+
+ final Set<String> lang1Stoplist = new LinkedHashSet<>();
+ final Set<String> lang2Stoplist = new LinkedHashSet<>();
+ final String lang1StoplistFile = keyValueArgs.remove("lang1Stoplist");
+ final String lang2StoplistFile = keyValueArgs.remove("lang2Stoplist");
+ if (lang1StoplistFile != null) {
+ lang1Stoplist.addAll(FileUtil.readLines(new File(lang1StoplistFile)));
+ }
+ if (lang2StoplistFile != null) {
+ lang2Stoplist.addAll(FileUtil.readLines(new File(lang2StoplistFile)));
+ }
+
+ String normalizerRules1 = keyValueArgs.remove("normalizerRules1");
+ String normalizerRules2 = keyValueArgs.remove("normalizerRules2");
+ if (normalizerRules1 == null) {
+ normalizerRules1 = lang1.getDefaultNormalizerRules();
+ }
+ if (normalizerRules2 == null) {
+ normalizerRules2 = lang2 == null ? null : lang2.getDefaultNormalizerRules();
+ }
+
+ final String dictOutFilename = keyValueArgs.remove("dictOut");
+ if (dictOutFilename == null) {
+ fatalError("--dictOut= must be specified.");
+ }
+
+ String dictInfo = keyValueArgs.remove("dictInfo");
+ if (dictInfo == null) {
+ fatalError("--dictInfo= must be specified.");
+ }
+ if (dictInfo.startsWith("@")) {
+ dictInfo = FileUtil.readToString(new File(dictInfo.substring(1)));
+ }
+
+ final String printFile = keyValueArgs.remove("print");
+
+ System.out.println("lang1=" + lang1);
+ System.out.println("lang2=" + lang2);
+ System.out.println("normalizerRules1=" + normalizerRules1);
+ System.out.println("normalizerRules2=" + normalizerRules2);
+ System.out.println("dictInfo=" + dictInfo);
+ System.out.println("dictOut=" + dictOutFilename);
+
+ final DictionaryBuilder dictionaryBuilder = new DictionaryBuilder(dictInfo, lang1, lang2, normalizerRules1, normalizerRules2, lang1Stoplist, lang2Stoplist);
+
+ for (int i = 0; i < 100; ++i) {
+ final String prefix = "input" + i;
+ if (keyValueArgs.containsKey(prefix)) {
+ final File file = new File(keyValueArgs.remove(prefix));
+ System.out.println("Processing: " + file);
+ String charsetName = keyValueArgs.remove(prefix + "Charset");
+ if (charsetName == null) {
+ charsetName = "UTF8";
+ }
+ final Charset charset = Charset.forName(charsetName);
+ String inputName = keyValueArgs.remove(prefix + "Name");
+ if (inputName == null) {
+ fatalError("Must specify human readable name for: " + prefix + "Name");
+ }
+ String pageLimitString = keyValueArgs.remove(prefix + "PageLimit");
+ if (pageLimitString == null) {
+ pageLimitString = "-1";
+ }
+ final int pageLimit = Integer.parseInt(pageLimitString);
+
+ final EntrySource entrySource = new EntrySource(dictionaryBuilder.dictionary.sources.size(), inputName, 0);
+ System.out.println();
+
+ String inputFormat = keyValueArgs.remove(prefix + "Format");
+ if ("tab_separated".equals(inputFormat)) {
+ final boolean flipColumns = "true".equals(keyValueArgs.remove(prefix + "FlipColumns"));
+ new DictFileParser(charset, flipColumns, DictFileParser.TAB, null, dictionaryBuilder).parse(file, entrySource, pageLimit);
+ } else if ("chemnitz".equals(inputFormat)) {
+ final boolean flipColumns = "true".equals(keyValueArgs.remove(prefix + "FlipColumns"));
+ new DictFileParser(charset, flipColumns, DictFileParser.DOUBLE_COLON, DictFileParser.PIPE, dictionaryBuilder).parse(file, entrySource, pageLimit);
+ } else if ("enwiktionary".equals(inputFormat)) {
