Project

General

Profile

Actions

Bug #1538

closed

Entity/Character Refeference Conversion Problems

Added by Dan Higgins over 20 years ago. Updated almost 20 years ago.

Status:
Resolved
Priority:
Immediate
Assignee:
Category:
metacat
Target version:
Start date:
04/28/2004
Due date:
% Done:

0%

Estimated time:
Bugzilla-Id:
1538

Description

In Morpho, we have run into some problem with the use of special, non-ascii
characters (like the 'degree' symbol or greek 'mu'). [Any character represented
by a byte with a decimal value > 127 is in this class of special characters.]
These characters have been copied from Word or PDF documents into Morpho fields
and then put into eml xml docs. Unfortunately, they are not not necessarily in
the correct format for xml documents and have caused parser problems.

The solution that was implemented in Morpho was to use entity/character
references. Any character with a value greater than 127 is written as '&#xxx;'
where 'xxx' is the decimal value of the character. On a Windows machine, the
'deg' symbol becomes '°' and 'mu' becomes 'μ'. XML parsers
automatically convert these character entities to the character for display, but
the conversion depends on the assumed character set.

The metacat problem is that when one submits a document containing such
character references (&#xxx;) and then reads the document back, one does not get
the character refenence, but rather the character itself! I assume this is due
to the XML parser. This is a violation of the idea that metacat should return
exactly the same data given it.

Morpho already handles this by converting back to character references any info
sent it by Metacat with character values greater than 127. But metacat actually
sends back the wrong character for some symbols! (e.g. a 'mu' becomes a '1/4'
symbol. I assume this is due to different character set assumption under linux
and windows. In any case, there is some data corruption here that we should
figureout how to avoid.

Actions #1

Updated by Matt Jones almost 20 years ago

Its not clear to me that we are handling non-ascii characters properly.
Internet Explorer and some other applications reject several documents that are
in metacat as invalid because of these non-ascii characters. We need to
determine exactly what the proper behavior is. We probably need to change the
xml declaration to indicate the proper character set encoding if non-ascii
characters are present. Andrea and Veronique reported several similar issues
with character problems in the most recent version of morpho (1.5.1).

Actions #2

Updated by Saurabh Garg almost 20 years ago

Changed the code of the normalize function in MetaCatUtil.java. Earlier code
was not taking care of characters above 123.
In DBSAXHandler.java, added call to normalize function before text is written
to db.

So now the data is normalized before it is stored in the DB. When the dp is
read again, the &#XXX; form is returned. It can be denormalized, if that is
considered appropriate.

Will close this bug after some more testing.

Actions #3

Updated by Saurabh Garg almost 20 years ago

Closing the bug.

Actions #4

Updated by Redmine Admin over 11 years ago

Original Bugzilla ID was 1538

Actions

Also available in: Atom PDF