r64309 MediaWiki - Code Review archive

Repository:	MediaWiki
Revision:	< r64308‎ \| r64309 \| r64310 >
Date:	20:24, 28 March 2010
Author:	daniel
Status:	deferred
Tags:
Comment:	refactored DAO model (untested, but compiles now)
Modified paths:	/trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/model/ConceptRelations.java (modified) (history) /trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/model/LocalConcept.java (modified) (history) /trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/model/TermReference.java (modified) (history) /trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/model/WikiWordConcept.java (modified) (history) /trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/rdf/LocalConceptSkosProperties.java (modified) (history) /trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/rdf/RdfOutput.java (modified) (history) /trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/schema/ConceptInfoStoreSchema.java (modified) (history) /trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/store/CalculatedProximityStore.java (modified) (history) /trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/store/DatabaseGlobalConceptStore.java (modified) (history) /trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/store/DatabaseLocalConceptStore.java (modified) (history) /trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/store/builder/DatabaseConceptInfoStoreBuilder.java (modified) (history) /trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/store/builder/DatabaseLocalConceptStoreBuilder.java (modified) (history) /trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/store/builder/DatabaseWikiWordConceptStoreBuilder.java (modified) (history) /trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/store/builder/DebugLocalConceptStoreBuilder.java (modified) (history) /trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/store/builder/LocalConceptStoreBuilder.java (modified) (history) /trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/store/builder/WikiWordConceptStoreBuilder.java (modified) (history)

Diff [purge]

Index: trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/store/CalculatedProximityStore.java
—	—	@@ -42,7 +42,7 @@
43	43	double prox = getProximity(centerFeatures, f.getFeatureVector());
44	44	if (prox<minProximity) continue;
45	45
46		~~- return newConcept(f.getId(), f.getName(), 1, prox);~~
	46	+ return newConcept(f.getId(), f.getConcept().getName(), f.getConcept().getType(), 1, prox);
47	47	} ;
48	48
49	49	return null;
Index: trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/store/DatabaseLocalConceptStore.java
—	—	@@ -31,9 +31,6 @@
32	32	import de.brightbyte.wikiword.schema.LocalConceptStoreSchema;
33	33	import de.brightbyte.wikiword.schema.LocalStatisticsStoreSchema;
34	34	import de.brightbyte.wikiword.schema.StatisticsStoreSchema;
35		~~-import de.brightbyte.wikiword.store.DatabaseWikiWordConceptStore.DatabaseConceptInfoStore;~~
36		~~-import de.brightbyte.wikiword.store.DatabaseWikiWordConceptStore.DatabaseStatisticsStore;~~
37		~~-import de.brightbyte.wikiword.store.DatabaseWikiWordConceptStore.DatabaseConceptInfoStore.ConceptFactory;~~
38	35
39	36	/**
40	37	* A LocalConceptStore implemented based upon a {@link de.brightbyte.db.DatabaseSchema} object,
—	—	@@ -295,7 +292,7 @@
296	293	}
297	294
298	295	@Override
299		~~- protected LocalConcept newConcept(Map<String, Object> m) {~~
	296	+ protected LocalConcept newConcept(Map<String, Object> m) throws PersistenceException {
300	297	int id = asInt(m.get("cId"));
301	298	String name = asString(m.get("cName"));
302	299	ConceptType type = corpus.getConceptTypes().getType(asInt(m.get("cType")));
—	—	@@ -317,7 +314,7 @@
318	315	LocalConcept[] outlinks = LocalConcept.parseList( asString(m.get("rOutlinks")), getConceptFactory(), ((ConceptInfoStoreSchema)database).outLinksReferenceListEntry );
319	316	LocalConcept[] broader = LocalConcept.parseList( asString(m.get("rBroader")), getConceptFactory(), ((ConceptInfoStoreSchema)database).broaderReferenceListEntry );
320	317	LocalConcept[] narrower = LocalConcept.parseList( asString(m.get("rNarrower")), getConceptFactory(), ((ConceptInfoStoreSchema)database).narrowerReferenceListEntry );
321		~~- TranslationReference[] langlinks = TranslationReference.parseList( asString(m.get("rLanglinks")), ((ConceptInfoStoreSchema)database).langlinkReferenceListEntry );~~
	318	+ LocalConcept[] langlinks = LocalConcept.parseList( asString(m.get("rLanglinks")), getConceptFactory(), ((ConceptInfoStoreSchema)database).langlinkReferenceListEntry );
322	319	LocalConcept[] similar = LocalConcept.parseList( asString(m.get("rSimilar")), getConceptFactory(), ((ConceptInfoStoreSchema)database).similarReferenceListEntry );
323	320	LocalConcept[] related = LocalConcept.parseList( asString(m.get("rRelated")), getConceptFactory(), ((ConceptInfoStoreSchema)database).relatedReferenceListEntry );
324	321	TermReference[] terms = TermReference.parseList( asString(m.get("dTerms")), getConceptFactory(), ((ConceptInfoStoreSchema)database).termReferenceListEntry );
Index: trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/store/DatabaseGlobalConceptStore.java
—	—	@@ -28,9 +28,6 @@
29	29	import de.brightbyte.wikiword.schema.ConceptInfoStoreSchema;
30	30	import de.brightbyte.wikiword.schema.GlobalConceptStoreSchema;
31	31	import de.brightbyte.wikiword.schema.StatisticsStoreSchema;
32		~~-import de.brightbyte.wikiword.store.DatabaseWikiWordConceptStore.DatabaseConceptInfoStore;~~
