Merge remote-tracking branch 'origin/main'

tika-app/pom.xml

@@ -83,12 +83,10 @@
     <dependency>
       <groupId>org.apache.logging.log4j</groupId>
       <artifactId>log4j-core</artifactId>
-      <version>${log4j2.version}</version>
     </dependency>
     <dependency>
       <groupId>org.apache.logging.log4j</groupId>
       <artifactId>log4j-slf4j2-impl</artifactId>
-      <version>${log4j2.version}</version>
     </dependency>
     <dependency>
       <groupId>org.slf4j</groupId>

tika-batch/pom.xml

@@ -49,12 +49,10 @@
     <dependency>
       <groupId>org.apache.commons</groupId>
       <artifactId>commons-compress</artifactId>
-      <version>${commons.compress.version}</version>
     </dependency>
     <dependency>
       <groupId>commons-cli</groupId>
       <artifactId>commons-cli</artifactId>
-      <version>${commons.cli.version}</version>
     </dependency>
 
     <dependency>
@@ -67,7 +65,6 @@
     <dependency>
       <groupId>org.slf4j</groupId>
       <artifactId>slf4j-simple</artifactId>
-      <version>${slf4j.version}</version>
       <scope>test</scope>
     </dependency>
   </dependencies>

tika-bundles/tika-bundle-standard/pom.xml

@@ -107,22 +107,19 @@
     <dependency>
       <groupId>org.osgi</groupId>
       <artifactId>org.osgi.core</artifactId>
-      <version>${osgi.core.version}</version>
       <scope>test</scope>
     </dependency>
 
     <!-- use non-log4j slf4j backend to prevent main classloader from loading log4j classes -->
     <dependency>
       <groupId>org.slf4j</groupId>
       <artifactId>slf4j-simple</artifactId>
-      <version>${slf4j.version}</version>
       <scope>test</scope>
     </dependency>
 
     <dependency>
       <groupId>org.glassfish.jaxb</groupId>
       <artifactId>jaxb-runtime</artifactId>
-      <version>${jaxb.version}</version>
       <scope>test</scope>
     </dependency>
     <dependency>

tika-core/src/main/java/org/apache/tika/detect/FileCommandDetector.java

@@ -108,7 +108,7 @@ public MediaType detect(InputStream input, Metadata metadata) throws IOException
 
         input.mark(maxBytes);
         try (TemporaryResources tmp = new TemporaryResources()) {
-            Path tmpFile = tmp.createTempFile();
+            Path tmpFile = tmp.createTempFile(metadata);
             Files.copy(new BoundedInputStream(maxBytes, input), tmpFile, REPLACE_EXISTING);
             return detectOnPath(tmpFile, metadata);
         } finally {

tika-core/src/main/java/org/apache/tika/extractor/ParserContainerExtractor.java

@@ -111,7 +111,7 @@ public void parse(
                 throws IOException, SAXException, TikaException {
             TemporaryResources tmp = new TemporaryResources();
             try {
-                TikaInputStream tis = TikaInputStream.get(stream, tmp);
+                TikaInputStream tis = TikaInputStream.get(stream, tmp, metadata);
 
                 // Figure out what we have to process
                 String filename = metadata.get(TikaCoreProperties.RESOURCE_NAME_KEY);

tika-core/src/main/java/org/apache/tika/extractor/ParsingEmbeddedDocumentExtractor.java

@@ -98,7 +98,8 @@ public void parseEmbedded(
 
         // Use the delegate parser to parse this entry
         try (TemporaryResources tmp = new TemporaryResources()) {
-            final TikaInputStream newStream = TikaInputStream.get(new CloseShieldInputStream(stream), tmp);
+            final TikaInputStream newStream =
+                    TikaInputStream.get(new CloseShieldInputStream(stream), tmp, metadata);
             if (stream instanceof TikaInputStream) {
                 final Object container = ((TikaInputStream) stream).getOpenContainer();
                 if (container != null) {

tika-core/src/main/java/org/apache/tika/io/FilenameUtils.java

@@ -19,6 +19,8 @@
 import java.util.HashSet;
 import java.util.Locale;
 
