tika PackageParser does not work with directories - apache-tika

Question

I am writing a class to recursively extract files from inside a zip file and produce them to a Kafka queue for further processing. My intent is to be able to extract files from multiple levels of zip. The code below is my implementation of the tika ContainerExtractor to do this.
import java.io.File;
import java.io.IOException;
import java.io.InputStream;
import java.util.ArrayList;
import java.util.Collections;
import java.util.HashSet;
import java.util.List;
import java.util.Set;
import java.util.Stack;
import org.apache.commons.lang.StringUtils;
import org.apache.tika.config.TikaConfig;
import org.apache.tika.detect.DefaultDetector;
import org.apache.tika.detect.Detector;
import org.apache.tika.exception.TikaException;
import org.apache.tika.extractor.ContainerExtractor;
import org.apache.tika.extractor.EmbeddedResourceHandler;
import org.apache.tika.io.TemporaryResources;
import org.apache.tika.io.TikaInputStream;
import org.apache.tika.metadata.Metadata;
import org.apache.tika.mime.MediaType;
import org.apache.tika.parser.AbstractParser;
import org.apache.tika.parser.AutoDetectParser;
import org.apache.tika.parser.ParseContext;
import org.apache.tika.parser.Parser;
import org.apache.tika.parser.pkg.PackageParser;
import org.xml.sax.ContentHandler;
import org.xml.sax.SAXException;
import org.xml.sax.helpers.DefaultHandler;
public class UberContainerExtractor implements ContainerExtractor {
/**
*
*/
private static final long serialVersionUID = -6636138154366178135L;
// statically populate SUPPORTED_TYPES
static {
Set<MediaType> supportedTypes = new HashSet<MediaType>();
ParseContext context = new ParseContext();
supportedTypes.addAll(new PackageParser().getSupportedTypes(context));
SUPPORTED_TYPES = Collections.unmodifiableSet(supportedTypes);
}
/**
* A stack that maintains the parent filenames for the recursion
*/
Stack<String> parentFileNames = new Stack<String>();
/**
* The default tika parser
*/
private final Parser parser;
/**
* Default tika detector
*/
private final Detector detector;
/**
* The supported container types into which we can recurse
*/
public final static Set<MediaType> SUPPORTED_TYPES;
/**
* The number of documents recursively extracted from the container and its
* children containers if present
*/
int extracted;
public UberContainerExtractor() {
this(TikaConfig.getDefaultConfig());
}
public UberContainerExtractor(TikaConfig config) {
this(new DefaultDetector(config.getMimeRepository()));
}
public UberContainerExtractor(Detector detector) {
this.parser = new AutoDetectParser(new PackageParser());
this.detector = detector;
}
public boolean isSupported(TikaInputStream input) throws IOException {
MediaType type = detector.detect(input, new Metadata());
return SUPPORTED_TYPES.contains(type);
}
#Override
public void extract(TikaInputStream stream, ContainerExtractor recurseExtractor, EmbeddedResourceHandler handler)
throws IOException, TikaException {
ParseContext context = new ParseContext();
context.set(Parser.class, new RecursiveParser(recurseExtractor, handler));
try {
Metadata metadata = new Metadata();
parser.parse(stream, new DefaultHandler(), metadata, context);
} catch (SAXException e) {
throw new TikaException("Unexpected SAX exception", e);
}
}
private class RecursiveParser extends AbstractParser {
/**
*
*/
private static final long serialVersionUID = -7260171956667273262L;
private final ContainerExtractor extractor;
private final EmbeddedResourceHandler handler;
private RecursiveParser(ContainerExtractor extractor, EmbeddedResourceHandler handler) {
this.extractor = extractor;
this.handler = handler;
}
public Set<MediaType> getSupportedTypes(ParseContext context) {
return parser.getSupportedTypes(context);
}
public void parse(InputStream stream, ContentHandler ignored, Metadata metadata, ParseContext context)
throws IOException, SAXException, TikaException {
TemporaryResources tmp = new TemporaryResources();
try {
TikaInputStream tis = TikaInputStream.get(stream, tmp);
// Figure out what we have to process
String filename = metadata.get(Metadata.RESOURCE_NAME_KEY);
MediaType type = detector.detect(tis, metadata);
if (extractor == null) {
// do nothing
} else {
// Use a temporary file to process the stream
File file = tis.getFile();
System.out.println("file is directory = " + file.isDirectory());
// Recurse and extract if the filetype is supported
if (SUPPORTED_TYPES.contains(type)) {
System.out.println("encountered a supported file:" + filename);
parentFileNames.push(filename);
extractor.extract(tis, extractor, handler);
parentFileNames.pop();
} else { // produce the file
List<String> parentFilenamesList = new ArrayList<String>(parentFileNames);
parentFilenamesList.add(filename);
String originalFilepath = StringUtils.join(parentFilenamesList, "/");
System.out.println("producing " + filename + " with originalFilepath:" + originalFilepath
+ " to kafka queue");
++extracted;
}
}
} finally {
tmp.dispose();
}
}
}
public int getExtracted() {
return extracted;
}
public static void main(String[] args) throws IOException, TikaException {
String filename = "/Users/rohit/Data/cd.zip";
File file = new File(filename);
TikaInputStream stream = TikaInputStream.get(file);
ContainerExtractor recursiveExtractor = new UberContainerExtractor();
EmbeddedResourceHandler resourceHandler = new EmbeddedResourceHandler() {
#Override
public void handle(String filename, MediaType mediaType, InputStream stream) {
// do nothing
}
};
recursiveExtractor.extract(stream, recursiveExtractor, resourceHandler);
stream.close();
System.out.println("extracted " + ((UberContainerExtractor) recursiveExtractor).getExtracted() + " files");
}
}
It works on multiple levels of zip as long as the files inside the zips are in a flat structure. for ex.
cd.zip
- c.txt
- d.txt
The code does not work if there the files in the zip are present inside a directory. for ex.
ab.zip
- ab/
- a.txt
- b.txt
While debugging I came across the following code snippet in the PackageParser
try {
ArchiveEntry entry = ais.getNextEntry();
while (entry != null) {
if (!entry.isDirectory()) {
parseEntry(ais, entry, extractor, xhtml);
}
entry = ais.getNextEntry();
}
} finally {
ais.close();
}
I tried to comment out the if condition but it did not work. Is there a reason why this is commented? Is there any way of getting around this?
I am using tika version 1.6

