1 | /**
|
---|
2 | * http://opennlp.apache.org/news/model-langdetect-183.html
|
---|
3 | * language detector model: http://opennlp.apache.org/models.html
|
---|
4 | * Use of Apache OpenNLP in general:
|
---|
5 | * http://opennlp.apache.org/docs/1.9.1/manual/opennlp.html#intro.api
|
---|
6 | * Use of OpenNLP for language detection:
|
---|
7 | * http://opennlp.apache.org/docs/1.9.1/manual/opennlp.html#tools.langdetect
|
---|
8 | *
|
---|
9 | * This code was based on the information and sample code at the above links and the links dispersed throughout this file.
|
---|
10 | */
|
---|
11 |
|
---|
12 | import java.io.*;
|
---|
13 | import opennlp.tools.langdetect.*;
|
---|
14 | import opennlp.tools.util.*;
|
---|
15 |
|
---|
16 | /**
|
---|
17 | * EXPORT OPENNLP_HOME environment variable to be your apache OpenNLP installation.
|
---|
18 | * Then, to compile this program:
|
---|
19 | * maori-lang-detection/src$ javac -cp ".:$OPENNLP_HOME/lib/opennlp-tools-1.9.1.jar" MaoriDetector.java
|
---|
20 | * To run this program, one of:
|
---|
21 | *
|
---|
22 | * maori-lang-detection/src$ java -cp ".:$OPENNLP_HOME/lib/*" MaoriDetector --help
|
---|
23 | *
|
---|
24 | * maori-lang-detection/src$ java -cp ".:$OPENNLP_HOME/lib/*" MaoriDetector --file <full/path/to/textfile>
|
---|
25 | *
|
---|
26 | * maori-lang-detection/src$ java -cp ".:$OPENNLP_HOME/lib/*" MaoriDetector -
|
---|
27 | * which expects text to stream in from standard input.
|
---|
28 | * If entering text manually, then remember to press Ctrl-D to indicate the usual end of StdIn.
|
---|
29 | *
|
---|
30 | * https://stackoverflow.com/questions/219585/including-all-the-jars-in-a-directory-within-the-java-classpath
|
---|
31 | * Also has information on how to run this class if it's in a Java package.
|
---|
32 | */
|
---|
33 | public class MaoriDetector {
|
---|
34 | /** The 3 letter language code for Maori in ISO 639-2 or ISO 639-3 */
|
---|
35 | public static final String MAORI_3LETTER_CODE = "mri";
|
---|
36 | public static final double DEFAULT_MINIMUM_CONFIDENCE = 0.50;
|
---|
37 |
|
---|
38 | /** Configurable: cut off minimum confidence value,
|
---|
39 | greater or equal to which determines that the best predicted language is acceptable to user of MaoriDetector. */
|
---|
40 | public final double MINIMUM_CONFIDENCE;
|
---|
41 | /** silentMode set to false means MaoriDetector won't print helpful messages while running. Set to true to run silently. */
|
---|
42 | public final boolean silentMode;
|
---|
43 |
|
---|
44 | /** Language Detection Model file for OpenNLP is expected to be at $OPENNLP_HOME/models/langdetect-183.bin */
|
---|
45 | private final String LANG_DETECT_MODEL_RELATIVE_PATH = "models" + File.separator + "langdetect-183.bin";
|
---|
46 | private LanguageDetector myCategorizer = null;
|
---|
47 |
|
---|
48 | /**
|
---|
49 | * String taken from our university website
|
---|
50 | * https://www.waikato.ac.nz/maori/
|
---|
51 | */
|
---|
52 | public static final String TEST_MRI_INPUT_TEXT = "Ko tÄnei te Whare WÄnanga o Waikato e whakatau nei i ngÄ iwi o te ao, ki roto i te riu o te awa e rere nei, ki runga i te whenua e hora nei, ki raro i te taumaru o ngÄ maunga whakaruru e tau awhi nei.";
|
---|
53 |
|
---|
54 | /** test input string for a negative result */
|
---|
55 | public static final String TEST_ENG_INPUT_TEXT = "The main program exits with -1 if an Exception occurred when attempting to detect the text's language";
|
---|
56 |
|
---|
57 |
|
---|
