Skip to main content

File Formats

The system supports two text-based graph file formats. Both the WASM browser parser and the Rust backend parser implement identical parsing logic, ensuring consistent results between client-side preview and server-side algorithm execution.

Edge List​

The most common format. Each line describes one directed edge.

Syntax: source target [weight]

# comment lines are ignored
% so are lines starting with %
0 1 2.5
0 2 1.0
1 3 1.5
2 3 0.5
3 4 2.0

Rules:

  • Fields are whitespace-separated (spaces or tabs)
  • Weight is optional — defaults to 1.0 if absent
  • Lines starting with # or % are skipped (common in SNAP dataset files)
  • Node IDs are integers (can be large and sparse — e.g., 0, 100, 999999)
  • Directed edges: 0 1 means edge from 0 to 1 only

Minimal example (unweighted):

0 1
1 2
2 3
3 0

SNAP-compatible:

The edge list format is compatible with Stanford SNAP datasets. Most SNAP files can be uploaded directly.

Adjacency List​

Each line describes one node and all of its outgoing neighbors.

Syntax: node: neighbor1[,weight1] neighbor2[,weight2] ...

0: 1,2.5 2,1.0
1: 3,1.5
2: 3,0.5
3: 4,2.0

Rules:

  • Node ID followed by colon, then space-separated neighbor entries
  • Each neighbor entry is neighborId or neighborId,weight
  • Weight defaults to 1.0 if omitted
  • Lines starting with # or % are skipped
  • Nodes with no outgoing edges can be listed with an empty neighbor list: 5:

Display Limits​

The browser WASM parser supports up to 3,000 display nodes for performance. If a file contains more:

  • The first 3,000 unique nodes encountered are rendered
  • Remaining nodes are counted but not displayed
  • The truncated: true flag is set in the parsed result
  • Server-side algorithm execution still processes all nodes regardless of display limit

Large File Tips​

For graphs with hundreds of thousands of edges:

  1. Edge list is faster to parse — simpler format, fewer string operations
  2. WASM parsing is ~10× faster than JavaScript — 100K edges parses in ~80ms
  3. Reduce to 3K nodes — pre-filter your graph to stay under the display limit for best visualization
  4. Algorithm execution has no size limit — the backend processes the full graph regardless

Format Detection​

The frontend does not auto-detect format — you must select "Edge List" or "Adjacency List" in the UI. Sending the wrong format to the API returns a parse error.

Sample Files​

Minimal directed graph (edge list):

0 1
1 2
2 3
3 1

This graph has an SCC: nodes 1, 2, 3 are strongly connected.

Weighted DAG (edge list):

0 1 1.0
0 2 4.0
1 2 2.0
1 3 5.0
2 3 1.0

Dijkstra from node 0: shortest path to node 3 is 0 → 1 → 2 → 3 with total weight 4.0.

Adjacency list equivalent:

0: 1,1.0 2,4.0
1: 2,2.0 3,5.0
2: 3,1.0