| Index: trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/schema/GlobalConceptStoreSchema.java |
| — | — | @@ -1,11 +1,9 @@ |
| 2 | 2 | package de.brightbyte.wikiword.schema; |
| 3 | 3 | |
| 4 | 4 | import java.sql.Connection; |
| 5 | | -import java.sql.ResultSet; |
| 6 | 5 | import java.sql.SQLException; |
| 7 | 6 | import java.util.ArrayList; |
| 8 | 7 | import java.util.Arrays; |
| 9 | | -import java.util.List; |
| 10 | 8 | import java.util.Map; |
| 11 | 9 | |
| 12 | 10 | import javax.sql.DataSource; |
| — | — | @@ -27,7 +25,7 @@ |
| 28 | 26 | protected RelationTable mergeTable; |
| 29 | 27 | protected RelationTable langprepTable; |
| 30 | 28 | |
| 31 | | - protected ConceptTypeSet conceptTypes; |
| | 29 | + private ConceptTypeSet conceptTypes; |
| 32 | 30 | protected TweakSet tweaks; |
| 33 | 31 | |
| 34 | 32 | public GlobalConceptStoreSchema(DatasetIdentifier dataset, Connection connection, TweakSet tweaks, boolean useFlushQueue) throws SQLException { |
| — | — | @@ -40,7 +38,7 @@ |
| 41 | 39 | init(tweaks); |
| 42 | 40 | } |
| 43 | 41 | |
| 44 | | - protected Corpus[] languages; |
| | 42 | + private Corpus[] languages; |
| 45 | 43 | |
| 46 | 44 | private void init(TweakSet tweaks) throws SQLException { |
| 47 | 45 | int nameSize = 32*255+32*16; //TODO: from tweaks! |
| — | — | @@ -100,7 +98,7 @@ |
| 101 | 99 | |
| 102 | 100 | //TODO: reference table (aka link) |
| 103 | 101 | |
| 104 | | - getLanguages(); //initialize knownlanguages, corpuses and content types# |
| | 102 | + //getLanguages(); //initialize knownlanguages, corpuses and content types# |
| 105 | 103 | } |
| 106 | 104 | |
| 107 | 105 | /** |
| — | — | @@ -166,6 +164,27 @@ |
| 167 | 165 | return null; |
| 168 | 166 | } |
| 169 | 167 | |
| | 168 | + protected Corpus[] getCorpuses(String[] languages) throws SQLException { |
| | 169 | + Corpus[] cc = new Corpus[languages.length]; |
| | 170 | + |
| | 171 | + int i = 0; |
| | 172 | + for (String l: languages) { |
| | 173 | + if (!getLanguageNames().containsKey(l)) { |
| | 174 | + throw new SQLException("bad corpus prefix: "+l+" is not a language name. Hint: check tweaks languages.*AsLanguage"); |
| | 175 | + } |
| | 176 | + |
| | 177 | + cc[i++] = Corpus.forName(getCollectionName(), l, tweaks); |
| | 178 | + } |
| | 179 | + |
| | 180 | + return cc; |
| | 181 | + } |
| | 182 | + |
| | 183 | + public Corpus[] setLanguages(String[] languages) throws SQLException { |
| | 184 | + Corpus[] cc = getCorpuses(languages); |
| | 185 | + setLanguages(cc); |
| | 186 | + return cc; |
| | 187 | + } |
| | 188 | + |
| 170 | 189 | public void setLanguages(Corpus[] languages) { |
| 171 | 190 | if (languages.length>32) throw new IllegalArgumentException("only up to 32 languages are supported!"); |
| 172 | 191 | |
| — | — | @@ -203,24 +222,19 @@ |
| 204 | 223 | |
| 205 | 224 | public Corpus[] getLanguages() throws SQLException { |
| 206 | 225 | if (languages!=null) return languages; |
| | 226 | + setLanguages( detectLanguages() ); |
| | 227 | + return languages; |
| | 228 | + } |
| | 229 | + |
| | 230 | + private Corpus[] allLanguages = null; |
| | 231 | + public Corpus[] detectLanguages() throws SQLException { |
| | 232 | + if (allLanguages!=null) return allLanguages; |
| 207 | 233 | |
| 208 | 234 | String[] ll = listPrefixes("resource"); |
| 209 | | - if (ll.length>32) throw new IllegalArgumentException("only up to 32 languages are supported! found "+ll.length+" prefixes: "+Arrays.toString(ll)); |
| 210 | | - |
| 211 | 235 | Arrays.sort(ll); //FIXME: sort by size! |
| 212 | | - Corpus[] cc = new Corpus[ll.length]; |
| 213 | 236 | |
| 214 | | - int i = 0; |
| 215 | | - for (String l: ll) { |
| 216 | | - if (!getLanguageNames().containsKey(l)) { |
| 217 | | - throw new SQLException("database inconsistency: encountered bad corpus prefix: "+l+" is not a language name. Hint: check tweaks languages.*AsLanguage"); |
| 218 | | - } |
| 219 | | - |
| 220 | | - cc[i++] = Corpus.forName(getCollectionName(), l, tweaks); |
| 221 | | - } |
| 222 | | - |
| 223 | | - setLanguages(cc); |
| 224 | | - return languages; |
