package com.hughes.android.dictionary.engine;
-import java.io.File;
-import java.io.PrintWriter;
-import java.io.RandomAccessFile;
-import java.util.ArrayList;
-import java.util.Collections;
import java.util.LinkedHashMap;
-import java.util.List;
import java.util.Map;
import junit.framework.TestCase;
+import com.hughes.android.dictionary.parser.wiktionary.WiktionaryLangs;
+
public class DictionaryBuilderMain extends TestCase {
- static final String INPUTS = "../DictionaryData/inputs/";
- static final String STOPLISTS = "../DictionaryData/inputs/stoplists/";
- static final String OUTPUTS = "../DictionaryData/outputs/";
+ static final String INPUTS = "data/inputs/";
+ static final String STOPLISTS = "data/inputs/stoplists/";
+ static final String OUTPUTS = "data/outputs/";
public static void main(final String[] args) throws Exception {
- final Map<String,String> isoToWikiName = new LinkedHashMap<String, String>(Language.isoCodeToWikiName);
+ // Builds all the dictionaries it can, outputs list to a text file.
+
+ final Map<String,String> isoToWikiName = new LinkedHashMap<String, String>(WiktionaryLangs.isoCodeToWikiName);
isoToWikiName.remove("EN");
isoToWikiName.remove("DE");
final Map<String,String> isoToDedication = new LinkedHashMap<String, String>();
isoToDedication.put("AF", "Afrikaans dictionary dedicated to Heiko and Mariëtte Horn.");
- isoToDedication.put("HR", "Croation dictionary dedicated to Ines Viskic and Miro Kresonja.");
+ isoToDedication.put("HR", "Croatian dictionary dedicated to Ines Viskic and Miro Kresonja.");
isoToDedication.put("NL", "Dutch dictionary dedicated to Mike LeBeau.");
// German handled in file.
isoToDedication.put("EL", "Greek dictionary dedicated to Noah Egge.");
isoToDedication.put("IT", "Italian dictionary dedicated to Carolina Tropini, my favorite stardust in the whole universe! Ti amo!");
- isoToDedication.put("JA", "Japanese dictionary dedicated to Akane Watanabe.");
isoToDedication.put("KO", "Korean dictionary dedicated to Ande Elwood--fall fashion und Fernsehturms!");
isoToDedication.put("PT", "Portuguese dictionary dedicated to Carlos Melo, one Tough Mudder.");
isoToDedication.put("RO", "Romanian dictionary dedicated to Radu Teodorescu.");
isoToStoplist.put("FR", "fr.txt");
final Map<String,String> isoToRegex = new LinkedHashMap<String, String>();
- isoToRegex.put("ZH", ".*Chinese.*|.*Mandarin.*|.*Cantonese.*");
+ // HACK: The missing "e" prevents a full match, causing "Cantonese" to be appended to the entries.
