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.android.dictionary.Language;
13 import com.hughes.util.Args;
14 import com.hughes.util.FileUtil;
22 --dictInfo=@dictInfo.txt
24 --input0=/Users/thadh/personal/quickDic/de-en-chemnitz.txt
27 --input0Format=chemnitz
29 --input1=/Users/thadh/personal/quickDic/dewiktionary-20100326-pages-articles.xml
30 --input1Name=wiktionary
31 --input1Format=wiktionary
33 --input2=/Users/thadh/personal/quickDic/de-en-dictcc.txt
35 --input2Charset=Cp1252
39 public class DictionaryBuilder {
41 final Dictionary dictionary;
43 final List<EntryData> entryDatas = new ArrayList<EntryData>();
45 final List<IndexBuilder> indexBuilders = new ArrayList<IndexBuilder>();
47 public DictionaryBuilder(final String dictInfo, final Language lang0, final Language lang1) {
48 dictionary = new Dictionary(dictInfo);
49 indexBuilders.add(new IndexBuilder(this, lang0.getSymbol(), lang0.getSymbol() + "->" + lang1.getSymbol(), lang0));
50 indexBuilders.add(new IndexBuilder(this, lang1.getSymbol(), lang1.getSymbol() + "->" + lang0.getSymbol(), lang1));
54 for (final IndexBuilder indexBuilder : indexBuilders) {
56 dictionary.indices.add(indexBuilder.index);
60 public static void main(final String[] args) throws IOException {
61 final Map<String,String> keyValueArgs = Args.keyValueArgs(args);
63 final Language lang1 = Language.lookup(keyValueArgs.remove("lang1"));
64 final Language lang2 = Language.lookup(keyValueArgs.remove("lang2"));
65 if (lang1 == null || lang2 == null) {
66 fatalError("--lang1= and --lang2= must both be specified.");
69 final String dictOutFilename = keyValueArgs.remove("dictOut");
70 if (dictOutFilename == null) {
71 fatalError("--dictOut= must be specified.");
74 String dictInfo = keyValueArgs.remove("dictInfo");
75 if (dictInfo == null) {
76 fatalError("--dictInfo= must be specified.");
78 if (dictInfo.startsWith("@")) {
79 dictInfo = FileUtil.readToString(new File(dictInfo.substring(1)));
82 final String printFile = keyValueArgs.remove("print");
84 System.out.println("lang1=" + lang1);
85 System.out.println("lang2=" + lang2);
86 System.out.println("dictInfo=" + dictInfo);
87 System.out.println("dictOut=" + dictOutFilename);
89 final DictionaryBuilder dictionaryBuilder = new DictionaryBuilder(dictInfo, lang1, lang2);
91 for (int i = 0; i < 100; ++i) {
92 final String prefix = "input" + i;
93 if (keyValueArgs.containsKey(prefix)) {
94 final File file = new File(keyValueArgs.remove(prefix));
95 System.out.println("Processing: " + file);
96 String charsetName = keyValueArgs.remove(prefix + "Charset");
97 if (charsetName == null) {
100 final Charset charset = Charset.forName(charsetName);
101 String inputName = keyValueArgs.remove(prefix + "Name");
102 if (inputName == null) {
103 fatalError("Must specify human readable name for: " + prefix + "Name");
106 String inputFormat = keyValueArgs.remove(prefix + "Format");
107 if ("dictcc".equals(inputFormat)) {
108 new DictFileParser(charset, false, DictFileParser.TAB, null, dictionaryBuilder, dictionaryBuilder.indexBuilders.toArray(new IndexBuilder[0]), null).parseFile(file);
109 } else if ("chemnitz".equals(inputFormat)) {
110 new DictFileParser(charset, false, DictFileParser.DOUBLE_COLON, DictFileParser.PIPE, dictionaryBuilder, dictionaryBuilder.indexBuilders.toArray(new IndexBuilder[0]), null).parseFile(file);
111 } else if ("wiktionary".equals(inputFormat)) {
112 throw new RuntimeException();
113 // new WiktionaryXmlParser(dict).parse(file);
115 fatalError("Invalid or missing input format: " + inputFormat);
118 final EntrySource entrySource = new EntrySource(dictionaryBuilder.dictionary.sources.size(), inputName);
119 dictionaryBuilder.dictionary.sources.add(entrySource);
120 System.out.println("Done: " + file + "\n\n");
124 dictionaryBuilder.build();
126 if (printFile != null) {
127 final PrintStream out = new PrintStream(new File(printFile));
128 dictionaryBuilder.dictionary.print(out);
132 System.out.println("Writing dictionary to: " + dictOutFilename);
133 final RandomAccessFile dictOut = new RandomAccessFile(dictOutFilename, "rw");
134 dictOut.setLength(0);
135 dictionaryBuilder.dictionary.write(dictOut);
138 if (!keyValueArgs.isEmpty()) {
139 System.err.println("WARNING: couldn't parse arguments: " + keyValueArgs);
145 private static void fatalError(String string) {
146 System.err.println(string);