Search in sources :

Example 1 with Lexicon

use of com.joliciel.jochre.lexicon.Lexicon in project jochre by urieli.

the class HarkavyLexiconReader method main.

public static void main(String[] args) throws Exception {
    long startTime = (new Date()).getTime();
    try {
        String command = args[0];
        if (command.equals("load")) {
            HarkavyLexiconReader reader = new HarkavyLexiconReader();
            File file = new File(args[1]);
            Writer variantWriter = null;
            if (args.length > 2) {
                File variantFile = new File(args[2]);
                variantFile.delete();
                variantWriter = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(variantFile, true), "UTF8"));
            }
            reader.setVariantWriter(variantWriter);
            try {
                TextFileLexicon lexicon = reader.read(file);
                if (args.length > 3) {
                    File lexiconFile = new File(args[3]);
                    lexicon.serialize(lexiconFile);
                }
            } finally {
                if (variantWriter != null)
                    variantWriter.close();
            }
        } else if (command.equals("deserialise")) {
            File memoryBaseFile = new File(args[1]);
            String word = args[2];
            Lexicon lexicon = TextFileLexicon.deserialize(memoryBaseFile);
            LOG.debug("Have entry " + word + ": " + lexicon.getFrequency(word));
        } else {
            throw new RuntimeException("Unknown command: " + command);
        }
    } finally {
        long endTime = (new Date()).getTime() - startTime;
        LOG.debug("Total runtime: " + ((double) endTime / 1000) + " seconds");
    }
}
Also used : FileOutputStream(java.io.FileOutputStream) TextFileLexicon(com.joliciel.jochre.lexicon.TextFileLexicon) Lexicon(com.joliciel.jochre.lexicon.Lexicon) TextFileLexicon(com.joliciel.jochre.lexicon.TextFileLexicon) OutputStreamWriter(java.io.OutputStreamWriter) File(java.io.File) Date(java.util.Date) BufferedWriter(java.io.BufferedWriter) Writer(java.io.Writer) OutputStreamWriter(java.io.OutputStreamWriter) BufferedWriter(java.io.BufferedWriter)

Example 2 with Lexicon

use of com.joliciel.jochre.lexicon.Lexicon in project jochre by urieli.

the class NameListReader method main.

public static void main(String[] args) throws Exception {
    long startTime = (new Date()).getTime();
    try {
        String command = args[0];
        if (command.equals("load")) {
            NameListReader reader = new NameListReader();
            File file = new File(args[1]);
            Writer variantWriter = null;
            if (args.length > 2) {
                File variantFile = new File(args[2]);
                variantFile.delete();
                variantWriter = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(variantFile, true), "UTF8"));
            }
            reader.setVariantWriter(variantWriter);
            reader.defaultAttribute = args[4];
            try {
                TextFileLexicon lexicon = reader.read(file);
                if (args.length > 3) {
                    File lexiconFile = new File(args[3]);
                    lexicon.serialize(lexiconFile);
                }
            } finally {
                if (variantWriter != null)
                    variantWriter.close();
            }
        } else if (command.equals("deserialise")) {
            File memoryBaseFile = new File(args[1]);
            Lexicon lexicon = TextFileLexicon.deserialize(memoryBaseFile);
            String[] words = new String[] { "חײמס", "חױמס" };
            for (String word : words) LOG.debug("Have entry " + word + ": " + lexicon.getFrequency(word));
        } else {
            throw new RuntimeException("Unknown command: " + command);
        }
    } finally {
        long endTime = (new Date()).getTime() - startTime;
        LOG.debug("Total runtime: " + ((double) endTime / 1000) + " seconds");
    }
}
Also used : FileOutputStream(java.io.FileOutputStream) TextFileLexicon(com.joliciel.jochre.lexicon.TextFileLexicon) Lexicon(com.joliciel.jochre.lexicon.Lexicon) TextFileLexicon(com.joliciel.jochre.lexicon.TextFileLexicon) OutputStreamWriter(java.io.OutputStreamWriter) File(java.io.File) Date(java.util.Date) BufferedWriter(java.io.BufferedWriter) Writer(java.io.Writer) OutputStreamWriter(java.io.OutputStreamWriter) BufferedWriter(java.io.BufferedWriter)

Example 3 with Lexicon

use of com.joliciel.jochre.lexicon.Lexicon in project jochre by urieli.

the class NiborskiLexiconReader method main.

public static void main(String[] args) throws Exception {
    long startTime = (new Date()).getTime();
    try {
        String command = args[0];
        if (command.equals("load")) {
            NiborskiLexiconReader reader = new NiborskiLexiconReader();
            File file = new File(args[1]);
            Writer variantWriter = null;
            if (args.length > 2) {
                File variantFile = new File(args[2]);
                variantFile.getParentFile().mkdirs();
                variantFile.delete();
                variantWriter = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(variantFile, true), "UTF8"));
            }
            reader.setVariantWriter(variantWriter);
            try {
                TextFileLexicon lexicon = reader.read(file);
                if (args.length > 3) {
                    File lexiconFile = new File(args[3]);
                    lexicon.serialize(lexiconFile);
                }
            } finally {
                if (variantWriter != null)
                    variantWriter.close();
            }
        } else if (command.equals("deserialise")) {
            File memoryBaseFile = new File(args[1]);
            Lexicon lexicon = TextFileLexicon.deserialize(memoryBaseFile);
            String[] words = new String[] { "אײגל", "אױגל" };
            for (String word : words) LOG.debug("Have entry " + word + ": " + lexicon.getFrequency(word));
        } else {
            throw new RuntimeException("Unknown command: " + command);
        }
    } finally {
        long endTime = (new Date()).getTime() - startTime;
        LOG.debug("Total runtime: " + ((double) endTime / 1000) + " seconds");
    }
}
Also used : FileOutputStream(java.io.FileOutputStream) TextFileLexicon(com.joliciel.jochre.lexicon.TextFileLexicon) Lexicon(com.joliciel.jochre.lexicon.Lexicon) TextFileLexicon(com.joliciel.jochre.lexicon.TextFileLexicon) OutputStreamWriter(java.io.OutputStreamWriter) File(java.io.File) Date(java.util.Date) BufferedWriter(java.io.BufferedWriter) Writer(java.io.Writer) OutputStreamWriter(java.io.OutputStreamWriter) BufferedWriter(java.io.BufferedWriter)

