Changeset 12884

Timestamp:

2006-09-28T10:44:01+12:00 (18 years ago)

Author:

kjdon

Message:

Accent folding patch thanks to Juan Grigera. parsing of stem/case/accent term modifiers now uses defines from mg_files.h

File:

• trunk/gsdl/src/mgpp/text/GSDLQueryParser.cpp (modified) (4 diffs)

trunk/gsdl/src/mgpp/text/GSDLQueryParser.cpp

@@ -141 +141 @@
 
 static unsigned long GetStemMethod(LexEl &el, int defaultStemMethod) {
-  // here expect el to contain some of c,s,i,u
-  // stem method 0 = c,u  00
-  // stem method 1 = i,u  01  - default for DL
-  // stem method 2 = c, s  10
-  // stem method 3 = i,s  11
-
+  // here expect el to contain some of c,s,i,u,f,a -- see mg_files.h CHAR_FLAG_STEM_* constants
   unsigned long stem = (unsigned long)defaultStemMethod;
       
@@ -152 +147 @@
   UCArray::const_iterator end = el.text.end();
 
-  unsigned char c1 = *here; 
-  if (!(c1 == 'c'|| c1 == 'i' || c1 == 'u' || c1 == 's')) 
-    return 4; // incorrect format
-  
-  ++here;
-  unsigned char c2 = 'a';
-  if (here !=end) {
-    c2 = *here;
-    if (!(c2 == 'c'|| c2 == 'i' || c2 == 'u' || c2 == 's')) 
-      return 4; // incorrect format
-  }
-  
-  if (c1 == 'i'|| c2=='i') stem |= 1; // set bit 0 to 1
-  if (c1 == 'c' || c2 == 'c') stem &=0xe; //set bit 0 to 0
-  if (c1 == 's'|| c2 == 's') stem |= 2; // set bit 1 to 1
-  if (c1 == 'u' || c2 =='u') stem &=0xd; // set bit 1 to 0
+  /* [JFG - Mar 06: Accent folding patch] */
+  /* Changed to use CHAR_FLAG_STEM* constants from mg_files.h */
+  while(here != end) {
+    unsigned char ch = *here; 
+    if (strchr (CHAR_FLAG_STEM_Validator, ch) == NULL)
+        return STEM_INVALID; // incorrect format
+
+    switch(ch) {
+        case CHAR_FLAG_STEM_CaseFold:       // ignore case (fold)
+            stem |= STEM_CaseFolding;
+            break;
+        case CHAR_FLAG_STEM_NoCaseFold:     // case sensitive 
+            stem &= (~STEM_CaseFolding);
+            break;
+        case CHAR_FLAG_STEM_Stemming:       // stem words
+            stem |= STEM_Stemming;
+            break;
+        case CHAR_FLAG_STEM_NoStemming:     // do not stem words
+            stem &= (~STEM_Stemming);
+            break;
+        case CHAR_FLAG_STEM_AccentFold:     // accent fold
+            stem |= STEM_AccentFolding;
+            break;
+        case CHAR_FLAG_STEM_NoAccentFold:   // do no accent folding
+            stem &= (~STEM_AccentFolding);
+            break;
+    };
+
+    ++here;     
+  }
   return stem;
 }
@@ -190 +199 @@
       if (ParseLexEl (here, end, stem) && stem.lexType == TermE) {
     termNode.stemMethod = GetStemMethod(stem, defaultStemMethod);
-    if (termNode.stemMethod == 4) { // error so backtrack
+    /* [JFG - Mar 06: Accent folding patch] */
+    /* use STEM_INVALID instead of hardcoded 4 */
+    if (termNode.stemMethod == STEM_INVALID) { // error so backtrack
       here = oldHere;
       termNode.stemMethod = (unsigned long)defaultStemMethod;
@@ -211 +222 @@
     
     if (partial_match) {
-      termNode.stemMethod |= 4; // set bit 2 to 1
-      termNode.stemMethod &=0xd; // set bit 1 to 0 // we dont have stemming on if doing partial matching.
+      /* [JFG - Mar 06: Accent folding patch] */
+      /* use STEM_PARTIAL_MATCH flag */
+      termNode.stemMethod |= STEM_PARTIAL_MATCH;    // set partial match flag
+      termNode.stemMethod &= (~STEM_Stemming);      // we dont have stemming on if doing partial matching.
     }
     oldHere = here;