Show OCR progress

OCR process can last for a while, so monitoring progress could help. Here is simple example how to do in console app.

Code

For setting terminal color there is need to download Termcolor is a header-only C++ library:

curl -O https://raw.githubusercontent.com/ikalnytskyi/termcolor/master/include/termcolor/termcolor.hpp

Example program looks like this:

tesseract_progress.cpp

    
#include <leptonica/allheaders.h>
#include <tesseract/baseapi.h>
#include <tesseract/ocrclass.h>

#include <thread>
#include <iostream>
#include <fstream>

#include "termcolor.hpp"


/*
 Show/hide console cursor
 Source:  https://github.com/p-ranav/indicators
 */

static inline void show_console_cursor(bool const show) {
#if defined(_MSC_VER)
    HANDLE out = GetStdHandle(STD_OUTPUT_HANDLE);

    CONSOLE_CURSOR_INFO cursorInfo;

    GetConsoleCursorInfo(out, &cursorInfo);
    cursorInfo.bVisible = show;
    SetConsoleCursorInfo(out, &cursorInfo);
#else
    std::fputs(show ? "\033[?25h" : "\033[?25l", stdout);
#endif
}

/*
  Simple progress bar
  */
void show_progress(int progress) {
    std::cout << termcolor::bold << termcolor::blue;  // termcolor.hpp
    std::cout << "\r" << "[" << std::string(progress / 2, (char)254u);
    std::cout << std::string(100 / 2 - progress / 2, ' ') << "]";
    std::cout << progress << "%";
    std::cout << termcolor::reset<< termcolor::reset;  // termcolor.hpp
    std::cout.flush();
}

/*
  Main loop
  */
int main() {
    const char       *filein, *fileout;
    // suppress leptonica error messages
    setMsgSeverity(L_SEVERITY_NONE);
    // Run tesseract process and its progress monitor in separate threads
    ETEXT_DESC *monitor = new ETEXT_DESC();
    show_console_cursor(false);
    filein = "f:/Project/tests/speccoll.png";
    fileout = "myfile.html";
    // OCR process
    auto job_ocr = [monitor, filein, fileout]() {
        tesseract::TessBaseAPI *api = new tesseract::TessBaseAPI();
        // suppress tesseract debug messages to avoid progress interuption
        api->SetVariable("debug_file", "/dev/null");
        Pix *image = pixRead(filein);
        api->Init(NULL, "eng");

        api->SetPageSegMode(tesseract::PSM_AUTO);
        api->SetImage(image);

        api->Recognize(monitor);

        // Write html output of OCR to file
        std::ofstream myfile(fileout);
        if (myfile.is_open()) {
            myfile << api->GetHOCRText(0);
        }
        myfile.close();
        pixDestroy(&image);
        api->End();
        delete api;
    };
    // Monitoring process
    auto job_monitor = [monitor]() {
        while (true) {
            show_progress(monitor[0].progress);
            if (monitor[0].progress == 100) {
                monitor[0].cancel_this = (void *)true;
                // to ensure 100% is shown
                show_progress(monitor[0].progress);
                break;
            }
        }
    };

    std::thread thread1(job_ocr);
    std::thread thread2(job_monitor);
    thread1.join();
    thread2.join();
    show_console_cursor(true);

    delete monitor;
    return 0;
}
		

Build

    
	"c:\Program Files (x86)\Microsoft Visual Studio\2019\Community\VC\Auxiliary\Build\vcvars64.bat" x64
	SET PATH=%PATH%;f:\win64\bin
	SET TESSDATA_PREFIX=f:\Project\tessdata
	cl /EHsc /Fe tesseract_progress.cpp /std:c11 /If:\win64\include /link /LIBPATH:f:/win64/lib tesseract41.lib leptonica-1.81.0.lib /machine:x64
   
   

PS: With tesseract 5 you need to use tesseract::ETEXT_DESC instead of ETEXT_DESC.