bool GMainEditor::forkProccesOCR_(pidID *pidIDArray,int ID, int maxFork){
string str;
pechaImg=LoadImageData(inputData.data["inputFile"],0);
int pidID=fork();
if (pidID < 0){
error((char*)"ERROR on fork");
return 0;
}
if (pidID == 0) {
str=substr(0,str.rfind("."),str);
str+=".html"; cout_<<str<<endl;
inputData.data["inputFileName"]=inputData.data["siteName"];
inputData.data["inputFileName"]+=substr(inputData.data["siteRoot"].size(),inputData.data["inputFile"]);
//readPageHTML();
cout<<"NEW START "<<inputData.data["inputFile"]<<endl;
pechaDataLoaded=0;
startOCR(pechaImg);
pidIDArray[ID].status=0;
cout<<"done "<<ID<<endl;
remove(inputData.data["log"].c_str());
exit(0);
}
pechaImg->destroy();
return 1;
}
void GMainEditor::forkProccesOCR(int pidID,GBitmap *pechaImgID_){
if (pidID < 0)
error((char*)"ERROR on fork");
if (pidID == 0) {
string str=inputData.data["inputFile"];
str=substr(0,(int)str.rfind("."),str);
str+=".html"; cout_<<str<<endl;
inputData.data["inputFileName"]=inputData.data["siteName"];
inputData.data["inputFileName"]+=substr((int)inputData.data["siteRoot"].size(),inputData.data["inputFile"]);
//readPageHTML();
cout_<<"NEW START "<<inputData.data["inputFile"]<<endl;
pechaDataLoaded=0;
startOCR(pechaImgID_);
//if(stat)writeLetterStat();
exit(0);
}
}
KOCRBase::KOCRBase( QWidget *parent, KSpellConfig *spellConfig,
KDialogBase::DialogType face )
:KDialogBase( face, i18n("Optical Character Recognition"),
User2|Close|User1, User1, parent,0, false, true,
KGuiItem( i18n("Start OCR" ), "launch",
i18n("Start the Optical Character Recognition process" )),
KGuiItem( i18n("Cancel" ), "stopocr",
i18n("Stop the OCR Process" ))),
m_animation(0L),
m_metaBox(0L),
m_imgHBox(0L),
m_previewPix(0L),
m_currImg(0L),
m_spellConfig(spellConfig),
m_wantSpellCfg(true),
m_userWantsSpellCheck(true),
m_cbWantCheck(0L),
m_gbSpellOpts(0L)
{
kdDebug(28000) << "OCR Base Dialog!" << endl;
// Layout-Boxes
KConfig *konf = KGlobal::config ();
KConfigGroupSaver gs( konf, CFG_OCR_KSPELL );
m_userWantsSpellCheck = konf->readBoolEntry(CFG_WANT_KSPELL, true);
/* Connect signals which disable the fields and store the configuration */
connect( this, SIGNAL( user1Clicked()), this, SLOT( writeConfig()));
connect( this, SIGNAL( user1Clicked()), this, SLOT( startOCR() ));
connect( this, SIGNAL( user2Clicked()), this, SLOT( stopOCR() ));
m_previewSize.setWidth(200);
m_previewSize.setHeight(300);
enableButton( User1, true ); /* start ocr */
enableButton( User2, false ); /* Cancel */
enableButton( Close, true );
}
void GMainEditor::startOCRBatch(){
imageEditor=(GImageEditor*)inputData.imageEditor;
fontEditor=(GFontEditor*)inputData.fontEditor;
logicProcessor=(GLogicProcessor*)inputData.logicProcessor;
if(!logicProcessor->dictionaryReady)logicProcessor->readDictionary();
aliKali=fontEditor->aliKali;
string strHeaderHTML,srcLine,str;
string path=inputData.data["tablePath"]+"/header.xml";
DIR *dir;
//int mode;
int i=0;
//читаем статистику использования букв книги
//readLetterStat();
//загружаем базу данных букв
#ifdef FORK
int maxFork=inputData.num_cores*0.75;
int pid;
pidID *pidIDArray;
if(inputData.fileList.size()>1){
int countFork=0;
MemoryFile *pidData_mf; //main file for conection with child process
//inputData.data["statPath"].c_str()
time_t seconds; seconds = time (NULL);
ostringstream out; out<<"/tmp/"<<seconds;
string path=out.str();
pidData_mf=MemoryFile::create(path.c_str(), MemoryFile::if_exists_keep_if_dont_exists_create);
pidData_mf->resize(sizeof(pidID)*maxFork);
pidIDArray=(pidID*)pidData_mf->data(); //array which can be share between processes.
