Ignored text before tables

[panagiotis-tsolakis]

In a sample of 2 pdfs that I converted into TEI with Grobid, I noticed that text lines preceding a table were sometimes dropped and did not appear in the final TEI file. The following screenshots come from two consecutive pages of a pdf file:

The text underlined in yellow was ignored by Grobid, as you can see in the TEI file. The text of the table was ignored as well by Grobid.

[lfoppiano]

Hi @panagiotis-tsolakis! Thanks for reporting this, could you attach here the pdf files, and let me know also which version of grobid did you use?

If you haven't used, I suggest you to try with the current master.

[panagiotis-tsolakis]

I used Grobid 0.8.1.

Here's the pdf file :

zwitter-vitez-etal-2022-extracting.pdf

[lfoppiano]

@panagiotis-tsolakis the text missing was a bug in 0.8.1, which has been fixed in the current master, so the text is not lost. The table, unfortunately is blended into the text, the recognition should hopefully improve with #963