[PR #6360] [MERGED] fix(javascript): report path data records as route candidates #7306

New issue

Closed

opened 2026-11-01 21:46:55 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-11-01 21:46:55 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/morluto/rea/pull/6360
Author: @Nikhi00718
Created: 10/22/2026
Status: ✅ Merged
Merged: 10/22/2026
Merged by: @morluto

Base: main ← Head: fix/javascript-route-property-confidence

📝 Commits (2)

5f2a3e8 fix(javascript): report path data records as route candidates
1e31386 fix(javascript): scan each route record's keys once

📊 Changes

5 files changed (+177 additions, -35 deletions)

View changed files

📝 src/application/javascript/JavaScriptArtifactGraphFindings.ts (+3 -1)
📝 src/domain/javascript/javascriptRouteScan.ts (+60 -21)
📝 src/domain/javascript/javascriptStaticAnalysis.routeProperties.test.ts (+63 -12)
📝 src/domain/javascript/javascriptStaticAnalysisTypes.ts (+4 -1)
📝 tests/boundary/filesystem/javascript-applications/javascriptArtifactGraphEndpoints.test.ts (+47 -0)

📄 Description

Summary

Fixes #6349.

analyze_javascript_application reported every string path/route object property as an exact, observed route endpoint, whatever object held it. Data records such as { type: "set", path: "updatedAt", value: 1 } therefore became exact routes, while a real app.get("/real-route", ...) in the same file was only a low-confidence network candidate. On the npm package @agentmemory/agentmemory@0.9.30 this gave 13 exact routes, none of them real.

Problem and expected behavior

Smallest trigger:

const patch = [{ type: "set", path: "updatedAt", value: 1 }];
const worktree = { path: "", head: "" };

Before: two route endpoints, mechanism: "property:path", no interpretation, projected as discover-static-endpoint / ast-static-analysis / observed / exact.

Violated invariant: an observed, exact finding should be backed by syntax that establishes it. A property name alone does not establish a route.

After: both are kept, but as interpretation: "candidate", projected as classify-static-endpoint-candidate / static-relationship-inference / inferred / low, with the limitation "Only the property name suggests a route; the enclosing object has no router fields." Router records such as { path: "settings", component: Settings } stay exact.

Change and scope

javascriptRouteScan.ts: the key scan in #object moves into a routeRecordShape helper with no change in behavior, cached once per object (#shape) so repeated path/route keys do not rescan the object. #property now receives its parent, and marks the legacy property:path / property:route finding as a candidate unless the parent is a router record: component, element, Component, children, or a typed _$ record. That is the same shape #object already uses to choose automatic route roots, so route recognition has one rule instead of two.
JavaScriptArtifactGraphFindings.ts: route candidates get a route-specific limitation. The existing network-candidate text talks about method receivers, which does not apply here.
javascriptStaticAnalysisTypes.ts: the interpretation doc comment now covers both kinds of candidate.

Not included:

analyze_web_bundle uses the same route scan, but its route finding schema (value, mechanism, location) has no confidence field. Its legacy route list is unchanged. Adding a field there changes a public output contract and should be a separate decision.
Router records that only have Angular/Vue/React Router keys outside this set (redirectTo, loadChildren, loadComponent, redirect, lazy) are now candidates rather than exact. They are still reported. Widening the shared shape would also change which objects #object treats as automatic route roots, so I left that alone.

Contribution scope: an ordinary fix inside the existing static JavaScript analysis boundary; maintainer scope decision not applicable. #6349 asked which of two contracts to use. This PR implements option 1 (demote by shape), because it removes no findings. Happy to switch to option 2 (drop non-router properties) if you prefer.

Contract and boundary impact

Semantic owner and earliest changed stage: JavaScriptRouteScan in src/domain/javascript, during static analysis.
Contribution scope fit and linked maintainer decision: in scope (static analysis fix); option choice asked in #6349.
CLI and MCP/tool-catalog contract: no schema change. Some route endpoints gain interpretation: "candidate" and are projected with candidate evidence, an existing shape already used for network candidates.
Provider, bridge, target-format, or platform compatibility: none.
Evidence, artifact, provenance, or reconstruction contract: affected route endpoints move from observed/exact to inferred/low; endpoint identity keys include the interpretation, so their node ids change.
Target selection, authority, upstream behavior, and cleanup details: not applicable.
Generated metadata, package, or installation impact: none.

Expected effects (check known effects; use documentation-only on its own).
Unchecked items do not establish absence; leave uncertainties visible.

