Backport changes from the grammar branch

[SXSI/XMLTree.git] / XMLTree.h

diff --git a/XMLTree.h b/XMLTree.h
index a3ac2ed..6d11d78 100644 (file)
--- a/XMLTree.h
+++ b/XMLTree.h
@@ -20,13 +20,12 @@
 \r
 #ifndef XMLTREE_H_\r
 #define XMLTREE_H_\r
+\r
+\r
 #include <unordered_set>\r
 #include <unordered_map>\r
+#include <sstream>\r
 #include "TextCollection/TextCollectionBuilder.h"\r
-#include <stdio.h>\r
-#include <stdlib.h>\r
-#include <cstring>\r
-\r
 \r
 #undef W\r
 #undef WW\r
@@ -77,6 +76,8 @@ typedef struct {
 #define PCDATA_TAG_ID 2\r
 #define ATTRIBUTE_DATA_OPEN_TAG "<@$>"\r
 #define ATTRIBUTE_DATA_TAG_ID 3\r
+#define CLOSING_TAG   "</>"\r
+#define CLOSING_TAG_ID 4\r
 #define DOCUMENT_CLOSE_TAG "/"\r
 #define ATTRIBUTE_CLOSE_TAG "/<@>"\r
 #define PCDATA_CLOSE_TAG "/<$>"\r
@@ -84,7 +85,7 @@ typedef struct {
 \r
 \r
 typedef std::unordered_set<int> TagIdSet;\r
-typedef std::unordered_map<string,int> TagIdMap;\r
+typedef std::unordered_map<std::string,int> TagIdMap;\r
 typedef TagIdMap::const_iterator TagIdMapIT;\r
 \r
 #define REGISTER_TAG(v,h,t) do { (h)->insert(std::make_pair((t),(v)->size()));\\r
@@ -95,6 +96,10 @@ typedef TagIdMap::const_iterator TagIdMapIT;
 #define NULLT_IF(x)  do { if (x) return NULLT; } while (0)\r
 \r
 \r
+\r
+\r
+\r
+\r
 class XMLTreeBuilder;\r
 \r
 class XMLTree {\r
@@ -107,7 +112,7 @@ class XMLTree {
    bp *Par;\r
  \r
    /** Mapping from tag identifer to tag name */  \r
-   vector<string> *TagName;\r
+   std::vector<std::string> *TagName;\r
    TagIdMap * tIdMap;\r
   \r
    /** Bit vector indicating with a 1 the positions of the non-empty texts. */\r
@@ -125,13 +130,47 @@ class XMLTree {
    bool disable_tc;\r
    \r
    FILE* stream;\r
-   int   stream_fd;\r
+   int   stream_fd; \r
+   std::string * buffer;\r
+   void myfputs(const char* s, FILE * fp){\r
+     buffer->append(s);\r
+     if (buffer->size() >= 100000){\r
+       fputs(buffer->c_str(),fp);\r
+       buffer->clear();\r
+     };\r
 \r
+   }\r
+   void myfputc(const char c, FILE*fp){\r
+     buffer->append(1,c);\r
+     if (buffer->size() >= 100000){\r
+       fputs(buffer->c_str(),fp);\r
+       buffer->clear();\r
+     };\r
+   }\r
+   void mybufferflush(FILE* fp){\r
+     fputs(buffer->c_str(), fp);\r
+     buffer->clear();\r
+   }\r
+\r
+   size_t myfprintf(const char* s, FILE * fp){\r
+     if (s == NULL)\r
+       return 0;\r
+     size_t i = buffer->size();\r
+     buffer->append(s);\r
+     size_t j = buffer->size();\r
+     if (buffer->size() >= 100000){\r
+       fputs(buffer->c_str(),fp);\r
+       buffer->clear();\r
+     };\r
+     return (j-i);\r
+   }\r
+\r
+   void PrintNode(treeNode n, int fd);\r
    /** Data structure constructors */\r
-   XMLTree(){;};\r
+   XMLTree(){ buffer = 0;};\r
 \r
    // non const pointer are freed by this method.\r
-  XMLTree( pb * const par, uint npar,  vector<string> * const TN,  TagIdMap * const tim, uint *empty_texts_bmp, TagType *tags,\r
+   XMLTree( pb * const par, uint npar,  std::vector<std::string> * const TN,  TagIdMap * const tim, uint *empty_texts_bmp, TagType *tags,\r
           TextCollection * const TC, bool dis_tc);\r
 \r
 public: \r
@@ -139,7 +178,7 @@ public:
    ~XMLTree();\r
    \r
    /** root(): returns the tree root. */\r
-   treeNode Root();\r
+   treeNode Root() { return 0; }\r
 \r
    /** Size() :  Number of parenthesis */\r
    unsigned int Size(){\r
@@ -234,7 +273,7 @@ public:
     *    if none.\r
     */\r
    treeNode NextElement(treeNode x);\r
-   \r
+\r
    /** PrevSibling(x): returns the previous sibling of node x, assuming it \r
     * exists. */\r
 \r
@@ -423,7 +462,7 @@ public:
    uchar* GetText(DocID d) {\r
      \r
        uchar * s = Text->GetText(d);\r
-       return (s[0] == 1 ? (uchar*)"" : s);\r
+       return (s[0] == 1 ? (s+1) : s);\r
    }\r
 \r
    /** GetText(i, j): returns the texts corresponding to documents with\r
@@ -456,8 +495,13 @@ public:
 \r
 \r
    /** Print procedure */\r
-   void Print(int fd,treeNode x);\r
+   void Print(int fd,treeNode x, bool no_text);\r
+   void Print(int fd,treeNode x) { Print(fd,x,false); }\r
 \r
 };\r
+\r
+\r
+\r
+\r
 #endif\r
 \r