enhanced timeXMLTree

git-svn-id: svn+ssh://idea.nguyen.vg/svn/sxsi/trunk/xpathcomp@253 3cdefd35-fc62-479d-8e8d-bae585ffb9ca

diff --git a/timeXMLTree.cpp b/timeXMLTree.cpp
index 68e7556..c3bd7f9 100644 (file)
--- a/timeXMLTree.cpp
+++ b/timeXMLTree.cpp
@@ -3,6 +3,7 @@
 #include "Utils.h"
 #include <sys/time.h>
 #include <time.h>
+#include <sys/stat.h> 
 
 using std::cout;
 using std::string;
@@ -191,29 +192,103 @@ unsigned int time_jump(XMLTree* tree, treeNode node,treeNode root){
 }
 
 
+int usage(char ** argv){
+  
+  std::cout << "usage : " << argv[0] << " [-d] [-s] file.{xml,.srx} tagname\n";
+  return 1;
 
+}
 
 
 int main(int argc, char ** argv){
   unsigned int count1,count2;
-  unsigned char * tagname = (unsigned char *) "keyword";
+  unsigned char * tagname;
+  string arg,filename,ext;
+  bool disable_tc = false; 
+  bool save = false;
+  bool srx;
+  XMLTree * tree;
+
+  int i = 1;
+  if ( i >= argc)
+    return usage(argv);
+  
+  arg = argv[i];
+  if (arg.compare("-d") == 0){
+    disable_tc = true;
+    i++;
+    if ( i >= argc)
+      return usage(argv);
+    arg = argv[i];
+  };
 
-  if (argc != 2){
-    std::cout << "Usage : " << argv[0] << " filename (without .srx)\n";
-    return 1;
+  if (arg.compare("-s") == 0){
+    save = true;
+    i++;
+    if ( i >= argc)
+      return usage(argv);
+    arg = argv[i];
   };
 
-  // The samplerate is not taken into account for loading anymore
-  XMLTree * tree = XMLTree::Load((unsigned char*) argv[1],64);
+
+  // The filename
+  if (arg.size() < 4)
+    return usage(argv);
+  
+  ext=(arg.substr(arg.size()-4,4));
+  if (ext.compare(".srx") == 0){
+    // must truncate
+    filename = arg.substr(0,arg.size()-4);
+
+    srx = true;
+  }
+  else if (ext.compare(".xml")==0) {
+    filename = arg;
+    srx = false;
+  }
+  else 
+    return usage(argv);
+  i++;
+
+  if (i >= argc)
+    return usage(argv);
+
+  tagname = (unsigned char*) argv[i];
   
+
+  
+  if (srx)     
+    // The samplerate is not taken into account for loading anymore
+    tree = XMLTree::Load((unsigned char*) filename.c_str(),64);
+  else {
+    try {
+      //filename, sampling factor, index empty texts, disable tc
+      XMLDocShredder shredder(filename.c_str(),64,false,disable_tc);      
+      shredder.processStartDocument("");  
+      shredder.parse();  
+      shredder.processEndDocument();         
+      tree = (XMLTree *) shredder.storageIfc_->returnDocument();
+      if (save){
+       filename = filename.substr(0,filename.size()-4).append(".srx");
+       struct stat stats;
+       int exists = stat(filename.c_str(),&stats);
+       if(exists == 0) { 
+         std::cout << "Warning : indexed file " << filename << " exists, not overwriting\n";
+       }
+       else {
+         tree->Save((unsigned char*) filename.substr(0,filename.size()-4).c_str());
+       };
+
+      };
+    }
+    catch (const std::exception& e){
+      std::cout << "Error during parsing : " << e.what() << "\n";
+      return 2;
+    };
+  };    
   traversal(tree,tree->Root(),tagname);
 
-  STARTTIMER();
-  count1 = time_traversal(tree,tree->Root());
-  STOPTIMER(FullTraversal);
 
-  count2 = time_jump(tree,tree->Root(),tree->Root());
-  STOPTIMER(JumpTraversal);
 
   PRINTSTATS(Tag);
   PRINTSTATS(FirstChild);
@@ -230,6 +305,20 @@ int main(int argc, char ** argv){
   PRINTSTATS(ParentNode);
   PRINTSTATS(PrevNode);
   std::cout << "\n";
+
+  if (target_tag == -1){
+    std::cout << "Warning:  tag " << tagname << " was not found in the document!\n"
+             << "Warning:  not timing traversal and jumping functions\n";
+    return 3;
+  };
+  
+  STARTTIMER();
+  count1 = time_traversal(tree,tree->Root());
+  STOPTIMER(FullTraversal);
+  
+  count2 = time_jump(tree,tree->Root(),tree->Root());
+  STOPTIMER(JumpTraversal);
+  
   std::cout << "Full traversal found " << count1 << " " << tagname << "  nodes\n";
   PRINTSTATS(FullTraversal);
   std::cout << "\n";
@@ -238,4 +327,5 @@ int main(int argc, char ** argv){
   
   
   return 0;
+  
 }