Re: commit: New method to repair importer mistakes.

>
>>
>> Martin,
>>
>> the new function prevents us to import quite a lot Word documents.
>> See e.g. bugzilla #8474.
>> We add lots of bogus empty blocks <p></p> now, e.g. in a table between
>> the cells.
>> You can easily see this by running abiword -t abw 8474.doc
>> and inspection of the resulting abw file.
>>
>> As this is a major regression in 2.2.4, maybe we should back this
>> out.
>>
>
> HI Robert,
> Thanks very much for working this. I will fix it right away.
>
> Martin
>

CVS: ----------------------------------------------------------------------
CVS: Enter Log. Lines beginning with `CVS:' are removed automatically
CVS:
CVS: Committing in .
CVS:
CVS: Modified Files:
CVS: src/text/fmt/xp/fv_View.cpp src/text/ptbl/xp/pd_Document.cpp
CVS: src/text/ptbl/xp/pd_Document.h
CVS: ----------------------------------------------------------------------
Fix horrible regression on document import.

This should be backported as soon it tests OK. I'll do it myslef by this
evening.

Sorry.

Martin

>> Best regards,
>>
>> Robert
>>
>> On Wed, 2005-02-09 at 00:47, msevior@physics.unimelb.edu.au wrote:
>>> CVS:
>>> ----------------------------------------------------------------------
>>> CVS: Enter Log. Lines beginning with `CVS:' are removed automatically
>>> CVS:
>>> CVS: Committing in .
>>> CVS:
>>> CVS: Modified Files:
>>> CVS: src/text/ptbl/xp/pd_Document.cpp
>>> CVS: src/text/ptbl/xp/pd_Document.h
>>> CVS:
>>> ----------------------------------------------------------------------
>>> New method to stop repair import mistakes. "appendBlockIfNeeded"
>>>
>>> This should be BAKCPORTED.
>>>
>>> This method simply looks at the last frag in the document. If it's a
>>> strux
>>> and it's not a Block, EndFootnote or EndEndnote, this method appends a
>>> block onto the PT.
>>>
>>> This method is called in appendSpan(..), appendObject(...),
>>> appendFmt(...)
>>> before ther span, object or fmt is appended onto the PT.
>>>
>>> This method will therefore correct the crashes we get when an importer
>>> screws up something and puts some content after a non-block strux.
>>>
>>> The cost is trivial, we simply grab a pointer and examine it.
>>>
>>> This method fixes a crash I got importing an RTF file because of an
>>> open
>>> Frame strux.
>>>
>>> Dom suggested we do something like this ages ago and we should have.
>>>
>>> Anyway, better late than never.
>>>
>>> Cheers
>>>
>>> Martin
>>>
>>
>>
>
>
Received on Mon Feb 28 01:32:47 2005