Opened 13 years ago
Closed 13 years ago
#3793 closed defect (fixed)
make VRT detection less fragile
Reported by: | sderle | Owned by: | warmerdam |
---|---|---|---|
Priority: | normal | Milestone: | 1.8.0 |
Component: | GDAL_Raster | Version: | svn-trunk |
Severity: | normal | Keywords: | |
Cc: |
Description
18:55 < Schuyler> why would GDAL refuse to recognize a VRT file that I generated programmatically? 18:56 < Schuyler> xmllint claims that it's clean XML 18:56 < FrankW> Because it is improperly formed, and doesn't start with <VRTRasterDataset> 18:56 < FrankW> err "<VRTDataset" 18:57 < Schuyler> so VRT isn't really XML? 18:57 < Schuyler> does indentation matter? 18:57 < FrankW> The VRT recognition stuff looks for <VRTDataset to be the first first characters. 18:57 < FrankW> if( (poOpenInfo->nHeaderBytes > 20 18:57 < FrankW> && EQUALN((const char *)poOpenInfo->pabyHeader,"<VRTDataset",11)) ) 18:57 < FrankW> return TRUE; 18:58 < FrankW> So it is a peculiar dependency.; 18:58 < FrankW> you likely have an encoding line first? 18:58 < Schuyler> ahh. 18:58 < Schuyler> I had a blank line first. 18:58 < FrankW> BTW, if you file a ticket I can make the format identification logic less fragile. 18:58 < Schuyler> thank you, Frank, that was it. 18:58 < Schuyler> *nod* 18:58 < Schuyler> how would you go about doing that? 18:59 < FrankW> I'd like just search for <VRTDataset anywhere in the first few hundred characters.
Note:
See TracTickets
for help on using tickets.
Fixed in r20944. It can accept <?xml version="1.0"?> prolog for example.