Example 4 with Lexicon

use of com.joliciel.jochre.lexicon.Lexicon in project jochre by urieli.

the class PlaceListReader method main.

public static void main(String[] args) throws Exception {
    long startTime = (new Date()).getTime();
    try {
        String command = args[0];
        if (command.equals("load")) {
            PlaceListReader reader = new PlaceListReader();
            File file = new File(args[1]);
            Writer variantWriter = null;
            if (args.length > 2) {
                File variantFile = new File(args[2]);
                variantFile.delete();
                variantWriter = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(variantFile, true), "UTF8"));
            }
            reader.setVariantWriter(variantWriter);
            try {
                TextFileLexicon lexicon = reader.read(file);
                if (args.length > 3) {
                    File lexiconFile = new File(args[3]);
                    lexicon.serialize(lexiconFile);
                }
            } finally {
                if (variantWriter != null)
                    variantWriter.close();
            }
        } else if (command.equals("deserialise")) {
            File memoryBaseFile = new File(args[1]);
            Lexicon lexicon = TextFileLexicon.deserialize(memoryBaseFile);
            String[] words = new String[] { "חײמס", "חױמס" };
            for (String word : words) LOG.debug("Have entry " + word + ": " + lexicon.getFrequency(word));
        } else {
            throw new RuntimeException("Unknown command: " + command);
        }
    } finally {
        long endTime = (new Date()).getTime() - startTime;
        LOG.debug("Total runtime: " + ((double) endTime / 1000) + " seconds");
    }
}
Also used : FileOutputStream(java.io.FileOutputStream) TextFileLexicon(com.joliciel.jochre.lexicon.TextFileLexicon) Lexicon(com.joliciel.jochre.lexicon.Lexicon) TextFileLexicon(com.joliciel.jochre.lexicon.TextFileLexicon) OutputStreamWriter(java.io.OutputStreamWriter) File(java.io.File) Date(java.util.Date) BufferedWriter(java.io.BufferedWriter) Writer(java.io.Writer) OutputStreamWriter(java.io.OutputStreamWriter) BufferedWriter(java.io.BufferedWriter)

Example 5 with Lexicon

use of com.joliciel.jochre.lexicon.Lexicon in project jochre by urieli.

the class JochreSession method readLexicon.

protected Lexicon readLexicon(File lexiconDir) {
    Lexicon myLexicon = null;
    if (lexiconDir.isDirectory()) {
        LexiconMerger lexiconMerger = new LexiconMerger();
        File[] lexiconFiles = lexiconDir.listFiles();
        for (File lexiconFile : lexiconFiles) {
            if (lexiconFile.getName().endsWith(".txt")) {
                TextFileLexicon textFileLexicon = new TextFileLexicon(lexiconFile, this.encoding);
                lexiconMerger.addLexicon(textFileLexicon);
            } else {
                Lexicon textFileLexicon = TextFileLexicon.deserialize(lexiconFile);
                lexiconMerger.addLexicon(textFileLexicon);
            }
        }
        myLexicon = lexiconMerger;
    } else {
        if (lexiconDir.getName().endsWith(".txt")) {
            TextFileLexicon textFileLexicon = new TextFileLexicon(lexiconDir, this.encoding);
            myLexicon = textFileLexicon;
        } else {
            Lexicon textFileLexicon = TextFileLexicon.deserialize(lexiconDir);
            myLexicon = textFileLexicon;
        }
    }
    return myLexicon;
}
Also used : LexiconMerger(com.joliciel.jochre.lexicon.LexiconMerger) FakeLexicon(com.joliciel.jochre.lexicon.FakeLexicon) TextFileLexicon(com.joliciel.jochre.lexicon.TextFileLexicon) Lexicon(com.joliciel.jochre.lexicon.Lexicon) TextFileLexicon(com.joliciel.jochre.lexicon.TextFileLexicon) File(java.io.File)

Aggregations

Lexicon (com.joliciel.jochre.lexicon.Lexicon)5 TextFileLexicon (com.joliciel.jochre.lexicon.TextFileLexicon)5 File (java.io.File)5 BufferedWriter (java.io.BufferedWriter)4 FileOutputStream (java.io.FileOutputStream)4 OutputStreamWriter (java.io.OutputStreamWriter)4 Writer (java.io.Writer)4 Date (java.util.Date)4 FakeLexicon (com.joliciel.jochre.lexicon.FakeLexicon)1 LexiconMerger (com.joliciel.jochre.lexicon.LexiconMerger)1