Read existing files or captured data.
Launch or attach to a process.
Inject instrumentation or run scripts/code.
Change process memory, registers, files, or configuration.
Contact network services.
Recover protected code or work around protections.
Documentation or usability only; no target interaction.
Not sure yet; investigation or maintainer guidance needed.

Evidence and regression coverage

Tests added or updated:
- src/domain/javascript/javascriptStaticAnalysis.routeProperties.test.ts: the existing case now uses router records (component/element) and still expects exact results. New cases cover children and typed _$ records staying exact, and patch ops, a { path, head } record and a bare { path: "/bare" } becoming candidates. A timed case checks that one object with 20,000 path keys stays under 2 s.
- tests/boundary/filesystem/javascript-applications/javascriptArtifactGraphEndpoints.test.ts: the graph projection gives a patch-op path inferred/low evidence with the route limitation, while a router record in the same file stays observed/exact.
Base reproduction: on fd5cd1b with the javascriptRouteScan.ts change stashed, the new candidate test fails (× reports path properties outside router records only as candidates); with the change it passes. The original report reproduced on rea-agents@6.7.0 (MCP) and 54f51f0 (#6349).
User-visible output: for the #6349 fixture, analyzeJavaScriptStaticSource(...).endpoints now returns updatedAt and "" with interpretation: "candidate"; /real-route is unchanged.
Remaining proof gaps: not re-run through the built CLI/MCP. Acceptance tests that start scripts/rea.mjs were not run (no build in this checkout). Real-world effect measured only on the #6349 fixture and the agentmemory bundle.

For evidence-bearing changes:

Observed, derived, and inferred claims remain distinguishable.
Artifact identity, source provenance, and failed attempts remain preserved.
Unsupported, incomplete, unavailable, or uncertain outcomes remain visible.

Validation performed

npx vitest run src/domain/javascript/javascriptStaticAnalysis.routeProperties.test.ts — 4 passed (repeated-keys case 564 ms).
Same file on 5f2a3e8 (before the per-object cache) — repeated-keys case failed at 82,255 ms.
Same file with the javascriptRouteScan.ts change stashed — 1 failed (the new candidate case), 2 passed.
npx vitest run tests/boundary/filesystem/javascript-applications/javascriptArtifactGraphEndpoints.test.ts — 3 passed.
npx vitest run src/domain/javascript src/application/javascript src/domain/webBundle tests/boundary/filesystem/javascript-applications, compared with the same run on base fd5cd1b — before: 4546 tests, 168 failed; after (1e31386): 4550 tests, 168 failed; no test changed from pass to fail. The 168 failures happen on base too: this Windows source checkout has no native addon ("REA Windows native controls are unavailable", see CONTRIBUTING "Windows source checkouts"), plus 4 ENOENT and 1 cancellation case in the same files.
npm run check:fast (typecheck, lint) — 2 tasks successful.
npx oxfmt --check on the changed files — correct format.

Compatibility, safety, and release

Breaking changes or migration steps: none in schema. Callers that treated every property:path route as exact will now see some of them as candidates with inferred/low evidence, and those endpoints get new node ids.
Real Hopper/Ghidra, browser, or OS coverage: not applicable; Windows 11, Node v24.15.0.
Package or release metadata impact: none.
Security, privacy, process, or containment review: no new reads, execution, or network access.

Author readiness

Human owner: @Nikhi00718

I reviewed the final diff and relevant existing code, understand the change and its edge cases, and take responsibility for the complete PR, including AI-assisted work.
I checked the evidence above, confirmed it supports the current change, and recorded meaningful validation gaps.
I declare this PR ready for initial maintainer review.