33		~~-import de.brightbyte.wikiword.store.DatabaseWikiWordConceptStore.DatabaseStatisticsStore;~~
34		~~-import de.brightbyte.wikiword.store.DatabaseWikiWordConceptStore.DatabaseConceptInfoStore.ConceptFactory;~~
35	32
36	33	/**
37	34	* A GlobalConceptStore implemented based upon a {@link de.brightbyte.db.DatabaseSchema} object,
—	—	@@ -268,18 +265,22 @@
269	266
270	267	Corpus[] languages = ((GlobalConceptStoreSchema)DatabaseGlobalConceptStore.this.database).getLanguages(langBits);
271	268
272		~~- GlobalConcept ref = new GlobalConcept(id, name, cardinality, relevance);~~
273		~~- GlobalConcept[] inlinks = GlobalConcept.parseList( asString(m.get("rInlinks")), ((ConceptInfoStoreSchema)database).inLinksReferenceListEntry );~~
274		~~- GlobalConcept[] outlinks = GlobalConcept.parseList( asString(m.get("rOutlinks")), ((ConceptInfoStoreSchema)database).outLinksReferenceListEntry );~~
275		~~- GlobalConcept[] broader = GlobalConcept.parseList( asString(m.get("rBroader")), ((ConceptInfoStoreSchema)database).broaderReferenceListEntry );~~
276		~~- GlobalConcept[] narrower = GlobalConcept.parseList( asString(m.get("rNarrower")), ((ConceptInfoStoreSchema)database).narrowerReferenceListEntry );~~
277		~~- TranslationReference[] langlinks = TranslationReference.parseList( asString(m.get("rLanglinks")), ((ConceptInfoStoreSchema)database).langlinkReferenceListEntry );~~
278		~~- GlobalConcept[] similar = GlobalConcept.parseList( asString(m.get("rSimilar")), ((ConceptInfoStoreSchema)database).similarReferenceListEntry );~~
279		~~- GlobalConcept[] related = GlobalConcept.parseList( asString(m.get("rRelated")), ((ConceptInfoStoreSchema)database).relatedReferenceListEntry );~~
	269	+ GlobalConcept concept = new GlobalConcept(getDatasetIdentifier(), id, type);
	270	+ concept.setCardinality(cardinality);
	271	+ concept.setRelevance(relevance);
280	272
	273	+ GlobalConcept[] inlinks = GlobalConcept.parseList( asString(m.get("rInlinks")), getConceptFactory(), ((ConceptInfoStoreSchema)database).inLinksReferenceListEntry );
	274	+ GlobalConcept[] outlinks = GlobalConcept.parseList( asString(m.get("rOutlinks")), getConceptFactory(), ((ConceptInfoStoreSchema)database).outLinksReferenceListEntry );
	275	+ GlobalConcept[] broader = GlobalConcept.parseList( asString(m.get("rBroader")), getConceptFactory(), ((ConceptInfoStoreSchema)database).broaderReferenceListEntry );
	276	+ GlobalConcept[] narrower = GlobalConcept.parseList( asString(m.get("rNarrower")), getConceptFactory(), ((ConceptInfoStoreSchema)database).narrowerReferenceListEntry );
	277	+ GlobalConcept[] langlinks = GlobalConcept.parseList( asString(m.get("rLanglinks")), getConceptFactory(), ((ConceptInfoStoreSchema)database).langlinkReferenceListEntry );
	278	+ GlobalConcept[] similar = GlobalConcept.parseList( asString(m.get("rSimilar")), getConceptFactory(), ((ConceptInfoStoreSchema)database).similarReferenceListEntry );
	279	+ GlobalConcept[] related = GlobalConcept.parseList( asString(m.get("rRelated")), getConceptFactory(), ((ConceptInfoStoreSchema)database).relatedReferenceListEntry );
	280	+
281	281	ConceptRelations<GlobalConcept> relations = new ConceptRelations<GlobalConcept>(broader, narrower, inlinks, outlinks, similar, related, langlinks);
	282	+ concept.setRelations(relations);
282	283
283		~~- return new GlobalConcept(ref, getDatasetIdentifier(), languages, type, DatabaseGlobalConceptStore.this, relations, null);~~
	284	+ return concept;
284	285	} catch (SQLException e) {
285	286	throw new PersistenceException(e);
286	287	}
Index: trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/schema/ConceptInfoStoreSchema.java
—	—	@@ -13,22 +13,24 @@
14	14	import de.brightbyte.wikiword.DatasetIdentifier;
15	15	import de.brightbyte.wikiword.TweakSet;
16	16	import de.brightbyte.wikiword.model.WikiWordConcept;
	17	+import static de.brightbyte.wikiword.model.WikiWordConcept.*;
17	18
18	19	public class ConceptInfoStoreSchema extends WikiWordStoreSchema {
19	20	public final String referenceSeparator; //info record separator
20	21	public final String referenceFieldSeparator; //info field separator
	22	+ public final String languagePrefixSeparator; //language prefix separator
21	23
22		~~- public final Pattern referenceSeparatorPattern; //info record separator~~
23		~~- public final Pattern referenceFieldSeparatorPattern; //info field separator~~
	24	+ public final Pattern conceptSeparatorPattern; //info record separator
	25	+ public final Pattern conceptFieldSeparatorPattern; //info field separator
	26	+ public final Pattern languagePrefixPattern; //language prefix separator
24	27
25		~~- public class ReferenceListEntrySpec extends WikiWordConcept.ListFormatSpec {~~
	28	+ public class ConceptListEntrySpec extends WikiWordConcept.ListFormatSpec {
26	29
27	30	public final String joinField;
28	31	public final String valueExpression;
29	32
30		~~- public ReferenceListEntrySpec(String field, String expression, boolean useId, boolean useName, boolean useCardinality, boolean useRelevance) {~~
31		~~- super(referenceSeparatorPattern, referenceFieldSeparatorPattern, useId, useName,~~
32		~~- useCardinality, useRelevance);~~
	33	+ public ConceptListEntrySpec(String field, String expression, int flags) {
	34	+ super(conceptSeparatorPattern, conceptFieldSeparatorPattern, languagePrefixPattern, flags);
33	35
34	36	this.joinField = field;
35	37	this.valueExpression = expression;
—	—	@@ -36,18 +38,18 @@
37	39
38	40	}
39	41
40		~~- public final ReferenceListEntrySpec langlinkReferenceListEntry;~~
