#367 Fault CharacterDataHandler if LF starts data

[Anonymous]

When a tag's data start with LF (0x0A), I get
XML_CharacterDataHandler with len=1 and s=0x0A
instead of the actual data that comes after the LF.

On the attached example I get the following calls to
XML_CharacterDataHandler :
- len =1, s=0x0A
- len=6, s="closed"
- len =1, s=0x0A
- len =1, s=0x0A
The last two calls for the function are problematic - I
don't get the actual data that comes after the LF.

The problem happens when the tag's data start with LF
but has more characters after the LF.

on Expat-1.95.8 created from
expat_win32bin_1_95_8.exe

[Nobody/Anonymous]

[Karl Waclawek]

Logged In: YES
user_id=290026

Your attached example does not have any LF directly
before or after the string "closed". Please clarify your problem.

[Nobody/Anonymous]

Logged In: NO

The attached example has LF (0x0A) in the following cases:
- before the text "sip:pep@example.com"
- before the text "Full state presence document"
In these 2 cases I receive XML_CharacterDataHandler with
len=1 and s=0x0A only. The function is not called for the
data AFTER the LF (In the example to "Full state presence
document").
Is there a way to overcome this problem?

[Mike Rosky]

Logged In: YES
user_id=1238831

I think that you (original sender) should examine your code,
when and for which cases you're calling chardata handler.
Please note that basically you have to (simplified) reset
chardata buffer at the start element point and accumulate
chardata value every time character data handler is invoked
until parser calls the end element handler for that element.

You can't suppose that chardata handler gets whole value
in one call - actually in your case it's called twice because of
CRLF, the first call returns CRLF (and eventually preceding
chars), the second the rest of chardatas. Chardatas can even
cross two source readings, which leads to same effect.

You can try it when you add a character data handler to
the outline.c example, where chardata handler just prints
out the current part enclosed in brackets or something.

Mike

[Karl Waclawek]

Logged In: YES
user_id=290026

Just to clarify - in Expat, a contiguous string of characters
does not necessarily have to be reported through exactly one
characterData() call-back. Often, line-breaks determine the
boundary between call-backs. In the attached example, the
character data for the <note> element will likely be reported
through three call-backs, as there are two line-breaks.

[Karl Waclawek]

Logged In: YES
user_id=290026

Closing this issue - no follow-up from poster.