Changeset 13477 for trunk/indexers/mgpp/text/mgpp_stem_idx.cpp

Timestamp:

2006-12-11T11:22:20+13:00 (17 years ago)

Author:

shaoqun

Message:

added code for accentfolding

File:

• trunk/indexers/mgpp/text/mgpp_stem_idx.cpp (modified) (3 diffs)

trunk/indexers/mgpp/text/mgpp_stem_idx.cpp

@@ -80 +80 @@
 
     // stem the word
-    stemmer (stemMethod, stemmerNum, mgWord);
+    mgpp_stemmer (stemMethod, stemmerNum, mgWord);
 
     // convert the result back to a UCArray
@@ -101 +101 @@
             int stemmerNum,
             unsigned long entriesPerBlock) {
+  
+  /* [JFG - Mar 06: Accent folding patch] */
   // Create appropriate stem index file
   FILE *stemDictFile = NULL;
-  if (stemMethod == 1) {
-    stemDictFile = create_file (filename, INVF_DICT_BLOCKED_1_SUFFIX,
-                "wb", MAGIC_STEM_1, MG_ABORT);
-  } else if (stemMethod == 2) {
-    stemDictFile = create_file (filename, INVF_DICT_BLOCKED_2_SUFFIX,
-                "wb", MAGIC_STEM_2, MG_ABORT);
-  } else if (stemMethod == 3) {
-    stemDictFile = create_file (filename, INVF_DICT_BLOCKED_3_SUFFIX,
-                "wb", MAGIC_STEM_3, MG_ABORT);
-  } else {
+  if (stemMethod >= STEM_MIN && stemMethod <= STEM_MAX) {
+    char *suffix = make_suffix (INVF_DICT_BLOCKED_SUFFIX_PAT, stemMethod, NULL);
+    stemDictFile = create_file (filename, suffix,
+                "wb", MAGIC_STEM_GEN(stemMethod + '0'), MG_ABORT);  
+  }
+  else {
     FatalError (1, "Unknown stem method %d", stemMethod);
   }
@@ -213 +211 @@
       break;
     case 'a':
-      stemmerNum = stemmernumber ((unsigned char *) optarg);
+      stemmerNum = mgpp_stemmernumber ((unsigned char *) optarg);
       break;
     case 'h':
     case '?':
       fprintf (stderr, "usage: %s [-d directory] "
-           "[-b entries-per-block] [-h] -s 1|2|3 "
-           "[-a stemmer-method] -f name\n", argv[0]);
+           "[-b entries-per-block] [-h] -s 1|2|3", argv[0]);
+#ifdef ENABLE_ACCENTFOLD
+      fprintf (stderr, "|4|5|6|7");
+#endif
+      fprintf (stderr, " [-a stemmer-method] -f name\n");
       exit (1);
     }
   }
   
-  if (stemMethod < 1 || stemMethod > 3)
-    FatalError (1, "Stem method must be 1, 2 or 3");
-
+  /* [JFG - Mar 06: Accent folding patch] */
+  if (stemMethod < STEM_MIN || stemMethod > STEM_MAX) 
+    FatalError (1, "Stem method must be between %d and %d", STEM_MIN, STEM_MAX);
+#ifndef ENABLE_ACCENTFOLD
+  if (stemMethod & STEM_AccentFolding) {
+    // accent folding not enabled
+    return -1;
+  }
+#endif
   // read in the dictionary and create the in memory dictionary
   StemMapDict stemDict;