1 // Copyright 2011 Google Inc. All Rights Reserved.
3 // Licensed under the Apache License, Version 2.0 (the "License");
4 // you may not use this file except in compliance with the License.
5 // You may obtain a copy of the License at
7 // http://www.apache.org/licenses/LICENSE-2.0
9 // Unless required by applicable law or agreed to in writing, software
10 // distributed under the License is distributed on an "AS IS" BASIS,
11 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12 // See the License for the specific language governing permissions and
13 // limitations under the License.
15 package com.hughes.android.dictionary.parser;
17 import java.io.BufferedReader;
19 import java.io.FileInputStream;
20 import java.io.IOException;
21 import java.io.InputStreamReader;
22 import java.nio.charset.Charset;
23 import java.util.Arrays;
24 import java.util.Collections;
25 import java.util.LinkedHashSet;
26 import java.util.List;
28 import java.util.logging.Logger;
29 import java.util.regex.Matcher;
30 import java.util.regex.Pattern;
32 import com.hughes.android.dictionary.engine.DictionaryBuilder;
33 import com.hughes.android.dictionary.engine.IndexedEntry;
34 import com.hughes.android.dictionary.engine.EntryTypeName;
35 import com.hughes.android.dictionary.engine.IndexBuilder;
36 import com.hughes.android.dictionary.engine.Language;
37 import com.hughes.android.dictionary.engine.PairEntry;
38 import com.hughes.android.dictionary.engine.PairEntry.Pair;
40 public class DictFileParser {
42 static final Logger logger = Logger.getLogger(DictFileParser.class.getName());
45 public static final Pattern TAB = Pattern.compile("\\t");
48 public static final Pattern DOUBLE_COLON = Pattern.compile(" :: ");
49 public static final Pattern PIPE = Pattern.compile("\\|");
51 static final Pattern SPACES = Pattern.compile("\\s+");
52 // static final Pattern DE_NOUN = Pattern.compile("([^ ]+) *\\{(m|f|n|pl)\\}");
53 // static final Pattern EN_VERB = Pattern.compile("^to ([^ ]+)");
55 static final Pattern BRACKETED = Pattern.compile("\\[([^]]+)\\]");
56 static final Pattern PARENTHESIZED = Pattern.compile("\\(([^)]+)\\)");
57 static final Pattern CURLY_BRACED = Pattern.compile("\\{([^}]+)\\}");
59 static final Pattern NON_CHAR_DASH = Pattern.compile("[^-'\\p{L}0-9]+");
60 public static final Pattern NON_CHAR = Pattern.compile("[^\\p{L}0-9]+");
62 static final Pattern TRIM_PUNC = Pattern.compile("^[^\\p{L}0-9]+|[^\\p{L}0-9]+$");
64 final Charset charset;
65 final boolean flipCols;
67 final Pattern fieldSplit;
68 final Pattern subfieldSplit;
70 final DictionaryBuilder dictBuilder;
71 final IndexBuilder[] langIndexBuilders;
72 final IndexBuilder bothIndexBuilder;
74 // final Set<String> alreadyDone = new HashSet<String>();
76 public DictFileParser(final Charset charset, boolean flipCols,
77 final Pattern fieldSplit, final Pattern subfieldSplit,
78 final DictionaryBuilder dictBuilder, final IndexBuilder[] langIndexBuilders,
79 final IndexBuilder bothIndexBuilder) {
80 this.charset = charset;
81 this.flipCols = flipCols;
82 this.fieldSplit = fieldSplit;
83 this.subfieldSplit = subfieldSplit;
84 this.dictBuilder = dictBuilder;
85 this.langIndexBuilders = langIndexBuilders;
86 this.bothIndexBuilder = bothIndexBuilder;
89 public void parseFile(final File file) throws IOException {
90 final BufferedReader reader = new BufferedReader(new InputStreamReader(new FileInputStream(file), charset));
93 while ((line = reader.readLine()) != null) {
94 if (count % 10000 == 0) {
95 logger.info("count=" + count + ", line=" + line);
102 private void parseLine(final String line) {
103 if (line.startsWith("#") || line.length() == 0) {
104 logger.info("Skipping comment line: " + line);
107 final String[] fields = fieldSplit.split(line);
108 if (fields.length != 2) {
109 logger.warning("Malformed line: " + line);
113 fields[0] = SPACES.matcher(fields[0]).replaceAll(" ").trim();
114 fields[1] = SPACES.matcher(fields[1]).replaceAll(" ").trim();
116 final String temp = fields[0];
117 fields[0] = fields[1];
121 final String[][] subfields = new String[2][];
122 if (subfieldSplit != null) {
123 subfields[0] = subfieldSplit.split(fields[0]);
124 subfields[1] = subfieldSplit.split(fields[1]);
125 if (subfields[0].length != subfields[1].length) {
126 logger.warning("Number of subfields doesn't match: " + line);
130 subfields[0] = new String[] { fields[0] };
131 subfields[1] = new String[] { fields[1] };
134 final PairEntry pairEntry = new PairEntry();
135 for (int i = 0; i < subfields[0].length; ++i) {
136 subfields[0][i] = subfields[0][i].trim();
137 subfields[1][i] = subfields[1][i].trim();
138 pairEntry.pairs.add(new Pair(subfields[0][i], subfields[1][i]));
140 final IndexedEntry entryData = new IndexedEntry(pairEntry);
142 for (int l = 0; l < 2; ++l) {
143 // alreadyDone.clear();
145 for (int j = 0; j < subfields[l].length; ++j) {
146 String subfield = subfields[l][j];
147 final IndexBuilder indexBuilder = langIndexBuilders[l];
148 if (indexBuilder.index.sortLanguage == Language.de) {
149 subfield = parseField_DE(indexBuilder, subfield, entryData, j);
150 } else if (indexBuilder.index.sortLanguage == Language.en) {
151 subfield = parseField_EN(indexBuilder, subfield, entryData, j);
153 parseFieldGeneric(indexBuilder, subfield, entryData, j, subfields[l].length);
158 private void parseFieldGeneric(final IndexBuilder indexBuilder, String field,
159 final IndexedEntry entryData, final int subfieldIdx, final int numSubFields) {
160 // remove bracketed and parenthesized stuff.
