Add config key graph parity fixture

Author: skooch

docs/gap-analysis.md

@@ -72,7 +72,7 @@ Intentionally deferred after Phase 7:
 - The remaining MCP work outside the completed daily-use slice, especially fuller Cypher parity.
 - Full Cypher parity beyond the broader day-to-day query subset now supporting node/edge reads, filtering, sorting, and counts.
 - Deeper usage/type-reference extraction parity and broader cross-language semantics beyond the current target daily-use slice.
-- Richer decorator/enrichment follow-ons, decorator-based route detection, `HANDLES`/`DATA_FLOWS` edge creation, and full config normalization surface. (Git-history coupling is now implemented; config-linking and route nodes are partially implemented.)
+- Richer decorator/enrichment follow-ons, broader route detection, and broader config normalization/linking surface. (Git-history coupling is now implemented; route nodes and config-linking both have strict shared graph-model fixture slices.)
 - Broader installer/self-update behavior beyond the current source-build-friendly Codex CLI / Claude Code support.
 - Idle-store and session-lifecycle extras beyond the current persistent-store + watcher runtime model.
 
@@ -127,7 +127,7 @@ Deferred or optional future slices:
   - git-history coupling — now implemented (subprocess `git log`, `FILE_CHANGES_WITH` edges)
   - long-tail edges — now implemented: `THROWS`/`RAISES` (JS/TS/TSX throw statements), decorator-backed `HANDLES`, and route-linked `DATA_FLOWS`; remaining or out-of-scope gaps: `OVERRIDE` (Go-only), `WRITES`/`READS` (not proven original-overlap by the current C reference fixture)
   - route nodes — partially implemented (stub and concrete URL/path/topic `Route` nodes, verified decorator-backed `Route`/`HANDLES`, strict shared route-linked `DATA_FLOWS`, and strict shared `ASYNC_CALLS`; broader framework coverage still open)
-  - config-linking — partially implemented (Strategy 1 key-symbol + Strategy 2 dependency-import; full config normalization surface not yet ported)
+  - config-linking — partially implemented (Strategy 1 key-symbol + Strategy 2 dependency-import; the first strict shared key-symbol normalization fixture now proves raw-key preservation and `CONFIGURES` query visibility, while dependency-import closure remains open)
   - richer decorator/enrichment promotion
   - current entrypoint: [graph-model-parity-plan.md](/Users/skooch/projects/codebase-memory-zig/docs/plans/in-progress/graph-model-parity-plan.md)
 - Productization beyond the current contract:
@@ -428,10 +428,10 @@ The Zig store has the schema (tables + indexes + pragmas) and opens in-memory DB
 | `pass_parallel` | 1,427 | MISSING | Thread pool orchestration |
 | `pass_similarity` | 505 (minhash.c) | MISSING | MinHash near-clone detection |
 | `pass_gitdiff` | ~200 | MISSING | Git diff → changed files/hunks |
-| `pass_route_nodes` | 742 | PARTIAL (stub and concrete URL/path Route nodes, verified decorator-backed `Route`/`HANDLES`, and Zig route-linked `DATA_FLOWS`) | HTTP route node creation, first handler association slice, and first data-flow bridge |
+| `pass_route_nodes` | 742 | PARTIAL (stub and concrete URL/path/topic Route nodes, verified decorator-backed `Route`/`HANDLES`, strict shared route-linked `DATA_FLOWS`, and strict shared `ASYNC_CALLS`) | HTTP/async route node creation, first handler association slice, and first data-flow bridge |
 | `pass_tests` | 285 | WORKS for the shared Python `TESTS` / `TESTS_FILE` slice | Test file/function tagging now verified on the local parity fixture; broader language breadth stays follow-on work |
 | `pass_enrichment` | ~200 | MISSING (deferred) | Decorator tag enrichment |
-| `pass_configlink` | ~200 | PARTIAL (Strategy 1 key-symbol + Strategy 2 dependency-import) | Config-code linking |
+| `pass_configlink` | ~200 | PARTIAL (Strategy 1 key-symbol + Strategy 2 dependency-import; first strict shared key-symbol fixture is locked) | Config-code linking |
 | `pass_githistory` | 514 | WORKS | Change coupling from git log |
 | `pipeline_incremental` | ~400 | MISSING (deferred) | Incremental re-indexing |
 

docs/plans/in-progress/graph-model-parity-plan.md

@@ -8,12 +8,16 @@ normalization surface.
 
