| Index: trunk/WikiWord/WikiWordProperties/.project |
| — | — | @@ -1,6 +1,6 @@ |
| 2 | 2 | <?xml version="1.0" encoding="UTF-8"?> |
| 3 | 3 | <projectDescription> |
| 4 | | - <name>WikiWordBuilder4Biography</name> |
| | 4 | + <name>WikiWordProperties</name> |
| 5 | 5 | <comment></comment> |
| 6 | 6 | <projects> |
| 7 | 7 | </projects> |
| Index: trunk/WikiWord/WikiWordProperties/src/main/java/de/brightbyte/wikiword/lifescience/wikis/WikiConfiguration_enwiki.java |
| — | — | @@ -153,7 +153,7 @@ |
| 154 | 154 | private static final String neuroNamesChars = "["+alphabeticChars+"]+-["+numericChars+"]+"; |
| 155 | 155 | |
| 156 | 156 | //TODO: exclude "Biography"... |
| 157 | | - public static final String lifeScienceJournalPattern = "(^|[ _])(Chem[a-z]*|Biol?[.a-z]*|Gen[eo][a-z]*|Med[a-z]*|Cell[a-z]*|DNA|RNA|Nucleic|EMBO|FEBS|Onco[a-z]*|Blood|Immono[a-z]*|Cancer|Virol[a-z]*|Med[a-z]*|Clin[a-z]*|Lancet|Nature|PLoS|Neuro[a-z]*|Zootaxa|JAMA|FASEB|Bacter[a-z]*|Mutat[a-z]*|Mol[a-z]*|Protein|Dermat[a-z]*|Pathol[a-z]*|Endocr[a-z]*|Microbio[a-z]*)($|[_ ])"; |
| | 157 | + //public static final String lifeScienceJournalPattern = "(^|[ _])(Chem[a-z]*|Biol?[.a-z]*|Gen[eo][a-z]*|Med[a-z]*|Cell[a-z]*|DNA|RNA|Nucleic|EMBO|FEBS|Onco[a-z]*|Blood|Immono[a-z]*|Cancer|Virol[a-z]*|Med[a-z]*|Clin[a-z]*|Lancet|Nature|PLoS|Neuro[a-z]*|Zootaxa|JAMA|FASEB|Bacter[a-z]*|Mutat[a-z]*|Mol[a-z]*|Protein|Dermat[a-z]*|Pathol[a-z]*|Endocr[a-z]*|Microbio[a-z]*)($|[_ ])"; |
| 158 | 158 | |
| 159 | 159 | |
| 160 | 160 | protected static DefaultTemplateParameterPropertySpec makeNamePropertySpec(String param, String prop, boolean multi, boolean space) { |
| — | — | @@ -219,8 +219,8 @@ |
| 220 | 220 | |
| 221 | 221 | propertyExtractors.add( new TemplateParameterExtractor( new ExactNameMatcher("Cite_journal"), |
| 222 | 222 | new DefaultTemplateParameterPropertySpec("journal", "journal") |
| 223 | | - .addNormalizer(punctuationStripPattern, "") |
| 224 | | - .setCondition(lifeScienceJournalPattern, 0, false) ) ); |
| | 223 | + .addNormalizer(punctuationStripPattern, "")/* |
| | 224 | + .setCondition(lifeScienceJournalPattern, 0, false)*/ ) ); |
| 225 | 225 | |
| 226 | 226 | TemplateParameterPropertySpec atcSpec = new AbstractTemplateParameterPropertySpec("ATC") { |
| 227 | 227 | private Matcher validator = Pattern.compile("["+upperAlphaNumericChars+"]+").matcher(""); |
| — | — | @@ -473,8 +473,8 @@ |
| 474 | 474 | makeNamePropertySpec("abbreviation", "Name", true, true), |
| 475 | 475 | makeNamePropertySpec("type", "RNA type", true, true), |
| 476 | 476 | new DefaultTemplateParameterPropertySpec("journal", "journal") |
| 477 | | - .addNormalizer(punctuationStripPattern, "") |
| 478 | | - .setCondition(lifeScienceJournalPattern, 0, false) |
| | 477 | + .addNormalizer(punctuationStripPattern, "") /* |
| | 478 | + .setCondition(lifeScienceJournalPattern, 0, false) */ |
| 479 | 479 | ) ); |
| 480 | 480 | |
| 481 | 481 | propertyExtractors.add( new TemplateParameterExtractor(new PatternNameMatcher("Infobox_chemical_analysis", 0, true), |
| Index: trunk/WikiWord/WikiWordProperties/pom.xml |
| — | — | @@ -4,7 +4,7 @@ |
| 5 | 5 |
|
| 6 | 6 | <modelVersion>4.0.0</modelVersion>
|
| 7 | 7 | <groupId>de.wikimedia</groupId>
|
| 8 | | - <artifactId>WikiWordBuilder4Biography</artifactId>
|
| | 8 | + <artifactId>WikiWordProperties</artifactId>
|
| 9 | 9 | <version>0.3</version>
|
| 10 | 10 |
|
| 11 | 11 | <dependencies>
|