Changeset 8919 for trunk/indexers/mg

Timestamp:

2005-01-20T16:06:52+13:00 (19 years ago)

Author:

kjdon

Message:

added some bits (copied from mgquery) to make paragraph searching work - it wasn't returning doc numbers, but was returning para numbers which are not in the gdbm database

File:

• trunk/indexers/mg/jni/MGWrapperImpl.c (modified) (3 diffs)

trunk/indexers/mg/jni/MGWrapperImpl.c

@@ -57 +57 @@
 #define MAX_INDEXES_CACHED 3
 
+/* copied from mgquery, needed to convert paragraph numbers to document numbers
+   for greenstone */
+#if defined(PARADOCNUM) ||  defined(NZDL)
+static int GetDocNumFromParaNum(query_data *qd, int paranum) {
+  int Documents = qd->td->cth.num_of_docs;
+  int *Paragraph = qd->paragraph;
+  int low = 1, high = Documents;
+  int mid = (low+high)/2;
+
+  while ((mid = (low+high)/2) >=1 && mid <= Documents)
+    {
+      if (paranum > Paragraph[mid])
+        low = mid+1;
+      else if (paranum <= Paragraph[mid-1])
+        high = mid-1;
+      else
+        return mid;
+    }
+  FatalError(1, "Bad paragraph number.\n");
+  return 0;
+}
+#endif
+
 
 /*********************************************
@@ -407 +430 @@
     rqi.MaxDocsToRetrieve = data->queryInfo->maxDocs;  /* Get only the desired number */
     rqi.MaxParasToRetrieve = rqi.MaxDocsToRetrieve;  /* OK? */
+    /* we may need to get more paragraphs to get enough documents. I copied the following from mgquery. it seems to work, not sure why - kjdon */
+    if (qd->id->ifh.InvfLevel == 3 && GetEnv ("maxparas")) {
+      rqi.MaxParasToRetrieve = atoi (GetEnv ("maxparas"));
+    }
+    
     rqi.Sort = 1;  /* Sort the query terms by frequency before ranking */
     rqi.AccumMethod = 'L';  /* Use a list when accumulating (has bugs though...) */
@@ -450 +478 @@
   /* Record the matching documents, but only the number requested */
   printf("Number of matching documents: %d\n", qd->DL->num);
+  
   for (i = 0; (i < qd->DL->num && i < data->queryInfo->maxDocs); i++) {
     int doc_num = qd->DL->DE[i].DocNum;
+#if defined(PARADOCNUM) || defined(NZDL)
+    if (qd->id->ifh.InvfLevel == 3) {
+      /* pararaph level, need to convert to doc level*/
+      doc_num = GetDocNumFromParaNum(qd, doc_num);
+    }
+#endif
     float doc_weight = qd->DL->DE[i].Weight;