|
|
|
/**
|
|
|
|
* LibraryProvider.java
|
|
|
|
* Copyright 2009 by Michael Peter Christen; mc@yacy.net, Frankfurt a. M., Germany
|
|
|
|
* first published 01.10.2009 on http://yacy.net
|
|
|
|
*
|
|
|
|
* This file is part of YaCy Content Integration
|
|
|
|
*
|
|
|
|
* This library is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
|
|
|
* version 2.1 of the License, or (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This library is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* Lesser General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU Lesser General Public License
|
|
|
|
* along with this program in the file lgpl21.txt
|
|
|
|
* If not, see <http://www.gnu.org/licenses/>.
|
|
|
|
*/
|
|
|
|
|
|
|
|
package net.yacy.document;
|
|
|
|
|
|
|
|
import java.io.BufferedReader;
|
|
|
|
import java.io.BufferedWriter;
|
|
|
|
import java.io.File;
|
|
|
|
import java.io.FileWriter;
|
|
|
|
import java.io.IOException;
|
|
|
|
import java.io.InputStream;
|
|
|
|
import java.io.InputStreamReader;
|
|
|
|
import java.io.PrintWriter;
|
|
|
|
import java.net.MalformedURLException;
|
|
|
|
import java.util.ArrayList;
|
|
|
|
import java.util.HashSet;
|
|
|
|
import java.util.List;
|
|
|
|
import java.util.Set;
|
|
|
|
import java.util.TreeSet;
|
|
|
|
import java.util.zip.ZipException;
|
|
|
|
import java.util.zip.ZipFile;
|
|
|
|
|
|
|
|
import net.yacy.cora.document.MultiProtocolURI;
|
|
|
|
import net.yacy.document.geolocalization.GeonamesLocation;
|
|
|
|
import net.yacy.document.geolocalization.OpenGeoDBLocation;
|
|
|
|
import net.yacy.document.geolocalization.OverarchingLocation;
|
|
|
|
import net.yacy.kelondro.logging.Log;
|
|
|
|
import net.yacy.kelondro.util.FileUtils;
|
|
|
|
|
|
|
|
public class LibraryProvider {
|
|
|
|
|
|
|
|
public static final char tagPrefix = '$';
|
|
|
|
public static final String path_to_source_dictionaries = "source";
|
|
|
|
public static final String path_to_did_you_mean_dictionaries = "didyoumean";
|
|
|
|
public static final String path_to_autotagging_dictionaries = "autotagging";
|
|
|
|
|
|
|
|
public static final String disabledExtension = ".disabled";
|
|
|
|
|
|
|
|
public static WordCache dymLib = new WordCache(null);
|
|
|
|
public static Autotagging autotagging = new Autotagging(null, tagPrefix);
|
|
|
|
public static OverarchingLocation geoLoc = new OverarchingLocation();
|
|
|
|
private static File dictSource = null;
|
|
|
|
private static File dictRoot = null;
|
|
|
|
|
|
|
|
public static enum Dictionary {
|
|
|
|
GEODB0(
|
|
|
|
"geo0",
|
|
|
|
"http://downloads.sourceforge.net/project/opengeodb/Data/0.2.5a/opengeodb-0.2.5a-UTF8-sql.gz" ),
|
|
|
|
GEODB1( "geo1", "http://fa-technik.adfc.de/code/opengeodb/dump/opengeodb-02624_2011-10-17.sql.gz" ),
|
|
|
|
GEON0( "geon0", "http://download.geonames.org/export/dump/cities1000.zip" ),
|
|
|
|
DRW0( "drw0", "http://www.ids-mannheim.de/kl/derewo/derewo-v-100000t-2009-04-30-0.1.zip" );
|
|
|
|
|
|
|
|
public String nickname, url, filename;
|
|
|
|
|
|
|
|
private Dictionary(final String nickname, final String url) {
|
|
|
|
try {
|
|
|
|
this.filename = new MultiProtocolURI(url).getFileName();
|
|
|
|
} catch ( final MalformedURLException e ) {
|
|
|
|
assert false;
|
|
|
|
}
|
|
|
|
this.nickname = nickname;
|
|
|
|
this.url = url;
|
|
|
|
}
|
|
|
|
|
|
|
|
public File file() {
|
|
|
|
return new File(dictSource, this.filename);
|
|
|
|
}
|
|
|
|
|
|
|
|
public File fileDisabled() {
|
|
|
|
return new File(dictSource, this.filename + disabledExtension);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* initialize the LibraryProvider as static class. This assigns default paths, and initializes the
|
|
|
|
* dictionary classes Additionally, if default dictionaries are given in the source path, they are
|
|
|
|