58 | public MaoriDetector(boolean silentMode) throws Exception {
|
---|
59 | this(silentMode, DEFAULT_MINIMUM_CONFIDENCE);
|
---|
60 | }
|
---|
61 |
|
---|
62 | public MaoriDetector(boolean silentMode, double min_confidence) throws Exception {
|
---|
63 | this.silentMode = silentMode;
|
---|
64 | this.MINIMUM_CONFIDENCE = min_confidence;
|
---|
65 |
|
---|
66 | // 1. Check we can find the Language Detect Model file in the correct location (check that $OPENNLP_HOME/models/langdetect-183.bin exists);
|
---|
67 | String langDetectModelPath = System.getenv("OPENNLP_HOME");
|
---|
68 | if(System.getenv("OPENNLP_HOME") == null) {
|
---|
69 | throw new Exception("\n\t*** Environment variable OPENNLP_HOME must be set to your Apache OpenNLP installation folder.");
|
---|
70 | }
|
---|
71 | langDetectModelPath = langDetectModelPath + File.separator + LANG_DETECT_MODEL_RELATIVE_PATH;
|
---|
72 | File langDetectModelBinFile = new File(langDetectModelPath);
|
---|
73 | if(!langDetectModelBinFile.exists()) {
|
---|
74 | throw new Exception("\n\t*** " + langDetectModelBinFile.getPath() + " doesn't exist."
|
---|
75 | + "\n\t*** Ensure the $OPENNLP_HOME folder contains a 'models' folder with the model file 'langdetect-183.bin' in it.");
|
---|
76 | }
|
---|
77 |
|
---|
78 |
|
---|
79 | // 2. Set up our language detector Model and the Categorizer for language predictions based on the Model.
|
---|
80 | // http://opennlp.apache.org/docs/1.9.1/manual/opennlp.html#intro.api
|
---|
81 | // https://docs.oracle.com/javase/tutorial/essential/exceptions/tryResourceClose.html
|
---|
82 | try (InputStream modelIn = new FileInputStream(langDetectModelPath)) {
|
---|
83 |
|
---|
84 | LanguageDetectorModel model = new LanguageDetectorModel(modelIn);
|
---|
85 |
|
---|
86 | // http://opennlp.apache.org/docs/1.9.1/manual/opennlp.html#tools.langdetect
|
---|
87 | this.myCategorizer = new LanguageDetectorME(model);
|
---|
88 | }/*catch(Exception e) {
|
---|
89 | e.printStackTrace();
|
---|
90 | }*/
|
---|
91 |
|
---|
92 | // instantiating function should handle critical exceptions. Constructors shouldn't.
|
---|
93 | }
|
---|
94 |
|
---|
95 | /**
|
---|
96 | * @return true if the input text is Maori (mri) with MINIMUM_CONFIDENCE levels of confidence (if set,
|
---|
97 | * else DEFAULT_MINIMUM_CONFIDENCE levels of confidence).
|
---|
98 | */
|
---|
99 | public boolean isTextInMaori(String text) {
|
---|
100 | return isTextInLanguage(MAORI_3LETTER_CODE, text);
|
---|
101 | }
|
---|
102 |
|
---|
103 | /** @param langCode is 3 letter language code, ISO 639-2/3
|
---|
104 | * https://www.loc.gov/standards/iso639-2/php/code_list.php
|
---|
105 | * https://en.wikipedia.org/wiki/ISO_639-3
|
---|
106 | * @return true if the input text is Maori (mri) with MINIMUM_CONFIDENCE levels of confidence (if set,
|
---|
107 | * else DEFAULT_MINIMUM_CONFIDENCE levels of confidence).
|
---|
108 | */
|
---|
109 | public boolean isTextInLanguage(String langCode, String text) {
|
---|
110 | // Get the most probable language
|
---|
111 | Language bestLanguage = myCategorizer.predictLanguage(text);
|
---|
112 | System.out.println("Best language: " + bestLanguage.getLang());
|
---|
113 | System.out.println("Best language confidence: " + bestLanguage.getConfidence());
|
---|
114 |
|
---|
115 | return (bestLanguage.getLang().equals(langCode) && bestLanguage.getConfidence() >= this.MINIMUM_CONFIDENCE);
|
---|
116 | }
|
---|
117 |
|
---|
118 |
|
---|
119 | /**
|
---|
120 | * Handle "smaller" textfiles/streams of text read in.