+import org.apache.tika.utils.StringUtils;
+
 
 public class FilenameUtils {
 
@@ -93,7 +95,7 @@ public static String normalize(final String name) {
     public static String getName(final String path) {
 
         if (path == null || path.length() == 0) {
-            return "";
+            return StringUtils.EMPTY;
         }
         int unix = path.lastIndexOf("/");
         int windows = path.lastIndexOf("\\");
@@ -102,8 +104,23 @@ public static String getName(final String path) {
         int colon = path.lastIndexOf(":");
         String cand = path.substring(Math.max(colon, Math.max(unix, windows)) + 1);
         if (cand.equals("..") || cand.equals(".")) {
-            return "";
+            return StringUtils.EMPTY;
         }
         return cand;
     }
+
+    /**
+     * This includes the period, e.g. ".pdf"
+     * @param path
+     * @return the suffix or an empty string if one could not be found
+     */
+    public static String getSuffixFromPath(String path) {
+        String n = getName(path);
+        int i = n.lastIndexOf(".");
+        //arbitrarily sets max extension length
+        if (i > -1 && n.length() - i < 6) {
+            return n.substring(i);
+        }
+        return StringUtils.EMPTY;
+    }
 }

tika-core/src/main/java/org/apache/tika/io/TemporaryResources.java

@@ -27,6 +27,9 @@
 import org.slf4j.LoggerFactory;
 
 import org.apache.tika.exception.TikaException;
+import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
+import org.apache.tika.utils.StringUtils;
 
 /**
  * Utility class for tracking and ultimately closing or otherwise disposing
@@ -52,7 +55,7 @@ public class TemporaryResources implements Closeable {
 
     /**
      * Sets the directory to be used for the temporary files created by
-     * the {@link #createTempFile()} method.
+     * the {@link #createTempFile(String)} method.
      *
      * @param tempFileDir temporary file directory,
      *                    or <code>null</code> for the system default
@@ -63,7 +66,7 @@ public void setTemporaryFileDirectory(Path tempFileDir) {
 
     /**
      * Sets the directory to be used for the temporary files created by
-     * the {@link #createTempFile()} method.
+     * the {@link #createTempFile(String)} method.
      *
      * @param tempFileDir temporary file directory,
      *                    or <code>null</code> for the system default
@@ -76,13 +79,15 @@ public void setTemporaryFileDirectory(File tempFileDir) {
     /**
      * Creates a temporary file that will automatically be deleted when
      * the {@link #close()} method is called, returning its path.
-     *
+     * @param suffix -- the suffix of the file if known, starting with "." as in ".pdf"
      * @return Path to created temporary file that will be deleted after closing
      * @throws IOException
      */
-    public Path createTempFile() throws IOException {
-        final Path path = tempFileDir == null ? Files.createTempFile("apache-tika-", ".tmp") :
-                Files.createTempFile(tempFileDir, "apache-tika-", ".tmp");
+    public Path createTempFile(String suffix) throws IOException {
+        String actualSuffix = StringUtils.isBlank(suffix) ? ".tmp" : suffix;
+
+        final Path path = tempFileDir == null ? Files.createTempFile("apache-tika-", actualSuffix) :
+                Files.createTempFile(tempFileDir, "apache-tika-", actualSuffix);
         addResource(() -> {
             try {
                 Files.delete(path);
@@ -95,16 +100,34 @@ public Path createTempFile() throws IOException {
         return path;
     }
 
+    public Path createTempFile() throws IOException {
+        return createTempFile(StringUtils.EMPTY);
+    }
+
+    /**
+     * Creates a temporary file that will automatically be deleted when
+     * the {@link #close()} method is called, returning its path.
+     *
+     * @return Path to created temporary file that will be deleted after closing
+     * @throws IOException
+     */
+    public Path createTempFile(Metadata metadata) throws IOException {
+        String resourceName = metadata.get(TikaCoreProperties.RESOURCE_NAME_KEY);
+        if (StringUtils.isBlank(resourceName)) {
+            return createTempFile(StringUtils.EMPTY);
+        }
+        return createTempFile(FilenameUtils.getSuffixFromPath(resourceName));
+    }
     /**
      * Creates and returns a temporary file that will automatically be
      * deleted when the {@link #close()} method is called.
      *
      * @return Created temporary file that'll be deleted after closing
      * @throws IOException
-     * @see #createTempFile()
+     * @see #createTempFile(String)
      */
     public File createTemporaryFile() throws IOException {
-        return createTempFile().toFile();
+        return createTempFile(StringUtils.EMPTY).toFile();
     }
 
