package com.hughes.android.dictionary.engine;
import java.io.File;
+import java.io.FileNotFoundException;
+import java.io.IOException;
import java.io.PrintWriter;
import java.io.RandomAccessFile;
import java.util.ArrayList;
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 {
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");
// isoToWikiName.keySet().retainAll(Arrays.asList("UK", "HR", "FI"));
- boolean go = true;
- isoToWikiName.clear();
+ //isoToWikiName.clear();
+ boolean go = false;
for (final String foreignIso : isoToWikiName.keySet()) {
- if (foreignIso.equals("SV")) {
+ if (foreignIso.equals("SK")) {
go = true;
}
if (!go) {
});
// 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();
+ printToText(dictFile);
} // foreignIso
+ final String dictFile = OUTPUTS + "DE-EN_chemnitz_enwiktionary.quickdic";
DictionaryBuilder.main(new String[] {
- "--dictOut=" + OUTPUTS + "DE-EN_chemnitz_enwiktionary.quickdic",
+ "--dictOut=" + dictFile,
"--lang1=DE",
"--lang2=EN",
String.format("--lang1Stoplist=%s", STOPLISTS + "de.txt"),
"--input3LangPattern=German",
"--input3LangCodePattern=de",
"--input3EnIndex=2",
-
});
-
+ printToText(dictFile);
+
+ }
+
+ static void printToText(final String dictFile) throws IOException {
+ 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();
}
}