|
---|
121 | * Return value is the same as for isTextInMaori(String text);
|
---|
122 | */
|
---|
123 | public boolean isTextInMaori(BufferedReader reader) throws Exception {
|
---|
124 | return isTextInLanguage(MAORI_3LETTER_CODE, reader);
|
---|
125 | }
|
---|
126 | /**
|
---|
127 | * Handle "smaller" textfiles/streams of text read in.
|
---|
128 | * Return value is the same as for isTextInLanguage(String langCode, String text);
|
---|
129 | */
|
---|
130 | public boolean isTextInLanguage(String langCode, BufferedReader reader) throws Exception {
|
---|
131 | // https://stackoverflow.com/questions/326390/how-do-i-create-a-java-string-from-the-contents-of-a-file
|
---|
132 |
|
---|
133 | StringBuilder text = new StringBuilder();
|
---|
134 | String line = null;
|
---|
135 |
|
---|
136 |
|
---|
137 | while((line = reader.readLine()) != null) { // readLine removes newline separator
|
---|
138 | text.append(line + "\n"); // add back (unix style) line ending
|
---|
139 | }
|
---|
140 | return isTextInLanguage(langCode, text.toString());
|
---|
141 | }
|
---|
142 |
|
---|
143 | /*
|
---|
144 | * Need better handling of "larger" textfiles/streams of text read in:
|
---|
145 | * what if multiple languages with high confidence every NUM_LINES read in?
|
---|
146 | * Does this mean the file is multi-lingual with each section dominated by a different language?
|
---|
147 | * How best to convey such information to the user?
|
---|
148 | */
|
---|
149 | /**
|
---|
150 | * Rudimentary attempt to deal with very large files.
|
---|
151 | * Return value is the same as for isTextInMaori(String text);
|
---|
152 | */
|
---|
153 | public boolean isLargeTextInMaori(BufferedReader reader) throws Exception {
|
---|
154 | return isLargeTextInLanguage(MAORI_3LETTER_CODE, reader);
|
---|
155 | }
|
---|
156 |
|
---|
157 | /**
|
---|
158 | * Rudimentary attempt to deal with very large files.
|
---|
159 | * Return value is the same as for isTextInLanguage(String langCode, String text);
|
---|
160 | */
|
---|
161 | public boolean isLargeTextInLanguage(String langCode, BufferedReader reader) throws Exception {
|
---|
162 | // https://stackoverflow.com/questions/326390/how-do-i-create-a-java-string-from-the-contents-of-a-file
|
---|
163 |
|
---|
164 | final int NUM_LINES = 100; // arbitrary 100 lines read, predict language, calculate confidence
|
---|
165 |
|
---|
166 | StringBuilder text = new StringBuilder();
|
---|
167 | String line = null;
|
---|
168 |
|
---|
169 | double cumulativeConfidence = 0;
|
---|
170 | int numLoops = 0;
|
---|
171 |
|
---|
172 | int i = 0;
|
---|
173 | String language = null;
|
---|
174 |
|
---|
175 | while((line = reader.readLine()) != null) { // readLine removes newline separator
|
---|
176 | text.append(line + "\n"); // add back (unix style) line ending
|
---|
177 |
|
---|
178 | i++; // read nth line of numLoop
|
---|
179 |
|
---|
180 |
|
---|
181 | if(i == NUM_LINES) { // arbitrary 100 lines read, predict language, calculate confidence
|
---|
182 |
|
---|
183 |
|
---|
184 | Language bestLanguage = myCategorizer.predictLanguage(text.toString());
|
---|
185 | if(language != null && !bestLanguage.getLang().equals(language)) { // predicted lang of current n lines not the same as predicted lang for prev n lines
|
---|
186 | System.err.println("**** WARNING: text seems to contain content in multiple languages or unable to consistently predict the same language.");
|
---|
187 | }
|
---|
188 | language = bestLanguage.getLang();
|
---|
189 | cumulativeConfidence += bestLanguage.getConfidence();
|
---|
190 |
|
---|
191 | System.err.println("Best predicted language for last " + NUM_LINES + " lines: " + language + "(confidence: " + bestLanguage.getConfidence() + ")");
|
---|
192 |
|
---|
193 | // finished analysing language of NUM_LINES of text
|
---|
194 | text = new StringBuilder();
|
---|
195 | i = 0;
|
---|
196 | numLoops++;
|
---|
197 | }
|
---|
198 | }
|
---|
199 |
|
---|
200 | // process any (remaining) text that was less than n NUM_LINES
