From: kim <kim@3cdefd35-fc62-479d-8e8d-bae585ffb9ca>
Date: Wed, 28 Apr 2010 06:40:29 +0000 (+0000)
Subject: Fix the printing some more
X-Git-Url: http://git.nguyen.vg/gitweb/?p=SXSI%2FXMLTree.git;a=commitdiff_plain;h=9c696a1b1a7034794b3768e4e1e40db86e87ebbb

Fix the printing some more


git-svn-id: svn+ssh://idea.nguyen.vg/svn/sxsi/trunk/XMLTree@799 3cdefd35-fc62-479d-8e8d-bae585ffb9ca
---

diff --git a/XMLTree.cpp b/XMLTree.cpp
index 9be500e..db9814c 100644
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -3,7 +3,7 @@
 #include <stack>
 #include "XMLTree.h"
 #include "timings.h"
-
+#include <errno.h>
 // functions to convert tag positions to the corresponding tree node and viceversa. 
 // These are implemented in order to be able to change the tree and Tags representations, 
 // without affecting the code so much.
@@ -53,7 +53,7 @@ static treeNode fast_first_child(bp *Par, treeNode x)
 
 static treeNode fast_next_sibling(bp* Par,treeNode x)
 {
-  x = fwd_excess(Par,x,0);
+  x = fast_find_close(Par,x)+1;
   return (fast_inspect(Par,x) == OP) ? x : NULLT;
 }
 
@@ -944,11 +944,15 @@ bool XMLTree::IsOpen(treeNode x) { return fast_inspect(Par,x); }
 //WARNING this uses directly the underlying implementation for plain text
 
 
-
 void XMLTree::Print(int fd,treeNode x, bool no_text){
   
   int newfd = dup(fd);
   stream = fdopen(newfd,"wa");
+  if (stream == 0){
+    perror(NULL);
+    return;
+  };
+
   if (buffer == 0)
     buffer = new string();
 
@@ -959,8 +963,8 @@ void XMLTree::Print(int fd,treeNode x, bool no_text){
   uchar * tagstr;
   range r = DocIds(x);
   treeNode first_idx;
-  treeNode first_text = (tag == PCDATA_TAG_ID ?  x : TaggedDescendant(x,PCDATA_TAG_ID));
-  treeNode first_att =  NULLT;//TaggedDesc(x,ATTRIBUTE_DATA_TAG_ID);
+  treeNode first_text = (tag == PCDATA_TAG_ID ?  x : ParentNode(r.min-1));
+  treeNode first_att =  NULLT;
   
   if (first_att  == NULLT)
   first_idx = first_text;
@@ -972,18 +976,18 @@ void XMLTree::Print(int fd,treeNode x, bool no_text){
    uchar * current_text=NULL;
    if (first_idx != NULLT)
    current_text = GetText(MyText(first_idx));
-   int read = 0;
+   size_t read = 0;
    std::vector<uchar*> st;
  while (n <= fin){
    if (fast_inspect(Par,n)){
      if (tag == PCDATA_TAG_ID  ) {       
-       //       myfputs((const char*) (GetText(MyTextUnsafe(n))),fp);
+
        if (no_text)
 	 myfputs("<$/>",fp);
        else{
-       read = fprintf(fp,"%s",(const char*) current_text);
-       current_text += (read + 1);
-       }
+	 read = myfprintf((const char*) current_text, fp);
+	 current_text += (read + 1);
+       };
        n+=2; // skip closing $
        tag = Tag(n);
       
@@ -1002,11 +1006,10 @@ void XMLTree::Print(int fd,treeNode x, bool no_text){
 	   while (fast_inspect(Par,n)){
 	     if (no_text) {
 	       myfputc('<',fp);
-	       const uchar * tmp = &(GetTagNameByRef(Tag(n))[3]);
-	       myfputs((const char*) tmp,fp);
+	       myfputs((const char*) &(GetTagNameByRef(Tag(n))[3]),fp);
 	       myfputc('>',fp);
 	       myfputs("<$@/></",fp);
-	       myfputs((const char*) tmp,fp);
+	       myfputs((const char*) &(GetTagNameByRef(Tag(n))[3]),fp);
 	       myfputc('>',fp);
 	       n+= 4;
 	     }
@@ -1015,9 +1018,8 @@ void XMLTree::Print(int fd,treeNode x, bool no_text){
 	       myfputs((const char*) &(GetTagNameByRef(Tag(n))[3]),fp);
 	       n++;
 	       myfputs("=\"",fp);
-	       read = fprintf(fp,"%s",(const char*) current_text);
+	       read = myfprintf((const char*) current_text,fp);
 	       current_text += (read + 1);
-	       //myfputs((const char*) GetText(MyTextUnsafe(n)),fp);
 	       myfputc('"',fp);
 	       n+=3;
 	     }
@@ -1049,8 +1051,8 @@ void XMLTree::Print(int fd,treeNode x, bool no_text){
      }while (!fast_inspect(Par,n) && !st.empty());
    tag=Tag(n);
  };
- //myfputc('\n',fp);
+ myfputc('\n',fp);
  mybufferflush(fp);
- fflush(fp);
+ //fflush(fp);
  fclose(fp);
 }
diff --git a/XMLTree.h b/XMLTree.h
index d0a5c73..ea7b778 100644
--- a/XMLTree.h
+++ b/XMLTree.h
@@ -162,6 +162,21 @@ class XMLTree {
      fputs(buffer->c_str(), fp);
      buffer->clear();
    }
+
+   size_t myfprintf(const char* s, FILE * fp){
+     if (s == NULL)
+       return 0;
+     size_t i = buffer->size();
+     buffer->append(s);
+     size_t j = buffer->size();
+     if (buffer->size() >= 100000){
+       fputs(buffer->c_str(),fp);
+       buffer->clear();
+     };
+     return (j-i);
+   }
+
+   void PrintNode(treeNode n, int fd);
    /** Data structure constructors */
    XMLTree(){ buffer = 0;};
 
@@ -458,7 +473,7 @@ public:
    uchar* GetText(DocID d) {
      
        uchar * s = Text->GetText(d);
-       return (s[0] == 1 ? (uchar*)"" : s);
+       return (s[0] == 1 ? (s+1) : s);
    }
 
    /** GetText(i, j): returns the texts corresponding to documents with
diff --git a/makefile b/makefile
index 2ba3549..4eae048 100644
--- a/makefile
+++ b/makefile
@@ -1,4 +1,4 @@
-FLAGS =-std=c++0x -O9 -I./libcds/includes/ -I. -fno-PIC
+FLAGS =-std=c++0x -O3 -I./libcds/includes/ -I. -fno-PIC
 
 LIBCDS_A=libcds/lib/libcds.a 
 OBJECTS_TCO= TextCollection/TextCollection.o TextCollection/TextCollectionBuilder.o TextCollection/TCImplementation.o TextCollection/Tools.o TextCollection/BitRank.o TextCollection/TextStorage.o