source: gs2-extensions/video-and-audio/trunk/src/opt/Terrier/ImageDocument.java

/**
 *  Adding support for Images in Terrier
 *  @author: John Thompson, jmt12, #9826509
 *
 *  The contents of this file are subject to the Mozilla Public License
 *  Version 1.1 (the "License"); you may not use this file except in
 *  compliance with the License. You may obtain a copy of the License at
 *  http://www.mozilla.org/MPL/
 *
 *  Software distributed under the License is distributed on an "AS IS"
 *  basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See
 *  the License for the specific language governing rights and limitations
 *  under the License.
 *
 *  Copyright (c) 2011 The University of Waikato. All Rights Reserved.
 */
package org.terrier.indexing;

import java.io.InputStream;
import java.io.StringReader;
import java.io.Reader;
import java.nio.charset.Charset;
import java.nio.file.Files;
import java.nio.file.Path;
import java.nio.file.Paths;
import java.security.MessageDigest;
import java.security.NoSuchAlgorithmException;
import java.util.Collections;
import java.util.Arrays;
import java.util.Map;
import java.util.Set;

import org.apache.log4j.Logger;
import org.terrier.indexing.StreamGobbler;
import org.terrier.indexing.tokenisation.TokenStream;
import org.terrier.indexing.tokenisation.Tokeniser;
import org.terrier.utility.ApplicationSetup;

public class ImageDocument
  implements Document
{
  /** A reference to the logger for messaging */
  protected static final Logger logger = Logger.getLogger(FileDocument.class);
  /** The map of properties (fields) for this document. */
  protected Map<String,String> properties;
  /** A reader built from a dummy text string. */
  protected Reader reader;
  /** A token stream produced by the configured tokeniser when feed the dummy
   *  reader.
   */
  protected TokenStream tokenizer;

  /** The preview filetype. **/
  protected final String preview_format = ApplicationSetup.getProperty("ImageDocument.preview_format", "jpg");
  /** The preview size (width). **/
  protected final String preview_width = ApplicationSetup.getProperty("ImageDocument.preview_width", "200");

  /** Default constructor. **/
  protected ImageDocument() {}

  /** Constructs an instance of the ImageDocument from the given input stream.
   *  @param docStream the input stream that reads the file.
   *  @param docProperties the initial properties (docno, filename)
   *  @param tok the tokeniser defined for this collection
   */
  public ImageDocument(InputStream istream, Map<String,String> default_properties, Tokeniser tok)
  {
    logger.info("ImageDocument::ImageDocument()");
    // Initialization from arguments
    this.properties = default_properties;

    // Set properties
    logger.info("ImageDocument - current properties");
    for (Map.Entry<String, String> entry : this.properties.entrySet())
    {
      logger.info(entry.getKey() + "=" + entry.getValue());
    }

    logger.info("ImageDocument - extracting properties");
    // A. Hardcoded properties
    this.properties.put("parser", "ImageDocument");
    this.properties.put("abstract", "This is an image so here is some dummy text to prevent indexer failing.");
    // B. Properties derived from filename
    // - A simple title for the document
    String filepath = this.properties.get("filename");
    String title = filepath.substring(filepath.lastIndexOf(System.getProperty("file.separator")) + 1);
    this.properties.put("title", title);
    String ext = filepath.substring(filepath.lastIndexOf(".") + 1);
    // - The name of the copy of the original document
    String target_filename = "doc." + ext;
    this.properties.put("source","doc." + ext);
    // - A unique associated directory. This gets a little tricky as we need
    //   to create the directory at the same time if an effort to promote
    //   synchronous behaviour
    String unique_id = this.generateHash(filepath);
    //   - we start with the first 4 characters
    int offset = 0;
    String assoc_filename = "D" + unique_id.substring(offset, offset + 4);
    //   - we add ".dir" as a suffix to the directory that actually contains
    //     files (so the non-suffixed version contains nested directories)
    Path assoc_path = Paths.get(ApplicationSetup.TERRIER_SHARE, "images", "assoc", assoc_filename + ".dir");
    //   - then we continue adding blocks of 4 characters until we get a
    //     directory that doesn't already exist
    while (assoc_path.toFile().exists() && offset < unique_id.length())
    {
      offset += 4;
      assoc_filename += System.getProperty("file.separator") + unique_id.substring(offset, offset + 4);
      assoc_path = Paths.get(ApplicationSetup.TERRIER_SHARE, "images", "assoc", assoc_filename + ".dir");
    }
    //   - still not unique? but run out of unique_id... time to complain
    if (assoc_path.toFile().exists())
    {
      logger.error("ImageDoument - can't determine unique assocfilepath");
      System.exit(0);
    }
    //   - create the directories quick... hopefully before someone else does
    assoc_path.toFile().mkdirs(); // bet there is a nice nio way to do this
    this.properties.put("assocfile", assoc_filename);