|
---|
201 | if(!text.toString().equals("")) {
|
---|
202 | text.append(line + "\n"); // add back (unix style) line ending
|
---|
203 | i++;
|
---|
204 |
|
---|
205 | Language bestLanguage = myCategorizer.predictLanguage(text.toString());
|
---|
206 |
|
---|
207 | if(language != null && !bestLanguage.getLang().equals(language)) { // predicted lang of current n lines not the same as predicted lang for prev n lines
|
---|
208 | System.err.println("**** WARNING: text seems to contain content in multiple languages or unable to consistently predict the same language.");
|
---|
209 | }
|
---|
210 | language = bestLanguage.getLang();
|
---|
211 | cumulativeConfidence += bestLanguage.getConfidence();
|
---|
212 | System.err.println("Best predicted language for final " + NUM_LINES + " lines: " + language + "(confidence: " + bestLanguage.getConfidence() + ")");
|
---|
213 | }
|
---|
214 |
|
---|
215 |
|
---|
216 | int totalLinesRead = numLoops * NUM_LINES + i; // not used
|
---|
217 | double avgConfidence = cumulativeConfidence/(numLoops + 1); // not quite the average as the text processed outside the loop may have fewer lines than NUM_LINES
|
---|
218 |
|
---|
219 |
|
---|
220 | return (language.equals(langCode) && avgConfidence >= this.MINIMUM_CONFIDENCE);
|
---|
221 | }
|
---|
222 |
|
---|
223 |
|
---|
224 |
|
---|
225 | /**
|
---|
226 | * Prints to STDOUT the predicted languages of the input text in order of descending confidence.
|
---|
227 | * Unused.
|
---|
228 | */
|
---|
229 | public void predictedLanguages(String text) {
|
---|
230 | // Get an array with the most probable languages
|
---|
231 |
|
---|
232 | Language[] languages = myCategorizer.predictLanguages(text);
|
---|
233 |
|
---|
234 | if(languages == null || languages.length <= 0) {
|
---|
235 | System.err.println("No languages predicted for the input text");
|
---|
236 | } else {
|
---|
237 | for(int i = 0; i < languages.length; i++) {
|
---|
238 | System.out.println("Language prediction " + i + ": " + languages[i]);
|
---|
239 | }
|
---|
240 | }
|
---|
241 |
|
---|
242 | }
|
---|
243 |
|
---|
244 | public static void printUsage() {
|
---|
245 | System.err.println("Run this program with:");
|
---|
246 | System.err.println("\t--help (-h)\tfor seeing this usage message again");
|
---|
247 | System.err.println("\t-\tto have input text read from STDIN (as always, hit Ctrl-D to mark end of text stream)");
|
---|
248 | System.err.println("\t--file (-f)\tto provide an input file path");
|
---|
249 | System.err.println("\t--silent (-s): optional, to run silently and just exit with exit value. [not yet implemented]");
|
---|
250 | System.err.println("\t--min-confidence (-c): optional, to override the default minimum confidence value (" + DEFAULT_MINIMUM_CONFIDENCE + ")");
|
---|
251 | System.err.println("\t\tof the predicted language that will be considered acceptable.");
|
---|
252 | System.err.println();
|
---|
253 | System.err.println("This program terminates with exit value:");
|
---|
254 | System.err.println("\t0 if the input text is in Maori");
|
---|
255 | System.err.println("\t1 if input text is not in Maori");
|
---|
256 | System.err.println();
|
---|
257 | System.err.println("\t-1 if the input arguments were wrong");
|
---|
258 | System.err.println("\t255(!) if an Exception occurred in instantiating the MaoriDetector when attempting to detect the text's language");
|
---|
259 | System.err.println("\t2 if the user asked to run this program with --help/-h.");
|
---|
260 | System.err.println();
|
---|
261 | }
|
---|
262 |
|
---|
263 | /**
|
---|
264 | * The main program exits with:
|
---|
265 | * 0 if text is in Maori;
|
---|
266 | * 1 if text is not in Maori;
|
---|
267 | *
|
---|
268 | * -1 if the input arguments were wrong
|
---|
269 | * 255(!) if an Exception occurred in instantiating the MaoriDetector when attempting to detect the text's language
|
---|
270 | * QTODO: why does the program exit value end up as 255 and not -1 when returnVal remains at -1 on Exception?
