[26186] | 1 | /**
|
---|
| 2 | * Adding support for Images in Terrier
|
---|
| 3 | * @author: John Thompson, jmt12, #9826509
|
---|
| 4 | *
|
---|
| 5 | * The contents of this file are subject to the Mozilla Public License
|
---|
| 6 | * Version 1.1 (the "License"); you may not use this file except in
|
---|
| 7 | * compliance with the License. You may obtain a copy of the License at
|
---|
| 8 | * http://www.mozilla.org/MPL/
|
---|
| 9 | *
|
---|
| 10 | * Software distributed under the License is distributed on an "AS IS"
|
---|
| 11 | * basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See
|
---|
| 12 | * the License for the specific language governing rights and limitations
|
---|
| 13 | * under the License.
|
---|
| 14 | *
|
---|
| 15 | * Copyright (c) 2011 The University of Waikato. All Rights Reserved.
|
---|
| 16 | */
|
---|
| 17 | package org.terrier.indexing;
|
---|
| 18 |
|
---|
| 19 | import java.io.InputStream;
|
---|
| 20 | import java.io.StringReader;
|
---|
| 21 | import java.io.Reader;
|
---|
[26214] | 22 | import java.nio.charset.Charset;
|
---|
[26186] | 23 | import java.nio.file.Files;
|
---|
| 24 | import java.nio.file.Path;
|
---|
| 25 | import java.nio.file.Paths;
|
---|
[26214] | 26 | import java.security.MessageDigest;
|
---|
| 27 | import java.security.NoSuchAlgorithmException;
|
---|
[26186] | 28 | import java.util.Collections;
|
---|
| 29 | import java.util.Arrays;
|
---|
| 30 | import java.util.Map;
|
---|
| 31 | import java.util.Set;
|
---|
| 32 |
|
---|
| 33 | import org.apache.log4j.Logger;
|
---|
[26190] | 34 | import org.terrier.indexing.StreamGobbler;
|
---|
[26186] | 35 | import org.terrier.indexing.tokenisation.TokenStream;
|
---|
| 36 | import org.terrier.indexing.tokenisation.Tokeniser;
|
---|
| 37 | import org.terrier.utility.ApplicationSetup;
|
---|
| 38 |
|
---|
| 39 | public class ImageDocument
|
---|
| 40 | implements Document
|
---|
| 41 | {
|
---|
| 42 | /** A reference to the logger for messaging */
|
---|
| 43 | protected static final Logger logger = Logger.getLogger(FileDocument.class);
|
---|
| 44 | /** The map of properties (fields) for this document. */
|
---|
| 45 | protected Map<String,String> properties;
|
---|
| 46 | /** A reader built from a dummy text string. */
|
---|
| 47 | protected Reader reader;
|
---|
| 48 | /** A token stream produced by the configured tokeniser when feed the dummy
|
---|
| 49 | * reader.
|
---|
| 50 | */
|
---|
| 51 | protected TokenStream tokenizer;
|
---|
| 52 |
|
---|
| 53 | /** The preview filetype. **/
|
---|
| 54 | protected final String preview_format = ApplicationSetup.getProperty("ImageDocument.preview_format", "jpg");
|
---|
| 55 | /** The preview size (width). **/
|
---|
| 56 | protected final String preview_width = ApplicationSetup.getProperty("ImageDocument.preview_width", "200");
|
---|
| 57 |
|
---|
| 58 | /** Default constructor. **/
|
---|
| 59 | protected ImageDocument() {}
|
---|
| 60 |
|
---|
| 61 | /** Constructs an instance of the ImageDocument from the given input stream.
|
---|
| 62 | * @param docStream the input stream that reads the file.
