pdplumber return empty string on importerror

[bosd]

Was running some tests, encountered following error when pdfplumber is not available.
This PR returns and empty value and let invoice2data fail gracefully.

Before:
invoice2data input.pdf --input-reader=pdfplumber

DEBUG:invoice2data.input.pdfplumber: Cannot import pdfplumber
Traceback (most recent call last):
  File "/home/emiel/.local/bin/invoice2data", line 8, in <module>
    sys.exit(main())
             ^^^^^^
  File "/home/emiel/.local/lib/python3.11/site-packages/invoice2data/main.py", line 312, in main
    res = extract_data(f.name, templates=templates, input_module=input_module)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/emiel/.local/lib/python3.11/site-packages/invoice2data/main.py", line 166, in extract_data
    extracted_str = input_module.to_text(invoicefile).decode("utf-8")
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/emiel/.local/lib/python3.11/site-packages/invoice2data/input/pdfplumber.py", line 29, in to_text
    with pdfplumber.open(path, laparams={"detect_vertical": True}) as pdf:
         ^^^^^^^^^^
UnboundLocalError: cannot access local variable 'pdfplumber' where it is not associated with a value

After:

ERROR:invoice2data.input.pdfplumber: Cannot import pdfplumber
ERROR:root: Failed to extract text from testin.pdf using invoice2data.input.pdfplumber

fixes #362

[rmilecki]

Returning empty string suggests that invoice was parsed but was empty.
If we want to return some value then make it None please.

If you take a look at pdftotext.py however, you'll see it raises EnvironmentError if pdftotext is missing. So returning None will make pdfplumber.py somehow incompatible with the pdftotext.py.

As for decision what is better: return None or raise EnvironmentError - I have no idea or preference.

[bosd]

@rmilecki I agree, we need to think of an solution for this.

Returning None conflicts with:

invoice2data/src/invoice2data/main.py

Line 89 in a5bdd50

extracted_str = input_module.to_text(invoicefile).decode("utf-8")

as Nonetype cannot be decoded
(Maybe we can remove the decode line? I assume it is a python2 leftover)

pdftotext might be a different story, as it is one of the default/main parsers.
So making everything fail when it's unavailable is not a big deal.

Is there a way to raise the error, only if the pdfplumber input module is called?
We don't want the whole lib to fail on this missing requirement.

This pr #491
is currently failing because of the missing pdfplumber.
(the test should not even run when it is unavailable but that's a different sunbject)
In that example, there should be an ImportError or EnvironmentError