Table data reader

The project has been created for one of my friends to help him fetch text presented in a grid from the application and it's screenshots. That application doesn't have any functions to export data and presents data at a custom grid widget and so data can't be read with WINAPI functions.

Prerequisites

The development may be performed in these two environments:

• Linux
• Windows

Prerequisites for Linux environment (Ubuntu 20.04 package list):

• g++ 9
• cmake
  • dpkg-dev (required to build DEB-package only)
  • cppcheck (required to static analysis only)
• libtesseract-dev
  • libopencv-dev
  • tesseract-ocr-eng
  • tesseract-ocr-rus
• googletest
  • googletest-tools
  • libgtest-dev

Prerequisites for Windows environment:

• CMake 3.16+
• Tesseract 4.1
  • tiff 4.1.0
    • libpng 1.6.37
    • zlib 1.2.11
    • jpeg 9d
  • OpenCV 4.1
    • laptonica 1.74.4
• GoogleTest 1.10
• cppcheck (only if you want to make static code analysis)
• NSIS 3.6 (only if you want to make package)

Visual Studio Code extensions:

• ms-vscode.cpptools
• ms-vscode.cmake-tools
• dbaeumer.vscode-eslint

Installing

Development under Linux performs using Visual Studio Code with the 'Remote Containers' extension. The environment is specified at .devcontainer/{devcontainer.json,Dockerfile}.

Development under Windows performs using Visual Studio Code, MinGW-w64-v7.0.0. To build the project under Windows you have to build (with MinGW) or install the libraries/packages accounted in the 'Prerequisites' section.

The great start point to build Tesseract with MinGW is here. A step by step series of examples that tell you how to get a development env running.

The working Windows batch file to setup environment for Visual Studio Code looks like:

set MSYS2_PATH_TYPE=inherit
set PATH=D:\bin\cmake\bin;D:\bin\mingw64\mingw64\bin;D:\bin\nsis-3.06.1;D:\bin\git\bin;%PATH%
set OpenCV_DIR=D:\bin\OpenCV-4.1.1-MinGW
set GTEST_ROOT=D:\bin\gtest
set Tesseract_DIR=D:\bin\tesseract-mingw\cmake
set Leptonica_DIR=D:\bin\tesseract-mingw\cmake
set TABREAD_EXTERNAL=D:/bin/tabread-external

The 'TABREAD_EXTERNAL' environment variable should point to the directory with Tesseract and MinGW shared libraries and data necessary for tabread executable. The directory structure is listed below:

.
+-- bin
|   +-- tessdata
|       +-- eng.traineddata
|       +-- equ.traineddata
|       +-- rus.traineddata
+-- +-- libgcc_s_seh-1.dll
+-- +-- libleptonica-1.74.4.dll
+-- +-- liblzma-5.dll
+-- +-- libopencv_core411.dll
+-- +-- libopencv_imgcodecs411.dll
+-- +-- libopencv_imgproc411.dll
+-- +-- libpng16.dll
+-- +-- libstdc++-6.dll
+-- +-- libtesseract41.dll
+-- +-- libtiff.dll
+-- +-- libwinpthread-1.dll
+-- +-- libzstd.dll

To check application, build it with Visual Studio Code an then run the command from the build directory:

tabread/tabread -i=../data/000.png -o=output.txt

to run with single threaded mode OCR or run:

tabread/tabread -i=../data/000.png -o=output.txt -th

for multithreaded (parrallel execution policy) OCR.

You may configure and then run 'package' target from the 'workspaceFolder' to generate installation bundle:

make -C build package

under Linux environment or

mingw-make -C package

under Windows environment.

Running the tests

Run tests from Visual Studio Code:

1. Open 'Command Palette' (Ctrl-Shift-P).
2. Type and then select 'CMake: Run Tests'.

Please note:

1. The 'version.h' is generated by CMake only at initialization stage (Ctrl-Shift-P, CMake: Configure).
2. Under Windows environment you should edit .vscode/settings.json and change configuration key name from "__cmake.generator" to "cmake.generator".

Deployment

1. Installation at Linux environment:

dpkg -i tabread_1.0.0_amd64.deb

2. Installation at Windows environment:
   Execute the installation bundle 'tabread-X.Y.Z.exe', where X.Y.Z is a package version.

Further work

1. Make GUI application (select widget/appliction or file, progress, etc.).
2. Read appliction's and filters' parameters from configuration file (especially for crop filter).
3. Improve grid detection algorithms.
4. Multiline text recognition.