Changeset 1873

Timestamp:

2001-01-30T12:08:49+13:00 (23 years ago)

Author:

paynter

Message:

Fixed a bug iin the Phrase extraction alogrithm that had the
"candidates" in the GetMinimalExpansions function sorted backwards.

Location:

trunk/gsdl/src/phind/generate

Files:

• phrase.cpp (modified) (9 diffs)
• phrase.h (modified) (1 diff)
• suffix.cpp (modified) (5 diffs)

trunk/gsdl/src/phind/generate/phrase.cpp

@@ -180 +180 @@
 // Does the phrase have a unique suffix/prefix extension?
 //
-// If uniqueSuffixExtensions is 1, then we do.
-// If it is > 1 then we do not.
-// If it is -1 then we have to check in the suffix array.
+// For suffix:
+//   If uniqueSuffixExtensions is 1, then phrase has some unique expansion. 
+//   If it is 0 then it does not.  If it is -1 then we don't know, and have
+//   to calculate it from the suffix array.  Same goes for prefixes.
 
 int Phrase::hasUniqueSuffixExtension() {
@@ -189 +190 @@
   if (uniqueSuffixExtension == -1) {
 
-    if (!suffixFound) {
-      findFirstAndLastSuffix();
-    }
-
-    /*    if (firstSuffixIndex == 56957) {
-      cout << "Testing suffix expansion: " << toString() << "\n";
-      cout << "mode: " << phraseMode << endl;
-    }
-    */
+    ensureSuffixFound();
 
     // pointers to the phrase's first and last occurances in te suffixArray
@@ -235 +228 @@
     ensurePrefixFound();
 
-    // pointers to the phrase's first and last occurances in te prefixArray
+    // pointers to the phrase's first and last occurances in the prefixArray
     symbol *fst = firstPrefix - length;
     symbol *lst = lastPrefix - length;
@@ -249 +242 @@
     // in STOPWORDS mode, make sure there is a unique next content symbol
     else {
-      // if (firstSuffixIndex == 56962) cout << "Testing 1: s" << *fst << " = s" << *lst << endl;
       uniquePrefixExtension = 0;
       while ((*fst == *lst) && (*fst > LASTDELIMITER)) {
@@ -266 +258 @@
 
 
-// Expand a phrase with a unique suffix/prefix extension by 1 symbol
+// Expand a phrase with a unique suffix/prefix by 1 symbol
 //
 // Note that in STOPWORDS mode a "unique extension" means a unique
@@ -274 +266 @@
 
 int Phrase::expandUniqueSuffixExtensionByOne() {
-  assert(forward);
   assert(suffixFound);
   assert(uniqueSuffixExtension == 1);
@@ -299 +290 @@
 
 int Phrase::expandUniquePrefixExtensionByOne() {
-  assert(back);
   assert(prefixFound);
   assert(uniquePrefixExtension == 1);
@@ -797 +787 @@
 // Compare the length of two phrases
 //
-// Given two phrases, return true if the first is shorter,
-// otherwise return false.  For use with the STL sort function.
+// Given two phrases, return true if the first is shorter/longer,
+// otherwise return false.  For use in various sort functions.
 
 bool isShorter(Phrase p1, Phrase p2) {
@@ -807 +797 @@
 }
 
-
+bool isLonger(Phrase p1, Phrase p2) {
+  if (p1.length > p2.length) {
+    return true;
+  }
+  return false;
+}
+
+

trunk/gsdl/src/phind/generate/phrase.h

@@ -126 +126 @@
 
 bool isShorter(Phrase p1, Phrase p2);
+bool isLonger(Phrase p1, Phrase p2);
 
 

trunk/gsdl/src/phind/generate/suffix.cpp

@@ -1 +1 @@
 /**********************************************************************
  *
- * suffix.h -- Extract the repeated phrases in the input using 
- *             suffix and prefix arrays.
+ * suffix.cpp -- Extract the repeated phrases in the input using 
+ *               suffix and prefix arrays.
  *
- * Copyright 2000 Gordon W. Paynter ([email protected])
+ * Copyright 2000 Gordon W. Paynter
  * Copyright 2000 The New Zealand Digital Library Project
  *
@@ -507 +507 @@
 
   // 2.2 Sort the candidates by phrase length
-  // sort(candidates.begin(), candidates.end(), isShorter);
-  make_heap(candidates.begin(), candidates.end(), isShorter);
-
+  make_heap(candidates.begin(), candidates.end(), isLonger);
 
   // 3. While candidates is non-empty, confirm the phrases it 
@@ -516 +514 @@
 
     // 3.1 Get next candidate
-    //Phrase c = candidates.front();
-    //candidates.erase(candidates.begin());
-    pop_heap(candidates.begin(), candidates.end(), isShorter);
+    pop_heap(candidates.begin(), candidates.end(), isLonger);
     Phrase c = candidates.back();
     candidates.pop_back();
@@ -538 +534 @@
     // cout << "expanding prefix " << c.toString() << "=> ";
     c.expandUniquePrefixExtensionByOne();
-    //candidates.push_back(c);
-    //sort(candidates.begin(), candidates.end(), isShorter);
     candidates.push_back(c);
-    push_heap(candidates.begin(), candidates.end(), isShorter);
+    push_heap(candidates.begin(), candidates.end(), isLonger);
      }
     
@@ -572 +566 @@
       else if (c.hasUniqueSuffixExtension()) {
     c.expandUniqueSuffixExtensionByOne();
-    //candidates.push_back(c);
-    //sort(candidates.begin(), candidates.end(), isShorter);
     candidates.push_back(c);
-    push_heap(candidates.begin(), candidates.end(), isShorter);
+    push_heap(candidates.begin(), candidates.end(), isLonger);
       }