AttributeError: 'PDFStream' object has no attribute 'replace'

[panoptikum]

Hello everybody,

At the moment I'm parsing tons of PDFs, but pdfminer.six fails on one of them. Any suggestions? I can open the PDF, but maybe pdfminer.six can't handle it properly. All the other PDFs origin from the same author/organization...

Traceback (most recent call last):
  File "/home/felix/anaconda3/bin/pdf2txt.py", line 136, in <module>
    if __name__ == '__main__': sys.exit(main())
  File "/home/felix/anaconda3/bin/pdf2txt.py", line 131, in main
    outfp = extract_text(**vars(A))
  File "/home/felix/anaconda3/bin/pdf2txt.py", line 63, in extract_text
    pdfminer.high_level.extract_text_to_fp(fp, **locals())
  File "/home/felix/anaconda3/lib/python3.6/site-packages/pdfminer/high_level.py", line 82, in extract_text_to_fp
    interpreter.process_page(page)    
  File "/home/felix/anaconda3/lib/python3.6/site-packages/pdfminer/pdfinterp.py", line 852, in process_page
    self.render_contents(page.resources, page.contents, ctm=ctm)
  File "/home/felix/anaconda3/lib/python3.6/site-packages/pdfminer/pdfinterp.py", line 862, in render_contents
    self.init_resources(resources)
  File "/home/felix/anaconda3/lib/python3.6/site-packages/pdfminer/pdfinterp.py", line 362, in init_resources
    self.fontmap[fontid] = self.rsrcmgr.get_font(objid, spec)
  File "/home/felix/anaconda3/lib/python3.6/site-packages/pdfminer/pdfinterp.py", line 212, in get_font
    font = self.get_font(None, subspec)
  File "/home/felix/anaconda3/lib/python3.6/site-packages/pdfminer/pdfinterp.py", line 203, in get_font
    font = PDFCIDFont(self, spec)
  File "/home/felix/anaconda3/lib/python3.6/site-packages/pdfminer/pdffont.py", line 658, in __init__
    self.cmap = CMapDB.get_cmap(name)
  File "/home/felix/anaconda3/lib/python3.6/site-packages/pdfminer/cmapdb.py", line 259, in get_cmap
    data = klass._load_data(name)
  File "/home/felix/anaconda3/lib/python3.6/site-packages/pdfminer/cmapdb.py", line 233, in _load_data
    name = name.replace("\0", "")
AttributeError: 'PDFStream' object has no attribute 'replace'

[tataganesh]

Is it possible for you to post the problematic PDF?

[panoptikum]

Yes.

Problematic PDF

OR

Problematic PDF 2

[marcelhekking]

I am encountering the same error

[skewed91]

Same error

[bimbocant]

I'm encountering the same error when trying to parse

[euarthuurr]

I am encountering the same error when I run on the Mac (OS 10.10.5), but this error doesn't happen when I'm using Windows 10.

[euarthuurr]

If I open file Problematic PDF and saved as PDF on Mac, pdfminer works correctly.
Problematic PDF works

[bimbocant]

I tried saving it as PDF on Windows 10. Still not working.

[bimbocant]

#228 works fine.

[pforero]

Hi! I am getting still the same issue. I have found that on the problematics PDFs the specs seemed to be mixed. The 'Encoding' spec is a PDFStream object, and the 'ToUnicode' spec is 'Identity-H'. I have also tried #228 but it does not resolve it, the final 'name' value is 'CM20' which does not Map to Identity-H/V.

I am not sure why for some PDFs (it is a reasonable amount of them, close to 10% of those I have) it mixes ToUnicode and Encoding. Hope you guys can have some further insight on the cause

EDIT: I have tried to work around by switching the Encoder with the ToUnicode... while in this manner it parses, all characters just show as (cid:75)(cid:101)(cid:121)....

EDIT2: I continued to go down the rabbit hole. So the difference between the PDFs where I get an error or no seem to be in the CIDSystemInfo Registry. If the Registry is Adobe all is good. If it is Actuate then the problems arise. Besides the ToUnicode and Encoding differences (which is the one that launches the error in this thread) in DescendantFonts, Actuate PDFs do not have a CIDToGIDMap, but instead have a DW (?) field. It seems strange that it can only handle PDFs built on Adobe and not on Actuate.

[tangxfei]

To modify the “cmapdb.py”（\Lib\site-packages\pdfminer\cmapdb.py）
Line:233
def _load_data(klass, name):
name = name.replace("\0","") #Modify to=> name=str(name).replace("\0","")

The problem can be solved.

[benzkji]

I am encountering the same error, on (probably) one of around 1000 files. Will investigate, report if re-saving it (I am on Linux) will help...

@tangxfei mind a PR ;-)

[igormp]

@benzkji there's already a PR at #228 that fixes this, but the project seems kinda dead

[vinayak-mehta]

@0xabu Looks like you have merge access since you merged #230. Please look into #228 and merge it :)

[vinayak-mehta]

@0xabu Do you also have access to push a new release to PyPI? cc: @ganeshtata

[0xabu]

@vinayak-mehta Someone (I think @goulu) merged some bugfix PRs and then added me to the org back in early 2017, but like you I just depend on pdfminer so I'm not comfortable (not to mention don't have time) taking on responsibility for it. I don't know anything about the PyPi package.

[pinnnip]

I have tried all the methods mentioned in all the related threads, it is still not working on most of my PDFs. Are there any more updates?

[pinnnip]

Some of output only has a string '\x0c' with no content from pdf file.

[pietermarsman]

Fixed by #283