r52895 MediaWiki - Code Review archive

Repository:	MediaWiki
Revision:	< r52894‎ \| r52895 \| r52896 >
Date:	09:39, 8 July 2009
Author:	daniel
Status:	deferred
Tags:
Comment:	basic ResourceType detection for it, pt
Modified paths:	/trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/analyzer/matcher/AnyNameMatcher.java (modified) (history) /trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/analyzer/sensor/HasTemplateLikeSensor.java (modified) (history) /trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/analyzer/sensor/HasTemplateSensor.java (modified) (history) /trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/wikis/WikiConfiguration_enwiki.java (modified) (history) /trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/wikis/WikiConfiguration_eswiki.java (modified) (history) /trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/wikis/WikiConfiguration_itwiki.java (modified) (history) /trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/wikis/WikiConfiguration_ptwiki.java (modified) (history)

Diff [purge]

Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/analyzer/matcher/AnyNameMatcher.java
—	—	@@ -11,6 +11,8 @@
12	12
13	13	public class AnyNameMatcher implements NameMatcher {
14	14
	15	+ public static final AnyNameMatcher instance = new AnyNameMatcher();
	16	+
15	17	public AnyNameMatcher() {
16	18	//noop
17	19	}
Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/analyzer/sensor/HasTemplateSensor.java
—	—	@@ -5,17 +5,18 @@
6	6
7	7	import java.util.Map;
8	8
	9	+import de.brightbyte.wikiword.analyzer.matcher.AnyNameMatcher;
9	10	import de.brightbyte.wikiword.analyzer.matcher.ExactNameMatcher;
10	11	import de.brightbyte.wikiword.analyzer.matcher.NameMatcher;
11	12
12	13	public class HasTemplateSensor<V> extends HasTemplateLikeSensor<V> {
13	14
14		~~- public HasTemplateSensor(V value, String pattern) {~~
15		~~- this(value, new ExactNameMatcher(pattern), null);~~
	15	+ public HasTemplateSensor(V value, String name) {
	16	+ this(value, name==null ? AnyNameMatcher.instance : new ExactNameMatcher(name), null);
16	17	}
17	18
18		~~- public HasTemplateSensor(V value, String pattern, String[] params) {~~
19		~~- this(value, new ExactNameMatcher(pattern), HasTemplateLikeSensor.<NameMatcher>paramKeyMap(params));~~
	19	+ public HasTemplateSensor(V value, String name, String... params) {
	20	+ this(value, name==null ? AnyNameMatcher.instance : new ExactNameMatcher(name), HasTemplateLikeSensor.<NameMatcher>paramKeyMap(params));
20	21	}
21	22
22	23	private HasTemplateSensor(V value, NameMatcher matcher, Map<String, NameMatcher> params) {
Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/analyzer/sensor/HasTemplateLikeSensor.java
—	—	@@ -10,6 +10,7 @@
11	11
12	12	import de.brightbyte.data.MultiMap;
13	13	import de.brightbyte.wikiword.analyzer.WikiPage;
	14	+import de.brightbyte.wikiword.analyzer.matcher.AnyNameMatcher;
14	15	import de.brightbyte.wikiword.analyzer.matcher.NameMatcher;
15	16	import de.brightbyte.wikiword.analyzer.matcher.PatternNameMatcher;
16	17	import de.brightbyte.wikiword.analyzer.template.TemplateExtractor;
—	—	@@ -23,16 +24,20 @@
24	25	//TODO: provide an OR mode, so this triggers if any param matches
25	26
26	27	public HasTemplateLikeSensor(V value, String pattern, int flags) {
27		~~- this(value, new PatternNameMatcher(pattern, flags \| Pattern.MULTILINE, false), null);~~
	28	+ this(value, pattern==null ? AnyNameMatcher.instance : new PatternNameMatcher(pattern, flags \| Pattern.MULTILINE, false), null);
28	29	}
29	30