Before merge

This PR addresses a concrete problem or an agreed enhancement.
Material scope and authority questions are resolved, and any required maintainer scope decision covers the final diff.
The PR has one focused outcome and the title follows type(scope): outcome.
Related issue is linked, or the reason for not linking one is stated above.
Tests cover changed observable behavior and meaningful failure paths.
Owning docs, contracts, and generated metadata are updated where needed.
User-visible CLI/MCP changes include representative output.
I checked the final diff for secrets, unrelated cleanup, and unsupported claims.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/morluto/rea/pull/6360 **Author:** [@Nikhi00718](https://github.com/Nikhi00718) **Created:** 10/22/2026 **Status:** ✅ Merged **Merged:** 10/22/2026 **Merged by:** [@morluto](https://github.com/morluto) **Base:** `main` ← **Head:** `fix/javascript-route-property-confidence` --- ### 📝 Commits (2) - [`5f2a3e8`](https://github.com/morluto/rea/commit/5f2a3e89cf23583ebf85b3546ae49f44db633c42) fix(javascript): report path data records as route candidates - [`1e31386`](https://github.com/morluto/rea/commit/1e31386e12bb42fe6c08896cf72f1c69a2ba8f05) fix(javascript): scan each route record's keys once ### 📊 Changes **5 files changed** (+177 additions, -35 deletions) <details> <summary>View changed files</summary> 📝 `src/application/javascript/JavaScriptArtifactGraphFindings.ts` (+3 -1) 📝 `src/domain/javascript/javascriptRouteScan.ts` (+60 -21) 📝 `src/domain/javascript/javascriptStaticAnalysis.routeProperties.test.ts` (+63 -12) 📝 `src/domain/javascript/javascriptStaticAnalysisTypes.ts` (+4 -1) 📝 `tests/boundary/filesystem/javascript-applications/javascriptArtifactGraphEndpoints.test.ts` (+47 -0) </details> ### 📄 Description ## Summary Fixes #6349. `analyze_javascript_application` reported every string `path`/`route` object property as an exact, observed `route` endpoint, whatever object held it. Data records such as `{ type: "set", path: "updatedAt", value: 1 }` therefore became exact routes, while a real `app.get("/real-route", ...)` in the same file was only a low-confidence `network` candidate. On the npm package `@agentmemory/agentmemory@0.9.30` this gave 13 exact routes, none of them real. ## Problem and expected behavior Smallest trigger: ```js const patch = [{ type: "set", path: "updatedAt", value: 1 }]; const worktree = { path: "", head: "" }; ``` Before: two `route` endpoints, `mechanism: "property:path"`, no `interpretation`, projected as `discover-static-endpoint` / `ast-static-analysis` / `observed` / `exact`. Violated invariant: an observed, exact finding should be backed by syntax that establishes it. A property name alone does not establish a route. After: both are kept, but as `interpretation: "candidate"`, projected as `classify-static-endpoint-candidate` / `static-relationship-inference` / `inferred` / `low`, with the limitation "Only the property name suggests a route; the enclosing object has no router fields." Router records such as `{ path: "settings", component: Settings }` stay exact. ## Change and scope - `javascriptRouteScan.ts`: the key scan in `#object` moves into a `routeRecordShape` helper with no change in behavior, cached once per object (`#shape`) so repeated `path`/`route` keys do not rescan the object. `#property` now receives its parent, and marks the legacy `property:path` / `property:route` finding as a candidate unless the parent is a router record: `component`, `element`, `Component`, `children`, or a typed `_$` record. That is the same shape `#object` already uses to choose automatic route roots, so route recognition has one rule instead of two. - `JavaScriptArtifactGraphFindings.ts`: route candidates get a route-specific limitation. The existing network-candidate text talks about method receivers, which does not apply here. - `javascriptStaticAnalysisTypes.ts`: the `interpretation` doc comment now covers both kinds of candidate. Not included: - `analyze_web_bundle` uses the same route scan, but its route finding schema (`value`, `mechanism`, `location`) has no confidence field. Its legacy route list is unchanged. Adding a field there changes a public output contract and should be a separate decision. - Router records that only have Angular/Vue/React Router keys outside this set (`redirectTo`, `loadChildren`, `loadComponent`, `redirect`, `lazy`) are now candidates rather than exact. They are still reported. Widening the shared shape would also change which objects `#object` treats as automatic route roots, so I left that alone. Contribution scope: an ordinary fix inside the existing static JavaScript analysis boundary; maintainer scope decision not applicable. #6349 asked which of two contracts to use. This PR implements option 1 (demote by shape), because it removes no findings. Happy to switch to option 2 (drop non-router properties) if you prefer. ## Contract and boundary impact - Semantic owner and earliest changed stage: `JavaScriptRouteScan` in `src/domain/javascript`, during static analysis. - Contribution scope fit and linked maintainer decision: in scope (static analysis fix); option choice asked in #6349. - CLI and MCP/tool-catalog contract: no schema change. Some `route` endpoints gain `interpretation: "candidate"` and are projected with candidate evidence, an existing shape already used for network candidates. - Provider, bridge, target-format, or