161 final StringBuilder bracketed = new StringBuilder();
162 final StringBuilder parenthesized = new StringBuilder();
165 while ((matcher = BRACKETED.matcher(field)).find()) {
166 bracketed.append(matcher.group(1)).append(" ");
167 field = matcher.replaceFirst(" ");
170 while ((matcher = PARENTHESIZED.matcher(field)).find()) {
171 parenthesized.append(matcher.group(1)).append(" ");
172 field = matcher.replaceFirst(" ");
175 field = SPACES.matcher(field).replaceAll(" ").trim();
177 // split words on non -A-z0-9, do them.
178 final String[] tokens = NON_CHAR_DASH.split(field);
180 final EntryTypeName entryTypeName;
181 if (numSubFields == 1) {
182 assert subfieldIdx == 0;
183 if (tokens.length == 1) {
184 entryTypeName = EntryTypeName.ONE_WORD;
185 } else if (tokens.length == 2) {
186 entryTypeName = EntryTypeName.TWO_WORDS;
187 } else if (tokens.length == 3) {
188 entryTypeName = EntryTypeName.THREE_WORDS;
189 } else if (tokens.length == 4) {
190 entryTypeName = EntryTypeName.FOUR_WORDS;
192 entryTypeName = EntryTypeName.FIVE_OR_MORE_WORDS;
195 assert numSubFields > 1;
196 if (subfieldIdx == 0) {
197 if (tokens.length == 1) {
198 entryTypeName = EntryTypeName.MULTIROW_HEAD_ONE_WORD;
200 entryTypeName = EntryTypeName.MULTIROW_HEAD_MANY_WORDS;
203 assert subfieldIdx > 0;
204 if (tokens.length == 1) {
205 entryTypeName = EntryTypeName.MULTIROW_TAIL_ONE_WORD;
207 entryTypeName = EntryTypeName.MULTIROW_TAIL_MANY_WORDS;
212 for (String token : tokens) {
213 token = TRIM_PUNC.matcher(token).replaceAll("");
214 if (/*!alreadyDone.contains(token) && */token.length() > 0) {
215 indexBuilder.addEntryWithTokens(entryData, Collections.singleton(token), entryTypeName);
216 // alreadyDone.add(token);
218 // also split words on dashes, do them, too.
219 if (token.contains("-")) {
220 final String[] dashed = token.split("-");
221 for (final String dashedToken : dashed) {
222 if (/*!alreadyDone.contains(dashedToken) && */dashedToken.length() > 0) {
223 indexBuilder.addEntryWithTokens(entryData, Collections.singleton(dashedToken), EntryTypeName.PART_OF_HYPHENATED);
228 } // if (!alreadyDone.contains(token)) {
229 } // for (final String token : tokens) {
231 // process bracketed stuff (split on spaces and dashes always)
232 final String[] bracketedTokens = NON_CHAR.split(bracketed.toString());
233 for (final String token : bracketedTokens) {
234 assert !token.contains("-");
235 if (/*!alreadyDone.contains(token) && */token.length() > 0) {
236 indexBuilder.addEntryWithTokens(entryData, Collections.singleton(token), EntryTypeName.BRACKETED);
240 // process paren stuff
241 final String[] parenTokens = NON_CHAR.split(parenthesized.toString());
242 for (final String token : parenTokens) {
243 assert !token.contains("-");
244 if (/*!alreadyDone.contains(token) && */token.length() > 0) {
245 indexBuilder.addEntryWithTokens(entryData, Collections.singleton(token), EntryTypeName.PARENTHESIZED);
251 private String parseField_DE(final IndexBuilder indexBuilder, String field,
252 final IndexedEntry entryData, final int subfieldIdx) {
254 // final Matcher matcher = DE_NOUN.matcher(field);
255 // while (matcher.find()) {
256 // final String noun = matcher.group(1);
257 //final String gender = matcher.group(2);
258 // if (alreadyDone.add(noun)) {
259 // System.out.println("Found DE noun " + noun + ", " + gender);
260 // final List<EntryData> entries = indexBuilder.getOrCreateEntries(noun, EntryTypeName.NOUN);
261 // entries.add(entryData);
265 // In English, curly braces are used for different tenses.
266 field = CURLY_BRACED.matcher(field).replaceAll(" ");
271 private String parseField_EN(final IndexBuilder indexBuilder, String field,
272 final IndexedEntry entryData, final int subfieldIdx) {
273 if (field.startsWith("to ")) {
274 field = field.substring(3);
279 public static final Set<String> tokenize(final String text, final Pattern pattern) {
280 final String[] split = pattern.split(text);
281 final Set<String> result = new LinkedHashSet<String>(Arrays.asList(split));