From fa8da7f89d3d8df8a40719afc212ceb47355d074 Mon Sep 17 00:00:00 2001
From: Michael Christen <mc@yacy.net>
Date: Sun, 8 Jan 2012 02:13:52 +0100
Subject: [PATCH] vocabularies are now also used as source for a did-you-mean
 computation

---
 source/net/yacy/document/Autotagging.java     | 4 ++++
 source/net/yacy/document/LibraryProvider.java | 4 ++++
 source/net/yacy/document/WordCache.java       | 6 ++++++
 3 files changed, 14 insertions(+)

diff --git a/source/net/yacy/document/Autotagging.java b/source/net/yacy/document/Autotagging.java
index 28d8b9caa..c3c220cdd 100644
--- a/source/net/yacy/document/Autotagging.java
+++ b/source/net/yacy/document/Autotagging.java
@@ -89,6 +89,10 @@ public class Autotagging {
         }
     }
 
+    public Set<String> allTags() {
+        return this.allTags.keySet();
+    }
+
     public void addDictionaries(Map<String, Dictionary> dictionaries) {
         for (Map.Entry<String, Dictionary> entry: dictionaries.entrySet()) {
             Vocabulary voc = new Vocabulary(entry.getKey(), entry.getValue());
diff --git a/source/net/yacy/document/LibraryProvider.java b/source/net/yacy/document/LibraryProvider.java
index 7074eeafa..37994fc03 100644
--- a/source/net/yacy/document/LibraryProvider.java
+++ b/source/net/yacy/document/LibraryProvider.java
@@ -32,6 +32,7 @@ import java.io.InputStreamReader;
 import java.io.PrintWriter;
 import java.net.MalformedURLException;
 import java.util.ArrayList;
+import java.util.HashSet;
 import java.util.List;
 import java.util.Set;
 import java.util.TreeSet;
@@ -111,8 +112,11 @@ public class LibraryProvider
         integrateOpenGeoDB();
         integrateGeonames();
         initAutotagging(tagPrefix);
+        Set<String> allTags = new HashSet<String>() ;
+        allTags.addAll(autotagging.allTags()); // we must copy this into a clone to prevent circularity
         autotagging.addLocalization(geoLoc);
         autotagging.addDictionaries(dymLib.getDictionaries());
+        WordCache.learn(allTags);
     }
 
     public static void integrateOpenGeoDB() {
diff --git a/source/net/yacy/document/WordCache.java b/source/net/yacy/document/WordCache.java
index a89d316f5..a6998c8d4 100644
--- a/source/net/yacy/document/WordCache.java
+++ b/source/net/yacy/document/WordCache.java
@@ -220,6 +220,12 @@ public class WordCache {
         }
     }
 
+    public static void learn(Set<String> wordset) {
+        for (String s: wordset) {
+            learn(new StringBuilder(s));
+        }
+    }
+
     /**
      * scan the input directory and load all dictionaries (again)
      */