Changeset 1683

Timestamp:

2000-11-20T21:24:33+13:00 (23 years ago)

Author:

paynter

Message:

Bug fix - all documents returned for phrases of more than one word were
being stored with an out-by-one error. The probem is that internally
documents are numbered 0 to N-1, but in MGPP they're numbered 1 to N. I
thought I'd fixed this, but had only updated the code for the
phrases-of-length-1 case.

File:

• trunk/gsdl/src/phind/generate/suffix.cpp (modified) (1 diff)

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

@@ -436 +436 @@
         phraseData << ";";
       }
-      phraseData << "d" << i; 
+      // Output the document number.  Note that here we've numbered the 
+      // N documents from 0 to N-1, but later they'll be 1-N.  Thus we
+      // add 1 to the document id when we output it.
+      phraseData << "d" << (i+1);
+      // Next, output the frequency with which the document occurs, but
+      // only if it is > 1.
       if (documentFrequency[i] > 1) {
         phraseData << "," << documentFrequency[i];