	42	+ public final ConceptListEntrySpec langlinkReferenceListEntry;
41	43
42		~~- public final ReferenceListEntrySpec termReferenceListEntry;~~
43		~~- public final ReferenceListEntrySpec broaderReferenceListEntry;~~
44		~~- public final ReferenceListEntrySpec narrowerReferenceListEntry;~~
45		~~- public final ReferenceListEntrySpec inLinksReferenceListEntry;~~
46		~~- public final ReferenceListEntrySpec outLinksReferenceListEntry;~~
47		~~- public final ReferenceListEntrySpec similarReferenceListEntry;~~
48		~~- public final ReferenceListEntrySpec relatedReferenceListEntry;~~
49		~~- public final ReferenceListEntrySpec related2ReferenceListEntry;~~
50		~~- public final ReferenceListEntrySpec featureReferenceListEntry;~~
51		~~- public final ReferenceListEntrySpec proximityReferenceListEntry;~~
	44	+ public final ConceptListEntrySpec termReferenceListEntry;
	45	+ public final ConceptListEntrySpec broaderReferenceListEntry;
	46	+ public final ConceptListEntrySpec narrowerReferenceListEntry;
	47	+ public final ConceptListEntrySpec inLinksReferenceListEntry;
	48	+ public final ConceptListEntrySpec outLinksReferenceListEntry;
	49	+ public final ConceptListEntrySpec similarReferenceListEntry;
	50	+ public final ConceptListEntrySpec relatedReferenceListEntry;
	51	+ public final ConceptListEntrySpec related2ReferenceListEntry;
	52	+ public final ConceptListEntrySpec featureReferenceListEntry;
	53	+ public final ConceptListEntrySpec proximityReferenceListEntry;
52	54
53	55	protected EntityTable conceptInfoTable;
54	56	protected EntityTable conceptDescriptionTable;
—	—	@@ -65,52 +67,54 @@
66	68
67	69	referenceSeparator = tweaks.getTweak("dbstore.cacheReferenceSeparator", "\u001E"); //ASCII Record Separator
68	70	referenceFieldSeparator = tweaks.getTweak("dbstore.cacheReferenceFieldSeparator", "\u001F"); //ASCII Field Separator
69		~~- referenceSeparatorPattern = Pattern.compile(referenceSeparator.replaceAll("[^$(){}\\[\\]\\\\]", "\\\\$0"));~~
70		~~- referenceFieldSeparatorPattern = Pattern.compile(referenceFieldSeparator.replaceAll("[^$(){}\\[\\]\\\\]", "\\\\$0"));~~
	71	+ languagePrefixSeparator = tweaks.getTweak("dbstore.languagePrefixSeparator", ":"); //ASCII Field Separator
	72	+ conceptSeparatorPattern = Pattern.compile(referenceSeparator.replaceAll("[^$(){}\\[\\]\\\\]", "\\\\$0"));
	73	+ conceptFieldSeparatorPattern = Pattern.compile(referenceFieldSeparator.replaceAll("[^$(){}\\[\\]\\\\]", "\\\\$0"));
	74	+ languagePrefixPattern = Pattern.compile(languagePrefixSeparator.replaceAll("[^$(){}\\[\\]\\\\]", "\\\\$0"));
71	75
72	76	langlinkReferenceListEntry =
73		~~- new ReferenceListEntrySpec("language, target", "concat(language, ':', target)",~~
74		~~- false, true, false, false);~~
	77	+ new ConceptListEntrySpec("language, target", "concat(language, ':', target)",
	78	+ LIST_FORMAT_USE_NAME \| LIST_FORMAT_USE_LANGUAGE_PREFIX );
75	79
76	80	termReferenceListEntry =
77		~~- new ReferenceListEntrySpec("term_text", fields("term_text", "freq"),~~
78		~~- false, true, true, false);~~
	81	+ new ConceptListEntrySpec("term_text", fields("term_text", "freq"),
	82	+ LIST_FORMAT_USE_NAME \| LIST_FORMAT_USE_CARDINALITY );
79	83
80	84	broaderReferenceListEntry =
81		~~- new ReferenceListEntrySpec("broad", cacheNames ? fields("broad", "broad_name", "if (lhs is null or lhs = 0, 0, 1/lhs)") : fields("broad", "if (lhs is null or lhs = 0, 0, 1/lhs)"),~~
82		~~- true, cacheNames, false, true);~~
	85	+ new ConceptListEntrySpec("broad", cacheNames ? fields("broad", "broad_name", "if (lhs is null or lhs = 0, 0, 1/lhs)") : fields("broad", "if (lhs is null or lhs = 0, 0, 1/lhs)"),
	86	+ LIST_FORMAT_USE_ID \| (cacheNames ? LIST_FORMAT_USE_NAME : 0) \| LIST_FORMAT_USE_RELEVANCE );
83	87
84	88	narrowerReferenceListEntry =
85		~~- new ReferenceListEntrySpec("narrow", cacheNames ? fields("narrow", "narrow_name", "if (lhs is null or lhs = 0, 0, 1/lhs)") : fields("narrow", "if (lhs is null or lhs = 0, 0, 1/lhs)"),~~
86		~~- true, cacheNames, false, true);~~
	89	+ new ConceptListEntrySpec("narrow", cacheNames ? fields("narrow", "narrow_name", "if (lhs is null or lhs = 0, 0, 1/lhs)") : fields("narrow", "if (lhs is null or lhs = 0, 0, 1/lhs)"),
	90	+ LIST_FORMAT_USE_ID \| (cacheNames ? LIST_FORMAT_USE_NAME : 0) \| LIST_FORMAT_USE_RELEVANCE);
87	91
88	92	inLinksReferenceListEntry =
89		~~- new ReferenceListEntrySpec("anchor", cacheNames ? fields("anchor", "anchor_name", "idf") : fields("anchor", "idf"),~~
90		~~- true, cacheNames, false, true);~~
	93	+ new ConceptListEntrySpec("anchor", cacheNames ? fields("anchor", "anchor_name", "idf") : fields("anchor", "idf"),
	94	+ LIST_FORMAT_USE_ID \| (cacheNames ? LIST_FORMAT_USE_NAME : 0) \| LIST_FORMAT_USE_RELEVANCE );
91	95
92	96	outLinksReferenceListEntry =
93		~~- new ReferenceListEntrySpec("target", cacheNames ? fields("target", "target_name", "idf") : fields("target", "idf"),~~
94		~~- true, cacheNames, false, true);~~
	97	+ new ConceptListEntrySpec("target", cacheNames ? fields("target", "target_name", "idf") : fields("target", "idf"),
	98	+ LIST_FORMAT_USE_ID \| (cacheNames ? LIST_FORMAT_USE_NAME : 0) \| LIST_FORMAT_USE_RELEVANCE );
95	99
96	100	similarReferenceListEntry =
97		~~- new ReferenceListEntrySpec("concept2", fields("concept2", "langmatch"), //TODO: frequency for similar from langref(!)~~
98		~~- true, false, true, false); //TODO: name?... in relation table?... //XXX: why no score~~
	101	+ new ConceptListEntrySpec("concept2", fields("concept2", "langmatch"), //TODO: frequency for similar from langref(!)
