Parse flat files describing a hierarchical data structure
Java
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
src
.travis.yml
README.md
pom.xml

README.md

README

Flatlinr parses simple flat file structures like CSV files and also really complex hierarchical data structures based on flat files.

git clone git://github.com/platzhaltr/flatlinr.git
cd flatlinr
mvn clean install

Usage

Assume the following example file

#Library
	- History
		- Greek;
		- Asia;
	- Language
		- Spanish;
		- English;

You want to break the file into the following structure

Library
	- Room
		- Shelf	

in which a library can have multiple rooms, and each room can have multiple shelves.

Flatlinr helps you to keep sane. Once you have defined the structure of the file, instantiate a FlatFileIterator, which will take care of traversing the tree for you. All you have to do is to use the iterator interface and react on the various record identifiers.

/**
 * Flatlinr interprets files as tree hierarchical structures.
 */
public class UsageExample {
	
	// define the structure of the file
	private final Node getRoot() {
		
		final Node root = 
				new Node("library")
				.add(new ConstantLeaf("#"))
				.add(new LineLeaf("name"));
			
		final Node room = 
			new Node("room")
			.add(new ConstantLeaf("\t- "))
			.add(new DelimitedLeaf("name", ';'));
		
		// you can also auto-convert delimited leafs via Features
		final Node shelf = 
			new Node("shelf")
			.add(new ConstantLeaf("\t\t- "))
			.add(new DelimitedLeaf("name", ';',
				Features.TRIM));

		room.addChild(shelf);
		root.addChild(room);

		return root;
	}

	public final void read(final File file) throws IOException {
		final FlatFileIterator iterator = new FlatFileIterator(getRoot(),
				new FileReader(file));

		while (iterator.hasNext()) {
			final Record record = iterator.next();
			if (record.getId().equals("library")) {
				System.out.println(record.get("name"));
			} else if (record.getId().equals("room")) {
				System.out.println(record.get("name"));
			} else if (record.getId().equals("shelf")) {
				System.out.println(record.get("name"));
			}
		}
	}
}

Traversal

flatlinr supports more complex structures

  • different delimiter per node
  • multiple children per node

The drawback is that the decision which node should be used for parsing the next line is hard. The decision can be defined in a TraversalStrategy. The DefaultTraversalStrategy uses these the following biases in order until it gives up. Each bias is based on the fact if the the next line starts with the first leaf of the current node.

  1. Bias towards the child if it starts with a constant leaf and the current node starts with a delimited leaf.
  2. Bias towards the current node itself. This allows for multiples instances.
  3. Bias towards the children
  4. Bias towards the next siblings
  5. Bias towards the prior siblings
  6. Bias towards the parent
  7. If direct parent doesn't match, recursively traverse the tree up with the parent as the currentNode and start again with rule 4.

If all fails IllegalStateException is thrown.

Notes

I also created Readr as a companion project. It creates java.io.Reader classes that help you clean up the files by removing or transforming certain lines.