    // Copy (symlink) the file into place in the shared directory
    Path source_path = Paths.get(properties.get("filename"));
    Path target_path = assoc_path.resolve(target_filename);
    if (target_path.toFile().exists())
    {
      logger.info("ImageDocument - removing existing (old) associated image");
      try
      {
        Files.delete(target_path);
      }
      catch (Exception e)
      {
        logger.error("Exception while deleting old image: ", e);
      }
    }
    logger.info("ImageDocument - symlinking image into assoc directory");
    try
    {
      Files.createSymbolicLink(target_path, source_path);
    }
    // not supported? We'll try copying below
    catch (UnsupportedOperationException ex)
    {
    }
    // All other exceptions can be fatal
    catch (Exception e)
    {
      logger.error("Exception while symlinking image: ", e);
    }
    // - copy if the file doesn't exist yet
    if (!target_path.toFile().exists())
    {
      logger.info("ImageDocument - symlink filaed, copying instead");
      try
      {
        Files.copy(source_path, target_path);
      }
      // Fatality!
      catch (Exception e)
      {
        logger.error("Exception while copying image: ", e);
      }
    }

    // Generate preview image
    logger.info("ImageDocument - generate preview image");
    try
    {
      String preview_filename = this.generatePreview(source_path, assoc_path);
      this.properties.put("preview",preview_filename);
    }
    catch (Exception e)
    {
      logger.error("Exception while generating preview image: ", e);
    }

    // Create a dummy reader around some dummy text and then tokenize it
    logger.info("ImageDocument - feed dummy text as token stream to indexer");
    try
    {
      this.reader = new StringReader(this.properties.get("abstract"));
      this.tokenizer = tok.tokenise(this.reader);
    }
    catch (Exception e)
    {
      logger.error("Exception while creating dummy text stream: ", e);
    }

    // Use OpenSIFT to generate a featureset (in Oxford format) for this image
    logger.info("ImageDocument - generate and record SIFT features");
    try
    {
      String sift_command[] = {
        "siftfeat",
        "-x",
        source_path.toString()
      };
      logger.info("ImageDocument - sift command: " + Arrays.toString(sift_command));
      Process sift_process = Runtime.getRuntime().exec(sift_command);
      // we'd usually send STDERR to /dev/null, but a streamgobbler is easier
      // in Java
      StreamGobbler sift_process_error_gobbler = new StreamGobbler(sift_process.getErrorStream());
      sift_process_error_gobbler.start();
      // the SIFT features, in Oxford format, will arrive from STDOUT
      BufferedReader sift_br = new BufferedReader(new InputStreamReader(sift_process.getInputStream()));
      String line;
      StringBuffer oxford_features;
      while ((line = sift_br.readLine()) != null)
      {
        oxford_features.append(line);
      }
      // this command blocks until process completes (emit return value) which
      // should be shortly after it emits the last line of SIFT feature data
      int sift_status = sift_process.waitFor();
      this.properties.put("sift", oxford_features.toString());
    }
    catch (Exception e)
    {
      logger.error("Exception while generating preview image: ", e);
    }

    logger.info("ImageDocument - Complete!");
  }
  /** ImageDocument(InputStream istream, Map<String,String> default_properties, Tokeniser tok) **/

