workspace validator: non-URI path

[bertsky]

From a workspace validate I got:

<report valid="false">
  <error>METS has no unique identifier</error>
  <error>Validation aborted with exception: Traceback (most recent call last):
  File "/data/ocr-d/ocrd_all/venv38/lib/python3.8/site-packages/ocrd_validators/workspace_validator.py", line 149, in _validate
    self._validate_mets_files()
  File "/data/ocr-d/ocrd_all/venv38/lib/python3.8/site-packages/ocrd_validators/workspace_validator.py", line 302, in _validate_mets_files
    scheme = f.url[0:f.url.index(':')]
ValueError: substring not found
</error>
</report>

The url in question simply was a relative file name, which obviously makes the URI validator crash.

This is very problematic for two reasons:

1. before ocrd differentiated between LOCTYPE=URL and OTHER, we created lots of data (including GT) with URL, despite being local paths – this now broken
2. in this case, the data was just created by current ocrd itself – via ocrd workspace add, because that implementation sets both local_filename and url to the local path

[mikegerber]

Yeah, this happens with a workspace built with ocrd workspace itself...

add seems to make, for example, this:

      <mets:file ID="XXX" MIMETYPE="image/jpeg">
        <mets:FLocat xlink:href="OCR-D-IMG/2812988X_1862-09-02_001.jpg" LOCTYPE="OTHER" OTHERLOCTYPE="FILE"/>
        <mets:FLocat xlink:href="OCR-D-IMG/2812988X_1862-09-02_001.jpg" LOCTYPE="URL"/>
      </mets:file>

[mikegerber]

1. Removing the FLocats with LOCTYPE="URL"
2. and making the image filename referenced in the PAGE XML consistent (I imported the XML)

fixes the validation at least (in the sense that it doesn't choke on exceptions itself).

[bertsky]

add also created this structMap:

that's also what I witnessed as prime problem in https://github.com/OCR-D/ocrd_tesserocr/issues/201. We need more diagnostics why and exactly when this is happening.

But this is a separate issue (has nothing to do with the validator).

[mikegerber]

But this is a separate issue (has nothing to do with the validator).

True, I'll open another GitHub issue for that, if you didn't already.

[bertsky]

I'll open another GitHub issue for that, if you didn't already.

No, please do!

[mikegerber]

No, please do!

Just for the sake of completeness: https://github.com/OCR-D/core/issues/1195