	102	+ LIST_FORMAT_USE_ID \| LIST_FORMAT_USE_CARDINALITY); //TODO: name?... in relation table?... //XXX: why no score
99	103
100	104	relatedReferenceListEntry =
101		~~- new ReferenceListEntrySpec("concept2", "concept2",~~
102		~~- true, false, false, false); //TODO: name?... in relation table?... //XXX: why no score~~
	105	+ new ConceptListEntrySpec("concept2", "concept2",
	106	+ LIST_FORMAT_USE_ID); //TODO: name?... in relation table?... //XXX: why no score
103	107
104	108	related2ReferenceListEntry =
105		~~- new ReferenceListEntrySpec("concept1","concept1",~~
106		~~- true, false, false, false); //TODO: name?... in relation table?... //XXX: why no score~~
	109	+ new ConceptListEntrySpec("concept1","concept1",
	110	+ LIST_FORMAT_USE_ID); //TODO: name?... in relation table?... //XXX: why no score
107	111
108	112	featureReferenceListEntry =
109		~~- new ReferenceListEntrySpec("target",fields("target", "weight"),~~
110		~~- true, false, false, true);~~
	113	+ new ConceptListEntrySpec("target",fields("target", "weight"),
	114	+ LIST_FORMAT_USE_ID \| LIST_FORMAT_USE_RELEVANCE );
111	115
112	116	proximityReferenceListEntry =
113		~~- new ReferenceListEntrySpec("target",fields("target", "proximity"),~~
114		~~- true, false, false, true);~~
	117	+ new ConceptListEntrySpec("target",fields("target", "proximity"),
	118	+ LIST_FORMAT_USE_ID \| LIST_FORMAT_USE_RELEVANCE );
115	119
116	120	init(tweaks, description);
117	121	}
Index: trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/model/LocalConcept.java
—	—	@@ -1,52 +1,19 @@
2	2	package de.brightbyte.wikiword.model;
3	3
4		~~-import java.text.Collator;~~
5		~~-import java.util.Comparator;~~
6		~~-import java.util.Locale;~~
7		-
8	4	import de.brightbyte.wikiword.ConceptType;
9	5	import de.brightbyte.wikiword.Corpus;
10	6
11	7	public class LocalConcept extends WikiWordConcept {
12	8
13		~~- public static class ByName implements Comparator<LocalConcept> {~~
14		~~- protected Collator collator;~~
15		-
16		~~- public ByName() {~~
17		~~- this((Collator)null);~~
18		~~- }~~
19		-
20		~~- public ByName(Locale locale) {~~
21		~~- this.collator = Collator.getInstance(locale);~~
22		~~- }~~
23		-
24		~~- public ByName(Collator collator) {~~
25		~~- this.collator = collator;~~
26		~~- }~~
27		-
28		~~- public int compare(LocalConcept a, LocalConcept b) {~~
29		~~- if (collator==null) return a.getName().compareTo(b.getName());~~
30		~~- else return collator.compare(a.getName(), b.getName());~~
31		~~- }~~
32		~~- };~~
33		-
34		~~- protected String name;~~
35	9	protected String definition;
36	10	protected String language;
37	11	protected WikiWordResource resource;
38	12
39	13	public LocalConcept(Corpus corpus, int id, ConceptType type, String name) {
40	14	super(corpus, id, type);
41		-
42		~~- this.name = name;~~
	15	+ setName(name);
43	16	}
44	17
45		~~- @Override~~
46		~~- public String toString() {~~
47		~~- if (name!=null) return "#"+id+":[["+name+"]]";~~
48		~~- else return "#"+id;~~
49		~~- }~~
50		-
51	18	public Corpus getCorpus() {
52	19	return (Corpus)getDatasetIdentifier();
53	20	}
—	—	@@ -79,13 +46,4 @@
80	47	this.language = language;
81	48	}
82	49
83		~~- public String getName() {~~
84		~~- return name;~~
85		~~- }~~
86		-
87		~~- public void setName(String name) {~~
88		~~- if (this.name!=null) throw new IllegalStateException("property already initialized");~~
89		~~- this.name = name;~~
90		~~- }~~
91		-
92	50	}
Index: trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/model/TermReference.java
—	—	@@ -1,7 +1,11 @@
2	2	package de.brightbyte.wikiword.model;
3	3
	4	+import de.brightbyte.util.PersistenceException;
	5	+import de.brightbyte.wikiword.model.WikiWordConcept.Factory;
	6	+import de.brightbyte.wikiword.schema.ConceptInfoStoreSchema.ConceptListEntrySpec;
4	7
5	8
	9	+
6	10	public class TermReference {
7	11
8	12	private String term;
—	—	@@ -67,4 +71,21 @@
68	72	return true;
69	73	}
70	74
	75	+ public static TermReference[] parseList(String s, Factory<LocalConcept> factory, ConceptListEntrySpec spec) throws PersistenceException {
	76	+ LocalConcept[] concepts = WikiWordConcept.parseList(s, factory, spec); //XXX: this is a terrible, terrible hack.
