stdin option doesn't work for psm 0 and 2

[catethos]

I am using the master branch from github (3.05.00dev) and facing some problems when using the stdin option in conjunction with the -psm 0 and -psm 2 options

cat <image file> | tesseract stdin stdout -psm 0

gives me the error

Error in fopenReadStream: file not found
Error in pixRead: image file not found: stdin
Cannot open input file: stdin

However, the command

cat <image file> | tesseract stdin stdout -psm 1

works as expected.

Do I miss something?

Regards,
Caleb

[zdenop]

At the moment stdin is supported only for OCR.
PSM_AUTO_ONLY (0) and PSM_OSD_ONLY (2) expects to read data based on filename

[jflesch]

Just so you know, if I'm not mistaken, it worked with Tesseract 3.03 and only broke with Tesseract 3.04. It is also very handy to script around Tesseract (avoid using temporary files) (see Pyocr for instance).

[zdenop]

@jflesch: I am currios: Why you call tesseract executable from python instead of using C-API? IMO it should be faster...

[jflesch]

When I started working on PyOCR, it was still Tesseract 2.x. At the time, if I remember correctly, there was only a C++ API and it seemed painfully hard to use (and binding on C++ API from Python is really annoying).
Also I did fork PyOCR from an already-existing python module (python-tesseract if I recall correctly).

Good to know there is a C API now. I'll have a look, see if I can bind on it easily. Thanks for the tip :)

[zdenop]

fixed with 7089c7b