blumia/deepin-ocr
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

feat: 删除设置识别语言包路径接口

Browse Source 
Description:  删除设置识别语言包接口，优化为内部根据系统语言设置

Log: 优化接口
Change-Id: I634d90297cd111a38139b2a710e1ea770faf1148
This commit is contained in:
wangcong 2021-06-15 13:22:21 +08:00 committed by liuminghang
parent d6b3771524
commit e73c254cb8
5 changed files with 148 additions and 94 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

BIN
assets/tesslangs/chi_sim_vert.traineddata Normal file
View File

Binary file not shown.

BIN
assets/tesslangs/chi_tra_vert.traineddata Normal file
View File

Binary file not shown.

2
deepin-ocr.pro
View File

@ -62,7 +62,7 @@ dbus_service.files=./com.deepin.Ocr.service
#Tesseract-ocr识别语言包
tesslangs.path=/usr/share/deepin-ocr/tesslangs
tesslangs.files=./assets/tesslangs/chi_sim.traineddata ./assets/tesslangs/chi_tra.traineddata ./assets/tesslangs/eng.traineddata
tesslangs.files=./assets/tesslangs/chi_sim.traineddata ./assets/tesslangs/chi_tra.traineddata ./assets/tesslangs/eng.traineddata ./assets/tesslangs/chi_sim_vert.traineddata ./assets/tesslangs/chi_tra_vert.traineddata
INSTALLS += target dbus_service translations tesslangs

