public class DictionaryBuilderTest extends TestCase {
public static final String TEST_INPUTS = "testdata/inputs/";
+ public static final String WIKISPLIT = "data/inputs/wikiSplit/";
public static final String WIKISPLIT_EN = "data/inputs/wikiSplit/en/";
public static final String STOPLISTS = "data/inputs/stoplists/";
public static final String GOLDENS = "testdata/goldens/";
"{{it-conj-arsi|lav}}\n" +
"{{it-conj-ersi|abbatt}}\n" +
"{{it-conj-iarsi|annoi}}\n" +
- "{{it-conj-carsi|coniug}\n" +
- "{{it-conj-ciarsi|affacc}\n" +
+ "{{it-conj-carsi|coniug}}\n" +
+ "{{it-conj-ciarsi|affacc}}\n" +
"{{it-conj-irsi|vest}}\n" +
"{{it-conj-irsi-b|fer}}\n" +
"{{it-conj-ursi|rid|essere}}\n" +
"{{it-conj-cire|ricuc|avere}}\n" +
"{{it-conj-iarsi-b|riavvi|essere}}" +
"{{it-conj-fare|putre|avere}}\n" +
- "{{it-conj-cirsi|cuc|essere}}\n"
+ "{{it-conj-cirsi|cuc|essere}}\n" +
+ "{{it-conj-ere|smett|avere|pastp=smesso|prem1s=smisi|prem3s=smise|prem3s2=''|prem3p=smisero|prem3p2=''}}\n" +
+ "{{term||[[cor#Latin|Cor]] [[Carolus#Latin|Carolī]]|Charles' heart}}\n" +
+ "{{term|sc=Grek|λόγος|tr=lógos||word}}\n" +
+ "{{term|verbo|verbō|for the word}}\n"
;
final DictionaryBuilder db = new DictionaryBuilder("", Language.en, Language.it, "", "", Collections.singleton("X"), Collections.singleton("X"));
- WholeSectionToHtmlParser parser = new WholeSectionToHtmlParser(db.indexBuilders.get(0), "EN", "IT");
+ WholeSectionToHtmlParser parser = new WholeSectionToHtmlParser(db.indexBuilders.get(0), null, "EN", "IT", "http://en.wiktionary.org/wiki/%s");
parser.title = "dummyTitle";
+ parser.entrySource = new EntrySource(0, "dummySource", 0);
parser.parseSection("dummyHeading", toParse);
db.build();
});
checkGolden(name, result);
}
-
+ //-----------------------------------------------------------------
+
+ public void testSingleLang_EN() throws Exception {
+ wiktionaryTestSingleLang("SingleLang_EN.quickdic", "EN", 100);
+ }
+
+ public void testSingleLang_DE() throws Exception {
+ wiktionaryTestSingleLang("SingleLang_DE.quickdic", "DE", 100);
+ }
+
+ public void testSingleLang_IT() throws Exception {
+ wiktionaryTestSingleLang("SingleLang_IT.quickdic", "IT", 100);
+ }
+
+ public void testSingleLang_FR() throws Exception {
+ wiktionaryTestSingleLang("SingleLang_FR.quickdic", "FR", 100);
+ }
+
+ public void wiktionaryTestSingleLang(final String name, final String langCode, final int pageLimit) throws Exception {
+ final File result = new File(TEST_OUTPUTS + name);
+ System.out.println("Writing to: " + result);
+ DictionaryBuilder.main(new String[] {
+ "--dictOut=" + result.getAbsolutePath(),
+ "--lang1=" + langCode,
+ "--lang1Stoplist=" + STOPLISTS + "empty.txt",
+ "--dictInfo=SomeWikiDataWholeSection",
+ "--input4=" + WIKISPLIT + langCode.toLowerCase() + "/" + langCode + ".data",
+ "--input4Name=" + name,
+ "--input4Format=" + WholeSectionToHtmlParser.NAME,
+ "--input4WiktionaryLang=" + langCode,
+ "--input4SkipLang=" + langCode,
+ "--input4TitleIndex=" + "1",
+ "--input4PageLimit=" + pageLimit,
+ "--print=" + result.getPath() + ".text",
+ });
+ checkGolden(name, result);
+ }
+
+ //-----------------------------------------------------------------
+
public void testWiktionary_IT_EN() throws Exception {