|
---|
| 63 | * @param docProperties the initial properties (docno, filename)
|
---|
| 64 | * @param tok the tokeniser defined for this collection
|
---|
| 65 | */
|
---|
| 66 | public ImageDocument(InputStream istream, Map<String,String> default_properties, Tokeniser tok)
|
---|
| 67 | {
|
---|
| 68 | logger.info("ImageDocument::ImageDocument()");
|
---|
| 69 | // Initialization from arguments
|
---|
| 70 | this.properties = default_properties;
|
---|
| 71 |
|
---|
| 72 | // Set properties
|
---|
[26208] | 73 | logger.info("ImageDocument - current properties");
|
---|
| 74 | for (Map.Entry<String, String> entry : this.properties.entrySet())
|
---|
| 75 | {
|
---|
| 76 | logger.info(entry.getKey() + "=" + entry.getValue());
|
---|
| 77 | }
|
---|
| 78 |
|
---|
[26186] | 79 | logger.info("ImageDocument - extracting properties");
|
---|
| 80 | // A. Hardcoded properties
|
---|
| 81 | this.properties.put("parser", "ImageDocument");
|
---|
| 82 | this.properties.put("abstract", "This is an image so here is some dummy text to prevent indexer failing.");
|
---|
| 83 | // B. Properties derived from filename
|
---|
[26214] | 84 | // - A simple title for the document
|
---|
| 85 | String filepath = this.properties.get("filename");
|
---|
| 86 | String title = filepath.substring(filepath.lastIndexOf(System.getProperty("file.separator")) + 1);
|
---|
[26186] | 87 | this.properties.put("title", title);
|
---|
[26214] | 88 | String ext = filepath.substring(filepath.lastIndexOf(".") + 1);
|
---|
| 89 | // - The name of the copy of the original document
|
---|
[26186] | 90 | String target_filename = "doc." + ext;
|
---|
| 91 | this.properties.put("source","doc." + ext);
|
---|
[26214] | 92 | // - A unique associated directory. This gets a little tricky as we need
|
---|
| 93 | // to create the directory at the same time if an effort to promote
|
---|
| 94 | // synchronous behaviour
|
---|
| 95 | String unique_id = this.generateHash(filepath);
|
---|
| 96 | // - we start with the first 4 characters
|
---|
| 97 | int offset = 0;
|
---|
| 98 | String assoc_filename = "D" + unique_id.substring(offset, offset + 4);
|
---|
| 99 | // - we add ".dir" as a suffix to the directory that actually contains
|
---|
| 100 | // files (so the non-suffixed version contains nested directories)
|
---|
| 101 | Path assoc_path = Paths.get(ApplicationSetup.TERRIER_SHARE, "images", "assoc", assoc_filename + ".dir");
|
---|
| 102 | // - then we continue adding blocks of 4 characters until we get a
|
---|
| 103 | // directory that doesn't already exist
|
---|
| 104 | while (assoc_path.toFile().exists() && offset < unique_id.length())
|
---|
[26207] | 105 | {
|
---|
[26214] | 106 | offset += 4;
|
---|
| 107 | assoc_filename += System.getProperty("file.separator") + unique_id.substring(offset, offset + 4);
|
---|
| 108 | assoc_path = Paths.get(ApplicationSetup.TERRIER_SHARE, "images", "assoc", assoc_filename + ".dir");
|
---|
| 109 | }
|
---|
| 110 | // - still not unique? but run out of unique_id... time to complain
|
---|
| 111 | if (assoc_path.toFile().exists())
|
---|
| 112 | {
|
---|
| 113 | logger.error("ImageDoument - can't determine unique assocfilepath");
|
---|
[26207] | 114 | System.exit(0);
|
---|
| 115 | }
|
---|
[26214] | 116 | // - create the directories quick... hopefully before someone else does
|
---|
| 117 | assoc_path.toFile().mkdirs(); // bet there is a nice nio way to do this
|
---|
[26186] | 118 | this.properties.put("assocfile", assoc_filename);
|
---|
| 119 |
|
---|
