SAX and invalid chars

[Christian]

Hello

My Problem is that I have to parse an XML file that contains som invalid
chars (i.e. 0x0E or 0x1E)

So running this normally will break parsing.
Though easy solution I could think of would be create a stream to pipe
the input through an filter lower bytes out.
The problem is that if my XML is not in windows-1252 but some other char
encoding I might break encoding by this.

Is there any patent solution to the problem?


Christian

 

[Arne Vajhøj]

My Problem is that I have to parse an XML file that contains som invalid
chars (i.e. 0x0E or 0x1E)

So running this normally will break parsing.
Though easy solution I could think of would be create a stream to pipe
the input through an filter lower bytes out.
The problem is that if my XML is not in windows-1252 but some other char
encoding I might break encoding by this.

Is there any patent solution to the problem?

Do the same as the XML parser.

Read the XML header and get encoding from there.

Arne

 

[Mike Schilling]

Do the same as the XML parser.

Read the XML header and get encoding from there.

Which is easy if you know that the XML file is in some superset of
ASCII, since the entrie XML header will then be in ASCII. It's
tricker if the XML file might be in any encoding at all (e.g. EBCDIC,
UTF-16, etc.) In the latter case, look at Appendix F
(http://www.w3.org/TR/REC-xml/#sec-guessing) for some useful tips.

 

[Christian]

Which is easy if you know that the XML file is in some superset of
ASCII, since the entrie XML header will then be in ASCII. It's
tricker if the XML file might be in any encoding at all (e.g. EBCDIC,
UTF-16, etc.) In the latter case, look at Appendix F
(http://www.w3.org/TR/REC-xml/#sec-guessing) for some useful tips.

Thx for your pointers..

Though the solution seems to be to heavy ... and as I am only expecting
utf-8 and windows-1252 I probably do with the hack of just removing the
bytes ... (and search the api now if there is some easy way to throw an
exception if none of these encodings are used..)

thx