163
src/tessocrutils/tessocrutils.cpp
View File

@ -39,12 +39,29 @@
*/
#include "tessocrutils.h"
/**
* @brief 语言包路径
*/
const QString TESSDATA_PATH = "/usr/share/deepin-ocr/tesslangs";
/**
* @brief TessOcrUtils::m_tessOcrUtils
*/
TessOcrUtils *TessOcrUtils::m_tessOcrUtils = nullptr;
TessOcrUtils::TessOcrUtils()
{
//设置语言包路径
setLanguagesPath(TESSDATA_PATH);
//设置当前默认使用的识别语言系统语言+英语
m_sLangs = getLanguages();
}
TessOcrUtils::~TessOcrUtils()
{
}
TessOcrUtils *TessOcrUtils::m_tessOcrUtils = nullptr;
TessOcrUtils *TessOcrUtils::instance()
{
@ -54,11 +71,11 @@ TessOcrUtils *TessOcrUtils::instance()
return m_tessOcrUtils;
}
//传入待识别图片的路径和想得到的返回结果类型，获取识别结果
RecognitionResult TessOcrUtils::getRecogitionResult(const QString imagePath,const ResultType resultType)
{
QString errorMessage = "";
ErrorCode errorCode = ErrorCode::UKNOW;
ErrorCode errorCode = ErrorCode::UNKNOWN;
RecognitionResult t_result;
Pix *p_image;
@ -66,7 +83,7 @@ RecognitionResult TessOcrUtils::getRecogitionResult(const QString imagePath,cons
{
//errorMesage = "不能传递空的图片路径！";
errorMessage = "Can't pass an empty image path!";
errorCode = ErrorCode::OCRPNULL;
errorCode = ErrorCode::OCR_P_NULL;
setResult(errorCode,errorMessage,resultType,t_result);
return t_result;
}
@ -77,7 +94,7 @@ RecognitionResult TessOcrUtils::getRecogitionResult(const QString imagePath,cons
{
//errorMesage = "图片不存在！imagePath: " + imagePath;
errorMessage = "Image does not exist! imagePath: " + imagePath;
errorCode = ErrorCode::OCRPNULL;
errorCode = ErrorCode::OCR_P_NULL;
setResult(errorCode,errorMessage,resultType,t_result);
return t_result;
}
@ -91,7 +108,7 @@ RecognitionResult TessOcrUtils::getRecogitionResult(const QString imagePath,cons
} catch (const std::logic_error &e) {
//errorMesage = "加载图片失败！" + QString(qExc.what());
errorMessage = "Failed to load picture! " + QString(e.what());
errorCode = ErrorCode::OCRLIF;
errorCode = ErrorCode::OCR_LI_F;
setResult(errorCode,errorMessage,resultType,t_result);
return t_result;
}
@ -99,22 +116,24 @@ RecognitionResult TessOcrUtils::getRecogitionResult(const QString imagePath,cons
return getRecogitionResult(p_image,resultType);
}
//传入待识别图片的路径，获取纯字符串的识别结果
RecognitionResult TessOcrUtils::getRecogitionResult(const QString imagePath)
{
return getRecogitionResult(imagePath,ResultType::ResultString);
return getRecogitionResult(imagePath,ResultType::RESULT_STRING);
}
//传入待识别图片和想得到的返回结果类型，获取识别结果
RecognitionResult TessOcrUtils::getRecogitionResult(QImage *image, const ResultType resultType)
{
QString errorMessage = "";
ErrorCode errorCode = ErrorCode::UKNOW;
ErrorCode errorCode = ErrorCode::UNKNOWN;
RecognitionResult t_result;
if(image->isNull())
{
//errorMesage = "不能传递空的图片路径！";
errorMessage = "Can't pass an empty image!";
errorCode = ErrorCode::OCRPNULL;
errorCode = ErrorCode::OCR_P_NULL;
setResult(errorCode,errorMessage,resultType,t_result);
return t_result;
}
@ -137,17 +156,51 @@ RecognitionResult TessOcrUtils::getRecogitionResult(QImage *image, const ResultT
//p_image->colormap->nalloc;
//p_image->colormap->n;
p_image->data = reinterpret_cast<l_uint32*>(image->bits());
//获取识别结果
return getRecogitionResult(p_image,resultType);
}
//传入待识别图片，获取纯字符串的识别结果
RecognitionResult TessOcrUtils::getRecogitionResult(QImage *image)
{
return getRecogitionResult(image,ResultType::ResultString);
return getRecogitionResult(image,ResultType::RESULT_STRING);
}
//获取系统当前的语言
Languages TessOcrUtils::getSystemLang()
{
QLocale locale;
if(locale.language() == QLocale::Language::Chinese){
//查询当前国别代码
QLocale::Country t_countryId = locale.country();
if(t_countryId == QLocale::Country::China){
return Languages::CHI_SIM;
}else {
return Languages::CHI_TRA;
}
}else {
return Languages::ENG;
}
}
//获取识别需要的使用的语言包
QString TessOcrUtils::getLanguages()
{
QString t_langs = "";
//当前系统语言
Languages t_systemLang = getSystemLang();
if(t_systemLang == Languages::ENG)
{
t_langs=getLangStr(t_systemLang);
}else{
t_langs=getLangStr(t_systemLang)+"+"+getLangStr(Languages::ENG);
}
return t_langs;
}
//设置识别需要的使用的语言包的路径
bool TessOcrUtils::setLanguagesPath(const QString langsPath)
{
if(langsPath.isNull() || langsPath.isEmpty())
@ -158,24 +211,22 @@ bool TessOcrUtils::setLanguagesPath(const QString langsPath)
return true;
}
//获取识别结果
