TesseractCore.cpp File Reference

Implements Tesseract OCR extraction helpers. More...

#include "text-recognition/pch.h"
#include "text-recognition/TesseractCore.h"
#include "core/ports/diagnostics/IDiagnostics.h"
#include "core/ports/infra/text-recognition/TextRecognitionTypes.h"
#include <tesseract/baseapi.h>
#include <leptonica/allheaders.h>
#include <sstream>
#include <fstream>
#include <algorithm>
#include <array>
#include <filesystem>
#include <cstdlib>
#include <system_error>

Include dependency graph for TesseractCore.cpp:

Go to the source code of this file.

Functions
static Pix *	pixFromBytes (const std::vector< uint8_t > &data)
static std::string	readEnvironmentVariable (const char *name)
static std::filesystem::path	executableDirectory ()
static std::string	resolveTessdataPath (const std::string &provided)
static tesseract::OcrEngineMode	toTesseractOem (ports::EngineMode mode)
static std::string	resolveLanguage (const ports::Settings &recognition)
static void	configureForStatements (tesseract::TessBaseAPI &ocr, const ports::Settings &recognition)
static void	applyPsm (tesseract::TessBaseAPI &ocr, int psm)
static ports::Text	recognizeTextFromBytes (const std::vector< uint8_t > &data, const std::string &tessdataPath, const ports::Settings &recognition, std::shared_ptr< core::ports::diagnostics::IDiagnostics > dbg)
static vector< ports::Word >	getWordsFromBytes (const std::vector< uint8_t > &data, const std::string &tessdataPath, const ports::Settings &recognition, std::shared_ptr< core::ports::diagnostics::IDiagnostics > dbg)

Detailed Description

Implements Tesseract OCR extraction helpers.

Definition in file TesseractCore.cpp.

Function Documentation

applyPsm()

void applyPsm ( tesseract::TessBaseAPI & ocr, int psm )

static

Definition at line 129 of file TesseractCore.cpp.

Here is the caller graph for this function:

configureForStatements()

void configureForStatements ( tesseract::TessBaseAPI & ocr, const ports::Settings & recognition )

static

Definition at line 121 of file TesseractCore.cpp.

Here is the caller graph for this function:

executableDirectory()

std::filesystem::path executableDirectory ( )

static

Definition at line 53 of file TesseractCore.cpp.

Here is the caller graph for this function:

getWordsFromBytes()

vector< ports::Word > getWordsFromBytes ( const std::vector< uint8_t > & data, const std::string & tessdataPath, const ports::Settings & recognition, std::shared_ptr< core::ports::diagnostics::IDiagnostics > dbg )

static

Definition at line 166 of file TesseractCore.cpp.

Here is the call graph for this function:

Here is the caller graph for this function:

pixFromBytes()

Pix * pixFromBytes ( const std::vector< uint8_t > & data )

static

Definition at line 27 of file TesseractCore.cpp.

Here is the caller graph for this function:

readEnvironmentVariable()

std::string readEnvironmentVariable ( const char * name )

static

Definition at line 33 of file TesseractCore.cpp.

Here is the caller graph for this function:

recognizeTextFromBytes()

ports::Text recognizeTextFromBytes ( const std::vector< uint8_t > & data, const std::string & tessdataPath, const ports::Settings & recognition, std::shared_ptr< core::ports::diagnostics::IDiagnostics > dbg )

static

Definition at line 137 of file TesseractCore.cpp.

Here is the call graph for this function:

Here is the caller graph for this function:

resolveLanguage()

std::string resolveLanguage ( const ports::Settings & recognition )

static

Definition at line 117 of file TesseractCore.cpp.

Here is the caller graph for this function:

resolveTessdataPath()

std::string resolveTessdataPath ( const std::string & provided )

static

Definition at line 67 of file TesseractCore.cpp.

Here is the call graph for this function:

Here is the caller graph for this function:

toTesseractOem()

tesseract::OcrEngineMode toTesseractOem ( ports::EngineMode mode )

static

Definition at line 103 of file TesseractCore.cpp.

Here is the caller graph for this function: