Add Google code for more unittests

[stweil]

Some additional changes of the Tesseract code base make integration of the new tests easier.
As an example, bitvector_test and some more tests were now added to the Tesseract test set.

[Shreeshrii]

Welcome addition!!
Thank you @jbreiden and @stweil .

[stweil]

Some of the new test files which are currently not activated use the Abseil API (https://abseil.io/, https://github.com/abseil/abseil-cpp). We have to decide whether we want to replace those calls, add a minimal implementation to Tesseract or require a full Abseil installation (maybe as a Git submodule, about 10 MB).

These tests are affected:

baseapi_test – done
baseapi_thread_test – done
fileio_test – done
imagedata_test – done
lang_model_test – done
mastertrainer_test – done
pango_font_info_test – done
paragraphs_test – done
stringrenderer_test – done
unicharcompress_test – done

[stweil]

These tests neither depend on the Abseil API nor on unavailable TIFF images, so might be the next candidates to get fixed and added to our test set:

commandlineflags_test – done
dawg_test – done
heap_test – done
indexmapbidi_test – done
intfeaturemap_test – done
ligature_table_test – done
linlsq_test – done
lstm_recode_test – done
lstm_squashed_test – done
networkio_test – done
normstrngs_test – done
nthitem_test – done
params_model_test – done
qrsequence_test – done
recodebeam_test – done
rect_test – done
scanutils_test – done
shapetable_test – done
stats_test – done
stridemap_test – done
tablefind_test – done
tablerecog_test – done
tabvector_test – done
tfile_test – done
unicharset_test – done
unichar_test – done
validate_grapheme_test – done
validate_indic_test – done
validate_khmer_test – done
validate_myanmar_test – done
validator_test – done

Still missing (2019-07-06):

pagesegmode_test – needs missing TIFF files
tatweel_test – needs files ara.* (from ara.traineddata?)

[Shreeshrii]

a full Abseil installation (maybe as a Git submodule, about 10 MB).

Abseil is an open-source collection of C++ code (compliant to C++11) designed to augment the C++ standard library.

We are already using Googletest and test (testdata) as submodules that are required only for testing, so I think it will be OK to add Abseil as a submodule too.

[amitdo]

qrsequence_test – needs unknown data type CycleTimer

It is mentioned here:

https://github.com/abseil/abseil-cpp/blob/87a4c07856e7dc69958019d47b2f02ae47746ec0/absl/base/internal/unscaledcycleclock.h

Also check
https://www.google.com/search?q=%22CycleTimer%22+%22google%22+%22apache%22+%22github%22

params_model_test – needs file eng.params_model

Try to extract a eng.traindata file that contains the data for the legacy engine.
eng.params_model should be there.

https://github.com/tesseract-ocr/tesseract/blob/5fdaa479da2c/doc/combine_tessdata.1.asc

needs unknown data type Array2D (class?)

Try to replace with GENERIC_2D_ARRAY

shapetable_test – needs unknown function StringPrintf

https://www.google.com/search?q=%22StringPrintf%22+%22google%22+%22apache%22

tatweel_test – needs util/utf8/public/unicodetext.h

https://github.com/tensorflow/models/search?q=%22unicodetext.h%22

dawg_test – needs util/process/subprocess.h

https://www.google.com/search?q=%22subprocess.h%22+%22google%22+%22apache%22

[stweil]

I'm just preparing the baseapi_test. It is the first test case which uses the Abseil API, and it also needs CycleTimer and LOG(). I am still struggling with the integration of Abseil into the build process, but hopefully will have finished that soon. Then adding more tests with similar requirements should be possible.

[jbreiden]

absl::StrFormat is the replacement for StringPrintf. Tests should work with the LOG lines removed, but the logging library is probably https://github.com/google/glog

[stweil]

Yes, the code is obviously using glog. Pull request #1980 includes a rudimentary implementation for LOG, so the Tesseract code does not need glog.

[stweil]

Latest Tesseract now has 40 unit tests (with lots of subtests) which pass successfully:

============================================================================
Testsuite summary for tesseract 4.0.0-225-gdaf6
============================================================================
# TOTAL: 40
# PASS:  40
# SKIP:  0
# XFAIL: 0
# FAIL:  0
# XPASS: 0
# ERROR: 0

[amitdo]

Good work. @stweil and @Shreeshrii, thank you!

How many tests are not activated?