Changeset 2807

Timestamp:

2001-10-30T15:49:34+13:00 (22 years ago)

Author:

kjm18

Message:

changed suffixCheck to bit array.

File:

• trunk/gsdl/src/phind/generate/suffix.cpp (modified) (7 diffs)

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

@@ -65 +65 @@
 check    *suffixCheck;
 
+// the length of the check array
+cellcount checkLength;
 
 // How many documents are in this collection?
@@ -91 +93 @@
 int pointerCompare(const void *, const void *);
 
-
+// some bit manipulation functions for the check arrays, defined below
+int getSuffixCheck(cellindex suff);
+void setSuffixCheck(cellindex suff);
+                                           
 // Functions for implementing "phrase memory".  These let us "remember"
 // each phrase that we've expanded without using too much memory.
@@ -137 +142 @@
   // Initialise the candidates, check array, and various variables.
   sort(candidates.begin(), candidates.end(), isShorter);
-  memset(suffixCheck, 0, sizeof(check)*inputLength);
-  //for (cellcount j = 0; j < inputLength; j++)
-  //  suffixCheck[j] = 0;
+  memset(suffixCheck, 0, sizeof(check)*checkLength);
   unsigned minimum_length = candidates.begin()->length;
   
@@ -153 +156 @@
     while (temp_phrase.length >= minimum_length && !shorter_found) {
       temp_phrase.ensureSuffixFound();
-      if (suffixCheck[temp_phrase.firstSuffixIndex] == 0)
+      //if (suffixCheck[temp_phrase.firstSuffixIndex] == 0)
+      if (getSuffixCheck(temp_phrase.firstSuffixIndex)==0)
     temp_phrase.shortenByOneAtPrefix();
       else
@@ -166 +170 @@
       candidate->ensureSuffixFound();
       for (cellcount k = candidate->firstSuffixIndex; k <= candidate->lastSuffixIndex; ++k)
-    suffixCheck[k] = candidate->length;
+    //suffixCheck[k] = candidate->length;
+    setSuffixCheck(k);
     }
   }
@@ -249 +254 @@
 }
 
+int getSuffixCheck(cellindex suff) {
+  cellindex cell = suff >> 3;
+  check remainder = suff & 0x07; // the last 3 bits
+  if (suffixCheck[cell]& (1 << remainder)) {
+    return 1;
+  }
+  return 0;
+}                                                                               
+void setSuffixCheck(cellindex suff) {
+  cellindex cell = suff >> 3;
+  check remainder = suff & 0x07; // the last 3 bits
+  suffixCheck[cell] |= (1 << remainder);
+
+}  
 
 // Read the clauses.numbers file into the "symbols" array.
@@ -786 +805 @@
   qsort(prefixArray, contentLength, sizeof(symbol *), prefixCompare);
 
-  suffixCheck = new check[contentLength];
+  checkLength = contentLength/8 + 1;
+  suffixCheck = new check[checkLength];
   if (suffixCheck == NULL) {
     cerr << "Suffix error: not enough memory to hold " << inputLength << " symbols." << endl;
     exit(2);
   }  
-  //for (cellcount j = 0; j < contentLength; j++)
-  //  suffixCheck[j] = 0;
-  memset(suffixCheck, 0, sizeof(check)*contentLength);
+  memset(suffixCheck, 0, sizeof(check)*checkLength);
 
   cout <<"\ngenerating the phrase hierarchy\n\n";