 ## Current Phase
 Phase 3 route-linked data flow is complete for the first shared public fixture,
-and the first shared async route caller fixture is now locked:
+the first shared async route caller fixture is locked, and Phase 4 now has its
+first strict config-key fixture:
 decorator route extraction creates `Route` nodes and `HANDLES` edges,
 route-registration calls preserve enough call metadata to emit `HANDLES` when
 the handler reference resolves, and the shared Python route fixture now proves a
 strict C/Zig `DATA_FLOWS` row through a `GET` route. The async fixture proves a
 broker-specific `Route` and `ASYNC_CALLS` row for a local `celery.delay` topic.
+The config fixture proves raw key preservation, dash/camel normalization, the
+`maxConnections -> max-connections` `CONFIGURES` row, and no false link for a
+short one-token config key.
 
 ## Current Codebase State
 - Complete and verified: advanced trace modes, risk labels, include-tests
@@ -36,11 +40,15 @@ broker-specific `Route` and `ASYNC_CALLS` row for a local `celery.delay` topic.
 - Newly implemented for async route coverage: Zig accepts async broker topics,
   preserves broker names for route QNs/properties, and the `graph-model-async`
   fixture proves the shared `enqueue_users -> users.refresh` `ASYNC_CALLS` row.
+- Newly locked for config-link coverage: the `graph-model-config` fixture proves
+  the first shared key-symbol normalization contract with a YAML
+  `max-connections` key linked from a camelCase Python symbol while preserving
+  the raw config key in query-visible output.
 - Still missing for graph-model parity: broader framework route coverage and
-  the full config normalization/linking surface beyond the current key-symbol
-  and dependency-import strategies.
-- Current full Zig-vs-C harness baseline after this route slice:
-  `165` comparisons, `92` strict matches, `21` diagnostic-only comparisons,
+  broader config-link coverage beyond the first strict key-symbol fixture,
+  especially dependency-import mismatch closure.
+- Current full Zig-vs-C harness baseline after this config slice:
+  `172` comparisons, `95` strict matches, `22` diagnostic-only comparisons,
   `10` mismatches, and `cli_progress: match`. The graph-model-related
   mismatches still include the existing `graph-enrichment-config-deps` and
   `graph-enrichment-http-calls` query rows.
@@ -67,7 +75,7 @@ broker-specific `Route` and `ASYNC_CALLS` row for a local `celery.delay` topic.
 - Modify: `testdata/interop/manifest.json`
 - Create: `testdata/interop/graph-model/routes/`
 - Create: `testdata/interop/graph-model/config/`
-- Create: `docs/plans/in-progress/graph-model-parity-progress.md` after approval
+- Modify: `docs/plans/in-progress/graph-model-parity-progress.md`
 
 ## Phases
 
@@ -77,7 +85,7 @@ broker-specific `Route` and `ASYNC_CALLS` row for a local `celery.delay` topic.
 - [x] Define fixture-backed acceptance rules for first-slice `HTTP_CALLS` route callers and route-linked `DATA_FLOWS` in Zig.
 - [x] Define fixture-backed acceptance rules for strict shared C/Zig `DATA_FLOWS`.
 - [x] Define fixture-backed acceptance rules for strict shared C/Zig `ASYNC_CALLS`.
-- [ ] Define fixture-backed acceptance rules for config normalization/linking.
+- [x] Define fixture-backed acceptance rules for first-slice config normalization/linking.
 - [x] Add a minimal Python route fixture under `testdata/interop/graph-model/routes/`.
 - [ ] Add minimal JavaScript/TypeScript route fixtures once the shared C/Zig public behavior is established for those registrations.
 - **Status:** partially complete
@@ -104,10 +112,11 @@ broker-specific `Route` and `ASYNC_CALLS` row for a local `celery.delay` topic.
 - **Status:** complete for the first async route caller slice
 