Answer 1

Tackling your question in reverse order:
Is there a reason why this is commented?
Entries in zip files are either directories or files. If files, they include the name of the directory they come from. As such, Tika doesn't need to do anything with the directories, all it needs to do is process the embedded files as and when they come up
The code does not work if there the files in the zip are present inside a directory. for ex. ab.zip - ab/ - a.txt - b.txt
You seem to be doing something wrong then. Tika's recursion and package parser handle zips with folders in them just fine!
To prove this, start with a zip file like this:
$ unzip -l ../tt.zip
Archive: ../tt.zip
Length Date Time Name
--------- ---------- ----- ----
0 2015-02-03 16:42 t/
0 2015-02-03 16:42 t/t2/
0 2015-02-03 16:42 t/t2/t3/
164404 2015-02-03 16:42 t/t2/t3/test.jpg
--------- -------
164404 4 files
Now, make us of the -z extraction flag of the Tika App, which causes Tika to extract out all of the embedded contents of a file. Run like that, and we get
$ java -jar tika-app-1.7.jar -z ../tt.zip
Extracting 't/t2/t3/test.jpg' (image/jpeg) to ./t/t2/t3/test.jpg
Then list the resulting directory, and we see
$ find . -type f
./t/t2/t3/Test.jpg
I can't see what's wrong with your code, but sadly for you we've shown that the problem is there, and not with Tika... You'd be best off reviewing the various examples of recursion that Tika provides, such as the Tika App tool and the Recursing Parser Wrapper, then re-write your code to be something simple based from those