databasegroup / apted Goto Github PK

Hi,

We're working a common file format for the output of tree differencing tools. The idea is that many tree differencing tools would have an option to output this format. This would:

• ease building tools on top of tree differencing
• ease scientific experiments to compare the effectiveness of each tool on specific use cases.

Here is a first draft of the TreeDiff file format: https://docs.google.com/document/d/15ITrZ4c0qdb4rGLfaMt8K9VPG1CkNp5QSgfOY6jzpEk/edit?usp=sharing (comments can be left in the Gdoc))

What do you think?

Best regards,

Jean-Rémy (@jrfaller), Matias (@martinezmatias) and Martin (@monperrus) (authors of Gumtree

Hi,

I see that in the "Customizing" section of your documentation, you mention that we can """Compares attribute .value of trees""". Therefore, the question is how should we input the tree which contains value at each node?

Thanks for your attention in advance!

Maybe an issue is not the best place for it, but I've ported APTED to Python and I thought it would be nice to let you know: https://github.com/JoaoFelipe/apted

Trees with node labels that contain '{' and '}' characters cannot be directly provided to this tool at the moment. It would nice to support a way to escape these characters.

One way would be to update LblTree.fromString method so that it updated the input by replacing, for instance "{" and "}' characters with "\opencurl" and "\closecurl". Also replace "\" characters, for instance, with "\slash". Then it can continue building the associated trees.

public static LblTree fromString(String s)
    {
        int treeID = FormatUtilities.getTreeID(s);
        // Implement escaping here
        s = s.substring(s.indexOf("{"), s.lastIndexOf("}") + 1);

I am trying to use Apted, but I think the NodeIndexer is not returning the proper node.
I have two trees were the only difference is that one node was deleted from the source tree.
String sourceTree = "{A{B{X}{Y}{F}}{C}}";
String destinationTree = "{A{B{X}{Y}}{C}}";

I compute the edit distance and create the edit mapping.

float ed = apted.computeEditDistance(source, dest);
LinkedList<int[]> em = apted.computeEditMapping();

I get a proper mapping saying that node 3 (3 as in the index of that node when indexed in post order) was deleted, so far so good.
The problem comes when I want to find which node is node 3 (which should be "{F}").
The way I am trying to do this is by creating a NodeIndexer and using method postL_to_node to look for node 3.
NodeIndexer<Node, CostModel> niSource = new NodeIndexer(source, myCM);
Node node = niSource.postL_to_node(3);

and I get that node 3 is "{B{X}{Y}{F}}"
which is not "{F}",the node deleted.
I would like to know if this is a bug in postL_to_node, or maybe there is an easier way to look for a particular node from the mapping (a function that I input 3 and gives me Node {F}).

Thanks in advance

problemWithNodeIndexer.txt