| 120 | // Copy (symlink) the file into place in the shared directory
|
---|
| 121 | Path source_path = Paths.get(properties.get("filename"));
|
---|
| 122 | Path target_path = assoc_path.resolve(target_filename);
|
---|
| 123 | if (target_path.toFile().exists())
|
---|
| 124 | {
|
---|
| 125 | logger.info("ImageDocument - removing existing (old) associated image");
|
---|
| 126 | try
|
---|
| 127 | {
|
---|
| 128 | Files.delete(target_path);
|
---|
| 129 | }
|
---|
| 130 | catch (Exception e)
|
---|
| 131 | {
|
---|
| 132 | logger.error("Exception while deleting old image: ", e);
|
---|
| 133 | }
|
---|
| 134 | }
|
---|
| 135 | logger.info("ImageDocument - symlinking image into assoc directory");
|
---|
| 136 | try
|
---|
| 137 | {
|
---|
| 138 | Files.createSymbolicLink(target_path, source_path);
|
---|
| 139 | }
|
---|
| 140 | // not supported? We'll try copying below
|
---|
| 141 | catch (UnsupportedOperationException ex)
|
---|
| 142 | {
|
---|
| 143 | }
|
---|
| 144 | // All other exceptions can be fatal
|
---|
| 145 | catch (Exception e)
|
---|
| 146 | {
|
---|
| 147 | logger.error("Exception while symlinking image: ", e);
|
---|
| 148 | }
|
---|
| 149 | // - copy if the file doesn't exist yet
|
---|
| 150 | if (!target_path.toFile().exists())
|
---|
| 151 | {
|
---|
| 152 | logger.info("ImageDocument - symlink filaed, copying instead");
|
---|
| 153 | try
|
---|
| 154 | {
|
---|
| 155 | Files.copy(source_path, target_path);
|
---|
| 156 | }
|
---|
| 157 | // Fatality!
|
---|
| 158 | catch (Exception e)
|
---|
| 159 | {
|
---|
| 160 | logger.error("Exception while copying image: ", e);
|
---|
| 161 | }
|
---|
| 162 | }
|
---|
| 163 |
|
---|
| 164 | // Generate preview image
|
---|
| 165 | logger.info("ImageDocument - generate preview image");
|
---|
| 166 | try
|
---|
| 167 | {
|
---|
| 168 | String preview_filename = this.generatePreview(source_path, assoc_path);
|
---|
| 169 | this.properties.put("preview",preview_filename);
|
---|
| 170 | }
|
---|
| 171 | catch (Exception e)
|
---|
| 172 | {
|
---|
| 173 | logger.error("Exception while generating preview image: ", e);
|
---|
| 174 | }
|
---|
| 175 |
|
---|
| 176 | // Create a dummy reader around some dummy text and then tokenize it
|
---|
| 177 | logger.info("ImageDocument - feed dummy text as token stream to indexer");
|
---|
| 178 | try
|
---|
| 179 | {
|
---|
| 180 | this.reader = new StringReader(this.properties.get("abstract"));
|
---|
| 181 | this.tokenizer = tok.tokenise(this.reader);
|
---|
| 182 | }
|
---|
| 183 | catch (Exception e)
|
---|
| 184 | {
|
---|
| 185 | logger.error("Exception while creating dummy text stream: ", e);
|
---|
| 186 | }
|
---|
[29648] | 187 |
|
---|
| 188 | // Use OpenSIFT to generate a featureset (in Oxford format) for this image
|
---|
| 189 | logger.info("ImageDocument - generate and record SIFT features");
|
---|
| 190 | try
|
---|
| 191 | {
|
---|
| 192 | String sift_command[] = {
|
---|
| 193 | "siftfeat",
|
---|
| 194 | "-x",
|
---|
| 195 | source_path.toString()
|
---|
| 196 | };
|
---|
| 197 | logger.info("ImageDocument - sift command: " + Arrays.toString(sift_command));
|
---|
| 198 | Process sift_process = Runtime.getRuntime().exec(sift_command);
|
---|
| 199 | // we'd usually send STDERR to /dev/null, but a streamgobbler is easier
|
---|
| 200 | // in Java
|
---|
| 201 | StreamGobbler sift_process_error_gobbler = new StreamGobbler(sift_process.getErrorStream());
|
---|
| 202 | sift_process_error_gobbler.start();
|
---|
| 203 | // the SIFT features, in Oxford format, will arrive from STDOUT