|
---|
271 | * 2 if the user asked to run this program with --help/-h.
|
---|
272 | */
|
---|
273 | public static void main(String args[]) {
|
---|
274 | int returnVal = -1;
|
---|
275 |
|
---|
276 | // 1. Check input arguments
|
---|
277 | boolean printUsage = false;
|
---|
278 | boolean readFromStdIn = false;
|
---|
279 | File inFile = null;
|
---|
280 | boolean runSilent = false;
|
---|
281 | double minConfidence = -1;
|
---|
282 |
|
---|
283 | for (int i = 0; !printUsage && i < args.length; i++) {
|
---|
284 |
|
---|
285 | // check for help first and quit after printing usage
|
---|
286 | if(args[i].equals("--help") || args[i].equals("-h")) {
|
---|
287 | printUsage = true;
|
---|
288 | returnVal = 2;
|
---|
289 | } else if(args[i].equals("--silent") || args[i].equals("-s")) {
|
---|
290 | runSilent = true;
|
---|
291 | } else if(args[i].equals("--min-confidence") || args[i].equals("-c")) {
|
---|
292 | i++;
|
---|
293 | if(i >= args.length) {
|
---|
294 | System.err.println("ERROR: No minimum confidence value provided with --min-confidence|-c flag.\n");
|
---|
295 | printUsage = true;
|
---|
296 | returnVal = -1;
|
---|
297 | } else {
|
---|
298 | try {
|
---|
299 | minConfidence = Double.parseDouble(args[i]);
|
---|
300 | if(minConfidence < 0 || minConfidence > 1) {
|
---|
301 | throw new NumberFormatException("Number out of range, must be between 0-1");
|
---|
302 | }
|
---|
303 | } catch(NumberFormatException nfe) {
|
---|
304 | System.err.println("ERROR: value for min-confidence is the wrong format or out of range. It must be a (decimal point) number between 0-1.\n");
|
---|
305 | printUsage = true;
|
---|
306 | returnVal = -1;
|
---|
307 | }
|
---|
308 | }
|
---|
309 | } else if(args[i].equals("-")) {
|
---|
310 | readFromStdIn = true;
|
---|
311 | //break; // don't bother continuing to check input arguments for any --file flag if we're told to read from stdin
|
---|
312 | } else if(args[i].equals("--file") || args[i].equals("-f")) {
|
---|
313 | i++;
|
---|
314 | if(i >= args.length) {
|
---|
315 | System.err.println("ERROR: No input file provided with --file|-f flag.\n");
|
---|
316 | printUsage = true;
|
---|
317 | returnVal = -1;
|
---|
318 | } else {
|
---|
319 | String filePath = args[i];
|
---|
320 | inFile = new File(filePath);
|
---|
321 | if(!inFile.isFile()) {
|
---|
322 | System.err.println("ERROR: Can't read text. Input file argument provided does not exist or is not a file.\n");
|
---|
323 | printUsage = true;
|
---|
324 | returnVal = -1;
|
---|
325 | }
|
---|
326 | }
|
---|
327 | } else { // unrecognised input argument
|
---|
328 | System.err.println("ERROR: Unrecognised " + i + "th argument to this program.\n");
|
---|
329 | printUsage = true;
|
---|
330 | returnVal = -1;
|
---|
331 | }
|
---|
332 | }
|
---|
333 |
|
---|
334 | if(!readFromStdIn && inFile == null) { // at least one input source must be provided
|
---|
335 | System.err.println("ERROR: must specify source to read text from, either STDIN (-) or input file (--file <file>).\n");
|
---|
336 | printUsage = true;
|
---|
337 | returnVal = -1;
|
---|
338 | }
|
---|
339 |
|
---|