for(int index=0;index<maxFork;index++){
//cout_<<"pidIDArray["<<index<<"].status="<<pidIDArray[index].status<<endl;
pidIDArray[index].status=0;
}
int ID=0;
while(i<inputData.fileList.size()){
cout<<"NEW file#1 "<<inputData.fileList[i]<<endl;
if( ( dir=opendir(inputData.fileList[i].c_str()))!=NULL){
i++; continue;
}
inputData.data["inputFile"]=inputData.fileList[i];
string path=inputData.data["inputFile"]; //проверяем есть ли такой распознаный файл
path=substr(0,(int)path.rfind("."),path);
string volume=path;
string fileIndex=fileName(path);
path+=".xml";
if(is_file(path)){i++; continue;}
if(!forkProccesOCR_(pidIDArray,ID,maxFork)){
cout<<"ERROR on fork return";
sleep(1); continue;
};
i++;if(i==inputData.fileList.size())break;
countFork++;
for(int index=0;index<maxFork;index++)cout<<pidIDArray[index].status<<" ";
cout<<endl;
ID=100; int status;
while(ID==100){
if(countFork<=maxFork){ //есть свободные слоты для новых процессов
for(int index=0;index<maxFork;index++){ //маркируем слот как занятый
//cout_<<"pidIDArray["<<index<<"].status="<<pidIDArray[index].status<<endl;
if(pidIDArray[index].status==0){
ID=index;
pidIDArray[index].status=1;
break;
}
}
}else{
int forkStatusCount=0;
for(int index=0;index<maxFork;index++){ //подсчитываем количество активных процессов
if(pidIDArray[index].status==1)forkStatusCount++;
}
if(forkStatusCount==countFork){sleep(1); continue;}; //ждем завершения процесса
wait(&status); //регистрируем с системе завершенный процесс
countFork--;
}
}
}
}else{
inputData.data["inputFile"]=inputData.fileList[0];
pechaImg=LoadImageData(inputData.data["inputFile"],0);
startOCR(pechaImg);
}
#else
cout<<"NO FORK";
while(i<inputData.fileList.size()){
if( ( dir=opendir(inputData.fileList[i].c_str()))!=NULL){
readDirectoryToArray(inputData.fileList, inputData.fileList[i],"img");
i++; continue;
}
GBitmap* pechaImg;
inputData.data["inputFile"]=inputData.fileList[i];
if(!is_file(inputData.data["inputFile"]))continue;
pechaImg=LoadImageData(inputData.data["inputFile"],0);
if(!pechaImg){cout_<<"no open file"<<inputData.data["inputFile"]<<endl; return;}
str=inputData.data["inputFile"];
str=substr(0,str.rfind("."),str);
str+=".html"; //cout_<<str<<endl;
inputData.data["inputFileName"]=inputData.data["siteName"];
inputData.data["inputFileName"]+=substr(inputData.data["siteRoot"].size(),inputData.data["inputFile"]);
//readPageHTML();
inputData.c_out.open(str.c_str());
pechaDataLoaded=0;
startOCR(pechaImg);
pechaImg->destroy();
inputData.c_out.close();
i++;
}
#endif
//drawLettersInCorrectionTable(DRAW_BASE);
cout_<<"COMPLETE"<<endl;
}//____________________________________________________________________________
void GMainEditor::startOCRSystem(){
string strHeaderHTML,srcLine,str;
string path=inputData.data["tablePath"]+"/header.xml";
DIR *dir;
//int mode;
int i=0;
uint freeMemory;
//читаем статистику использования букв книги
//readLetterStat();
int maxProcess=inputData.num_cores*0.85;
pidID *pidIDArray;
if(inputData.data["pathDB"]!=""){
inputData.data["inputFile"]=inputData.fileList[0];
pechaImg=LoadImageData(inputData.data["inputFile"],0);
startOCR(pechaImg);
MemoryFile *pidData_mf=MemoryFile::create(inputData.data["pathDB"].c_str(),
MemoryFile::if_exists_keep_if_dont_exists_create);
int ID=atoi(inputData.data["ID"].c_str());
pidIDArray=(pidID*)pidData_mf->data(); //array which can be share between processes.