 ### Phase 4: Finish Config-Link Normalization
-- [ ] Extend config-key extraction and matching only for original-overlap config patterns with fixture evidence.
-- [ ] Preserve raw config keys while adding the minimum normalized lookup metadata needed for stable links.
-- [ ] Add regression coverage for config-key normalization, dependency-import matching, deduplication, and query visibility.
-- **Status:** pending approval
+- [x] Add the first strict shared key-symbol fixture for dash/camel config normalization.
+- [x] Verify raw config keys remain query-visible while normalized matching creates stable links.
+- [ ] Extend config-key extraction and matching only for additional original-overlap config patterns with fixture evidence.
+- [ ] Add regression coverage for dependency-import matching, deduplication, and broader query visibility.
+- **Status:** in progress
 
 ### Phase 5: Verify and Reclassify
 - [x] Run `zig build`, `zig build test`, `bash scripts/run_interop_alignment.sh --zig-only`, and focused graph-model fixture queries.

docs/plans/in-progress/graph-model-parity-progress.md

@@ -39,12 +39,24 @@
   - Added strict manifest assertions for `__route__celery__users.refresh` and `enqueue_users -> users.refresh` over `ASYNC_CALLS`.
   - Regenerated `graph-model-async` golden coverage.
 
+### Phase 4: Finish Config-Link Normalization
+- **Status:** in progress for the first shared key-symbol slice
+- Actions:
+  - Re-read the original C `pass_configlink` behavior and compared it with Zig's `runConfigLinkPass` and `normalizeConfigName`.
+  - Probed a focused YAML/Python fixture against both implementations and confirmed the shared public rows before adding repo fixtures.
+  - Added `testdata/interop/graph-model/config/` with `maxConnections`, `max_connections`, `max-connections`, and `short`.
+  - Added strict manifest assertions that prove the raw `max-connections` config key, the normalized `maxConnections -> max-connections` `CONFIGURES` row, and no false `CONFIGURES` row for `short`.
+  - Regenerated `graph-model-config` golden coverage.
+
 ### Verification
 - `zig build` -> passed
 - `zig build test` -> passed
-- `bash scripts/run_interop_alignment.sh --update-golden` -> passed, 21/21 golden snapshots updated
-- `bash scripts/run_interop_alignment.sh --zig-only` -> passed, 21/21 golden comparison
-- `bash scripts/run_interop_alignment.sh` -> passed with 21 fixtures, 165 comparisons, 92 strict matches, 21 diagnostic-only comparisons, 10 known mismatches, and `cli_progress: match`
+- `python3 -m json.tool testdata/interop/manifest.json >/dev/null` -> passed
+- `bash scripts/run_interop_alignment.sh --update-golden` -> passed, 22/22 golden snapshots updated
+- `bash scripts/run_interop_alignment.sh --zig-only` -> passed, 22/22 golden comparison
+- `bash scripts/run_interop_alignment.sh` -> passed with 22 fixtures, 172 comparisons, 95 strict matches, 22 diagnostic-only comparisons, 10 known mismatches, and `cli_progress: match`
+- `git diff --check` -> passed
+- `command -v zlint` -> blocked; `zlint` is not installed in this environment
 