wiktionaryTestWithLangToEn("wiktionary.it_en.quickdic", "IT", "it.txt",
- "EN.data", "enwiktionary.english", "Italian", "it");
+ "EN.data", "enwiktionary.english", "Italian", "it", 1000);
}
public void testWiktionary_ZH_EN() throws Exception {
wiktionaryTestWithLangToEn("wiktionary.zh_en.quickdic", "ZH", "empty.txt",
// These missing "e" prevents a complete match, forcing the name to be printed
- "EN.data", "enwiktionary.english", "Chinese|Mandarin|Cantones", "zh");
+ "EN.data", "enwiktionary.english", "Chinese|Mandarin|Cantones", "zh", 1000);
}
public void testWiktionary_DE_EN() throws Exception {
wiktionaryTestWithLangToEn("wiktionary.de_en.quickdic", "DE", "de.txt",
- "EN.data", "enwiktionary.english", "German", "it");
+ "EN.data", "enwiktionary.english", "German", "de", 1000);
}
public void testWiktionary_IT_IT() throws Exception {
wiktionaryTestWithLangToEn("wiktionary.it_it.quickdic", "IT", "it.txt",
- "IT.data", "enwiktionary.italian", "Italian", "it");
+ "IT.data", "enwiktionary.italian", "Italian", "it", 1000);
}
// French
public void testWiktionary_FR_FR() throws Exception {
wiktionaryTestWithLangToEn("wiktionary.fr_fr.quickdic", "FR", "fr.txt",
- "FR.data", "enwiktionary.french", "French", "fr");
+ "FR.data", "enwiktionary.french", "French", "fr", 1000);
}
// Arabic
public void testWiktionary_AR_AR() throws Exception {
+ // Arabic is really big for some reason, use fewer pages.
wiktionaryTestWithLangToEn("wiktionary.ar_ar.quickdic", "AR", "empty.txt",
- "AR.data", "enwiktionary.arabic", "Arabic", "ar");
+ "AR.data", "enwiktionary.arabic", "Arabic", "ar", 200);
}
// Chinese
public void testWiktionary_ZH_ZH() throws Exception {
wiktionaryTestWithLangToEn("wiktionary.zh_zh.quickdic", "ZH", "empty.txt",
// These missing "e" prevents a complete match, forcing the name to be printed.
- "ZH.data", "enwiktionary.chinese", "Chinese|Mandarin|Cantones", "zh");
+ "ZH.data", "enwiktionary.chinese", "Chinese|Mandarin|Cantones", "zh", 1000);
}
// German
public void testWiktionary_DE_DE() throws Exception {
wiktionaryTestWithLangToEn("wiktionary.de_de.quickdic", "DE", "de.txt",
- "DE.data", "enwiktionary.german", "German", "it");
+ "DE.data", "enwiktionary.german", "German", "de", 1000);
}
// Thai
public void testWiktionary_TH_TH() throws Exception {
wiktionaryTestWithLangToEn("wiktionary.th_th.quickdic", "TH", "empty.txt",
// These missing "e" prevents a complete match, forcing the name to be printed.
- "TH.data", "enwiktionary.thai", "Thai", "th");
+ "TH.data", "enwiktionary.thai", "Thai", "th", 1000);
}
public void wiktionaryTestWithLangToEn(final String name, final String lang1,
final String stoplist, final String data, final String dictName,
- final String langPattern, final String langCode) throws Exception {
+ final String langPattern, final String langCode, int pageLimit) throws Exception {
final File result = new File(TEST_OUTPUTS + name);
System.out.println("Writing to: " + result);
final String type = data.equals("EN.data") ? "EnToTranslation" : "EnForeign";
"--input4LangPattern=" + langPattern,
"--input4LangCodePattern=" + langCode,
"--input4EnIndex=2",
- "--input4PageLimit=1000",
+ "--input4PageLimit=" + pageLimit,
"--print=" + result.getPath() + ".text",
});
projects / DictionaryPC.git / blobdiff