     /**

tika-core/src/main/java/org/apache/tika/io/TikaInputStream.java

@@ -119,6 +119,10 @@ public class TikaInputStream extends TaggedInputStream {
     private int consecutiveEOFs = 0;
     private byte[] skipBuffer;
 
+    //suffix of the file if known. This is used to create temp files
+    //with the right suffixes. This should include the initial . as in ".doc"
+    private String suffix = null;
+
     /**
      * Creates a TikaInputStream instance. This private constructor is used
      * by the static factory methods based on the available information.
@@ -131,13 +135,15 @@ private TikaInputStream(Path path) throws IOException {
         this.path = path;
         this.tmp = new TemporaryResources();
         this.length = Files.size(path);
+        this.suffix = FilenameUtils.getSuffixFromPath(path.getFileName().toString());
     }
 
     private TikaInputStream(Path path, TemporaryResources tmp, long length) throws IOException {
         super(new BufferedInputStream(Files.newInputStream(path)));
         this.path = path;
         this.tmp = tmp;
         this.length = length;
+        this.suffix = FilenameUtils.getSuffixFromPath(path.getFileName().toString());
     }
 
     /**
@@ -154,6 +160,8 @@ private TikaInputStream(File file) throws FileNotFoundException {
         this.path = file.toPath();
         this.tmp = new TemporaryResources();
         this.length = file.length();
+        this.suffix = FilenameUtils.getSuffixFromPath(path.getFileName().toString());
+
     }
 
     /**
@@ -168,11 +176,12 @@ private TikaInputStream(File file) throws FileNotFoundException {
      * @param tmp    tracker for temporary resources associated with this stream
      * @param length total length of the stream, or -1 if unknown
      */
-    private TikaInputStream(InputStream stream, TemporaryResources tmp, long length) {
+    private TikaInputStream(InputStream stream, TemporaryResources tmp, long length, String suffix) {
         super(stream);
         this.path = null;
         this.tmp = tmp;
         this.length = length;
+        this.suffix = suffix;
     }
 
     /**
@@ -215,7 +224,7 @@ public static boolean isTikaInputStream(InputStream stream) {
      * @return a TikaInputStream instance
      * @since Apache Tika 0.10
      */
-    public static TikaInputStream get(InputStream stream, TemporaryResources tmp) {
+    public static TikaInputStream get(InputStream stream, TemporaryResources tmp, Metadata metadata) {
         if (stream == null) {
             throw new NullPointerException("The Stream must not be null");
         }
@@ -227,17 +236,28 @@ public static TikaInputStream get(InputStream stream, TemporaryResources tmp) {
             if (!(stream.markSupported())) {
                 stream = new BufferedInputStream(stream);
             }
-            return new TikaInputStream(stream, tmp, -1);
+            return new TikaInputStream(stream, tmp, -1, getExtension(metadata));
         }
     }
 
+    /**
+     * @deprecated use {@link TikaInputStream#get(InputStream, TemporaryResources, Metadata)}
+     * @param stream
+     * @param tmp
+     * @return
+     */
+    @Deprecated
+    public static TikaInputStream get(InputStream stream, TemporaryResources tmp) {
+        return get(stream, tmp, null);
+    }
+
     /**
      * Casts or wraps the given stream to a TikaInputStream instance.
      * This method can be used to access the functionality of this class
      * even when given just a normal input stream instance.
      * <p>
      * Use this method instead of the
-     * {@link #get(InputStream, TemporaryResources)} alternative when you
+     * {@link #get(InputStream, TemporaryResources, Metadata)} alternative when you
      * <em>do</em> explicitly close the returned stream. The recommended
      * access pattern is:
      * <pre>
@@ -254,7 +274,7 @@ public static TikaInputStream get(InputStream stream, TemporaryResources tmp) {
      * @return a TikaInputStream instance
      */
     public static TikaInputStream get(InputStream stream) {
-        return get(stream, new TemporaryResources());
+        return get(stream, new TemporaryResources(), null);
     }
 
