package com.hughes.android.dictionary.engine;
import java.io.File;
+import java.io.FileNotFoundException;
import java.io.IOException;
import java.io.PrintStream;
import java.io.RandomAccessFile;
public class DictionaryBuilderTest extends TestCase {
- public static final String TEST_INPUTS = "../DictionaryData/testdata/inputs/";
- public static final String WIKISPLIT = "../DictionaryData/inputs/wikiSplit/";
- public static final String GOLDENS = "../DictionaryData/testdata/goldens/";
+ public static final String TEST_INPUTS = "testdata/inputs/";
+ public static final String WIKISPLIT = "data/inputs/wikiSplit/en/";
+ public static final String STOPLISTS = "data/inputs/stoplists/";
+ public static final String GOLDENS = "testdata/goldens/";
- public static final String TEST_OUTPUTS = "../DictionaryData/testdata/outputs/";
- public static final String OUTPUTS = "../DictionaryData/outputs/";
+ public static final String TEST_OUTPUTS = "testdata/outputs/";
+
+ public void testWiktionary_IT_EN() throws Exception {
+ wiktionaryTestWithLangToEn("wiktionary.it_en.quickdic", "IT", "it.txt",
+ "EN.data", "enwiktionary.english", "Italian", "it");
+ }
+
+ 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");
+ }
+
+ public void testWiktionary_DE_EN() throws Exception {
+ wiktionaryTestWithLangToEn("wiktionary.de_en.quickdic", "DE", "de.txt",
+ "EN.data", "enwiktionary.english", "German", "it");
+ }
+
+ public void testWiktionary_IT_IT() throws Exception {
+ wiktionaryTestWithLangToEn("wiktionary.it_it.quickdic", "IT", "it.txt",
+ "IT.data", "enwiktionary.italian", "Italian", "it");
+ }
+
+ // French
+ public void testWiktionary_FR_FR() throws Exception {
+ wiktionaryTestWithLangToEn("wiktionary.fr_fr.quickdic", "FR", "fr.txt",
+ "FR.data", "enwiktionary.french", "French", "fr");
+ }
+
+
+ // Arabic
+ public void testWiktionary_AR_AR() throws Exception {
+ wiktionaryTestWithLangToEn("wiktionary.ar_ar.quickdic", "AR", "empty.txt",
+ "AR.data", "enwiktionary.arabic", "Arabic", "ar");
+ }
+
+ // 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");
+ }
- public void testWiktionaryItalian() throws Exception {
- final File result = new File(TEST_OUTPUTS + "wiktionary.it.quickdic");
+ // German
+ public void testWiktionary_DE_DE() throws Exception {
+ wiktionaryTestWithLangToEn("wiktionary.de_de.quickdic", "DE", "de.txt",
+ "DE.data", "enwiktionary.german", "German", "it");
+ }
+
+ 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 File result = new File(TEST_OUTPUTS + name);
System.out.println("Writing to: " + result);
+ final String type = data.equals("EN.data") ? "EnToTranslation" : "EnForeign";
DictionaryBuilder.main(new String[] {
"--dictOut=" + result.getAbsolutePath(),
- "--lang1=IT",
+ "--lang1=" + lang1,
"--lang2=EN",
+ "--lang1Stoplist=" + STOPLISTS + stoplist,
+ "--lang2Stoplist=" + STOPLISTS + "en.txt",
"--dictInfo=SomeWikiData",
- /*
- "--input3=" + WIKISPLIT + "english.data",
- "--input3Name=enwiktionary.english",
- "--input3Format=enwiktionary",
- "--input3LangPattern=Italian",
- "--input3LangCodePattern=it",
- "--input3EnIndex=2",
- "--input3PageLimit=1000",
-*/
- "--input4=" + WIKISPLIT + "italian.data",
- "--input4Name=enwiktionary.italian",
+ "--input4=" + WIKISPLIT + data,
+ "--input4Name=" + dictName,
"--input4Format=enwiktionary",
- "--input4LangPattern=Italian",
- "--input4LangCodePattern=it",
+ "--input4WiktionaryType=" + type,
+ "--input4LangPattern=" + langPattern,
+ "--input4LangCodePattern=" + langCode,
"--input4EnIndex=2",
"--input4PageLimit=1000",
- "--print=" + result.getName() + ".text",
+ "--print=" + result.getPath() + ".text",
});
- // Check it once:
- assertFilesEqual(GOLDENS + "wiktionary.it_it.quickdic.text", result.getName() + ".text");
-
- // Check it again.
- final Dictionary dict = new Dictionary(new RandomAccessFile(result.getAbsolutePath(), "r"));
- final PrintStream out = new PrintStream(new File(result.getName() + ".text"));
- dict.print(out);
- out.close();
-
- assertFilesEqual(GOLDENS + "wiktionary.it_it.quickdic.text", result.getName() + ".text");
+ checkGolden(name, result);
}
-
public void testGermanCombined() throws Exception {
- final File result = new File(TEST_OUTPUTS + "de-en.quickdic");
+ final String name = "de-en.quickdic";
+ final File result = new File(TEST_OUTPUTS + name);
System.out.println("Writing to: " + result);
DictionaryBuilder.main(new String[] {
"--dictOut=" + result.getAbsolutePath(),
"--input2=" + TEST_INPUTS + "de-en_dictcc_simulated",
"--input2Name=dictcc",
"--input2Charset=UTF8",
- "--input2Format=dictcc",
+ "--input2Format=tab_separated",
- "--print=" + result.getName() + ".text",
+ "--print=" + result.getPath() + ".text",
});
+ checkGolden(name, result);
+ }
+
+ private void checkGolden(final String dictName, final File dictFile)
+ throws IOException, FileNotFoundException {
// Check it once:
- assertFilesEqual(GOLDENS + "de-en.quickdic.text", result.getName() + ".text");
-
+ assertFilesEqual(GOLDENS + dictName + ".text", dictFile.getPath() + ".text");
+
// Check it again.
- final Dictionary dict = new Dictionary(new RandomAccessFile(result.getAbsolutePath(), "r"));
- final PrintStream out = new PrintStream(result.getName() + ".text");
+ final Dictionary dict = new Dictionary(new RandomAccessFile(dictFile.getAbsolutePath(), "r"));
+ final PrintStream out = new PrintStream(new File(dictFile.getPath() + ".text"));
dict.print(out);
out.close();
-
- assertFilesEqual(GOLDENS + "de-en.quickdic.text", result.getName() + ".text");
+ assertFilesEqual(GOLDENS + dictName + ".text", dictFile.getPath() + ".text");
}
-
void assertFilesEqual(final String expected, final String actual) throws IOException {
final String expectedString = FileUtil.readToString(new File(expected));
final String actualString = FileUtil.readToString(new File(actual));