tests: add tesseract training data "tessdata", still trying to enhance OCR reliability in UI tests
--- a/tests/ide_tests/sikuliberemiz.py Mon Dec 05 15:53:25 2022 +0100
+++ b/tests/ide_tests/sikuliberemiz.py Thu Dec 15 14:45:52 2022 +0100
@@ -12,9 +12,9 @@
beremiz_path = os.environ["BEREMIZPATH"]
python_bin = os.environ.get("BEREMIZPYTHONPATH", "/usr/bin/python")
-
opj = os.path.join
+tessdata_path = os.environ["TESSDATAPATH"]
class KBDShortcut:
"""Send shortut to app by calling corresponding methods.
@@ -199,6 +199,7 @@
Returns:
Sikuli App class instance
"""
+ sikuli.OCR.Options().dataPath(tessdata_path)
sikuli.OCR.Options().oem(0)
self.screenshotnum = 0
--- a/tests/tools/Docker/Dockerfile Mon Dec 05 15:53:25 2022 +0100
+++ b/tests/tools/Docker/Dockerfile Thu Dec 15 14:45:52 2022 +0100
@@ -69,7 +69,17 @@
twisted nevow autobahn click opcua \
wxPython==4.1.1
-# Point to python binary test scripts will use
+RUN set -xe && \
+ cd /home/$UNAME && mkdir tessdata && \
+ wget -q https://github.com/tesseract-ocr/tessdata/archive/refs/tags/4.1.0.tar.gz \
+ -O tessdata.tar.gz && \
+ echo 89e25c7c40a59be7195422a01f57fcb2 tessdata.tar.gz | md5sum -c && \
+ tar --strip-components=1 -C tessdata -x -v -z -f tessdata.tar.gz && \
+ rm tessdata.tar.gz
+
+ENV TESSDATAPATH /home/$UNAME/tessdata
+
+# Points to python binary that test will use
ENV BEREMIZPYTHONPATH /home/$UNAME/beremizenv/bin/python
# easy to remember 'do_tests' alias to invoke main makefile