30		~~- public HasTemplateLikeSensor(V value, String pattern, int flags, String[] params) {~~
31		~~- this(value, new PatternNameMatcher(pattern, flags \| Pattern.MULTILINE, false), HasTemplateLikeSensor.<NameMatcher>paramKeyMap(params));~~
	31	+ public HasTemplateLikeSensor(V value, String pattern, int flags, String... params) {
	32	+ this(value, pattern==null ? AnyNameMatcher.instance : new PatternNameMatcher(pattern, flags \| Pattern.MULTILINE, false), HasTemplateLikeSensor.<NameMatcher>paramKeyMap(params));
32	33	}
33	34
34	35	public HasTemplateLikeSensor(V value, NameMatcher matcher, Map<String, NameMatcher> params) {
35	36	super(value);
	37	+ if (matcher==null) throw new NullPointerException();
	38	+
	39	+
36	40	this.matcher = matcher;
	41	+ this.params = params;
37	42	}
38	43
39	44	@Override
—	—	@@ -77,6 +82,7 @@
78	83
79	84	protected static <V> Map<String, V> paramKeyMap(String[] params) {
80	85	if (params==null) return null;
	86	+ if (params.length==0) return null;
81	87
82	88	HashMap<String, V> m = new HashMap<String, V>();
83	89	for (String p: params) {
Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/wikis/WikiConfiguration_ptwiki.java
—	—	@@ -1,11 +1,19 @@
2	2	package de.brightbyte.wikiword.wikis;
3	3
	4	+import de.brightbyte.wikiword.ResourceType;
4	5	import de.brightbyte.wikiword.analyzer.WikiConfiguration;
	6	+import de.brightbyte.wikiword.analyzer.sensor.HasCategoryLikeSensor;
	7	+import de.brightbyte.wikiword.analyzer.sensor.HasTemplateLikeSensor;
	8	+import de.brightbyte.wikiword.analyzer.sensor.NamespaceSensor;
5	9
6	10	public class WikiConfiguration_ptwiki extends WikiConfiguration {
7	11
8	12	public WikiConfiguration_ptwiki() {
9		~~- //TODO: configure for no.wikipedia.org~~
	13	+ resourceTypeSensors.add( new NamespaceSensor<ResourceType>(ResourceType.LIST, 102)); // 102 = Anexo
	14	+ resourceTypeSensors.add( new HasCategoryLikeSensor<ResourceType>(ResourceType.LIST, "^(Listas\|\u00cdndice)(_\|$)", 0));
	15	+ resourceTypeSensors.add( new HasTemplateLikeSensor<ResourceType>(ResourceType.BAD, "^(Elimina\u00e7\u00e3o_r\u00e1pida\|Er1\|Apagar2\|VDA2\|Copyright2)$", 0));
	16	+
	17	+ //resourceTypeSensors.add( new HasTemplateLikeSensor<ResourceType>(ResourceType.BAD, "^ambox$", 0, ....));
10	18	}
11	19
12	20	}
Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/wikis/WikiConfiguration_eswiki.java
—	—	@@ -5,6 +5,7 @@
6	6	import de.brightbyte.wikiword.analyzer.WikiConfiguration;
7	7	import de.brightbyte.wikiword.analyzer.sensor.HasSectionSensor;
8	8	import de.brightbyte.wikiword.analyzer.sensor.HasTemplateLikeSensor;
	9	+import de.brightbyte.wikiword.analyzer.sensor.HasTemplateSensor;
9	10	import de.brightbyte.wikiword.analyzer.sensor.NamespaceSensor;
10	11
11	12	public class WikiConfiguration_eswiki extends WikiConfiguration {
—	—	@@ -16,10 +17,17 @@
17	18
18	19	resourceTypeSensors.add( new HasTemplateLikeSensor<ResourceType>(ResourceType.BAD, "^(Destruir\|Copyvio\|Plagio\|CdbM?\|SRA\|Sin_?relevancia\|Irrelevante\|Autotrad\|RobotDestruir\|Prob\|Publicidad\|Infraesbozo)$", 0));
19	20
20		~~- conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.PERSON, "^(Fica_de_(persona\|actor\|artista\|autoridad)\|BD\|NF\|Sucesi\u00f3n)$", 0));~~
	21	+ conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.PERSON, "^(BD\|NF\|Sucesi\u00f3n)$", 0));
	22	+ conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.PERSON, "^(Fica_de_.+)$", 0, "fechanac"));
	23	+ conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.PERSON, "^(Fica_de_.+)$", 0, "fechamuerte"));
	24	+
21	25	conceptTypeSensors.add( new HasSectionSensor<ConceptType>(ConceptType.PERSON, "Biograf\u00eda"));