     /**
@@ -301,7 +321,7 @@ public static TikaInputStream get(byte[] data) {
     public static TikaInputStream get(byte[] data, Metadata metadata) {
         metadata.set(Metadata.CONTENT_LENGTH, Integer.toString(data.length));
         return new TikaInputStream(new UnsynchronizedByteArrayInputStream(data), new TemporaryResources(),
-                data.length);
+                data.length, getExtension(metadata));
     }
 
     /**
@@ -414,7 +434,7 @@ public static TikaInputStream get(InputStreamFactory factory) throws IOException
      */
     public static TikaInputStream get(InputStreamFactory factory, TemporaryResources tmp)
             throws IOException {
-        TikaInputStream stream = get(factory.getInputStream(), tmp);
+        TikaInputStream stream = get(factory.getInputStream(), tmp, null);
         stream.streamFactory = factory;
         return stream;
     }
@@ -451,6 +471,7 @@ public static TikaInputStream get(Blob blob) throws SQLException {
      * @throws SQLException if BLOB data can not be accessed
      */
     public static TikaInputStream get(Blob blob, Metadata metadata) throws SQLException {
+
         long length = -1;
         try {
             length = blob.length();
@@ -465,8 +486,17 @@ public static TikaInputStream get(Blob blob, Metadata metadata) throws SQLExcept
             return get(blob.getBytes(1, (int) length), metadata);
         } else {
             return new TikaInputStream(new BufferedInputStream(blob.getBinaryStream()),
-                    new TemporaryResources(), length);
+                    new TemporaryResources(), length,
+                    getExtension(metadata));
+        }
+    }
+
+    private static String getExtension(Metadata metadata) {
+        if (metadata == null) {
+            return StringUtils.EMPTY;
         }
+        String name = metadata.get(TikaCoreProperties.RESOURCE_NAME_KEY);
+        return FilenameUtils.getSuffixFromPath(name);
     }
 
     /**
@@ -570,7 +600,7 @@ public static TikaInputStream get(URL url, Metadata metadata) throws IOException
         }
 
         return new TikaInputStream(new BufferedInputStream(connection.getInputStream()),
-                new TemporaryResources(), length);
+                new TemporaryResources(), length, getExtension(metadata));
     }
 
     /**
@@ -678,7 +708,7 @@ public Path getPath(int maxBytes) throws IOException {
             if (position > 0) {
                 throw new IOException("Stream is already being read");
             } else {
-                Path tmpFile = tmp.createTempFile();
+                Path tmpFile = tmp.createTempFile(suffix);
                 if (maxBytes > -1) {
                     try (InputStream lookAhead = new LookaheadInputStream(this, maxBytes)) {
                         Files.copy(lookAhead, tmpFile, REPLACE_EXISTING);

tika-core/src/main/java/org/apache/tika/parser/AutoDetectParser.java

@@ -141,7 +141,7 @@ public void parse(InputStream stream, ContentHandler handler, Metadata metadata,
         }
         TemporaryResources tmp = new TemporaryResources();
         try {
-            TikaInputStream tis = TikaInputStream.get(stream, tmp);
+            TikaInputStream tis = TikaInputStream.get(stream, tmp, metadata);
 
             //figure out if we should spool to disk
             maybeSpool(tis, autoDetectParserConfig, metadata);

tika-core/src/main/java/org/apache/tika/parser/CompositeParser.java

@@ -287,7 +287,7 @@ public void parse(InputStream stream, ContentHandler handler, Metadata metadata,
             context.set(ParseRecord.class, parserRecord);
         }
         try {
-            TikaInputStream taggedStream = TikaInputStream.get(stream, tmp);
+            TikaInputStream taggedStream = TikaInputStream.get(stream, tmp, metadata);
             TaggedContentHandler taggedHandler =
                     handler != null ? new TaggedContentHandler(handler) : null;
             String parserClassname = ParserUtils.getParserClassname(parser);