1 package com.hughes.android.dictionary.engine;
4 import java.io.IOException;
5 import java.io.PrintStream;
6 import java.io.RandomAccessFile;
7 import java.nio.charset.Charset;
8 import java.util.ArrayList;
12 import com.hughes.util.Args;
13 import com.hughes.util.FileUtil;
21 --dictInfo=@dictInfo.txt
23 --input0=/Users/thadh/personal/quickDic/de-en-chemnitz.txt
26 --input0Format=chemnitz
28 --input1=/Users/thadh/personal/quickDic/dewiktionary-20100326-pages-articles.xml
29 --input1Name=wiktionary
30 --input1Format=wiktionary
32 --input2=/Users/thadh/personal/quickDic/de-en-dictcc.txt
34 --input2Charset=Cp1252
38 public class DictionaryBuilder {
40 final Dictionary dictionary;
42 final List<EntryData> entryDatas = new ArrayList<EntryData>();
44 final List<IndexBuilder> indexBuilders = new ArrayList<IndexBuilder>();
46 public DictionaryBuilder(final String dictInfo, final Language lang0, final Language lang1) {
47 dictionary = new Dictionary(dictInfo);
48 indexBuilders.add(new IndexBuilder(this, lang0.getSymbol(), lang0.getSymbol() + "->" + lang1.getSymbol(), lang0, false));
49 indexBuilders.add(new IndexBuilder(this, lang1.getSymbol(), lang1.getSymbol() + "->" + lang0.getSymbol(), lang1, true));
53 for (final IndexBuilder indexBuilder : indexBuilders) {
55 dictionary.indices.add(indexBuilder.index);
59 public static void main(final String[] args) throws IOException {
60 final Map<String,String> keyValueArgs = Args.keyValueArgs(args);
62 final Language lang1 = Language.lookup(keyValueArgs.remove("lang1"));
63 final Language lang2 = Language.lookup(keyValueArgs.remove("lang2"));
64 if (lang1 == null || lang2 == null) {
65 fatalError("--lang1= and --lang2= must both be specified.");
68 final String dictOutFilename = keyValueArgs.remove("dictOut");
69 if (dictOutFilename == null) {
70 fatalError("--dictOut= must be specified.");
73 String dictInfo = keyValueArgs.remove("dictInfo");
74 if (dictInfo == null) {
75 fatalError("--dictInfo= must be specified.");
77 if (dictInfo.startsWith("@")) {
78 dictInfo = FileUtil.readToString(new File(dictInfo.substring(1)));
81 final String printFile = keyValueArgs.remove("print");
83 System.out.println("lang1=" + lang1);
84 System.out.println("lang2=" + lang2);
85 System.out.println("dictInfo=" + dictInfo);
86 System.out.println("dictOut=" + dictOutFilename);
88 final DictionaryBuilder dictionaryBuilder = new DictionaryBuilder(dictInfo, lang1, lang2);
90 for (int i = 0; i < 100; ++i) {
91 final String prefix = "input" + i;
92 if (keyValueArgs.containsKey(prefix)) {
93 final File file = new File(keyValueArgs.remove(prefix));
94 System.out.println("Processing: " + file);
95 String charsetName = keyValueArgs.remove(prefix + "Charset");
96 if (charsetName == null) {
99 final Charset charset = Charset.forName(charsetName);
100 String inputName = keyValueArgs.remove(prefix + "Name");
101 if (inputName == null) {
102 fatalError("Must specify human readable name for: " + prefix + "Name");
105 String inputFormat = keyValueArgs.remove(prefix + "Format");
106 if ("dictcc".equals(inputFormat)) {
107 new DictFileParser(charset, false, DictFileParser.TAB, null, dictionaryBuilder, dictionaryBuilder.indexBuilders.toArray(new IndexBuilder[0]), null).parseFile(file);
108 } else if ("chemnitz".equals(inputFormat)) {
109 new DictFileParser(charset, false, DictFileParser.DOUBLE_COLON, DictFileParser.PIPE, dictionaryBuilder, dictionaryBuilder.indexBuilders.toArray(new IndexBuilder[0]), null).parseFile(file);
110 } else if ("wiktionary".equals(inputFormat)) {
111 throw new RuntimeException();
112 // new WiktionaryXmlParser(dict).parse(file);
114 fatalError("Invalid or missing input format: " + inputFormat);
117 final EntrySource entrySource = new EntrySource(dictionaryBuilder.dictionary.sources.size(), inputName);
118 dictionaryBuilder.dictionary.sources.add(entrySource);
119 System.out.println("Done: " + file + "\n\n");
123 dictionaryBuilder.build();
125 if (printFile != null) {
126 final PrintStream out = new PrintStream(new File(printFile));
127 dictionaryBuilder.dictionary.print(out);
131 System.out.println("Writing dictionary to: " + dictOutFilename);
132 final RandomAccessFile dictOut = new RandomAccessFile(dictOutFilename, "rw");
133 dictOut.setLength(0);
134 dictionaryBuilder.dictionary.write(dictOut);
137 if (!keyValueArgs.isEmpty()) {
138 System.err.println("WARNING: couldn't parse arguments: " + keyValueArgs);
144 private static void fatalError(String string) {
145 System.err.println(string);