[FEATURE] Tika with Tesseract OCR

[notaus123]

Hi there,

I installed tika as standalone-Server and added OCR Support with tesseract.
(https://cwiki.apache.org/confluence/display/tika/tikaocr)

In ext/tika/Classes/Service/Tika/ServerService.php I had to add several headers to enable ocr-ing (would be great if this could become configurable btw)

OCRing is working, but my log is full of warnings:
Core: Error handler (BE): PHP Warning: file_get_contents(http://tika.xxx:9998/tika): failed to open stream: Connection refused in /var/www/html_relaunch/site/releases/initial/typo3conf/ext/tika/Classes/Service/Tika/ServerService.php line 238

It seems the file_get_contents(..) is getting a timout. Tried to increase the php.ini setting for socket timeout, no change at all.
When disabling the additional ocr-headers, it's running fine.

Manually calling the tika server with ocr-header works fine.

Maybe you have an idea?

[dkd-kaehm]

@notaus123 Thanks for reporting.
Could you please prepare a pull-request with mentioned headers and tell us about the TYPO3, Solr-Server, Tika-Server, EXT-Solr version stack?

[dkd-kaehm]

I cant reproduce the issue with logs "file_get_contents(http://tika.xxx:9998/tika): failed to open stream" on EXT:tika v10.0.0.
The EXT:tika v10 introduced the configurable PSR-18 HTTP client from TYPO3.

No headers are required to extract texts from images. That works already with tika "full" docker image.
See the significant change in:

TYPO3-Solr/solr-ddev-site@dda2afd#diff-0af25d60a5d43139a91176ff71c3c0bc162408b1df0736dd6e4c0f701cbc750eR5