Motivation

Processing of huge XML files can become cumbersome if your hardware is limited.

“Parsing a sample 20 MB XML document^[1] containing Wikipedia document abstracts into a DOM tree using the Xerces library roughly consumes about 100 MB of RAM. Other document model implementations^[2] such as Saxon’s TinyTree are more memory efficient; parsing the same document in Saxon consumes about 50 MB of memory. These numbers will vary with document contents, but generally the required memory scales linearly with document size, and is typically a single-digit multiple of the file size on disk.”

Probst, Martin. “Processing Arbitrarily Large XML using a Persistent DOM.” 2010. https://www.balisage.net/Proceedings/vol5/html/Probst01/BalisageVol5-Probst01.html

A good way to deal with huge files is to split them into smaller ones. But sometimes you don’t have that option.

Here is where Random Access comes into play. While random access of binary files is well supported by standard Java tools, this is not true for higher-order text-based formats like XML.

The Plan

Find proper access points, by taking XML structure into account.
Translate character offsets to byte offsets (take encoding into account)

This sounds straightforward.

Existing Libraries

The StAX library offers streaming access to XML data without the need of loading a complete DOM model into memory. The library comes with an XMLStreamReader offering a method getLocation().getCharacterOffset() .

But unfortunately this will only return character offsets. In order to access the file with standard java readers we need byte offsets. UTF-8 uses variable lengths for encoding characters. This means that we have to reread the whole file from the beginning to calculate the byte offset from character offset. This seems not acceptable.

Solution

In the following I will introduce a solution, based on a generated XML parser using ANTLR4.

We will use the parser to walk through the XML file. While the parser is doing it’s work it will spit out byte offsets whenever a certain criteria is fulfilled (in the example we will search for XML-Elements with the name ‘page’).
I will use the byte offsets to access the XML file and to read portions of XML into a Java bean using JAXB.

The Following works very well on a ~17GB Wikipedia dump/20170501/dewiki-20170501-pages-articles-multistream.xml.bz2 . I still had to increase heap size using -xX6GB but compared to a DOM approach this looks much more acceptable.

1. Get XML Grammar

cd /tmp
git clone https://github.com/antlr/grammars-v4

2. Generate Parser

cd /tmp/grammars-v4/xml/
mvn clean install

3. Copy Generated Java files to your Project

cp -r target/generated-sources/antlr4 /path/to/your/project/gen

4. Hook in with a Listener to collect character offsets

package stack43366566;

import java.util.ArrayList;
import java.util.List;

import org.antlr.v4.runtime.ANTLRFileStream;
import org.antlr.v4.runtime.CommonTokenStream;
import org.antlr.v4.runtime.tree.ParseTreeWalker;

import stack43366566.gen.XMLLexer;
import stack43366566.gen.XMLParser;
import stack43366566.gen.XMLParser.DocumentContext;
import stack43366566.gen.XMLParserBaseListener;

public class FindXmlOffset {

    List<Integer> offsets = null;
    String searchForElement = null;

    public class MyXMLListener extends XMLParserBaseListener {
        public void enterElement(XMLParser.ElementContext ctx) {
            String name = ctx.Name().get(0).getText();
            if (searchForElement.equals(name)) {
                offsets.add(ctx.start.getStartIndex());
            }
        }
    }

    public List<Integer> createOffsets(String file, String elementName) {
        searchForElement = elementName;
        offsets = new ArrayList<>();
        try {
            XMLLexer lexer = new XMLLexer(new ANTLRFileStream(file));
            CommonTokenStream tokens = new CommonTokenStream(lexer);
            XMLParser parser = new XMLParser(tokens);
            DocumentContext ctx = parser.document();
            ParseTreeWalker walker = new ParseTreeWalker();
            MyXMLListener listener = new MyXMLListener();
            walker.walk(listener, ctx);
            return offsets;
        } catch (Exception e) {
            throw new RuntimeException(e);
        }
    }

    public static void main(String[] arg) {
        System.out.println("Search for offsets.");
        List<Integer> offsets = new FindXmlOffset().createOffsets("/tmp/dewiki-20170501-pages-articles-multistream.xml",
                        "page");
        System.out.println("Offsets: " + offsets);
    }

}

5. Result

Prints:

Offsets: [2441, 10854, 30257, 51419 ….

6. Read from Offset Position

To test the code I’ve written class that reads in each wikipedia page to a java object

@JacksonXmlRootElement
class Page {
 public Page(){};
 public String title;
}

using basically this code

private Page readPage(Integer offset, String filename) {
        try (Reader in = new FileReader(filename)) {
            in.skip(offset);
            ObjectMapper mapper = new XmlMapper();
             mapper.configure(DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES, false);
            Page object = mapper.readValue(in, Page.class);
            return object;
        } catch (Exception e) {
            throw new RuntimeException(e);
        }
    }

Download

Find complete example on github.

Direct Accessing XML with Java