* translated into the input format inside the DATA/DICTIONARIES directory
|
|
|
|
*
|
|
|
|
* @param pathToSource
|
|
|
|
* @param pathToDICTIONARIES
|
|
|
|
*/
|
|
|
|
public static void initialize(final File rootPath) {
|
|
|
|
dictSource = new File(rootPath, path_to_source_dictionaries);
|
|
|
|
if ( !dictSource.exists() ) {
|
|
|
|
dictSource.mkdirs();
|
|
|
|
}
|
|
|
|
dictRoot = rootPath;
|
|
|
|
|
|
|
|
// initialize libraries
|
|
|
|
integrateDeReWo();
|
|
|
|
initDidYouMean();
|
|
|
|
integrateOpenGeoDB();
|
|
|
|
integrateGeonames();
|
|
|
|
initAutotagging(tagPrefix);
|
|
|
|
Set<String> allTags = new HashSet<String>() ;
|
|
|
|
allTags.addAll(autotagging.allTags()); // we must copy this into a clone to prevent circularity
|
|
|
|
autotagging.addPlaces(geoLoc);
|
|
|
|
//autotagging.addDictionaries(dymLib.getDictionaries()); // strange results with this: normal word lists are 'too full'
|
|
|
|
WordCache.learn(allTags);
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void integrateOpenGeoDB() {
|
|
|
|
final File geo1 = Dictionary.GEODB1.file();
|
|
|
|
final File geo0 = Dictionary.GEODB0.file();
|
|
|
|
if ( geo1.exists() ) {
|
|
|
|
if ( geo0.exists() ) {
|
|
|
|
geo0.renameTo(Dictionary.GEODB0.fileDisabled());
|
|
|
|
}
|
|
|
|
geoLoc.addLocalization(Dictionary.GEODB1.nickname, new OpenGeoDBLocation(geo1, false));
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
if ( geo0.exists() ) {
|
|
|
|
geoLoc.addLocalization(Dictionary.GEODB0.nickname, new OpenGeoDBLocation(geo0, false));
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void integrateGeonames() {
|
|
|
|
final File geon = Dictionary.GEON0.file();
|
|
|
|
if ( geon.exists() ) {
|
|
|
|
geoLoc.addLocalization(Dictionary.GEON0.nickname, new GeonamesLocation(geon));
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void initDidYouMean() {
|
|
|
|
final File dymDict = new File(dictRoot, path_to_did_you_mean_dictionaries);
|
|
|
|
if ( !dymDict.exists() ) {
|
|
|
|
dymDict.mkdirs();
|
|
|
|
}
|
|
|
|
dymLib = new WordCache(dymDict);
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void initAutotagging(char prefix) {
|
|
|
|
final File autotaggingPath = new File(dictRoot, path_to_autotagging_dictionaries);
|
|
|
|
if ( !autotaggingPath.exists() ) {
|
|
|
|
autotaggingPath.mkdirs();
|
|
|
|
}
|
|
|
|
autotagging = new Autotagging(autotaggingPath, prefix);
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void removeDeReWo() {
|
|
|
|
final File dymDict = new File(dictRoot, path_to_did_you_mean_dictionaries);
|
|
|
|
final File derewoInput = LibraryProvider.Dictionary.DRW0.file();
|
|
|
|
final File derewoOutput = new File(dymDict, derewoInput.getName() + ".words");
|
|
|
|
FileUtils.deletedelete(derewoOutput);
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void integrateDeReWo() {
|
|
|
|
// translate input files (once..)
|
|
|
|
final File dymDict = new File(dictRoot, path_to_did_you_mean_dictionaries);
|
|
|
|
if ( !dymDict.exists() ) {
|
|
|
|
dymDict.mkdirs();
|
|
|
|
}
|
|
|
|
final File derewoInput = LibraryProvider.Dictionary.DRW0.file();
|
|
|
|
final File derewoOutput = new File(dymDict, derewoInput.getName() + ".words");
|
|
|
|
if ( !derewoOutput.exists() && derewoInput.exists() ) {
|
|
|
|
// create the translation of the derewo file (which is easy in this case)
|
|
|
|
final ArrayList<String> derewo = loadDeReWo(derewoInput, true);
|
|
|
|
try {
|
|
|
|
writeWords(derewoOutput, derewo);
|
|
|
|
} catch ( final IOException e ) {
|
|
|
|
Log.logException(e);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
private static ArrayList<String> loadList(final File file, String comment, boolean toLowerCase) {
|
|
|
|
final ArrayList<String> list = new ArrayList<String>();
|
|
|
|
if (!file.exists()) return list;
|
|
|
|
BufferedReader reader = null;
|
|
|
