Extend --spreadsheet to handle invisible lines

I would be nice to support this kind of tables:

$ wget https://static.healthcare.siemens.com/siemens_hwem-hwem_ssxa_websites-context-root/wcm/idc/groups/public/@global/@services/documents/download/mdaw/mtiz/~edisp/x500_1_0_conformance_statement-00074148.pdf
$ java -jar $TABULA --spreadsheet  -p 26 x500_1_0_conformance_statement-00074148.pdf

Since this is a multiline table, the option --spreadsheet is required here. Without --spreadsheet I can extract some stuff, but it is then difficult to join cell text by hand. It would be nice if --spreadsheet would not return an empty output.

[jeremybmerrill]

Unfortunately, there are no invisible lines in this PDF, just white spaces. The “stream”/no-spreadsheet algorithm is built to use these white spaces; the spreadsheet/“lattice” algorithm needs actual line primitives in the PDF to work. There is, unfortunately, no way I know of to transform white spaces into lines in the general case into lines with sufficient confidence to use the lattice algorithm: even though our eyes can make the distinction easily, the computer can often see white lines when we know there aren’t any, e.g. “rivers”.

So, unfortunately, that PDF lacks lines, so the spreadsheet algorithm will not work on it. The “stream” mode output is the best you’re going to get. I know it’s a pain to try to combine cells, but we have not been able to come up with a general purpose method for combining them.

I’ve attached the table we’re talking about: ultrasound-26.pdf

[malaterre]

Technically I want Table 6 Ultrasound Image and Ultrasound Retired Image IOD Attributes (page 24-29), Table 7 Ultrasound MultiFrame and Ultrasound MultiFrame Retired Image IOD Attributes (page 30-36), Table 8 Secondary Capture Image IOD Attributes (page 37-41) and Table 9 Comprehensive SR IOD Attributes (page 41-43).