+ fix sync. bug in multithreading indexer

src/common/kiwix/indexer.cpp

@@ -85,10 +85,6 @@ namespace kiwix {
     zim::Article currentArticle;
 
     while (currentOffset < lastOffset) {
-      if (self->getVerboseFlag()) {
-	std::cout << "Extracting article with offset: " << currentOffset << std::endl;
-      }
-
       currentArticle = zimHandler->getArticle(currentOffset);
       
       if (!currentArticle.isRedirect()) {
@@ -379,13 +375,13 @@ namespace kiwix {
   }
 
   void Indexer::setArticleCount(const unsigned int articleCount) {
-    pthread_mutex_lock(&articleCountMutex); 
+    pthread_mutex_lock(&articleCountMutex);
     this->articleCount = articleCount;
     pthread_mutex_unlock(&articleCountMutex); 
   }
 
   unsigned int Indexer::getArticleCount() {
-    pthread_mutex_lock(&articleCountMutex); 
+    pthread_mutex_lock(&articleCountMutex);
     unsigned int retVal = this->articleCount;
     pthread_mutex_unlock(&articleCountMutex);
     return retVal;
@@ -410,17 +406,28 @@ namespace kiwix {
       std::cout << "Indexing of '" << zimPath << "' starting..." <<std::endl;
     }
 
+    this->setArticleCount(0);
     this->setProgression(0);
     this->setZimPath(zimPath);
     this->setIndexPath(indexPath);
-    
+
     pthread_mutex_lock(&threadIdsMutex); 
     this->articleExtractorRunning(true);
     pthread_create(&(this->articleExtractor), NULL, Indexer::extractArticles, (void*)this);
     pthread_detach(this->articleExtractor);
+
+    while(this->isArticleExtractorRunning() && this->getArticleCount() == 0) {
+#ifdef _WIN32
+    Sleep(100);
+#else
+    usleep(100000);
+#endif
+    }
+    
     this->articleParserRunning(true);
     pthread_create(&(this->articleParser), NULL, Indexer::parseArticles, (void*)this);
     pthread_detach(this->articleParser);
+
     this->articleIndexerRunning(true);
     pthread_create(&(this->articleIndexer), NULL, Indexer::indexArticles, (void*)this);
     pthread_detach(this->articleIndexer);

src/common/kiwix/reader.cpp

@@ -48,8 +48,8 @@ namespace kiwix {
       this->firstArticleOffset = this->zimFileHandler->getNamespaceBeginOffset('A');
       this->lastArticleOffset = this->zimFileHandler->getNamespaceEndOffset('A');
       this->currentArticleOffset = this->firstArticleOffset;
-      this->articleCount = this->zimFileHandler->getNamespaceCount('A');
-      this->mediaCount = this->zimFileHandler->getNamespaceCount('I');
+      this->nsACount = this->zimFileHandler->getNamespaceCount('A');
+      this->nsICount = this->zimFileHandler->getNamespaceCount('I');
     }
 
     /* initialize random seed: */
@@ -100,7 +100,7 @@ namespace kiwix {
     unsigned int counter = 0;
     
     if (counterMap.empty()) {
-      counter = this->articleCount;
+      counter = this->nsACount;
     } else {
       std::map<std::string, unsigned int>::const_iterator it = counterMap.find("text/html");
       if (it != counterMap.end())
@@ -115,7 +115,7 @@ namespace kiwix {
     unsigned int counter = 0;
 
     if (counterMap.empty())
-      counter = this->mediaCount;
+      counter = this->nsICount;
     else {
       std::map<std::string, unsigned int>::const_iterator it;
 
@@ -169,7 +169,7 @@ namespace kiwix {
   /* Return an URL from a title*/
   string Reader::getRandomPageUrl() {
     zim::size_type idx = this->firstArticleOffset + 
-      (zim::size_type)((double)rand() / ((double)RAND_MAX + 1) * this->articleCount); 
+      (zim::size_type)((double)rand() / ((double)RAND_MAX + 1) * this->nsACount); 
     zim::Article article = zimFileHandler->getArticle(idx);
 
     return article.getLongUrl().c_str();

src/common/kiwix/reader.h

@@ -74,8 +74,8 @@ namespace kiwix {
     zim::size_type firstArticleOffset;
     zim::size_type lastArticleOffset;
     zim::size_type currentArticleOffset;
-    zim::size_type articleCount;
-    zim::size_type mediaCount;
+    zim::size_type nsACount;
+    zim::size_type nsICount;
     
     std::vector<std::string> suggestions;
     std::vector<std::string>::iterator suggestionsOffset;