[Graphbolt] Offline script to convert from COO to CSC sampling graph.

[frozenbugs]

🔨Work Item

Such that it can be load directly to CSC sampling graph. IMPORTANT:

• This template is only for dev team to track project progress. For feature request or bug report, please use the corresponding issue templates.
• DO NOT create a new work item if the purpose is to fix an existing issue or feature request. We will directly use the issue in the project tracker.

Project tracker: https://github.com/orgs/dmlc/projects/2

Description

Depending work items or issues

[keli-wen]

I've noticed that there's a CPU version implementation in DGL Code.

I wonder if we need to follow this code? My understanding is that if our only requirement is to convert a single COO to CSC, this might be sufficient.

[Rhett-Ying]

could we offer a utility API to convert COO(which could be homogeneous or heterogenous, unsorted) to a sorted CSCSamplingGraph? could we construct a DGLGraph first form coo data and convert via dgl.to_homogeneous() to obtain csc matrix?

[keli-wen]

The PR: [Graphbolt] Add the preprocess_ondisk_dataset function. implemented a preprocess_ondisk_dataset() function:

• Receive the input_config_path (a YAML file), extract the data from the graph field, and utilize the original data in .csv/.npy format to create a CSCSamplingGraph.
• Convert all torch format data into numpy format.
• Return the processed YAML file path.