	77	+ TermReference[] terms = new TermReference[concepts.length];
	78	+
	79	+ for (int i=0; i<terms.length; i++) {
	80	+ WikiWordConcept dummy = concepts[i];
	81	+
	82	+ String term = dummy.getName(); //UGHA!
	83	+ double score = dummy.getCardinality();
	84	+
	85	+ WikiWordConcept target = factory.newInstance(dummy.getId(), null, dummy.getType());
	86	+ terms[i] = new TermReference(term, target, score);
	87	+ }
	88	+
	89	+ return terms;
	90	+ }
	91	+
71	92	}
Index: trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/model/WikiWordConcept.java
—	—	@@ -1,7 +1,9 @@
2	2	package de.brightbyte.wikiword.model;
3	3
	4	+import java.text.Collator;
4	5	import java.util.Arrays;
5	6	import java.util.Comparator;
	7	+import java.util.Locale;
6	8	import java.util.regex.Pattern;
7	9
8	10	import de.brightbyte.data.measure.Measure;
—	—	@@ -10,24 +12,61 @@
11	13	import de.brightbyte.wikiword.DatasetIdentifier;
12	14
13	15	public abstract class WikiWordConcept {
14		~~- protected DatasetIdentifier dataset;~~
15		~~- protected ConceptType type;~~
	16	+ public static class ByName implements Comparator<WikiWordConcept> {
	17	+ protected Collator collator;
	18	+
	19	+ public ByName() {
	20	+ this((Collator)null);
	21	+ }
	22	+
	23	+ public ByName(Locale locale) {
	24	+ this.collator = Collator.getInstance(locale);
	25	+ }
	26	+
	27	+ public ByName(Collator collator) {
	28	+ this.collator = collator;
	29	+ }
	30	+
	31	+ public int compare(WikiWordConcept a, WikiWordConcept b) {
	32	+ if (collator==null) return a.getName().compareTo(b.getName());
	33	+ else return collator.compare(a.getName(), b.getName());
	34	+ }
	35	+ };
16	36
17		~~- protected int id;~~
	37	+ private DatasetIdentifier dataset;
	38	+ private ConceptType type;
	39	+
	40	+ private int id;
	41	+ private String name;
18	42
19		~~- protected int cardinality = 1;~~
20		~~- protected double relevance = 1;~~
	43	+ private int cardinality = 1;
	44	+ private double relevance = 1;
21	45
22		~~- protected ConceptRelations relations;~~
23		~~- protected ConceptFeatures features;~~
24		~~- protected TermReference[] terms;~~
	46	+ private ConceptRelations relations;
	47	+ private ConceptFeatures features;
	48	+ private TermReference[] terms;
25	49
26	50	public WikiWordConcept(DatasetIdentifier dataset, int id, ConceptType type) {
27	51	this.id = id;
28	52	this.dataset = dataset;
29	53	this.type = type;
30	54	}
	55	+
	56	+ @Override
	57	+ public String toString() {
	58	+ if (name!=null) return "#"+id+":[["+name+"]]";
	59	+ else return "#"+id;
	60	+ }
	61	+
	62	+ public String getName() {
	63	+ return name;
	64	+ }
31	65
	66	+ public void setName(String name) {
	67	+ if (this.name!=null) throw new IllegalStateException("property already initialized");
	68	+ this.name = name;
	69	+ }
	70	+
32	71	public DatasetIdentifier getDatasetIdentifier() {
33	72	return dataset;
34	73	}
—	—	@@ -108,13 +147,7 @@
109	148
110	149	return id == other.id;
111	150	}
112		-
113	151
114		~~- @Override~~
115		~~- public String toString() {~~
116		~~- return "#"+id;~~
117		~~- }~~
118		-
119	152	public static final Measure<WikiWordConcept> theCardinality = new Measure<WikiWordConcept>(){
120	153	public double measure(WikiWordConcept r) {
121	154	return r.getCardinality();
—	—	@@ -144,30 +177,43 @@
145	178	else return (int)(rb-ra) +1;
146	179	}
147	180	};
148		-
	181	+
	182	+ public static final Comparator<WikiWordConcept> byName = new ByName();
	183	+
149	184	public static interface Factory<T extends WikiWordConcept> {
150	185	public T newInstance(int id, String name, ConceptType type) throws PersistenceException;
151	186	public T newInstance(int id, String name, ConceptType type, int card, double rel) throws PersistenceException;
152	187	public T[] newArray(int size);
153	188	}
154	189
	190	+ public static final int LIST_FORMAT_USE_ID = 0x0001;
	191	+ public static final int LIST_FORMAT_USE_NAME = 0x0001;
	192	+ public static final int LIST_FORMAT_USE_CARDINALITY = 0x0010;
	193	+ public static final int LIST_FORMAT_USE_RELEVANCE = 0x0020;
	194	+ public static final int LIST_FORMAT_USE_LANGUAGE_PREFIX= 0x0100;
	195	+
155	196	public static class ListFormatSpec {
156	197	public final boolean useId;
157	198	public final boolean useName;
158	199	public final boolean useCardinality;
159	200	public final boolean useRelevance;
	201	+ public final boolean useLanguagePrefix;