+ final String type = keyValueArgs.remove(prefix + "WiktionaryType");
+ final Pattern langPattern = Pattern.compile(keyValueArgs.remove(prefix + "LangPattern"), Pattern.CASE_INSENSITIVE);
+ final Pattern langCodePattern = Pattern.compile(keyValueArgs.remove(prefix + "LangCodePattern"));
+ final int enIndex = Integer.parseInt(keyValueArgs.remove(prefix + "EnIndex")) - 1;
+
+ if (enIndex < 0 || enIndex >= 2) {
+ fatalError("Must be 1 or 2: " + prefix + "EnIndex");
+ }
+ final Parser parser;
+ if ("EnToTranslation".equals(type)) {
+ parser = new EnToTranslationParser(dictionaryBuilder.indexBuilders.get(enIndex), dictionaryBuilder.indexBuilders.get(1-enIndex),
+ langPattern, langCodePattern, enIndex != 0);
+ } else if ("EnForeign".equals(type)) {
+ parser = new EnForeignParser(dictionaryBuilder.indexBuilders.get(enIndex), dictionaryBuilder.indexBuilders.get(1-enIndex),
+ langPattern, langCodePattern, enIndex != 0);
+ } else if ("EnEnglish".equals(type)) {
+ parser = new EnForeignParser(dictionaryBuilder.indexBuilders.get(enIndex), dictionaryBuilder.indexBuilders.get(enIndex),
+ langPattern, langCodePattern, true);
+ } else {
+ fatalError("Invalid WiktionaryType (use EnToTranslation or EnForeign or EnEnglish): " + type);
+ return;
+ }
+ parser.parse(file, entrySource, pageLimit);
+ } else if (EnTranslationToTranslationParser.NAME.equals(inputFormat)) {
+ final String code1 = keyValueArgs.remove(prefix + "LangPattern1");
+ final String code2 = keyValueArgs.remove(prefix + "LangPattern2");
+ if (code1 == null || code2 == null) {
+ fatalError("Must specify LangPattern1 and LangPattern2.");
+ return;
+ }
+ final Pattern codePattern1 = Pattern.compile(code1, Pattern.CASE_INSENSITIVE);
+ final Pattern codePattern2 = Pattern.compile(code2, Pattern.CASE_INSENSITIVE);
+ new EnTranslationToTranslationParser(dictionaryBuilder.indexBuilders, new Pattern[] {codePattern1, codePattern2}).parse(file, entrySource, pageLimit);
+ } else if (WholeSectionToHtmlParser.NAME.equals(inputFormat)) {
+ final int titleIndex = Integer.parseInt(keyValueArgs.remove(prefix + "TitleIndex")) - 1;
+ final String wiktionaryLang = keyValueArgs.remove(prefix + "WiktionaryLang");
+ final String webUrlTemplate = keyValueArgs.remove(prefix + "WebUrlTemplate");
+ String skipLang = keyValueArgs.remove(prefix + "SkipLang");
+ if (skipLang == null) skipLang = "";
+ new WholeSectionToHtmlParser(dictionaryBuilder.indexBuilders.get(titleIndex), null, wiktionaryLang, skipLang, webUrlTemplate).parse(file, entrySource, pageLimit);
+ } else {
+ fatalError("Invalid or missing input format: " + inputFormat);
+ }
+
+ dictionaryBuilder.dictionary.sources.add(entrySource);
+ System.out.println("Done: " + file + "\n\n");
+ }
+ }
+
+ dictionaryBuilder.build();
+ // Drop indexBuilders to free RAM
+ dictionaryBuilder.indexBuilders.clear();
+
+ if (printFile != null) {
+ final PrintStream out = new PrintStream(new File(printFile));
+ dictionaryBuilder.dictionary.print(out);
+ out.close();
+ }
+
+ System.out.println("Writing dictionary to: " + dictOutFilename);
+ final RandomAccessFile dictOut = new RandomAccessFile(dictOutFilename, "rw");
+ dictOut.setLength(0);
+ dictionaryBuilder.dictionary.write(dictOut);
+ dictOut.close();
+
+ if (!keyValueArgs.isEmpty()) {
+ System.err.println("WARNING: couldn't parse arguments: " + keyValueArgs);
+ System.exit(1);
+ }
+