  /** Returns true when the end of the document has been reached, and there
   *  are no other terms to be retrieved from it.
   *  @return boolean true if there are no more terms in the document, otherwise
   *          it returns false.
   */
  public boolean endOfDocument()
  {
    return !this.tokenizer.hasNext();
  }
  /** endOfDocument() **/

  /** Use ImageMagick to generate a preview image.
   *  @pre assumes you have ImageMagick installed and available on Path
   *  @pre uses member variables preview_format and preview_width
   *  @return the filename of the preview image (within the assoc directory)
   */
  private String generatePreview(Path source_path, Path assoc_path)
    throws Exception
  {
    String preview_filename = "preview." + this.preview_format;
    Path preview_path = assoc_path.resolve(preview_filename);
    String convert_command[] = {
      "convert",
      source_path.toString(),
      "-resize",
      this.preview_width + "x",
      preview_path.toString()
    };
    logger.info("ImageDocument - convert command: " + Arrays.toString(convert_command));
    Process convert_process = Runtime.getRuntime().exec(convert_command);
    // Gobble up the streams to prevent them hanging the process when buffers
    // are full
    StreamGobbler convert_process_error_gobbler = new StreamGobbler(convert_process.getErrorStream());
    convert_process_error_gobbler.start();
    StreamGobbler convert_process_input_gobbler = new StreamGobbler(convert_process.getInputStream());
    convert_process_input_gobbler.start();
    // Let the conversion finish
    int convert_status = convert_process.waitFor();
    if (convert_status != 0 || !preview_path.toFile().exists())
    {
      throw new Exception("Convert command failed (exit status: " + convert_status + ")");
    }
    return preview_filename;
  }
  /** generatePreview(Path, Path) **/

  /** Returns the underlying map of all the properties defined by this Document.
   *  @since 1.1.0
   */
  public Map<String,String> getAllProperties()
  {
    return this.properties;
  }
  /** getAllProperties() **/

  /** Returns a list of the fields the current term appears in.
   * @return HashSet a set of the terms that the current term appears in.
   */
  public Set<String> getFields()
  {
    // Returns null because there is no support for fields with file documents.
    return Collections.emptySet();
  }
  /** getFields() **/

  /** Gets the next term of the document.
   *  <B>NB:</B>Null string returned from getNextTerm() should
   *  be ignored. They do not signify the lack of any more terms.
   *  endOfDocument() should be used to check that.
   *  @return String the next term of the document. Null returns should be
   *          ignored.
   */
  public String getNextTerm()
  {
    return this.tokenizer.next();
  }
  /** getNextTerm() **/

  /** Allows access to a named property of the Document. Examples might be URL,
   *  filename etc.
   *  @param name Name of the property. It is suggested, but not required that
   *         this name should not be case insensitive.
   *  @since 1.1.0
   */
  public String getProperty(String name)
  {
    return this.properties.get(name.toLowerCase());
  }
  /** getProperty(String name) **/

  /** Returns a Reader object so client code can tokenise the document
   * or deal with the document itself. Examples might be extracting URLs,
   * language detection. */
  public Reader getReader()
  {
    return this.reader;
  }
  /** getReader() **/

  /**
   */
  private String generateHash(String string)
  {
    StringBuffer sb = new StringBuffer();
    try
    {
      final MessageDigest message_digest = MessageDigest.getInstance("MD5");
      message_digest.reset();
      message_digest.update(string.getBytes(Charset.forName("UTF8")));
      final byte[] result_bytes = message_digest.digest();
      for (int i = 0; i < result_bytes.length; ++i)
      {
        sb.append(Integer.toHexString((result_bytes[i] & 0xFF) | 0x100).substring(1,3));
      }
    }
    catch (NoSuchAlgorithmException e)
    {
      System.err.println("Exception: " + e);
      System.exit(0);
    }
    return sb.toString();
  }
  /** generateHash(String) **/
}