340 | if(readFromStdIn && inFile != null) { // this program can't be asked to read from stdin and from an input file
|
---|
341 | System.err.println("ERROR: instructed to read from both STDIN and from an input file. Not possible.\n");
|
---|
342 | printUsage = true;
|
---|
343 | returnVal = -1;
|
---|
344 | }
|
---|
345 |
|
---|
346 | if(printUsage) {
|
---|
347 | if(!runSilent || returnVal == 2) { // if expressly asked for help or not running silent
|
---|
348 | printUsage();
|
---|
349 | }
|
---|
350 | System.exit(returnVal);
|
---|
351 | }
|
---|
352 |
|
---|
353 | try {
|
---|
354 | MaoriDetector maoriTextDetector = null;
|
---|
355 | if(minConfidence == -1) {
|
---|
356 | maoriTextDetector = new MaoriDetector(runSilent);
|
---|
357 | } else {
|
---|
358 | maoriTextDetector = new MaoriDetector(runSilent, minConfidence);
|
---|
359 | }
|
---|
360 |
|
---|
361 | //boolean textIsInMaori = maoriTextDetector.isTextInMaori(TEST_MRI_INPUT_TEXT);
|
---|
362 | boolean textIsInMaori = false;
|
---|
363 |
|
---|
364 | // Using try with resources, https://docs.oracle.com/javase/tutorial/essential/exceptions/tryResourceClose.html
|
---|
365 | if(inFile != null) {
|
---|
366 | System.err.println("Reading text from file " + inFile.getPath());
|
---|
367 | try (BufferedReader reader = new BufferedReader(new FileReader(inFile))) {
|
---|
368 | textIsInMaori = maoriTextDetector.isTextInMaori(reader);
|
---|
369 | } // let outer try deal with any file/reading exceptions
|
---|
370 | }
|
---|
371 | else if (readFromStdIn) {
|
---|
372 | System.err.println("Waiting to read text from STDIN... (press Ctrl-D when done entering text)>");
|
---|
373 | try (BufferedReader reader = new BufferedReader(new InputStreamReader(System.in))) {
|
---|
374 | textIsInMaori = maoriTextDetector.isTextInMaori(reader);
|
---|
375 | } // let outer try deal with any file/reading exceptions
|
---|
376 | }
|
---|
377 |
|
---|
378 | if(textIsInMaori) {
|
---|
379 | returnVal = 0;
|
---|
380 | } else {
|
---|
381 | returnVal = 1;
|
---|
382 | }
|
---|
383 |
|
---|
384 | } catch(Exception e) {
|
---|
385 | e.printStackTrace();
|
---|
386 |
|
---|
387 | } finally {
|
---|
388 | System.err.println("Exitting program with returnVal " + returnVal + "...\n");
|
---|
389 | System.exit(returnVal);
|
---|
390 | }
|
---|
391 | }
|
---|
392 |
|
---|
393 | // test hardcoded string
|
---|
394 | public static void oldMain(String args[]) {
|
---|
395 | int returnVal = -1;
|
---|
396 | boolean silentMode = false;
|
---|
397 |
|
---|
398 | try {
|
---|
399 | MaoriDetector maoriTextDetector = new MaoriDetector(silentMode);
|
---|
400 |
|
---|
401 | boolean textIsInMaori = maoriTextDetector.isTextInMaori(TEST_MRI_INPUT_TEXT);
|
---|
402 | if(textIsInMaori) {
|
---|
403 | returnVal = 0;
|
---|
404 | } else {
|
---|
405 | returnVal = 1;
|
---|
406 | }
|
---|
407 |
|
---|
408 | } catch(Exception e) {
|
---|
409 | e.printStackTrace();
|
---|
410 | } finally {
|
---|
411 | System.err.println("Exitting program with returnVal " + returnVal + "...\n");
|
---|
412 | System.exit(returnVal);
|
---|
413 | }
|
---|
414 | }
|
---|
415 |
|
---|
416 |
|
---|
417 | }
|
---|