|
---|
| 204 | BufferedReader sift_br = new BufferedReader(new InputStreamReader(sift_process.getInputStream()));
|
---|
| 205 | String line;
|
---|
| 206 | StringBuffer oxford_features;
|
---|
| 207 | while ((line = sift_br.readLine()) != null)
|
---|
| 208 | {
|
---|
| 209 | oxford_features.append(line);
|
---|
| 210 | }
|
---|
| 211 | // this command blocks until process completes (emit return value) which
|
---|
| 212 | // should be shortly after it emits the last line of SIFT feature data
|
---|
| 213 | int sift_status = sift_process.waitFor();
|
---|
| 214 | this.properties.put("sift", oxford_features.toString());
|
---|
| 215 | }
|
---|
| 216 | catch (Exception e)
|
---|
| 217 | {
|
---|
| 218 | logger.error("Exception while generating preview image: ", e);
|
---|
| 219 | }
|
---|
| 220 |
|
---|
[26186] | 221 | logger.info("ImageDocument - Complete!");
|
---|
| 222 | }
|
---|
| 223 | /** ImageDocument(InputStream istream, Map<String,String> default_properties, Tokeniser tok) **/
|
---|
| 224 |
|
---|
| 225 | /** Returns true when the end of the document has been reached, and there
|
---|
| 226 | * are no other terms to be retrieved from it.
|
---|
| 227 | * @return boolean true if there are no more terms in the document, otherwise
|
---|
| 228 | * it returns false.
|
---|
| 229 | */
|
---|
| 230 | public boolean endOfDocument()
|
---|
| 231 | {
|
---|
| 232 | return !this.tokenizer.hasNext();
|
---|
| 233 | }
|
---|
| 234 | /** endOfDocument() **/
|
---|
| 235 |
|
---|
| 236 | /** Use ImageMagick to generate a preview image.
|
---|
| 237 | * @pre assumes you have ImageMagick installed and available on Path
|
---|
| 238 | * @pre uses member variables preview_format and preview_width
|
---|
| 239 | * @return the filename of the preview image (within the assoc directory)
|
---|
| 240 | */
|
---|
| 241 | private String generatePreview(Path source_path, Path assoc_path)
|
---|
| 242 | throws Exception
|
---|
| 243 | {
|
---|
| 244 | String preview_filename = "preview." + this.preview_format;
|
---|
| 245 | Path preview_path = assoc_path.resolve(preview_filename);
|
---|
| 246 | String convert_command[] = {
|
---|
| 247 | "convert",
|
---|
| 248 | source_path.toString(),
|
---|
| 249 | "-resize",
|
---|
| 250 | this.preview_width + "x",
|
---|
| 251 | preview_path.toString()
|
---|
| 252 | };
|
---|
| 253 | logger.info("ImageDocument - convert command: " + Arrays.toString(convert_command));
|
---|
| 254 | Process convert_process = Runtime.getRuntime().exec(convert_command);
|
---|
| 255 | // Gobble up the streams to prevent them hanging the process when buffers
|
---|
| 256 | // are full
|
---|
| 257 | StreamGobbler convert_process_error_gobbler = new StreamGobbler(convert_process.getErrorStream());
|
---|
| 258 | convert_process_error_gobbler.start();
|
---|
| 259 | StreamGobbler convert_process_input_gobbler = new StreamGobbler(convert_process.getInputStream());
|
---|
| 260 | convert_process_input_gobbler.start();
|
---|
| 261 | // Let the conversion finish
|
---|
| 262 | int convert_status = convert_process.waitFor();
|
---|
| 263 | if (convert_status != 0 || !preview_path.toFile().exists())
|
---|
| 264 | {
|
---|
| 265 | throw new Exception("Convert command failed (exit status: " + convert_status + ")");
|
---|
| 266 | }
|
---|
| 267 | return preview_filename;
|
---|
| 268 | }
|
---|
| 269 | /** generatePreview(Path, Path) **/
|
---|
| 270 |
|
---|
| 271 | /** Returns the underlying map of all the properties defined by this Document.
