Thanks Andi! I'm working towards that direction. I'm wondering how the checksum is calculated. That's the last two bytes in the header...
Thanks for helping! Bing 2014-05-29 4:04 GMT+08:00 Andreas Beeker <andreas.bee...@gmx.de>: > Hi Bing, > > the wmf code is in a branch, because I'd like to commit (save) it without > interfering with the trunk ... and it's still far from being finished. > Not sure, how the github synchronisation works, but I guess "it" only > fetches the trunk. > > And for the header: I couldn't find the reference on a quick search, but - > I think - the wmf pictures are > always saved without the placeable header. At least when I've > implemented/tested the wmf code in HssfWorkbook.addPicture, it only worked, > when I've removed that header - maybe it's different with hwpf ... > So I guess JWord is simply making a header up - either it uses standard > values for the bounding box or it reads one of the window records [1]. > > Anyways, I would simply try to append the 22 bytes of a working picture to > the pictures with the missing magic code 0x9AC6CDD7 (little endian) and see > if your post-processing is ok with that. > > Andi > > > [1] http://svn.apache.org/repos/asf/poi/branches/wmf_render/ > src/scratchpad/src/org/apache/poi/hwmf/record/WmfWindowing.java > > > On 28.05.2014 18:34, Bing Ran wrote: > >> Hi Andreas, >> >> Thanks for answer. >> >> The raw data was acquired from overriding the >> AbstractWordConverter.processingImage()... >> in the hwpf package, by calling picture.getContent(). I cannot >> immediately figure >> out how to reset the header after reading your code reference. >> >> BTW, I was using a local compile of the POI modules from Github. Is the >> code considered out of date? I could not find the hwmf package in the >> github code. >> >> Thanks >> >> Bing >> >> > > --------------------------------------------------------------------- > To unsubscribe, e-mail: user-unsubscr...@poi.apache.org > For additional commands, e-mail: user-h...@poi.apache.org > >
