From 20daa92b65bc0e81db1978aeb2f6d6d3d7a277da Mon Sep 17 00:00:00 2001
From: kelson42 <kelson42@users.sourceforge.net>
Date: Wed, 28 Mar 2012 14:59:57 +0000
Subject: [PATCH] + start of the work on the multithreading indexer

---
 src/common/kiwix/indexer.cpp | 66 ++++++++++++++++++++++++++++++++++--
 src/common/kiwix/indexer.h   | 24 +++++++++----
 2 files changed, 81 insertions(+), 9 deletions(-)

diff --git a/src/common/kiwix/indexer.cpp b/src/common/kiwix/indexer.cpp
index c2941cd91..02e67a275 100644
--- a/src/common/kiwix/indexer.cpp
+++ b/src/common/kiwix/indexer.cpp
@@ -59,18 +59,76 @@ namespace kiwix {
     /* Compute few things */
     this->articleCount = this->zimFileHandler->getNamespaceCount('A');
     this->stepSize = (float)this->articleCount / (float)100;
+
+    /* Thread mgmt */
+    this->runningStatus = 0;
   }
-  
+
+  void *Indexer::extractArticles(void *ptr) {
+    kiwix::Indexer *self = (kiwix::Indexer *)ptr;
+    self->incrementRunningStatus();
+    unsigned int startOffset = self->zimFileHandler->getNamespaceBeginOffset('A');
+    unsigned int endOffset = self->zimFileHandler->getNamespaceEndOffset('A');
+
+    /* Goes trough all articles */
+    unsigned int currentOffset = startOffset;
+    zim::Article currentArticle;
+
+    while (currentOffset <= endOffset) {
+      /* Test if the thread should be cancelled */
+      pthread_testcancel();
+
+      /* Redirects are not indexed */
+      do {
+	currentArticle = self->zimFileHandler->getArticle(currentOffset++);
+      } while (currentArticle.isRedirect() && currentOffset++ != endOffset);
+
+      cout << currentArticle.getTitle() << endl;
+
+    }
+
+    self->decrementRunningStatus();
+    pthread_exit(NULL);
+    return NULL;
+  }
+
+  void *Indexer::parseArticles(void *ptr) {
+    pthread_exit(NULL);
+    return NULL;
+  }
+
+  void *Indexer::writeIndex(void *ptr) {
+    pthread_exit(NULL);
+    return NULL;
+  }
+
   bool Indexer::start() {
+    pthread_create(&(this->articleExtracter), NULL, Indexer::extractArticles, ( void *)this);
+    pthread_detach(this->articleExtracter);
+    cout << "end" << endl;
+
     return true;
   }
 
   bool Indexer::stop() {
-    return true;
+      
+      return true;
+  }
+
+  void Indexer::incrementRunningStatus() {
+    this->runningStatus++;
+  }
+
+  void Indexer::decrementRunningStatus() {
+    this->runningStatus--;
+  }
+
+  unsigned int Indexer::getRunningStatus() {
+    return this->runningStatus;
   }
 
   bool Indexer::isRunning() {
-    return true;
+    return this->runningStatus > 0;
   }
 
   void Indexer::setCurrentArticleOffset(unsigned int offset) {
@@ -82,6 +140,8 @@ namespace kiwix {
   }
 
   unsigned int Indexer::getProgression() {
+    unsigned int progression = 0;
+    return progression;
   }
 
   /* Read the file containing the stopwords */
diff --git a/src/common/kiwix/indexer.h b/src/common/kiwix/indexer.h
index f97dc3075..31a68c744 100644
--- a/src/common/kiwix/indexer.h
+++ b/src/common/kiwix/indexer.h
@@ -27,6 +27,7 @@
 #include <sstream>
 
 #include <xapian.h>
+#include <pthread.h>
 #include <unaccent.h>
 #include <zim/file.h>
 #include <zim/article.h>
@@ -42,7 +43,23 @@ namespace kiwix {
   public:
     Indexer(const string &zimFilePath);
     bool indexNextPercent(const bool &verbose = false);
+    bool setZimFilePath(const string &zimFilePath);
+    bool start();
+    bool stop();
+    bool isRunning();
+    unsigned int getProgression();
+
+  private:
+    pthread_t articleExtracter, articleParser, indexWriter;
+    static void *extractArticles(void *ptr);
+    static void *parseArticles(void *ptr);
+    static void *writeIndex(void *ptr);
     
+    unsigned int runningStatus;
+    void incrementRunningStatus();
+    void decrementRunningStatus();
+    unsigned int getRunningStatus();
+
   protected:
     virtual void indexNextPercentPre() = 0;
     virtual void indexNextArticle(const string &url, 
@@ -56,14 +73,9 @@ namespace kiwix {
     virtual void indexNextPercentPost() = 0;
     virtual void stopIndexing() = 0;
 
-    /* General */
-    bool setZimFilePath(const string &zimFilePath);
-    bool start();
-    bool stop();
-    bool isRunning();
+    /* Article offset */
     void setCurrentArticleOffset(unsigned int offset);
     unsigned int getCurrentArticleOffset();
-    unsigned int getProgression();
 
     /* ZIM file handling */
     zim::File* zimFileHandler;