RecognitionResult TessOcrUtils::getRecogitionResult(Pix * image,ResultType resultType)
{
QString errorMessage = "";
ErrorCode errorCode = ErrorCode::UKNOW;
ErrorCode errorCode = ErrorCode::UNKNOWN;
QString result = "";
RecognitionResult t_result;
if(!isExistsResultType(resultType))
{
//errorMesage = "结果类型不存在 resultType: " + QString::number(resultType);
errorMessage = "The result type does not exist! resultType: " + QString::number(resultType);
errorCode = ErrorCode::OCRRTNULL;
errorCode = ErrorCode::OCR_RT_NULL;
setResult(errorCode,errorMessage,resultType,t_result);
return t_result;
}
tesseract::TessBaseAPI *t_Tesseract = new tesseract::TessBaseAPI();
try{
//实例化tesseract
t_Tesseract = new tesseract::TessBaseAPI();
@ -184,45 +235,40 @@ RecognitionResult TessOcrUtils::getRecogitionResult(Pix * image,ResultType resul
{
//errorMesage = "初始化 Tesseract 失败！" + QString::number(resultType);
errorMessage = "Could not initialize tesseract! Tesseract couldn't load any languages!";
errorCode = ErrorCode::OCRINIF;
errorCode = ErrorCode::OCR_INI_F;
setResult(errorCode,errorMessage,resultType,t_result);
return t_result;
}
}catch(const std::logic_error &e){
//errorMesage = "初始化 Tesseract 失败！Tesseract 内部错误！" + QString(qExc.what());
errorMessage = "Could not initialize tesseract! Tesseract internal error! " + QString(e.what());
errorCode = ErrorCode::OCRINIF;
errorCode = ErrorCode::OCR_INI_F;
setResult(errorCode,errorMessage,resultType,t_result);
return t_result;
}
//为Tesseract设置待识别图片
t_Tesseract->SetImage(image);
try {
//识别图片
switch (resultType) {
case ResultType::ResultXML:
//result = QString(t_Tesseract->GetAltoText(0));
break;
case ResultType::ResultHTML:
case ResultType::RESULT_HTML:
result = QString(t_Tesseract->GetHOCRText(0));
break;
case ResultType::ResultString:
case ResultType::RESULT_STRING:
result = QString(t_Tesseract->GetUTF8Text());
break;
}
} catch (const std::logic_error &e) {
//errorMesage = "识别图片失败！" + QString(qExc.what());
errorMessage = "Image recognition failed! " + QString(e.what());
errorCode = ErrorCode::OCRLIF;
errorCode = ErrorCode::OCR_LI_F;
setResult(errorCode,errorMessage,resultType,t_result);
return t_result;
}
t_Tesseract->End();
delete t_Tesseract;
pixDestroy(&image);
t_result.flag = true;
t_result.message = errorMessage;
t_result.errorCode = ErrorCode::OK;
@ -231,9 +277,9 @@ RecognitionResult TessOcrUtils::getRecogitionResult(Pix * image,ResultType resul
return t_result;
}
//设置返回结果,内部使用
void TessOcrUtils::setResult(ErrorCode errCode, const QString errMessage,const ResultType resultType,RecognitionResult &result)
{
result.flag = false;
result.message = errMessage;
result.errorCode = errCode;
@ -241,34 +287,32 @@ void TessOcrUtils::setResult(ErrorCode errCode, const QString errMessage,const R
result.result = "";
}
bool TessOcrUtils::setLanguages(const QList<Languages> langs)
{
QString t_langs = "";
if(langs.isEmpty())
{
return false;
}
int i;
for ( i = 0;i<langs.length();i++) {
Languages lang = langs[i];
if(!isExistsLanguage(lang))
{
return false;
}
if((langs.length()-1) == i){
t_langs+=getLangStr(lang);
}else{
t_langs+=getLangStr(lang) + "+";
}
}
m_sLangs = t_langs;
return true;
}
//设置当前使用的语言包
//bool TessOcrUtils::setLanguages(const QList<Languages> langs)
//{
// QString t_langs = "";
// if(langs.isEmpty())
// {
// return false;
// }
// int i;
// for ( i = 0;i<langs.length();i++) {
// Languages lang = langs[i];
// if(!isExistsLanguage(lang))
// {
// return false;
// }
// if((langs.length()-1) == i){
// t_langs+=getLangStr(lang);
// }else{
// t_langs+=getLangStr(lang) + "+";
// }
// }
// m_sLangs = t_langs;
// return true;
//}
//根据传入的枚举，获取相应的语言包字符串
QString TessOcrUtils::getLangStr(Languages lang)
{
QString langStr = "";
@ -287,24 +331,23 @@ QString TessOcrUtils::getLangStr(Languages lang)
return langStr;
}
//判断指定的结果类型是否存在
bool TessOcrUtils::isExistsResultType(ResultType resultType)
{
bool flag = false;
switch (resultType)
{
case ResultType::ResultString:
case ResultType::RESULT_STRING:
flag = true;
break;
case ResultType::ResultHTML:
flag = true;
break;
case ResultType::ResultXML:
case ResultType::RESULT_HTML:
flag = true;
break;
}
return flag;
}
//判断指定的语言包类型是否存在
bool TessOcrUtils::isExistsLanguage(Languages lang)
{
bool flag = false;

