fix(context): make ContextGraph.add_edge idempotent by deduping on edge_id

[pravit-amp]

Fixes #922

ContextGraph._add_internal_edge appended every edge unconditionally, so adding the same edge twice stored two copies sharing one content-derived edge_id. This inflated stats()["edge_count"], pushed density() past 1.0, and made re-ingest double the edge set on every run.

Changes

• Added an edge_id -> ContextEdge index (_edge_index), mirroring how self.nodes dedupes by node ID.
• _add_internal_edge now returns False when the edge_id already exists, before touching edges, edge_type_index, or _adjacency, and before firing the mutation callback (no phantom ADD_EDGE audit events).
• The two state-reset paths (load and clear()) also clear the new index.
• Four regression tests: repeat add_edge is a no-op, parallel edges with distinct attributes are preserved, re-ingest via build_from_entities_and_relationships stays at one edge, and clear() resets the dedupe index.

Notes

• This implements the silent no-op option from the issue discussion (option 1). Happy to switch to update-in-place if maintainers prefer.
• This also resolves the interaction with feat(context): add ContextGraph Markdown round-trip #852: ordinary re-ingested graphs no longer trip that PR's duplicate-edge-ID export guard.

Testing

• Full tests/context/ suite passes (484 tests).
• All repro scenarios from bug(context): ContextGraph.add_edge has no dedupe — identical edges are stored repeatedly under one shared edge ID, and re-ingest doubles the edge set #922 now produce expected output (1 edge, density 0.5, re-ingest stays at 1).

[qodo-code-review]

ⓘ Qodo reviews are paused because the subscription is no longer active. Ask your workspace admin to reactivate the subscription to resume reviews. Manage billing

[qodo-free-for-open-source-projects]

PR Summary by Qodo

Make ContextGraph.add_edge idempotent by deduping edges on edge_id

🐞 Bug fix 🧪 Tests 🕐 10-20 Minutes

AI Description

• Add an internal edge_id index to prevent duplicate edge storage.
• Make re-adding an existing edge_id a silent no-op (no index/audit mutations).
• Add regression tests covering idempotency, parallel edges, re-ingest, and clear().

Diagram

graph TD
  A["ContextGraph.add_edge"] --> B["Resolve edge_id"] --> C["_add_internal_edge"] --> D{ "edge_id exists?" }
  D -->|Yes| I["Return False"]
  D -->|No| E[("Edge storage")]
  E --> F["edge_type_index + adjacency"]
  E --> G["mutation_callback"]
  H["load_from_file / clear"] --> E

Loading

High-Level Assessment

The following are alternative approaches to this PR:

1. Update-in-place on duplicate edge_id (upsert semantics)

• ➕ Supports refreshing metadata/weight on re-add without creating new edges
• ➕ Can simplify callers that naturally re-submit edges during re-ingest
• ➖ Ambiguous behavior for an API named add_edge; re-add would mutate existing state
• ➖ Harder to define and audit (ADD vs UPDATE) and can mask upstream duplication bugs

2. Deduplicate by (source_id, target_id, edge_type) instead of edge_id

• ➕ Aligns with a common mental model of 'one relationship per pair/type'
• ➖ Would collapse legitimate parallel edges that differ by attributes (e.g., confidence, validity window)
• ➖ Requires defining equality rules across optional fields; higher risk of breaking behavior

3. Make duplicate-edge behavior configurable (no-op vs upsert vs error)

• ➕ Allows different downstream expectations without changing core logic later
• ➖ Adds configuration complexity and expands the test matrix and documentation surface

Recommendation: Keep the current silent no-op dedupe keyed by the content-derived edge_id. It directly addresses the reported duplication/density/stat inflation and prevents phantom audit events, while still allowing parallel edges when attributes differ (validated by tests). If update semantics are needed later, add an explicit update_edge/upsert_edge API rather than overloading add_edge.

Files changed (2) +64 / -0

Bug fix (1) +9 / -0

context_graph.pyAdd _edge_index and short-circuit duplicate internal edge inserts +9/-0

Add _edge_index and short-circuit duplicate internal edge inserts

• Introduces an internal _edge_index (edge_id -> ContextEdge) and uses it in _add_internal_edge to return False before mutating edges, edge_type_index, adjacency, or emitting mutation callbacks when an existing edge_id is re-added. Ensures load_from_file() and clear() also clear _edge_index to avoid stale dedupe state after resets.

semantica/context/context_graph.py

Tests (1) +55 / -0

test_context.pyAdd regression tests for edge dedupe and reset behavior +55/-0

Add regression tests for edge dedupe and reset behavior

• Adds tests asserting add_edge is idempotent for identical edges, parallel edges with distinct attributes remain distinct, repeated build_from_entities_and_relationships doesn’t duplicate edges, and clear() resets the dedupe index.

tests/context/test_context.py

[qodo-free-for-open-source-projects]

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (0) 📎 Requirement gaps (0)

Great, no issues found!

Qodo reviewed your code and found no material issues that require review

Tip of the day

💡 Did you know, you can enable the Remediation agent and Qodo fixes findings in a dedicated fix PR

More tips ↗ | Customize Qodo ↗ | Qodo docs ↗