| | 237 | + allLanguages = getCorpuses(ll); |
| | 238 | + return allLanguages; |
| 225 | 239 | } |
| 226 | 240 | |
| 227 | 241 | @Override |
| Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/builder/BuildThesaurus.java |
| — | — | @@ -71,21 +71,15 @@ |
| 72 | 72 | String lang = args.getStringOption("languages", null); |
| 73 | 73 | if (lang!=null) { |
| 74 | 74 | String[] ll = lang.split("[,;/|\\s+]+"); |
| 75 | | - languages = new Corpus[ll.length]; |
| 76 | | - |
| 77 | | - int i = 0; |
| 78 | | - for (String l: ll) { |
| 79 | | - languages[i++] = Corpus.forName(getConfiguredCollectionName(), l, tweaks); |
| 80 | | - } |
| | 75 | + ((GlobalConceptStoreBuilder)this.conceptStore).setLanguages(ll); |
| 81 | 76 | } |
| 82 | 77 | |
| 83 | | - if (languages==null) { |
| 84 | | - languages = ((GlobalConceptStoreBuilder)this.conceptStore).detectLanguages(); |
| 85 | | - } |
| | 78 | + languages = ((GlobalConceptStoreBuilder)this.conceptStore).getLanguages(); |
| | 79 | + } else { |
| | 80 | + ((GlobalConceptStoreBuilder)this.conceptStore).setLanguages(languages); |
| 86 | 81 | } |
| 87 | 82 | |
| 88 | 83 | info("Using languages: "+Arrays.toString(languages)); |
| 89 | | - this.conceptStore.setLanguages(languages); |
| 90 | 84 | |
| 91 | 85 | if (agenda.beginTask("BuildThesaurus.run", "importConcepts")) { |
| 92 | 86 | this.conceptStore.importConcepts(); |
| Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/store/builder/GlobalConceptStoreBuilder.java |
| — | — | @@ -20,5 +20,7 @@ |
| 21 | 21 | public void setLanguages(Corpus[] languages); |
| 22 | 22 | |
| 23 | 23 | public int getMaxConceptId() throws PersistenceException; |
| | 24 | + |
| | 25 | + public void setLanguages(String[] languages) throws PersistenceException; |
| 24 | 26 | |
| 25 | 27 | } |
| \ No newline at end of file |
| Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/store/builder/DatabaseGlobalConceptStoreBuilder.java |
| — | — | @@ -59,7 +59,6 @@ |
| 60 | 60 | protected TweakSet tweaks; |
| 61 | 61 | |
| 62 | 62 | protected int idOffsetGranularity; |
| 63 | | - private Corpus[] languages; |
| 64 | 63 | |
| 65 | 64 | /** |
| 66 | 65 | * Constructs a DatabaseWikiStore, soring information from/about the given Corpus |
| — | — | @@ -136,7 +135,7 @@ |
| 137 | 136 | //------------------------------- |
| 138 | 137 | public Corpus[] detectLanguages() throws PersistenceException { |
| 139 | 138 | try { |
| 140 | | - Corpus[] languages = ((GlobalConceptStoreSchema)database).getLanguages(); |
| | 139 | + Corpus[] languages = ((GlobalConceptStoreSchema)database).detectLanguages(); |
| 141 | 140 | return languages; |
| 142 | 141 | } catch (SQLException e) { |
| 143 | 142 | throw new PersistenceException(e); |
| — | — | @@ -145,15 +144,24 @@ |
| 146 | 145 | |
| 147 | 146 | public void setLanguages(Corpus[] languages) { |
| 148 | 147 | if (languages==null) throw new NullPointerException(); |
| 149 | | - if (this.languages!=null) throw new IllegalStateException("languages already set"); |
| 150 | | - //XXX: set languages in GlobalConceptStoreSchema too? |
| 151 | | - |
| 152 | | - this.languages = languages; |
| | 148 | + ((GlobalConceptStoreSchema)database).setLanguages(languages); |
| 153 | 149 | } |
| 154 | 150 | |
| | 151 | + public void setLanguages(String[] languages) throws PersistenceException { |
| | 152 | + if (languages==null) throw new NullPointerException(); |
| | 153 | + try { |
| | 154 | + ((GlobalConceptStoreSchema)database).setLanguages(languages); |
| | 155 | + } catch (SQLException e) { |
| | 156 | + throw new PersistenceException(e); |
| | 157 | + } |
| | 158 | + } |
| | 159 | + |
| 155 | 160 | public Corpus[] getLanguages() throws PersistenceException { |
| 156 | | - if (languages==null) languages = detectLanguages(); |
| 157 | | - return languages; |
| | 161 | + try { |
| | 162 | + return ((GlobalConceptStoreSchema)database).getLanguages(); |
| | 163 | + } catch (SQLException e) { |
| | 164 | + throw new PersistenceException(e); |
| | 165 | + } |
| 158 | 166 | } |
| 159 | 167 | |
| 160 | 168 | public int getNextIdOffset() throws PersistenceException { |