF\\U0001FFFE\\U0001FFFF\\U0002FFFE\\U0002FFFF\\U0003FFFE\\U0003FFFF\\U0004FFFE\\U0004FFFF\\U0005FFFE\\U0005FFFF\\U0006FFFE\\U0006FFFF\\U0007FFFE\\U0007FFFF\\U0008FFFE\\U0008FFFF\\U0009FFFE\\U0009FFFF\\U000AFFFE\\U000AFFFF\\U000BFFFE\\U000BFFFF\\U000CFFFE\\U000CFFFF\\U000DFFFE\\U000DFFFF\\U000EFFFE\\U000EFFFF\\U000FFFFE\\U000FFFFF\\U0010FFFE\\U0010FFFF]"',
48, 55, 'illegal Unicode character')

It looks like Jython (via Java) is enforcing valid unicode in the
literal while standard Python is not.
msg4761 (view) Author: Jim Baker (zyasoft) Date: 2009-05-30.00:45:57
This is a fundamental design decision: we do not allow for isolated half 
surrogates in Jython, since we use the same underlying representation as 
Java, UTF-16, for our unicode strings. In Jython, unicode is just a 
wrapper around java.lang.String.

Wikipedia succinctly describes the issue here: "All possible code points 
from U+0000 through U+10FFFF, except for the surrogate code points 
U+D800–U+DFFF (which are not characters), are uniquely mapped by UTF-16 
regardless of the code point's current or future character assignment or 
use." (http://en.wikipedia.org/wiki/UTF-16).

So the workaround is to special case for Jython \uD8000-\uDFFF, instead 
of using a regex as in msg4625.

Similar considerations would apply for other Unicode usage in CPython, 
notably UCS2 vs UCS4.

A similar problem was seen in Pygments, http://dev.pocoo.org/projects/pygments/ticket/358
History
Date User Action Args
2009-05-30 00:45:58zyasoftsetstatus: open -> closed
resolution: wont fix
messages: + msg4761
2009-05-29 01:42:19pjenveysetassignee: zyasoft
nosy: + zyasoft
2009-05-01 13:12:39dmbaggettcreate