|
try {
|
|
|
|
reader = new BufferedReader(new InputStreamReader(new FileInputStream(file)));
|
|
|
|
String line;
|
|
|
|
while ((line = reader.readLine()) != null) {
|
|
|
|
line = line.trim();
|
|
|
|
if ((line.length() > 0) && (!(line.startsWith(comment)))) {
|
|
|
|
list.add((toLowerCase) ? line.trim().toLowerCase() : line.trim());
|
|
|
|
}
|
|
|
|
}
|
|
|
|
reader.close();
|
|
|
|
} catch (final IOException e) {
|
|
|
|
Log.logException(e);
|
|
|
|
} finally {
|
|
|
|
if (reader != null) try { reader.close(); } catch (final Exception e) {}
|
|
|
|
}
|
|
|
|
return list;
|
|
|
|
}
|
|
|
|
*/
|
|
|
|
|
|
|
|
private static Set<String> sortUnique(final List<String> list) {
|
|
|
|
final Set<String> s = new TreeSet<String>();
|
|
|
|
for ( final String t : list ) {
|
|
|
|
s.add(t);
|
|
|
|
}
|
|
|
|
return s;
|
|
|
|
}
|
|
|
|
|
|
|
|
private static void writeWords(final File f, final ArrayList<String> list) throws IOException {
|
|
|
|
final Set<String> s = sortUnique(list);
|
|
|
|
final PrintWriter w = new PrintWriter(new BufferedWriter(new FileWriter(f)));
|
|
|
|
for ( final String t : s ) {
|
|
|
|
w.println(t);
|
|
|
|
}
|
|
|
|
w.close();
|
|
|
|
}
|
|
|
|
|
|
|
|
private static ArrayList<String> loadDeReWo(final File file, final boolean toLowerCase) {
|
|
|
|
final ArrayList<String> list = new ArrayList<String>();
|
|
|
|
|
|
|
|
// get the zip file entry from the file
|
|
|
|
InputStream derewoTxtEntry;
|
|
|
|
try {
|
|
|
|
final ZipFile zip = new ZipFile(file);
|
|
|
|
/*
|
|
|
|
final Enumeration<? extends ZipEntry> i = zip.entries();
|
|
|
|
while (i.hasMoreElements()) {
|
|
|
|
final ZipEntry e = i.nextElement();
|
|
|
|
System.out.println("loadDeReWo: " + e.getName());
|
|
|
|
}
|
|
|
|
*/
|
|
|
|
derewoTxtEntry = zip.getInputStream(zip.getEntry("derewo-v-100000t-2009-04-30-0.1"));
|
|
|
|
} catch ( final ZipException e ) {
|
|
|
|
Log.logException(e);
|
|
|
|
return list;
|
|
|
|
} catch ( final IOException e ) {
|
|
|
|
Log.logException(e);
|
|
|
|
return list;
|
|
|
|
}
|
|
|
|
|
|
|
|
BufferedReader reader = null;
|
|
|
|
try {
|
|
|
|
reader = new BufferedReader(new InputStreamReader(derewoTxtEntry, "UTF-8"));
|
|
|
|
String line;
|
|
|
|
|
|
|
|
// read until text starts
|
|
|
|
while ( (line = reader.readLine()) != null ) {
|
|
|
|
if ( line.startsWith("# -----") ) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
// read empty line
|
|
|
|
line = reader.readLine();
|
|
|
|
|
|
|
|
// read lines
|
|
|
|
int p;
|
|
|
|
//int c;
|
|
|
|
String w;
|
|
|
|
while ( (line = reader.readLine()) != null ) {
|
|
|
|
line = line.trim();
|
|
|
|
p = line.indexOf(' ', 0);
|
|
|
|
if ( p > 0 ) {
|
|
|
|
//c = Integer.parseInt(line.substring(p + 1));
|
|
|
|
//if (c < 1) continue;
|
|
|
|
w =
|
|
|
|
(toLowerCase) ? line.substring(0, p).trim().toLowerCase() : line
|
|
|
|
.substring(0, p)
|
|
|
|
.trim();
|
|
|
|
if ( w.length() < 4 ) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
list.add(w);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
reader.close();
|
|
|
|
} catch ( final IOException e ) {
|
|
|
|
Log.logException(e);
|
|
|
|
} finally {
|
|
|
|
if ( reader != null ) {
|
|
|
|
try {
|
|
|
|
reader.close();
|
|
|
|
} catch ( final Exception e ) {
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return list;
|
|
|
|
}
|
|
|
|
|
|
|
|
public static void main(final String[] args) {
|
|
|
|
final File here = new File("dummy").getParentFile();
|
|
|
|
initialize(new File(here, "DATA/DICTIONARIES"));
|
|
|
|
System.out.println("dymDict-size = " + dymLib.size());
|
|
|
|
final Set<StringBuilder> r = dymLib.recommend(new StringBuilder("da"));
|
|
|
|
for ( final StringBuilder s : r ) {
|
|
|
|
System.out.println("$ " + s);
|
|
|
|
}
|
|
|
|
System.out.println("recommendations: " + r.size());
|
|
|
|
}
|
|
|
|
}
|