pidIDArray[ID].status=0;
cout<<"done OCR";
inputData.log<<" done OCR"<<endl;
inputData.log.close();
remove(inputData.data["log"].c_str());
exit(0);
}
if(inputData.fileList.size()>1){
int countProcess=0;
MemoryFile *pidData_mf; //main file for inter proccess communications
//inputData.data["statPath"].c_str()
time_t seconds; seconds = time (NULL);
ostringstream out; out<<"/tmp/"<<seconds;
string pathDB=out.str();
pidData_mf=MemoryFile::create(pathDB.c_str(), MemoryFile::if_exists_keep_if_dont_exists_create);
pidData_mf->resize(sizeof(pidID)*maxProcess);
pidIDArray=(pidID*)pidData_mf->data(); //array which can be share between processes.
for(int index=0;index<maxProcess;index++){
//cout_<<"pidIDArray["<<index<<"].status="<<pidIDArray[index].status<<endl;
pidIDArray[index].status=0;
}
int ID=0;
while(i<inputData.fileList.size()){
if( ( dir=opendir(inputData.fileList[i].c_str()))!=NULL){
i++; continue;
}
inputData.data["inputFile"]=inputData.fileList[i];
string next="";
if(i<inputData.fileList.size()-2){
next=inputData.fileList[i+1];
next=str_replace(".tif",".html",next);
next=str_replace(".jpg",".html",next);
}
string path=inputData.data["inputFile"]; //проверяем есть ли такой распознаный файл
path=substr(0,(int)path.rfind("."),path);
string volume=path;
string fileIndex=fileName(path);
path+=".html";
if(is_file(path)){i++; continue;}
while(1){
//проверяем загрузку системы
str=run("vm_stat");
vector<string>vm=explode("\n", str);
str=str_replace("Pages free:", "", vm[1]);
str=str_replace(" ", "", str);
str=str_replace(".", "", str);
freeMemory=atoi(str.c_str());
freeMemory=(freeMemory*4.096)/1000;
//cout<<"@freeMemory:"<<freeMemory<<endl;
if(freeMemory<10){
cout<<"critical memory loading error";
run("killall OCRLib");
exit(0);
}
if(freeMemory>500)break;
cout<<"@ no free memory for process. freeMemory:"<<freeMemory<<endl;
sleep(10);
}
cout<<"NEW file#1 "<<inputData.fileList[i]<<endl;
ostringstream out;
out<<inputData.data["rootApp"]<<" \"xml=<fileList>"<<inputData.data["inputFile"]<<
"</fileList><ocrData>"<<inputData.data["ocrData"]<<"</ocrData>"<<
"<ocrLn>"<<inputData.data["ocrLn"]<<"</ocrLn>"<<"<pathDB>"<<pathDB<<"</pathDB>"<<
"<ID>"<<ID<<"</ID><nextPage>"<<next<<"</nextPage><scale>"<<inputData.data["scale"]<<"</scale>\" &";
string cmd=out.str();
//cout<<cmd; exit(0);
system(cmd.c_str());
//sleep(1); //continue;
i++;if(i==inputData.fileList.size())break;
countProcess++;
for(int index=0;index<maxProcess;index++)cout<<pidIDArray[index].status<<" ";
cout<<endl;
ID=100; int status;
while(ID==100){
if(countProcess<=maxProcess){ //есть свободные слоты для новых процессов
for(int index=0;index<maxProcess;index++){ //маркируем слот как занятый
//cout_<<"pidIDArray["<<index<<"].status="<<pidIDArray[index].status<<endl;
if(pidIDArray[index].status==0){
ID=index;
pidIDArray[index].status=1;
break;
}
}
}else{
int processStatusCount=0;
for(int index=0;index<maxProcess;index++){ //подсчитываем количество активных процессов
if(pidIDArray[index].status==1)processStatusCount++;
}
if(processStatusCount==countProcess){sleep(1); continue;}; //ждем завершения процесса
countProcess--;
}
}
}
}else{
i=0;
//while(i<inputData.fileList.size()){
// if( ( dir=opendir(inputData.fileList[i].c_str()))!=NULL){
// i++; continue;
// }
inputData.data["inputFile"]=inputData.fileList[0];
pechaImg=LoadImageData(inputData.data["inputFile"],0);
startOCR(pechaImg);
// i++;
//}
}
}
