HP Scan invalid Length workaround

[mfazekas]

Fixes: #191

It looks that some version of HP Scan produces invalid pdf files. Where the length value would point ot of the file.

8 0 obj
<</Length 2200
/Type
/Metadata
/Subtype 
/XML>>
stream
endstream
endobj

I've added a new relaxed: option to allow read such files.
Not sure if it's any better than #209. More complicated for sure.
Also if we'd need the new option for this?

[RBIII]

Any idea when this might merge? I added it to my local and it seems to be working

[juliolinarez]

Works for me. Nice solution

[osvaldoalvaradodev]

Any idea when they merge into the main branch of this?

[mfazekas]

@boazsegev I've now rebased the pr to current master

[boazsegev]

Does this actually solve cases where Length is invalid but still within the file's range?

Wouldn't it be better to test for endsteam and if it's not there search for it while ignoring the Length data?

[mfazekas]

@boazsegev thanks much for looking into it.

I've updated my PR:

• instead of relaxed the option name is check_stream_length
• if check_stream_length is enabled we'll check if the Length would point out of the file, or there is no following endstream. And in those cases we'll search for endstream and ignore Length. See advance_pos_with_length_check: (https://github.com/boazsegev/combine_pdf/pull/215/files#diff-aef1b3a880fd7ed0b5166faa2124139a38a4e9456a8bc1f8d8927a2a47e7abc3R516-R538)