From 164adea5e18f69c6afae6772ab8cb4444816d236 Mon Sep 17 00:00:00 2001
From: nvalimak <nvalimak@3cdefd35-fc62-479d-8e8d-bae585ffb9ca>
Date: Fri, 29 May 2009 14:40:05 +0000
Subject: [PATCH] Approx pattern matching functions

git-svn-id: svn+ssh://idea.nguyen.vg/svn/sxsi/trunk/XMLTree@422 3cdefd35-fc62-479d-8e8d-bae585ffb9ca
---
 XMLTree.h | 50 ++++++++++++++++++++++++++++++++++++++++----------
 1 file changed, 40 insertions(+), 10 deletions(-)

diff --git a/XMLTree.h b/XMLTree.h
index 275b9bc..ee44d32 100644
--- a/XMLTree.h
+++ b/XMLTree.h
@@ -40,7 +40,7 @@
 #include <static_sequence.h>
 using SXSI::TextCollection;
 using SXSI::TextCollectionBuilder;
-
+using SXSI::TextStorage;
 
 // this constant is used to efficiently compute the child operation in the tree
 #define OPTD 10
@@ -325,6 +325,20 @@ public:
    TextCollection::document_result LessThan(uchar const *s) {
       return Text->LessThan(s);
    }
+
+   /** KMismatches (s): returns document identifiers for the texts that
+    * contain occurrence of string s with at most K mismatches. */
+   TextCollection::document_result KMismatches(uchar const *s, unsigned K) {
+       return Text->KMismatches(s, K);
+   }
+
+   /** KErrors (s): returns document identifiers for the texts that
+    * contain occurrence of string s with at most K errors. 
+    * The accepted "errors" are insertions, deletions and mutations of chars.
+    */
+   TextCollection::document_result KErrors(uchar const *s, unsigned K) {
+       return Text->KErrors(s, K);
+   }
    
    /** IsPrefix(x): returns true if there is a text prefixed by string s. */
    bool IsPrefix(uchar const *s) {
@@ -385,18 +399,34 @@ public:
    }
    
    /** GetText(d): returns the text corresponding to document with
-    * id d. */
-   uchar* GetText(DocID d) {
-     uchar * s = Text->GetText(d);
-     return (s[0] == 1 ? (uchar*)"" : s);
+    * id d. 
+    *
+    * Implementation of GetText() may or may NOT 
+    * require you to free() the pointer that is returned.
+    * Call DeleteText() for each pointer returned by GetText()
+    * to avoid possible memory leaks.
+    */
+   uchar* GetText(DocID d) const {
+     return Text->GetText(d);
+   }
+
+   /**
+    * Free the pointer returned by GetText().
+    */
+   void DeleteText(uchar *text) const {
+       Text->DeleteText(text);
    }
 
    /** GetText(i, j): returns the texts corresponding to documents with
-    * ids i, i+1, ..., j. Texts are separated by '\0' character.  */
-   //   uchar* GetText(DocID i, DocID j) {
-   //  uchar * s = Text->GetText(i, j);
-   // return (s[0] == 1 ? (uchar*)"" : s);
-   //}
+    * ids i, i+1, ..., j. Texts are separated by '\0' character.  
+    *
+    * Call DeleteText() for each pointer returned by GetText()
+    * to avoid possible memory leaks.
+    */
+   uchar* GetText(DocID i, DocID j) const {
+     return Text->GetText(i, j);
+   }
+
 
    TextCollection *getTextCollection() {
       return Text;
-- 
2.17.1