No text lines in ALTO output

[stweil]

The conversion of an example file from PAGE to ALTO creates an ALTO file without text lines:

java -jar PATH/PageConverter.jar -source-xml FILE_0063_OCR-D-OCR-TESSEROCR.xml -target-xml FILE_0063_OCR-D-OCR-TESSEROCR-ALTO.xml -convert-to ALTO

Is ALTO support still incomplete?

[chris1010010]

Hi, what format is the input file and what does it contain?

[stweil]

The input is the PAGE XML file https://digi.bib.uni-mannheim.de/~stweil/FILE_0063_OCR-D-OCR-TESSEROCR.xml (link was also given above). It contains the layout information and OCR results for a single page from one of our books.

[chris1010010]

Okay, that explains it. Words (STRINGs) are mandatory in ALTO. So because there are no words in the PAGE XML, the ALTO exporter cannot add the text lines.

[stweil]

@kba, @bertsky, does that mean that we have to change ocrd-tesserocr-recognize to produce PAGE XML with words?

[stweil]

@chris1010010, thank you for your explanation.

[wrznr]

@stweil No. You can set textequiv_level to word.

[bertsky]

For anyone who traps into this fallacy from another angle (besides the special case ocrd-tesserocr-recognize):
Consider using https://github.com/kba/page-to-alto, which can do much more conversions than the PRImA library, including dummy printspace, dummy lines and dummy regions, if needed.