On 10/24/20 9:37 AM, Philippe Mathieu-Daudé wrote:
On 10/24/20 8:40 AM, Thomas Huth wrote:
On 24/10/2020 08.35, Thomas Huth wrote:
On 21/10/2020 12.50, Philippe Mathieu-Daudé wrote:
We are going to reuse the tesseract OCR code.
Create a new tesseract_ocr() helper and use it.
Signed-off-by: Philippe Mathieu-Daudé <f4...@amsat.org>
---
tests/acceptance/machine_m68k_nextcube.py | 21 +++++----------------
tests/acceptance/tesseract_utils.py | 18 ++++++++++++++++++
2 files changed, 23 insertions(+), 16 deletions(-)
...
Would it make sense to completely hide the tesseract version handling in
this new tesseract_utils.py file now, so that the tests themselves do
not
have to worry about this anymore
The problem is the recognized strings differ between versions,
see in tests/acceptance/machine_m68k_nextcube.py:
lines = tesseract_ocr(screenshot_path, tesseract_version=3)
text = '\n'.join(lines)
self.assertIn('Backplane', text)
self.assertIn('Ethernet address', text)
and:
lines = tesseract_ocr(screenshot_path, tesseract_version=4)
text = '\n'.join(lines)
self.assertIn('Testing the FPU, SCC', text)
self.assertIn('System test failed. Error code', text)
self.assertIn('Boot command', text)
self.assertIn('Next>', text)
Yes, good idea.
(i.e. would it be possible to merge
test_bootrom_framebuffer_ocr_with_tesseract_v3 and
test_bootrom_framebuffer_ocr_with_tesseract_v4 into one single test
that way?)
If I've got that right, there is also now a proper release 4 of
Tesseract,
so maybe we can simply scratch the testing with version 3 now?
Good to know, I'll have a look. Thanks!
Thomas