77
src/tessocrutils/tessocrutils.h
View File

@ -57,28 +57,29 @@
* @brief 返回结果字符串的类型
*/
enum ResultType{
ResultString = 1, //纯字符串结果
ResultHTML = 2, //HTML文本
ResultXML = 3 //XML文本
RESULT_STRING = 1, //纯字符串结果
RESULT_HTML = 2, //HTML文本
UNKNOWN_TYPE = -1 //XML文本
};
/**
* @brief 返回识别过程中的错误码
*/
enum ErrorCode{
OK = 1, //成功
UKNOW = -1, //未知错误
OCRPNULL = 101, //文件路径为空
OCRRTNULL = 102, //结果字符串类型不存在
OCRINIF = 103, //ocr三方库初始化失败
OCRLIF = 104, //OCR加载图片失败
OCRRIF = 105 //OCR识别图片失败
OK = 1, //成功
UNKNOWN = -1, //未知错误
OCR_P_NULL = 101, //文件路径为空
OCR_RT_NULL = 102, //结果字符串类型不存在
OCR_INI_F = 103, //ocr三方库初始化失败
OCR_LI_F = 104, //OCR加载图片失败
OCR_RI_F = 105 //OCR识别图片失败
};
/**
* @brief 现今支持的语言包类型
*/
enum Languages{
UNKNOWN_LAN = -1, //未知语言
CHI_SIM=1, //简体中文
CHI_TRA=2, //繁体中文
ENG=3 //英文
@ -117,23 +118,22 @@ struct RecognitionResult{
RecognitionResult(){
flag = false;
message.clear();
errorCode = ErrorCode::UKNOW;
resultType = ResultType::ResultString;
errorCode = ErrorCode::UNKNOWN;
resultType = ResultType::RESULT_STRING;
result.clear();
}
};
/**
* @brief ocr接口工具
* 使用方法:
* 1.设置语言包路径 TessOcrUtils::instance()->setLanguagesPath(langPath);
* 2.设置用来识别的语言包 TessOcrUtils::instance()->setLanguages(langs);
* 3.获取识别结果 TessOcrUtils::instance()->getRecogitionResult(t_image);
* 使用方法: *
* 1.获取识别结果 TessOcrUtils::instance()->getRecogitionResult(t_image);
*/
class TessOcrUtils
{
public:
TessOcrUtils();
~TessOcrUtils();
static TessOcrUtils *instance();
/**
@ -166,23 +166,22 @@ public:
*/
RecognitionResult getRecogitionResult(QImage *image);
/**
* @brief 设置当前使用的语言包
* @param 需要使用的语言包
* @return 是否成功
*/
bool setLanguages(const QList<Languages> langs);
/**
* @brief 设置当前使用的语言包路径
* @param 语言包路径
* @return
*/
bool setLanguagesPath(const QString langsPath);
private :
/**
* @brief 设置当前使用的语言包
* @param 需要使用的语言包
* @return 是否成功
*/
//bool setLanguages(const QList<Languages> langs);
/**
* @brief 设置当前使用的语言包路径
* @param 语言包路径
* @return
*/
bool setLanguagesPath(const QString langsPath);
/**
* @brief 判断指定的结果类型是否存在
* @param 指定的类型
@ -204,6 +203,19 @@ private :
*/
QString getLangStr(Languages lang);
/**
* @brief 获取系统当前的语言
* @return
*/
Languages getSystemLang();
/**
* @brief 获取当前使用的语言包
* @param 需要使用的语言包
* @return
*/
QString getLanguages();
/**
* @brief 获取纯字符串的识别结果
* @param 需识别的图片
@ -220,9 +232,8 @@ private :
*/
void setResult(ErrorCode errCode, const QString errMessage,const ResultType resultType,RecognitionResult &result);
/**
* @brief 语言包路径m_image
* @brief 语言包路径
*/
QString m_sTessdataPath;