 ## Errors
 | Timestamp | Error | Attempt | Resolution |

docs/port-comparison.md

@@ -132,7 +132,7 @@ It is intentionally not a wish list. It describes:
 | LSP hybrid type resolution | Present for Go/C/C++ in original | Not implemented | `Deferred` | No | The original README explicitly calls out LSP-style hybrid type resolution; Zig does not ship that layer. |
 | Test tagging pass | Yes | Yes for the verified shared Python fixture slice | `Near parity` | Yes | Zig now derives `TESTS` and `TESTS_FILE` from shared filename and call-edge rules on the parity fixture without reopening broader language-specific enrichment work. |
 | Git history coupling | Yes | Implemented | `Near parity` | Yes | Zig pass uses subprocess `git log` (no libgit2); creates `FILE_CHANGES_WITH` edges with `co_changes` and `coupling_score` properties. |
-| Config linking / config normalization | Yes | Partial | `Partial` | No | Zig now implements Strategy 1 (key-symbol) and Strategy 2 (dependency-import matching) but not the full config normalization surface. |
+| Config linking / config normalization | Yes | Partial | `Partial` | No | Zig now implements Strategy 1 (key-symbol) and Strategy 2 (dependency-import matching); the first strict shared key-symbol normalization fixture locks raw-key query visibility and `maxConnections -> max-connections`, but dependency-import mismatch closure and broader normalization coverage remain open. |
 | Route-node creation / cross-service graph | Yes | Partial | `Partial` | No | Zig now classifies call edges as `HTTP_CALLS`/`ASYNC_CALLS`, creates stub and concrete URL/path/topic `Route` nodes, emits verified decorator-backed `HANDLES`, and has strict shared route-linked `DATA_FLOWS` plus async topic fixtures. Broader framework coverage remains open. |
 | Infra scanning (`Docker`, `K8s`, Terraform, etc.) | Yes in the original codebase | Not ported | `Cut` | No | Zig intentionally excludes the infra-scan family from the current port target. |
 | OTLP traces | Stubbed | Not ported | `N/A` | No | Not a meaningful implemented-vs-implemented gap. |
@@ -146,7 +146,7 @@ This section compares what kinds of graph entities the two systems are built to
 | Core code graph (`Project`, `Folder`, `File`, `Module`, `Class`, `Function`, `Method`, `Interface`, `Enum`) | Yes | Yes | `Near parity` | Yes | These are the backbone of the shipped Zig graph. |
 | Core code edges (`CONTAINS_*`, `DEFINES`, `DEFINES_METHOD`, `CALLS`, `USAGE`) | Yes | Yes | `Near parity` | Yes | All are part of the current Zig daily-use contract. |
 | `SIMILAR_TO` | Yes | Yes | `Near parity` | Yes | Landed in Zig Phase 6. |
-| Shared `CONFIGURES` contract | Yes | Yes for the verified shared fixture slice | `Near parity` | Yes | The parity fixtures now compare the overlapping `CONFIGURES` row directly; broader config-link systems remain deferred elsewhere. |
+| Shared `CONFIGURES` contract | Yes | Yes for the verified shared fixture slices | `Near parity` | Yes | The parity fixtures now compare overlapping `CONFIGURES` rows directly, including the graph-model key-symbol normalization case; broader config-link systems remain deferred elsewhere. |
 | Internal serving architecture | Graph-centric serving path | Hybrid internal serving path: SQLite graph core, FTS5 lexical index, optional SCIP sidecar overlay, and query router | `Near parity` | Yes | This is an internal implementation improvement in the Zig port; it deliberately preserves the interoperable MCP surface rather than creating a new client contract. |
 | Route graph (`Route`, `HTTP_CALLS`, `ASYNC_CALLS`, `HANDLES`, route-linked data flows) | Yes | Partial | `Partial` | No | Zig now creates `HTTP_CALLS` and `ASYNC_CALLS` edges, concrete URL/path/topic `Route` nodes, verified decorator-backed `HANDLES`, strict shared route-linked `DATA_FLOWS`, and strict shared async topic caller rows. Broader framework coverage remains open. |
 | Resource / infra graph (`Resource`, K8s/Kustomize entities) | Yes | Not shipped | `Cut` | No | Intentionally outside the Zig scope. |
@@ -226,7 +226,7 @@ If someone asks “what still separates the Zig port from the original?”, the
 | No full Cypher parity | Some advanced graph query patterns remain C-only. |
 | Partial route / cross-service graph stack | Zig now emits verified decorator-backed `HANDLES`, strict shared route-linked `DATA_FLOWS`, and strict shared async topic caller rows, but broader framework coverage remains open. |
 | No LSP-assisted hybrid resolution | Some higher-fidelity call/type resolution paths remain original-only. |
-| No full config normalization | Git-history coupling is implemented, config linking has dependency-import matching, decorator-backed `HANDLES` is verified, strict shared route-linked `DATA_FLOWS` is covered, and the first async topic route fixture is covered. The remaining graph-model gap is broader framework route coverage and the full config normalization surface. |
+| No full config normalization | Git-history coupling is implemented, config linking has dependency-import matching and a strict shared key-symbol normalization fixture, decorator-backed `HANDLES` is verified, strict shared route-linked `DATA_FLOWS` is covered, and the first async topic route fixture is covered. The remaining graph-model gap is broader framework route coverage plus dependency-import and broader config-normalization closure. |
 | No UI subsystem | The original can run a graph visualization UI; the Zig port intentionally cannot. |
 | Much narrower installer ecosystem | The original configures 10 agents plus hooks/instructions; Zig currently configures 2 agents and only MCP entries. |
 