160	202	public final Pattern referenceSeparator;
161	203	public final Pattern fieldSeparator;
	204	+ public final Pattern prefixSeparator;
162	205	public final int width;
163	206
164		~~- public ListFormatSpec(Pattern referenceSeparator, Pattern fieldSeparator, boolean useId, boolean useName, boolean useCardinality, boolean useRelevance) {~~
	207	+ public ListFormatSpec(Pattern referenceSeparator, Pattern fieldSeparator, Pattern prefixSeparator, int flags) {
165	208	this.referenceSeparator = referenceSeparator;
166	209	this.fieldSeparator = fieldSeparator;
167		~~- this.useId = useId;~~
168		~~- this.useName = useName;~~
169		~~- this.useCardinality = useCardinality;~~
170		~~- this.useRelevance = useRelevance;~~
	210	+ this.prefixSeparator = prefixSeparator;
171	211
	212	+ useId = (flags & LIST_FORMAT_USE_ID) > 0;
	213	+ useName = (flags & LIST_FORMAT_USE_NAME) > 0;
	214	+ useCardinality = (flags & LIST_FORMAT_USE_CARDINALITY) > 0;
	215	+ useRelevance = (flags & LIST_FORMAT_USE_RELEVANCE) > 0;
	216	+ useLanguagePrefix = (flags & LIST_FORMAT_USE_LANGUAGE_PREFIX) > 0;
	217	+
172	218	int w = 0;
173	219	if (useId) w++;
174	220	if (useName) w++;
—	—	@@ -208,8 +254,19 @@
209	255	if (rr.length>=i+1 && spec.useName) name = rr[i++];
210	256	if (rr.length>=i+1 && spec.useCardinality) cardinality = rr[i].length()==0 ? -1 : Integer.parseInt(rr[i++]);
211	257	if (rr.length>=i+1 && spec.useRelevance) relevance = rr[i].length()==0 ? -1 : Double.parseDouble(rr[i++]);
	258	+ String lang = null;
212	259
213		~~- a[c++] = factory.newInstance(id, name, null, cardinality, relevance);~~
	260	+ if (spec.useLanguagePrefix) {
	261	+ String[] nn = spec.prefixSeparator.split(name, 2);
	262	+ if (nn.length>1) {
	263	+ name = nn[1];
	264	+ lang = nn[0];
	265	+ }
	266	+ }
	267	+
	268	+ T t = factory.newInstance(id, name, null, cardinality, relevance);
	269	+ if (lang!=null && t instanceof LocalConcept) ((LocalConcept)t).setLanguage(lang);
	270	+ a[c++] = t;
214	271	}
215	272
216	273	if (c<a.length) { //should not happen, but might, if values get truncated
—	—	@@ -220,7 +277,7 @@
221	278
222	279	if (spec.useRelevance) Arrays.sort(a, byRelevance);
223	280	else if (spec.useCardinality) Arrays.sort(a, byCardinality);
224		~~- //else if (spec.useName) Arrays.sort(a, byName);~~
	281	+ else if (spec.useName) Arrays.sort(a, byName);
225	282
226	283	return a;
227	284	}
Index: trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/model/ConceptRelations.java
—	—	@@ -1,6 +1,5 @@
2	2	package de.brightbyte.wikiword.model;
3	3
4		~~-import java.util.Map;~~
5	4
6	5	public class ConceptRelations<R extends WikiWordConcept> {
7	6
—	—	@@ -10,7 +9,7 @@
11	10	protected R[] narrower;
12	11	protected R[] similar;
13	12	protected R[] related;
14		~~- protected Map<String, LocalConcept> langlinks;~~
	13	+ protected R[] langlinks;
15	14
16	15	//TODO: inlinks, outlinks, coocc, co-coocc
17	16
—	—	@@ -20,7 +19,7 @@
21	20	R[] outlinks,
22	21	R[] similar,
23	22	R[] related,
24		~~- Map<String, LocalConcept> langlinks) {~~
	23	+ R[] langlinks) {
25	24
26	25	if (inlinks==null) throw new NullPointerException();
27	26	if (outlinks==null) throw new NullPointerException();
—	—	@@ -43,7 +42,7 @@
44	43	return broader;
45	44	}
46	45
47		~~- public Map<String, LocalConcept> getLanglinks() {~~
	46	+ public R[] getLanglinks() {
48	47	return langlinks;
49	48	}
50	49
Index: trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/rdf/LocalConceptSkosProperties.java
—	—	@@ -49,7 +49,7 @@
50	50	TermReference[] tt = concept.getTerms();
51	51	for (TermReference t: tt) {
52	52	//if (t.getName().equals(name)) continue; //NOTE: for prefLabel
53		~~- setLiteralProperty(about, skos.altLabel, t.getName(), lang);~~
	53	+ setLiteralProperty(about, skos.altLabel, t.getTerm(), lang);
54	54	}
55	55
56	56	//TODO: idf, lhs
Index: trunk/WikiWord/WikiWord/src/main/java/de/brightbyte/wikiword/rdf/RdfOutput.java
—	—	@@ -27,7 +27,7 @@
28	28
29	29	@SuppressWarnings("unchecked")
30	30	public RdfOutput(WikiWordIdentifiers identifiers, String platform, Writer writer, String format, DatasetIdentifier ds) throws RdfException, PersistenceException {
31		~~- this(identifiers, (RdfPlatform<V, R, A, W>)RdfPlatforms.newPlatform(platform));~~
	31	+ this(identifiers, (RdfPlatform<V, R, A, W>)(Object)RdfPlatforms.newPlatform(platform)); //XXX: ugly insane cast to avoid some compilers complaining.