+ isoToRegex.put("ZH", "Chinese|Mandarin|Cantones");
- boolean go = false;
- isoToWikiName.clear();
+// isoToWikiName.keySet().retainAll(Arrays.asList("UK", "HR", "FI"));
+ //isoToWikiName.clear();
+ boolean go = false;
for (final String foreignIso : isoToWikiName.keySet()) {
- if (foreignIso.equals("GA")) {
+ if (foreignIso.equals("BO")) {
go = true;
}
if (!go) {
continue;
}
- final String dictFile = String.format(OUTPUTS + "/EN-%s_enwiktionary.quickdic", foreignIso);
+ final String dictFile = String.format("%s/EN-%s_enwiktionary.quickdic", OUTPUTS, foreignIso);
System.out.println("building dictFile: " + dictFile);
if (!isoToStoplist.containsKey(foreignIso)) {
isoToDedication.put(foreignIso, "");
}
if (!isoToRegex.containsKey(foreignIso)) {
- isoToRegex.put(foreignIso, ".*" + isoToWikiName.get(foreignIso) + ".*");
+ isoToRegex.put(foreignIso, isoToWikiName.get(foreignIso));
}
DictionaryBuilder.main(new String[] {
String.format("--lang2=%s", foreignIso),
String.format("--lang1Stoplist=%s", STOPLISTS + isoToStoplist.get("EN")),
String.format("--lang2Stoplist=%s", STOPLISTS + isoToStoplist.get(foreignIso)),
- String.format("--dictInfo=(EN)Wikitionary-based EN-%s dictionary. %s", foreignIso, isoToDedication.get(foreignIso)),
+ String.format("--dictInfo=(EN)Wikitionary-based EN-%s dictionary.\n\n%s", foreignIso, isoToDedication.get(foreignIso)),
- "--input2=" + INPUTS + "enWikiSplit/" + foreignIso + ".data",
+ "--input2=" + INPUTS + "eikiSplit/en/" + foreignIso + ".data",
"--input2Name=enwiktionary." + foreignIso,
"--input2Format=enwiktionary",
"--input2LangPattern=" + isoToRegex.get(foreignIso),
"--input2LangCodePattern=" + foreignIso.toLowerCase(),
"--input2EnIndex=1",
- "--input3=" + INPUTS + "enWikiSplit/EN.data",
+ "--input3=" + INPUTS + "wikiSplit/en/EN.data",
"--input3Name=enwiktionary.english",
"--input3Format=enwiktionary",
"--input3LangPattern=" + isoToRegex.get(foreignIso),
});
- // Print the entries for diffing.
- final RandomAccessFile raf = new RandomAccessFile(new File(dictFile), "r");
- final Dictionary dict = new Dictionary(raf);
- final PrintWriter textOut = new PrintWriter(new File(dictFile + ".text"));
- final List<PairEntry> sorted = new ArrayList<PairEntry>(dict.pairEntries);
- Collections.sort(sorted);
- for (final PairEntry pairEntry : sorted) {
- textOut.println(pairEntry.getRawText(false));
- }
- textOut.close();
- raf.close();
-
} // foreignIso
+ final String dictFile = String.format("%s/DE-EN_chemnitz_enwiktionary.quickdic", OUTPUTS);
DictionaryBuilder.main(new String[] {
- "--dictOut=" + OUTPUTS + "DE-EN_chemnitz_enwiktionary.quickdic",
+ "--dictOut=" + dictFile,
"--lang1=DE",
"--lang2=EN",
- String.format("--lang1Stoplist=%s", STOPLISTS + isoToStoplist.get("DE")),
- String.format("--lang2Stoplist=%s", STOPLISTS + isoToStoplist.get("EN")),
+ String.format("--lang1Stoplist=%s", STOPLISTS + "de.txt"),
+ String.format("--lang2Stoplist=%s", STOPLISTS + "en.txt"),
"--dictInfo=@" + INPUTS + "de-en_chemnitz_enwiktionary.info",
- "--input1=" + INPUTS + "de-en_chemnitz.txt",
- "--input1Name=chemnitz",
- "--input1Charset=UTF8",
- "--input1Format=chemnitz",
+ "--input4=" + INPUTS + "de-en_chemnitz.txt",
+ "--input4Name=chemnitz",
+ "--input4Charset=UTF8",
+ "--input4Format=chemnitz",
- "--input2=" + INPUTS + "enWikiSplit/DE.data",
+ "--input2=" + INPUTS + "wikiSplit/en/DE.data",
"--input2Name=enwiktionary.DE",
"--input2Format=enwiktionary",
- "--input2LangPattern=" + isoToRegex.get("DE"),
+ "--input2LangPattern=German",
"--input2LangCodePattern=de",
"--input2EnIndex=2",
- "--input3=" + INPUTS + "enWikiSplit/EN.data",
+ "--input3=" + INPUTS + "wikiSplit/en/EN.data",
"--input3Name=enwiktionary.english",
"--input3Format=enwiktionary",
- "--input3LangPattern=" + isoToRegex.get("DE"),
+ "--input3LangPattern=German",
"--input3LangCodePattern=de",
"--input3EnIndex=2",
-
});
-
+
}
-
+
}
projects / DictionaryPC.git / blobdiff