import junit.framework.TestCase;
-import com.hughes.android.dictionary.parser.enwiktionary.EnWiktionaryLangs;
+import com.hughes.android.dictionary.parser.wiktionary.WiktionaryLangs;
public class DictionaryBuilderMain extends TestCase {
static final String INPUTS = "data/inputs/";
static final String STOPLISTS = "data/inputs/stoplists/";
static final String OUTPUTS = "data/outputs/";
-
- static final String VERSION_SUFFIX = "v002";
-
-
+
public static void main(final String[] args) throws Exception {
// Builds all the dictionaries it can, outputs list to a text file.
- final Map<String,String> isoToWikiName = new LinkedHashMap<String, String>(EnWiktionaryLangs.isoCodeToWikiName);
+ final Map<String,String> isoToWikiName = new LinkedHashMap<String, String>(WiktionaryLangs.isoCodeToWikiName);
isoToWikiName.remove("EN");
isoToWikiName.remove("DE");
// isoToWikiName.keySet().retainAll(Arrays.asList("UK", "HR", "FI"));
//isoToWikiName.clear();
- boolean go = true;
+ boolean go = false;
for (final String foreignIso : isoToWikiName.keySet()) {
- if (foreignIso.equals("SK")) {
+ if (foreignIso.equals("BO")) {
go = true;
}
if (!go) {
continue;
}
- final String dictFile = String.format("%s/EN-%s_enwiktionary.%s.quickdic", OUTPUTS, foreignIso, VERSION_SUFFIX);
+ final String dictFile = String.format("%s/EN-%s_enwiktionary.quickdic", OUTPUTS, foreignIso);
System.out.println("building dictFile: " + dictFile);
if (!isoToStoplist.containsKey(foreignIso)) {
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),
} // foreignIso
- final String dictFile = String.format("%s/DE-EN_chemnitz_enwiktionary.%s.quickdic", OUTPUTS, VERSION_SUFFIX);
+ final String dictFile = String.format("%s/DE-EN_chemnitz_enwiktionary.quickdic", OUTPUTS);
DictionaryBuilder.main(new String[] {
"--dictOut=" + dictFile,
"--lang1=DE",
"--input4Charset=UTF8",
"--input4Format=chemnitz",
- "--input2=" + INPUTS + "enWikiSplit/DE.data",
+ "--input2=" + INPUTS + "wikiSplit/en/DE.data",
"--input2Name=enwiktionary.DE",
"--input2Format=enwiktionary",
"--input2LangPattern=German",
"--input2LangCodePattern=de",
"--input2EnIndex=2",
- "--input3=" + INPUTS + "enWikiSplit/EN.data",
+ "--input3=" + INPUTS + "wikiSplit/en/EN.data",
"--input3Name=enwiktionary.english",
"--input3Format=enwiktionary",
"--input3LangPattern=German",