32	32	init(this.platform.newWriter(writer, format), ds);
33	33	}
34	34
Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/store/builder/WikiWordConceptStoreBuilder.java
—	—	@@ -2,7 +2,6 @@
3	3
4	4	import de.brightbyte.util.PersistenceException;
5	5	import de.brightbyte.wikiword.model.WikiWordConcept;
6		~~-import de.brightbyte.wikiword.model.WikiWordConceptReference;~~
7	6	import de.brightbyte.wikiword.store.WikiWordConceptStore;
8	7	import de.brightbyte.wikiword.store.WikiWordConceptStoreBase;
9	8
—	—	@@ -12,6 +11,6 @@
13	12	public ConceptInfoStoreBuilder<T> getConceptInfoStoreBuilder() throws PersistenceException;
14	13	public ProximityStoreBuilder getProximityStoreBuilder() throws PersistenceException;
15	14
16		~~- public WikiWordConceptStore<T, ? extends WikiWordConceptReference<T>> getConceptStore() throws PersistenceException;~~
	15	+ public WikiWordConceptStore<T> getConceptStore() throws PersistenceException;
17	16
18	17	}
Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/store/builder/DebugLocalConceptStoreBuilder.java
—	—	@@ -19,8 +19,6 @@
20	20	import de.brightbyte.wikiword.ExtractionRule;
21	21	import de.brightbyte.wikiword.ResourceType;
22	22	import de.brightbyte.wikiword.model.LocalConcept;
23		~~-import de.brightbyte.wikiword.model.LocalConceptReference;~~
24		~~-import de.brightbyte.wikiword.model.WikiWordConceptReference;~~
25	23	import de.brightbyte.wikiword.schema.AliasScope;
26	24	import de.brightbyte.wikiword.store.GroupNameTranslator;
27	25	import de.brightbyte.wikiword.store.WikiWordConceptStore;
—	—	@@ -834,11 +832,11 @@
835	833	log("+ storeWarning: rcId="+rcId+", problem="+problem+", details="+details);
836	834	}
837	835
838		~~- public WikiWordConceptStore<LocalConcept, WikiWordConceptReference<LocalConcept>> getConceptStore() throws PersistenceException {~~
	836	+ public WikiWordConceptStore<LocalConcept> getConceptStore() throws PersistenceException {
839	837	return null; //XXX...
840	838	}
841	839
842		~~- public DataSet<LocalConceptReference> listUnknownConcepts() throws PersistenceException {~~
	840	+ public DataSet<LocalConcept> listUnknownConcepts() throws PersistenceException {
843	841	return null; //XXX...
844	842	}
845	843
—	—	@@ -846,7 +844,7 @@
847	845	//noop
848	846	}
849	847
850		~~- public int processUnknownConcepts(CursorProcessor<LocalConceptReference> processor) throws PersistenceException {~~
	848	+ public int processUnknownConcepts(CursorProcessor<LocalConcept> processor) throws PersistenceException {
851	849	//noop
852	850	return 0;
853	851	}
Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/store/builder/DatabaseLocalConceptStoreBuilder.java
—	—	@@ -1,7 +1,5 @@
2	2	package de.brightbyte.wikiword.store.builder;
3	3
4		~~-import static de.brightbyte.db.DatabaseUtil.asString;~~
5		-
6	4	import java.io.File;
7	5	import java.sql.Connection;
8	6	import java.sql.ResultSet;
—	—	@@ -38,7 +36,6 @@
39	37	import de.brightbyte.wikiword.TweakSet;
40	38	import de.brightbyte.wikiword.builder.NameMaps;
41	39	import de.brightbyte.wikiword.model.LocalConcept;
42		~~-import de.brightbyte.wikiword.model.LocalConceptReference;~~
43	40	import de.brightbyte.wikiword.schema.AliasScope;
44	41	import de.brightbyte.wikiword.schema.ConceptInfoStoreSchema;
45	42	import de.brightbyte.wikiword.schema.LocalConceptStoreSchema;
—	—	@@ -1453,16 +1450,18 @@
1454	1451	return new DatabaseLocalConceptStore((LocalConceptStoreSchema)database, tweaks);
1455	1452	}
1456	1453
1457		~~- protected static DatabaseDataSet.Factory<LocalConceptReference> localConceptReferenceFactory = new DatabaseDataSet.Factory<LocalConceptReference>() {~~
1458		~~- public LocalConceptReference newInstance(ResultSet row) throws SQLException, PersistenceException {~~
	1454	+ protected DatabaseDataSet.Factory<LocalConcept> localConceptReferenceFactory = new DatabaseDataSet.Factory<LocalConcept>() {
	1455	+ public LocalConcept newInstance(ResultSet row) throws SQLException, PersistenceException {
1459	1456	int id = row.getInt("id");
1460		~~- String name = asString(row.getObject("name"));~~
	1457	+ String name = DatabaseUtil.asString(row.getObject("name"));
	1458	+ //FIXME: type?!
