RecursionError: maximum recursion depth exceeded

[jjhuff]

Example url: http://www.accessdata.fda.gov/drugsatfda_docs/label/2019/021107s029lbl.pdf

Backtrace:

  File "amplion/utils/pdf.py", line 29, in extract_pdf_text_from_fp
    interpreter.process_page(page)
  File "/usr/local/lib/python3.7/site-packages/pdfminer/pdfinterp.py", line 896, in process_page
    self.device.end_page(page)
  File "/usr/local/lib/python3.7/site-packages/pdfminer/converter.py", line 49, in end_page
    self.cur_item.analyze(self.laparams)
  File "/usr/local/lib/python3.7/site-packages/pdfminer/layout.py", line 733, in analyze
    group.analyze(laparams)
  File "/usr/local/lib/python3.7/site-packages/pdfminer/layout.py", line 501, in analyze
    LTTextGroup.analyze(self, laparams)
  File "/usr/local/lib/python3.7/site-packages/pdfminer/layout.py", line 350, in analyze
    obj.analyze(laparams)
  File "/usr/local/lib/python3.7/site-packages/pdfminer/layout.py", line 501, in analyze
    LTTextGroup.analyze(self, laparams)
  File "/usr/local/lib/python3.7/site-packages/pdfminer/layout.py", line 350, in analyze
    obj.analyze(laparams)
  File "/usr/local/lib/python3.7/site-packages/pdfminer/layout.py", line 501, in analyze
    LTTextGroup.analyze(self, laparams)
  File "/usr/local/lib/python3.7/site-packages/pdfminer/layout.py", line 350, in analyze
    obj.analyze(laparams)
  File "/usr/local/lib/python3.7/site-packages/pdfminer/layout.py", line 501, in analyze
    LTTextGroup.analyze(self, laparams)
  File "/usr/local/lib/python3.7/site-packages/pdfminer/layout.py", line 350, in analyze
    obj.analyze(laparams)
  File "/usr/local/lib/python3.7/site-packages/pdfminer/layout.py", line 501, in analyze
    LTTextGroup.analyze(self, laparams)
  File "/usr/local/lib/python3.7/site-packages/pdfminer/layout.py", line 350, in analyze
    obj.analyze(laparams)
  File "/usr/local/lib/python3.7/site-packages/pdfminer/layout.py", line 501, in analyze
    LTTextGroup.analyze(self, laparams)
.....

[pietermarsman]

Hi @jjhuff, thanks for raising this issue!

What version of pdfminer.six are you using? And what command are you using?

[jjhuff]

I was using 20200124
The code is more or less this, but I believe it should repo with pdf2txt

    laparams = LAParams()
    rsrcmgr = PDFResourceManager()
    with io.StringIO() as outfp:
        device = TextConverter(rsrcmgr, outfp, laparams=laparams)
        interpreter = PDFPageInterpreter(rsrcmgr, device)
        for page in PDFPage.get_pages(fp):
            interpreter.process_page(page)
        device.close()
        text = outfp.getvalue()

[pietermarsman]

I can replicate this using pdf2txt.py. The problem happens on page 5.

You can fix your problem by increasing the line_margin. Setting it to 2.0 fixes it for me. This merges more lines into text-boxes and makes the next analyse step (the one with the recursion error) easier.

I'm not sure what causes the problem so I would like to keep this issue open until we figure out if there is something we can do about it.

[metalogueur]

Just ran into that same bug using version 20200726... Will wait for the issue to be solved... Thanks for your hard work!

[jonaswinkler]

I've also encountered this issue on a document after calling extract_text on version 20201018

  File "/home/jonas/.local/share/virtualenvs/paperless-RxYopU_u/lib/python3.9/site-packages/pdfminer/high_level.py", line 121, in extract_text
    interpreter.process_page(page)
  File "/home/jonas/.local/share/virtualenvs/paperless-RxYopU_u/lib/python3.9/site-packages/pdfminer/pdfinterp.py", line 896, in process_page
    self.device.end_page(page)
  File "/home/jonas/.local/share/virtualenvs/paperless-RxYopU_u/lib/python3.9/site-packages/pdfminer/converter.py", line 50, in end_page
    self.cur_item.analyze(self.laparams)
  File "/home/jonas/.local/share/virtualenvs/paperless-RxYopU_u/lib/python3.9/site-packages/pdfminer/layout.py", line 814, in analyze
    group.analyze(laparams)
  File "/home/jonas/.local/share/virtualenvs/paperless-RxYopU_u/lib/python3.9/site-packages/pdfminer/layout.py", line 575, in analyze
    LTTextGroup.analyze(self, laparams)
  File "/home/jonas/.local/share/virtualenvs/paperless-RxYopU_u/lib/python3.9/site-packages/pdfminer/layout.py", line 362, in analyze
    obj.analyze(laparams)
  File "/home/jonas/.local/share/virtualenvs/paperless-RxYopU_u/lib/python3.9/site-packages/pdfminer/layout.py", line 575, in analyze
    LTTextGroup.analyze(self, laparams)
  File "/home/jonas/.local/share/virtualenvs/paperless-RxYopU_u/lib/python3.9/site-packages/pdfminer/layout.py", line 362, in analyze
    obj.analyze(laparams)
  File "/home/jonas/.local/share/virtualenvs/paperless-RxYopU_u/lib/python3.9/site-packages/pdfminer/layout.py", line 575, in analyze
    LTTextGroup.analyze(self, laparams)

I believe that this is the same issue and the different line numbers are caused by changes somewhere else in that file.

[rain01]

Recently I have started seeing this same issue in my error logs.
I am using pdfminer.six==20211012 with:

<LAParams: char_margin=2.0, line_margin=0.1, word_margin=0.1 all_texts=True>

Increasing line_margin helped as a temporary solution, any idea if there is a better way?

[pietermarsman]

Fixed in the current version.

The issue was partially fixed by @0xabu in #659. After this PR pdf2txt.py run indefinitely on page 5.

This issue was then completely fixed by @jwyawney in #689. I guess by ignoring empty characters the layout algorithm does generate less LTTextLine and therefore the grouping is easier. Or something along that lines.

[Rumpelcita]

Hi, I'm still running into this issue on version 20220524 of pdfminer.six. I can't provide the pdf that causes the issue since it needs to be anonymized, but these are the LAParams we're using for pdfminer:

LAParams(
    line_overlap= 0.5,
    char_margin= 1.1,
    word_margin= 0.2,
    line_margin= 0.5,
    boxes_flow= 0.5,
    all_texts=True,
        )

Traceback:

RecursionError: maximum recursion depth exceeded
(245 additional frame(s) were not displayed)
...
  File "pdfminer/layout.py", line 705, in analyze
    super().analyze(laparams)
  File "pdfminer/layout.py", line 439, in analyze
    obj.analyze(laparams)
  File "pdfminer/layout.py", line 705, in analyze
    super().analyze(laparams)
  File "pdfminer/layout.py", line 439, in analyze
    obj.analyze(laparams)
  File "pdfminer/layout.py", line 439, in analyze
    obj.analyze(laparams)

maximum recursion depth exceeded