From 65b4015f03ecafe891a28657e63d36348e523ca2 Mon Sep 17 00:00:00 2001
From: kelson42 <kelson42@users.sourceforge.net>
Date: Sat, 7 Apr 2012 14:09:47 +0000
Subject: [PATCH] + new indexer

---
 src/common/kiwix/indexer.cpp     | 24 ++++++++++++++++++++----
 src/common/kiwix/reader.cpp      |  7 +++++--
 src/common/kiwix/xapianIndexer.h |  1 -
 3 files changed, 25 insertions(+), 7 deletions(-)

diff --git a/src/common/kiwix/indexer.cpp b/src/common/kiwix/indexer.cpp
index e114c20f3..0babb0042 100644
--- a/src/common/kiwix/indexer.cpp
+++ b/src/common/kiwix/indexer.cpp
@@ -61,7 +61,9 @@ namespace kiwix {
 
     /* Get the number of article to index */
     kiwix::Reader reader(self->getZimPath());
-    self->setArticleCount(reader.getArticleCount());
+    unsigned int articleCount = reader.getArticleCount();
+    self->setArticleCount(articleCount);
+    cout << "Article Count from reader: " << articleCount << endl;
 
     /* Goes trough all articles */
     zim::File *zimHandler = reader.getZimFileHandler();
@@ -184,7 +186,14 @@ namespace kiwix {
 
     indexerToken token;
     unsigned indexedArticleCount = 0;
-    unsigned int stepSize = ((self->getArticleCount() / 100) < 1 ? 1 : (self->getArticleCount() / 100));
+    unsigned int articleCount = self->getArticleCount();
+    unsigned int currentProgression = self->getProgression();
+    float stepSize = articleCount / 100;
+
+    cout << "Article count: " << articleCount << endl;
+    cout << "Progression step size: " << stepSize << endl;
+    cout << "Curent progression: " << currentProgression << endl;
+
     self->indexingPrelude(self->getIndexPath()); 
 
     while (self->popFromToIndexQueue(token)) {
@@ -198,10 +207,16 @@ namespace kiwix {
 		  token.wordCount
 		  );
       
-      if (++indexedArticleCount % stepSize == 0) {
-	self->setProgression(self->getProgression() + 1);
+      indexedArticleCount += 1;
+
+      /* Update the progression counter (in percent) */
+      if ((unsigned int)((float)indexedArticleCount/(float)articleCount*100) > currentProgression) {
+	self->setProgression((unsigned int)((float)indexedArticleCount/(float)articleCount*100));
+	currentProgression = self->getProgression();
+	cout << indexedArticleCount << " articles indexed, that means a progression of " << currentProgression << endl;
       }
 
+      /* Make a hard-disk flush every 10.000 articles */
       if (indexedArticleCount % 10000 == 0) {
 	self->flush();
       }
@@ -211,6 +226,7 @@ namespace kiwix {
     }
     self->setProgression(100);
     self->indexingPostlude();
+    sleep(1);
     self->articleIndexerRunning(false);
     pthread_exit(NULL);
     return NULL;
diff --git a/src/common/kiwix/reader.cpp b/src/common/kiwix/reader.cpp
index 2d92a1c7f..2386f8eef 100644
--- a/src/common/kiwix/reader.cpp
+++ b/src/common/kiwix/reader.cpp
@@ -79,6 +79,7 @@ namespace kiwix {
     string counterUrl = "/M/Counter";
 
     this->getContentByUrl(counterUrl, content, contentLength, mimeType);
+    cout << "Counter URL: " << content << endl;
     stringstream ssContent(content);
 
     while(getline(ssContent, item,  ';')) {
@@ -99,12 +100,14 @@ namespace kiwix {
     std::map<std::string, unsigned int> counterMap = this->parseCounterMetadata();
     unsigned int counter = 0;
     
-    if (counterMap.empty())
+    if (counterMap.empty()) {
       counter = this->articleCount;
-    else {
+      cout << "Article count from offsets: " << counter << endl;
+    } else {
       std::map<std::string, unsigned int>::const_iterator it = counterMap.find("text/html");
       if (it != counterMap.end())
 	counter = it->second;
+      cout << "Article count from metadata: " << counter << endl;
     }
     
     return counter;
diff --git a/src/common/kiwix/xapianIndexer.h b/src/common/kiwix/xapianIndexer.h
index 635fc80de..06fadbe7a 100644
--- a/src/common/kiwix/xapianIndexer.h
+++ b/src/common/kiwix/xapianIndexer.h
@@ -31,7 +31,6 @@ namespace kiwix {
     
   public:
     XapianIndexer();
-    ~XapianIndexer();
     
   protected:
     void indexingPrelude(const string &indexPath);