22	26	conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.LIFEFORM, "^(Taxobox\|Fica_de_(tax\u00f3n))$", 0));
23	27	conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.PLACE, "^(Ficha_de_localidad.*)$", 0));
	28	+ conceptTypeSensors.add( new HasTemplateSensor<ConceptType>(ConceptType.PLACE, "coord"));
	29	+ conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.PLACE, "^(Fica_de_.+)$", 0, "coor"));
	30	+ conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.PLACE, "^(Fica_de_.+)$", 0, "mapa"));
	31	+ conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.PLACE, "^(Fica_de_.+)$", 0, "poblaci\u00f3n"));
24	32
25	33	//TODO: number, date, event, work...
26	34	}
Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/wikis/WikiConfiguration_itwiki.java
—	—	@@ -1,11 +1,16 @@
2	2	package de.brightbyte.wikiword.wikis;
3	3
	4	+import de.brightbyte.wikiword.ResourceType;
4	5	import de.brightbyte.wikiword.analyzer.WikiConfiguration;
	6	+import de.brightbyte.wikiword.analyzer.sensor.HasCategoryLikeSensor;
	7	+import de.brightbyte.wikiword.analyzer.sensor.HasTemplateLikeSensor;
5	8
6	9	public class WikiConfiguration_itwiki extends WikiConfiguration {
7	10
8	11	public WikiConfiguration_itwiki() {
9		~~- //TODO: configure for no.wikipedia.org~~
	12	+ //resourceTypeSensors.add( new TitleSensor<ResourceType>(ResourceType.LIST, "^Elenco_", 0));
	13	+ resourceTypeSensors.add( new HasCategoryLikeSensor<ResourceType>(ResourceType.LIST, "^(Liste\|Cronologie\|Cronologia)(_\|$)", 0));
	14	+ resourceTypeSensors.add( new HasTemplateLikeSensor<ResourceType>(ResourceType.BAD, "^(Cancellazione\|Cancella_subito\|Cancelcopy)$", 0));
10	15	}
11	16
12	17	}
Index: trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/wikis/WikiConfiguration_enwiki.java
—	—	@@ -149,8 +149,13 @@
150	150	conceptTypeSensors.add( new HasTemplateSensor<ConceptType>(ConceptType.TIME, "Day", null));
151	151
152	152	conceptTypeSensors.add( new HasCategoryLikeSensor<ConceptType>(ConceptType.WORK, "(^\|_)(statue\|work\|play\|album\|song\|painting\|opera\|novel\|musical\|novel\|composition)s?(_\|$)", Pattern.CASE_INSENSITIVE));
	153	+ conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.EVENT, "^Infobox_.*$", Pattern.CASE_INSENSITIVE, "artist"));
	154	+ conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.EVENT, "^Infobox_.*$", Pattern.CASE_INSENSITIVE, "author"));
	155	+ conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.EVENT, "^Infobox_.*$", Pattern.CASE_INSENSITIVE, "composer"));
	156	+
153	157	conceptTypeSensors.add( new HasCategoryLikeSensor<ConceptType>(ConceptType.EVENT, "(^\|_)(event\|war\|battle\|siege\|treaties\|flood\|famine\|fire\|conflict\|crisis\|disaster\|riot\|assasination\|execution\|crime)s?(_\|$)", Pattern.CASE_INSENSITIVE));
154		~~- conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.EVENT, "^Infobox_Military_Conflict$", Pattern.CASE_INSENSITIVE));~~
	158	+ conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.EVENT, "^Infobox_.*$", Pattern.CASE_INSENSITIVE, "date"));
	159	+ conceptTypeSensors.add( new HasTemplateLikeSensor<ConceptType>(ConceptType.EVENT, "^Infobox_.*$", Pattern.CASE_INSENSITIVE, "year"));
155	160
156	161	conceptTypeSensors.add( new HasTemplateSensor<ConceptType>(ConceptType.LIFEFORM, "Taxobox", null));
157	162	conceptTypeSensors.add( new HasCategoryLikeSensor<ConceptType>(ConceptType.NUMBER, "^Integers$\|(^N\|_n)umbers$", 0));

Status & tagging log

18:28, 1 August 2009 😂 (talk | contribs) changed the status of r52895 [removed: new added: deferred]