platform compatibility: none. - Evidence, artifact, provenance, or reconstruction contract: affected route endpoints move from observed/exact to inferred/low; endpoint identity keys include the interpretation, so their node ids change. - Target selection, authority, upstream behavior, and cleanup details: not applicable. - Generated metadata, package, or installation impact: none. Expected effects (check known effects; use documentation-only on its own). Unchecked items do not establish absence; leave uncertainties visible. - [x] Read existing files or captured data. - [ ] Launch or attach to a process. - [ ] Inject instrumentation or run scripts/code. - [ ] Change process memory, registers, files, or configuration. - [ ] Contact network services. - [ ] Recover protected code or work around protections. - [ ] Documentation or usability only; no target interaction. - [ ] Not sure yet; investigation or maintainer guidance needed. ## Evidence and regression coverage - Tests added or updated: - `src/domain/javascript/javascriptStaticAnalysis.routeProperties.test.ts`: the existing case now uses router records (`component`/`element`) and still expects exact results. New cases cover `children` and typed `_$` records staying exact, and patch ops, a `{ path, head }` record and a bare `{ path: "/bare" }` becoming candidates. A timed case checks that one object with 20,000 `path` keys stays under 2 s. - `tests/boundary/filesystem/javascript-applications/javascriptArtifactGraphEndpoints.test.ts`: the graph projection gives a patch-op `path` inferred/low evidence with the route limitation, while a router record in the same file stays observed/exact. - Base reproduction: on `fd5cd1b` with the `javascriptRouteScan.ts` change stashed, the new candidate test fails (`× reports path properties outside router records only as candidates`); with the change it passes. The original report reproduced on `rea-agents@6.7.0` (MCP) and `54f51f0` (#6349). - User-visible output: for the #6349 fixture, `analyzeJavaScriptStaticSource(...).endpoints` now returns `updatedAt` and `""` with `interpretation: "candidate"`; `/real-route` is unchanged. - Remaining proof gaps: not re-run through the built CLI/MCP. Acceptance tests that start `scripts/rea.mjs` were not run (no build in this checkout). Real-world effect measured only on the #6349 fixture and the agentmemory bundle. For evidence-bearing changes: - [x] Observed, derived, and inferred claims remain distinguishable. - [x] Artifact identity, source provenance, and failed attempts remain preserved. - [x] Unsupported, incomplete, unavailable, or uncertain outcomes remain visible. ## Validation performed - `npx vitest run src/domain/javascript/javascriptStaticAnalysis.routeProperties.test.ts` — 4 passed (repeated-keys case 564 ms). - Same file on 5f2a3e8 (before the per-object cache) — repeated-keys case failed at 82,255 ms. - Same file with the `javascriptRouteScan.ts` change stashed — 1 failed (the new candidate case), 2 passed. - `npx vitest run tests/boundary/filesystem/javascript-applications/javascriptArtifactGraphEndpoints.test.ts` — 3 passed. - `npx vitest run src/domain/javascript src/application/javascript src/domain/webBundle tests/boundary/filesystem/javascript-applications`, compared with the same run on base `fd5cd1b` — before: 4546 tests, 168 failed; after (1e31386): 4550 tests, 168 failed; no test changed from pass to fail. The 168 failures happen on base too: this Windows source checkout has no native addon ("REA Windows native controls are unavailable", see CONTRIBUTING "Windows source checkouts"), plus 4 ENOENT and 1 cancellation case in the same files. - `npm run check:fast` (typecheck, lint) — 2 tasks successful. - `npx oxfmt --check` on the changed files — correct format. ## Compatibility, safety, and release - Breaking changes or migration steps: none in schema. Callers that treated every `property:path` route as exact will now see some of them as candidates with inferred/low evidence, and those endpoints get new node ids. - Real Hopper/Ghidra, browser, or OS coverage: not applicable; Windows 11, Node v24.15.0. - Package or release metadata impact: none. - Security, privacy, process, or containment review: no new reads, execution, or network access. ## Author readiness Human owner: @Nikhi00718 - [x] I reviewed the final diff and relevant existing code, understand the change and its edge cases, and take responsibility for the complete PR, including AI-assisted work. - [x] I checked the evidence above, confirmed it supports the current change, and recorded meaningful validation gaps. - [x] I declare this PR ready for initial maintainer review. ## Before merge - [ ] This PR addresses a concrete problem or an agreed enhancement. - [ ] Material scope and authority questions are resolved, and any required maintainer scope decision covers the final diff. - [ ] The PR has one focused outcome and the title follows `type(scope): outcome`. - [ ] Related issue is linked, or the reason for not linking one is stated above. - [ ] Tests cover changed observable behavior and meaningful failure paths. - [ ] Owning docs, contracts, and generated metadata are updated where needed. - [ ] User-visible CLI/MCP changes include representative output. - [ ] I checked the final diff for secrets, unrelated cleanup, and unsupported claims. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-11-01 21:46:55 +00:00