docs/zig-port-plan.md

@@ -567,9 +567,9 @@ Current state: the broader daily-use MCP surface is now live, the runtime uses a
 - [x] Git history pass (change coupling)
 - [ ] Route node creation and handler/data-flow links (partial: stub and concrete URL/path/topic `Route` nodes, verified decorator-backed `HANDLES`, strict shared route-linked `DATA_FLOWS`, and strict shared `ASYNC_CALLS` are implemented; broader framework coverage remains open)
 - [x] Test tagging
-- [ ] Config-code linking (partial: key-symbol and dependency-import strategies are implemented)
+- [ ] Config-code linking (partial: key-symbol and dependency-import strategies are implemented; first strict shared key-symbol normalization fixture is locked)
 - [ ] Decorator enrichment
-Current state: M4 is no longer a blank backlog. Git-history coupling, test tagging, route stubs, concrete URL/path/topic route callers, decorator-backed `HANDLES`, strict shared route-linked `DATA_FLOWS`, strict shared `ASYNC_CALLS`, service-pattern call classification, partial config linking, and long-tail `THROWS` / `RAISES` edges are implemented. The active graph-model parity entrypoint is [graph-model-parity-plan.md](/Users/skooch/projects/codebase-memory-zig/docs/plans/in-progress/graph-model-parity-plan.md), which tracks the remaining broader route handler coverage and config-normalization work.
+Current state: M4 is no longer a blank backlog. Git-history coupling, test tagging, route stubs, concrete URL/path/topic route callers, decorator-backed `HANDLES`, strict shared route-linked `DATA_FLOWS`, strict shared `ASYNC_CALLS`, service-pattern call classification, partial config linking with a strict shared key-symbol normalization fixture, and long-tail `THROWS` / `RAISES` edges are implemented. The active graph-model parity entrypoint is [graph-model-parity-plan.md](/Users/skooch/projects/codebase-memory-zig/docs/plans/in-progress/graph-model-parity-plan.md), which tracks the remaining broader route handler coverage and dependency-import/config-normalization work.
 
 ## Completion Summary
 

testdata/interop/golden/graph-model-config.json

@@ -0,0 +1,68 @@
+{
+  "delete_project": [],
+  "detect_changes_count": 0,
+  "fixture_id": "graph-model-config",
+  "get_architecture": [],
+  "get_code_snippet": [],
+  "get_graph_schema": [],
+  "index_repository": {
+    "edges_actual": 10,
+    "edges_min": 2,
+    "nodes_actual": 9,
+    "nodes_min": 4
+  },
+  "index_status": [],
+  "list_projects": [
+    "config"
+  ],
+  "manage_adr": [],
+  "query_graph": [
+    {
+      "columns": [
+        "n.name"
+      ],
+      "rows": [
+        [
+          "max-connections"
+        ]
+      ]
+    },
+    {
+      "columns": [
+        "a.name",
+        "b.name"
+      ],
+      "rows": [
+        [
+          "maxConnections",
+          "max-connections"
+        ]
+      ]
+    },
+    {
+      "columns": [
+        "a.name",
+        "b.name"
+      ],
+      "rows": []
+    }
+  ],
+  "search_code": [],
+  "search_graph": [],
+  "tools_list": [
+    "delete_project",
+    "detect_changes",
+    "get_architecture",
+    "get_code_snippet",
+    "get_graph_schema",
+    "index_repository",
+    "index_status",
+    "list_projects",
+    "manage_adr",
+    "query_graph",
+    "search_code",
+    "search_graph",
+    "trace_call_path"
+  ],
+  "trace_call_path": []
+}

testdata/interop/graph-model/config/app.py

@@ -0,0 +1,7 @@
+def maxConnections():
+    return 10
+
+
+def max_connections():
+    return 20
+

testdata/interop/graph-model/config/settings.yaml

@@ -0,0 +1,3 @@
+max-connections: 100
+short: 1
+