Ticket #296: scriptindex.cc.spelling.diff

scriptindex.cc

@@ -95 +95 @@
 const char * action_names[] = {
     "bad", "new",
     "boolean", "date", "field", "hash", "index", "indexnopos", "load", "lower",
-    "truncate", "unhtml", "unique", "value", "weight"
+    "spell", "truncate", "unhtml", "unique", "value", "weight"
 };
 
 // For debugging:
@@ -106 +106 @@
     typedef enum {
         BAD, NEW,
         BOOLEAN, DATE, FIELD, HASH, INDEX, INDEXNOPOS, LOAD, LOWER,
-        TRUNCATE, UNHTML, UNIQUE, VALUE, WEIGHT
+        SPELL, TRUNCATE, UNHTML, UNIQUE, VALUE, WEIGHT
     } type;
 private:
     type action;
@@ -233 +233 @@
                             code = Action::LOAD;
                         }
                         break;
+                    case 's':
+                        if (action == "spell") {
+                            code = Action::SPELL;
+                        }
+                        break;
                     case 't':
                         if (action == "truncate") {
                             code = Action::TRUNCATE;
@@ -374 +379 @@
             switch (action) {
                 case Action::HASH:
                 case Action::LOWER:
+                case Action::SPELL:
                 case Action::TRUNCATE:
                 case Action::UNHTML:
                     done = false;
@@ -429 +435 @@
         Xapian::docid docid = 0;
         map<string, list<string> > fields;
         bool seen_content = false;
+        bool add_spelling = false;
         while (!line.empty()) {
             // Cope with files from MS Windows (\r\n end of lines).
             // Trim multiple \r characters, since that seems the best way
@@ -458 +465 @@
                 line.resize(last + 1);
                 value += line;
             }
+            // don't add words to the spelling dictionary by default, so clear this flag if it was set in a previous iteration
+                add_spelling = false;
+            if(indexer.set_flags((Xapian::TermGenerator::flags)0, (Xapian::TermGenerator::flags)(~0)) & indexer.FLAG_SPELLING)
+                indexer.set_flags((Xapian::TermGenerator::flags)0, (Xapian::TermGenerator::flags)(~indexer.FLAG_SPELLING));
 
             const vector<Action> &v = index_spec[field];
             string old_value = value;
@@ -486 +497 @@
                         }
                         break;
                     case Action::INDEX:
+                    if(add_spelling)
+                    {
+                        indexer.set_database(database);
+                        indexer.set_flags(indexer.FLAG_SPELLING, (Xapian::TermGenerator::flags)(~indexer.FLAG_SPELLING));
+                    }
                         indexer.index_text(value,
                                            i->get_num_arg(),
                                            i->get_string_arg());
                         break;
                     case Action::INDEXNOPOS:
+                    if(add_spelling)
+                    {
+                        indexer.set_database(database);
+                        indexer.set_flags(indexer.FLAG_SPELLING, (Xapian::TermGenerator::flags)(~indexer.FLAG_SPELLING));
+                    }
                         // No positional information so phrase searching
                         // won't work.  However, the database will use much
                         // less diskspace.
@@ -534 +555 @@
                     case Action::TRUNCATE:
                         utf8_truncate(value, i->get_num_arg());
                         break;
+                    case Action::SPELL:
+                        add_spelling = true;
+                        break;
                     case Action::UNHTML: {
                         MyHtmlParser p;
                         try {