Fix a bug in TaggedFoll

[SXSI/XMLTree.git] / XMLTree.h

diff --git a/XMLTree.h b/XMLTree.h
index 759aff8..7c08fcc 100644 (file)
--- a/XMLTree.h
+++ b/XMLTree.h
@@ -26,6 +26,7 @@
 #include <stdlib.h>\r
 #include <cstring>\r
 \r
+//KIM : OJO\r
 //clash between TextCollection/Tools.h and libcds/includes/basics.h\r
 #undef W\r
 #undef WW\r
@@ -61,6 +62,13 @@ typedef struct {
 } range;\r
 \r
 \r
+//KIM : OJO\r
+// I know this class implements the working draft that we have but the logics seem flawed here...\r
+// We should have two classes. One XMLTreeBuilder and one XMLTree.\r
+// XMLTreeBuilder would have OpenDocument, NewOpenTag,... and CloseDocument would return an XMLTree\r
+// XMLTree would have only an initialized structure. If find it really ugly to check (!finished) or (!initialized)\r
+// in every function (FirstChild....).\r
+\r
 class XMLTree {\r
    /** Balanced parentheses representation of the tree */\r
    bp *Par;\r
@@ -96,7 +104,17 @@ class XMLTree {
    int parArraySize;\r
    int ntagnames;\r
    unsigned int *empty_texts_aux;\r
+\r
+   // KIM : OJO\r
+   // I added those two. The TagName array should always contains two special tags\r
+   // <@> for attributes and <$> for PCDATA.\r
+   // <$> can never be in a document (since we handle the text differently)\r
+   // but <@> can be returned by the parser. This boolean is needed for the construction\r
+   // of the Tag bitmap to know if <@> must be taken into account or not\r
    bool found_attributes;\r
+\r
+   // KIM : OJO\r
+   // Allows to disable the TextCollection for benchmarkin purposes\r
    bool disable_tc;\r
    \r
 public:\r
@@ -226,6 +244,7 @@ public:
     * (i.e. everything is considered an empty text *)\r
     * Returns a non-zero value upon success, NULLT in case of \r
     * error. */\r
+\r
    int OpenDocument(bool empty_texts, int sample_rate_text, bool dtc);\r
 \r
    /** CloseDocument(): finishes the construction of the data structure for \r
@@ -265,6 +284,19 @@ public:
    unsigned char *GetTagName(TagType tagid);\r
 \r
 \r
+   // OJO\r
+   /** GetTagName(tagid): returns the tag name of a given tag identifier.     \r
+    *  The result is just a reference and should not be freed by the caller.\r
+    */\r
+   const unsigned char *GetTagNameByRef(TagType tagid);\r
+\r
+   //OJO\r
+   /** RegisterTag adds a new tag to the tag collection this is needed\r
+    * if the query contains a tag which is not in the document, we need\r
+    * to give this new tag a fresh id and store it somewhere. A logical\r
+    * choice is here.\r
+    * We might want to use a hashtable instead of an array though.\r
+    */\r
    TagType RegisterTag(unsigned char *tagname);\r
 \r
    bool EmptyText(DocID i) {\r