[gdal-dev] Read PDF with OGR

[Even Rouault]

Le mercredi 19 octobre 2016 17:09:03, Andreas Oxenstierna a écrit :
> Hi list
> 
> I want to attach georeferencing to existing vector PDF files that lacks
> georef info.
> 
> This works without any issues using gdal:
> gdal_translate -of PDF -a_srs EPSG:3006 -a_ullr 718500 6359200 726500
> 6369100  <infile> <outfile> --config GDAL_PDF_DPI 300

gdal_translate will not attach georeferencing to existing PDF file, but it will 
create a new one, re-rasterizing its content.

Use rather gdal_edit.py : http://gdal.org/gdal_edit.html
That will just add the geospatial metadata and not change the rest of the PDF 
content.
This will also work for "vector" PDF files (although your example is already 
georeferenced)

> but all attempts with ogr fails, even if the PDF driver is listed:
> Unable to open datasource `k.pdf' with the following drivers.
>    -> PDF
> 
> I tested with https://www.nps.gov/hfc/carto/PDF/ANTImap1.pdf with the
> same result, i.e. GDAL can read the vector PDF file but not OGR
> ogrinfo --format pdf returns all available options etc.
> 

The OGR PDF driver only recognize by default very specific formulations of 
vector PDF that use the StructTreeRoot objects of the PDF specification, which 
is not the case for this sample.

There was a feature actually not yet documented. Done now :
"""
If there is no such logical structure, the driver will not try to interpret 
the vector content of the PDF, unless you defined the 
OGR_PDF_READ_NON_STRUCTURED configuration option to YES."""

So try:

ogrinfo ANTImap1.pdf --config OGR_PDF_READ_NON_STRUCTURED YES

Note that the conversion from PDF vectors to OGR geometries can be rough for 
some PDF objects.

> I would like to use ogr2ogr so vector geospatial PDF's can be created.
> Is this possible?

Yes. 
And you can combine raster + vector as explained in the tutorial at 
http://latuviitta.org/documents/Geospatial_PDF_maps_from_OSM_with_GDAL.pdf 
(last link of the PDF driver documentation page)

> The two sample links in the bottom of http://www.gdal.org/frmt_pdf.html
> are broken.

I've removed/updated them

Even

-- 
Spatialys - Geospatial professional services
http://www.spatialys.com