1461	1459
1462		~~- return new LocalConceptReference(id, name, -1, -1);~~
	1460	+ LocalConcept concept = new LocalConcept(getCorpus(), id, null, name);
	1461	+ return concept;
1463	1462	}
1464	1463	};
1465	1464
1466		~~- public int processUnknownConcepts(final CursorProcessor<LocalConceptReference> processor) throws PersistenceException {~~
	1465	+ public int processUnknownConcepts(final CursorProcessor<LocalConcept> processor) throws PersistenceException {
1467	1466	String sql = "SELECT * FROM "+conceptTable.getSQLName();
1468	1467	String where = "type = "+ConceptType.UNKNOWN.getCode();
1469	1468
—	—	@@ -1483,7 +1482,7 @@
1484	1483	return ds;
1485	1484	}*/
1486	1485
1487		~~- public DataSet<LocalConceptReference> listUnknownConcepts() throws PersistenceException {~~
	1486	+ public DataSet<LocalConcept> listUnknownConcepts() throws PersistenceException {
1488	1487	String sql = "SELECT * FROM "+conceptTable.getSQLName();
1489	1488	String where = "type = "+ConceptType.UNKNOWN.getCode();
1490	1489
Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/store/builder/DatabaseConceptInfoStoreBuilder.java
—	—	@@ -12,7 +12,7 @@
13	13	import de.brightbyte.wikiword.model.WikiWordConcept;
14	14	import de.brightbyte.wikiword.schema.ConceptInfoStoreSchema;
15	15	import de.brightbyte.wikiword.schema.WikiWordConceptStoreSchema;
16		~~-import de.brightbyte.wikiword.schema.ConceptInfoStoreSchema.ReferenceListEntrySpec;~~
	16	+import de.brightbyte.wikiword.schema.ConceptInfoStoreSchema.ConceptListEntrySpec;
17	17
18	18	public abstract class DatabaseConceptInfoStoreBuilder<T extends WikiWordConcept>
19	19	extends DatabaseWikiWordStoreBuilder
—	—	@@ -118,7 +118,7 @@
119	119	protected int buildConceptPropertyCache(
120	120	final DatabaseTable cacheTable, final String cacheIdField,
121	121	final String propertyField, final String realtion, final String relConceptField,
122		~~- final ReferenceListEntrySpec spec, final boolean append, final String threshold,~~
	122	+ final ConceptListEntrySpec spec, final boolean append, final String threshold,
123	123	final int chunkFactor) throws PersistenceException {
124	124
125	125	final DatabaseTable relationTable = conceptStore.getDatabaseAccess().getTable(realtion);
Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/store/builder/DatabaseWikiWordConceptStoreBuilder.java
—	—	@@ -24,7 +24,6 @@
25	25	import de.brightbyte.wikiword.ConceptType;
26	26	import de.brightbyte.wikiword.TweakSet;
27	27	import de.brightbyte.wikiword.model.WikiWordConcept;
28		~~-import de.brightbyte.wikiword.model.WikiWordConceptReference;~~
29	28	import de.brightbyte.wikiword.schema.WikiWordConceptStoreSchema;
30	29	import de.brightbyte.wikiword.store.DatabaseWikiWordConceptStore;
31	30
—	—	@@ -425,16 +424,16 @@
426	425
427	426	protected abstract DatabaseStatisticsStoreBuilder newStatisticsStoreBuilder() throws SQLException, PersistenceException;
428	427	protected abstract DatabaseConceptInfoStoreBuilder<T> newConceptInfoStoreBuilder() throws SQLException, PersistenceException;
429		~~- protected abstract DatabaseWikiWordConceptStore<T, ? extends WikiWordConceptReference<T>> newConceptStore() throws SQLException, PersistenceException;~~
	428	+ protected abstract DatabaseWikiWordConceptStore<T> newConceptStore() throws SQLException, PersistenceException;
430	429	protected abstract DatabaseProximityStoreBuilder newProximityStoreBuilder() throws SQLException;
431	430
432	431	private DatabaseStatisticsStoreBuilder statsStore;
433	432	private DatabaseProximityStoreBuilder proximityStore;
434	433	private DatabaseConceptInfoStoreBuilder<T> infoStore;
435	434
436		~~- private DatabaseWikiWordConceptStore<T, ? extends WikiWordConceptReference<T>> conceptStore;~~
	435	+ private DatabaseWikiWordConceptStore<T> conceptStore;
437	436
438		~~- public DatabaseWikiWordConceptStore<T, ? extends WikiWordConceptReference<T>> getConceptStore() throws PersistenceException {~~
	437	+ public DatabaseWikiWordConceptStore<T> getConceptStore() throws PersistenceException {
439	438	try {
440	439	if (conceptStore==null) conceptStore = newConceptStore();
441	440	return conceptStore;
Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/store/builder/LocalConceptStoreBuilder.java
—	—	@@ -2,15 +2,14 @@
3	3
4	4	import java.util.Date;
5	5
6		~~-import de.brightbyte.data.cursor.DataSet;~~
7	6	import de.brightbyte.data.cursor.CursorProcessor;
	7	+import de.brightbyte.data.cursor.DataSet;
8	8	import de.brightbyte.util.PersistenceException;
9	9	import de.brightbyte.wikiword.ConceptType;
10	10	import de.brightbyte.wikiword.Corpus;
11	11	import de.brightbyte.wikiword.ExtractionRule;
12	12	import de.brightbyte.wikiword.ResourceType;
13	13	import de.brightbyte.wikiword.model.LocalConcept;
14		~~-import de.brightbyte.wikiword.model.LocalConceptReference;~~
15	14	import de.brightbyte.wikiword.schema.AliasScope;
16	15
17	16	/**
—	—	@@ -98,8 +97,8 @@
99	98	public Corpus getCorpus();
100	99
101	100	public void resetTermsForUnknownConcepts() throws PersistenceException;
102		~~- public DataSet<LocalConceptReference> listUnknownConcepts() throws PersistenceException;~~
103		~~- public int processUnknownConcepts(CursorProcessor<LocalConceptReference> processor) throws PersistenceException;~~
	101	+ public DataSet<LocalConcept> listUnknownConcepts() throws PersistenceException;
	102	+ public int processUnknownConcepts(CursorProcessor<LocalConcept> processor) throws PersistenceException;
104	103
105	104	public TextStoreBuilder getTextStoreBuilder() throws PersistenceException;
106	105	public PropertyStoreBuilder getPropertyStoreBuilder() throws PersistenceException;

Status & tagging log

16:43, 29 March 2010 MaxSem (talk | contribs) changed the status of r64309 [removed: new added: deferred]