|
---|
| 272 | * @since 1.1.0
|
---|
| 273 | */
|
---|
| 274 | public Map<String,String> getAllProperties()
|
---|
| 275 | {
|
---|
| 276 | return this.properties;
|
---|
| 277 | }
|
---|
| 278 | /** getAllProperties() **/
|
---|
| 279 |
|
---|
| 280 | /** Returns a list of the fields the current term appears in.
|
---|
| 281 | * @return HashSet a set of the terms that the current term appears in.
|
---|
| 282 | */
|
---|
| 283 | public Set<String> getFields()
|
---|
| 284 | {
|
---|
| 285 | // Returns null because there is no support for fields with file documents.
|
---|
| 286 | return Collections.emptySet();
|
---|
| 287 | }
|
---|
| 288 | /** getFields() **/
|
---|
| 289 |
|
---|
| 290 | /** Gets the next term of the document.
|
---|
| 291 | * <B>NB:</B>Null string returned from getNextTerm() should
|
---|
| 292 | * be ignored. They do not signify the lack of any more terms.
|
---|
| 293 | * endOfDocument() should be used to check that.
|
---|
| 294 | * @return String the next term of the document. Null returns should be
|
---|
| 295 | * ignored.
|
---|
| 296 | */
|
---|
| 297 | public String getNextTerm()
|
---|
| 298 | {
|
---|
| 299 | return this.tokenizer.next();
|
---|
| 300 | }
|
---|
| 301 | /** getNextTerm() **/
|
---|
| 302 |
|
---|
| 303 | /** Allows access to a named property of the Document. Examples might be URL,
|
---|
| 304 | * filename etc.
|
---|
| 305 | * @param name Name of the property. It is suggested, but not required that
|
---|
| 306 | * this name should not be case insensitive.
|
---|
| 307 | * @since 1.1.0
|
---|
| 308 | */
|
---|
| 309 | public String getProperty(String name)
|
---|
| 310 | {
|
---|
| 311 | return this.properties.get(name.toLowerCase());
|
---|
| 312 | }
|
---|
| 313 | /** getProperty(String name) **/
|
---|
| 314 |
|
---|
| 315 | /** Returns a Reader object so client code can tokenise the document
|
---|
| 316 | * or deal with the document itself. Examples might be extracting URLs,
|
---|
| 317 | * language detection. */
|
---|
| 318 | public Reader getReader()
|
---|
| 319 | {
|
---|
| 320 | return this.reader;
|
---|
| 321 | }
|
---|
| 322 | /** getReader() **/
|
---|
[26214] | 323 |
|
---|
| 324 | /**
|
---|
| 325 | */
|
---|
| 326 | private String generateHash(String string)
|
---|
| 327 | {
|
---|
| 328 | StringBuffer sb = new StringBuffer();
|
---|
| 329 | try
|
---|
| 330 | {
|
---|
| 331 | final MessageDigest message_digest = MessageDigest.getInstance("MD5");
|
---|
| 332 | message_digest.reset();
|
---|
| 333 | message_digest.update(string.getBytes(Charset.forName("UTF8")));
|
---|
| 334 | final byte[] result_bytes = message_digest.digest();
|
---|
| 335 | for (int i = 0; i < result_bytes.length; ++i)
|
---|
| 336 | {
|
---|
| 337 | sb.append(Integer.toHexString((result_bytes[i] & 0xFF) | 0x100).substring(1,3));
|
---|
| 338 | }
|
---|
| 339 | }
|
---|
| 340 | catch (NoSuchAlgorithmException e)
|
---|
| 341 | {
|
---|
| 342 | System.err.println("Exception: " + e);
|
---|
| 343 | System.exit(0);
|
---|
| 344 | }
|
---|
| 345 | return sb.toString();
|
---|
| 346 | }
|
---|
| 347 | /** generateHash(String) **/
|
---|
[26186] | 348 | }
|
---|
| 349 |
|
---|