@opencodehub/ingestion 0.1.0

package/dist/pipeline/types.d.ts

@@ -0,0 +1,235 @@
+/**
+ * Shared types for the pipeline runner and phase modules.
+ *
+ * The pipeline is a static DAG: each phase declares its dependencies by name,
+ * the runner validates (duplicate / missing / cycle) and sorts phases
+ * topologically, then executes in order. Phases communicate via a typed
+ * output map keyed by phase name; each phase only sees outputs of phases it
+ * declared as dependencies (dep isolation prevents hidden coupling).
+ *
+ * The runner plus scan/structure/parse are the core; markdown, routes,
+ * tools, orm, crossFile, mro, communities, and processes extend the set
+ * without touching this file.
+ */
+import type { CodeRelation, GraphNode, KnowledgeGraph } from "@opencodehub/core-types";
+/** Runtime context supplied to each phase. */
+export interface PipelineContext {
+    /** Absolute repository root path. */
+    readonly repoPath: string;
+    /** Normalized pipeline options. */
+    readonly options: PipelineOptions;
+    /** Single shared mutable graph accumulator populated across phases. */
+    readonly graph: KnowledgeGraph;
+    /** Outputs emitted by phases that completed before the caller. */
+    readonly phaseOutputs: ReadonlyMap<string, unknown>;
+    /** Optional progress sink; thrown errors from the callback are swallowed. */
+    readonly onProgress?: (ev: ProgressEvent) => void;
+}
+/**
+ * Minimal projection of a prior-run graph sufficient for the incremental-scope
+ * phase to compute the import-closure walk. We intentionally keep this
+ * narrower than a full {@link KnowledgeGraph} so callers can materialise
+ * it cheaply from persisted storage (DuckDB rows, sidecar JSON, etc.) without
+ * hydrating every node/edge kind in the graph.
+ *
+ * All arrays carry repo-relative posix paths (matching `ScannedFile.relPath`).
+ * The inputs must be pre-deduplicated; no validation is performed downstream.
+ */
+export interface PreviousGraph {
+    /**
+     * Relative file paths present in the prior index, each paired with the
+     * content-addressed sha256 captured on the last successful scan. Paths
+     * absent from the current scan are silently ignored; paths whose hash
+     * differs from the current scan are treated as "changed" and seed the
+     * closure walk.
+     */
+    readonly files: readonly {
+        readonly relPath: string;
+        readonly contentSha: string;
+    }[];
+    /**
+     * IMPORTS edges at file granularity — `importer` depends on `target`. The
+     * shape mirrors the edges written by the parse phase (`File -> File`). Edges
+     * whose endpoints are outside `files` are ignored by the closure walk.
+     */
+    readonly importEdges: readonly {
+        readonly importer: string;
+        readonly target: string;
+    }[];
+    /**
+     * Heritage neighbourhoods per file. Used to grow the closure by one hop of
+     * EXTENDS/IMPLEMENTS ancestors + descendants so an edit to a base class
+     * refreshes its subtype MROs. Empty array is valid (no heritage known).
+     */
+    readonly heritageEdges: readonly {
+        readonly childFile: string;
+        readonly parentFile: string;
+    }[];
+    /**
+     * Full prior-run node set. When present, the four active incremental
+     * phases (crossFile/mro/communities/processes) carry forward any
+     * non-closure node from the prior run so incremental mode produces a
+     * byte-identical graph hash vs a fresh full run at the same commit.
+     * `undefined` keeps the passive contract where incremental-scope emits
+     * the closure hint but the downstream phases re-compute from scratch.
+     */
+    readonly nodes?: readonly GraphNode[];
+    /**
+     * Full prior-run edge set. Paired with {@link nodes}; see its docstring
+     * for activation semantics. Active incremental phases use this snapshot
+     * to carry forward edges whose both endpoints lie outside the closure
+     * unchanged. Edges whose endpoints are now missing from the current
+     * graph are silently dropped on carry-forward.
+     */
+    readonly edges?: readonly CodeRelation[];
+}
+/** Operator knobs accepted by the top-level `runIngestion` entry. */
+export interface PipelineOptions {
+    /** Re-run phases even if a prior run's artifacts look current. */
+    readonly force?: boolean;
+    /**
+     * Snapshot of the prior index, supplied when the orchestrator is invoked in
+     * incremental mode. When present the {@link INCREMENTAL_SCOPE_PHASE_NAME}
+     * phase walks the IMPORTS closure and emits the set of files that need
+     * re-processing. When absent (or when `--force` is set), the phase declares
+     * a full reindex. See `packages/ingestion/src/pipeline/phases/incremental-scope.ts`.
+     */
+    readonly incrementalFrom?: PreviousGraph;
+    /**
+     * Refuse to spawn any child that hits the network. Phase authors are
+     * expected to honor this by avoiding network IO; the orchestrator does
+     * not sandbox — it simply threads the flag through.
+     */
+    readonly offline?: boolean;
+    /** Extra chatter on the progress callback (implementation-defined). */
+    readonly verbose?: boolean;
+    /** Skip `git` invocations during scan when we are not in a git repo. */
+    readonly skipGit?: boolean;
+    /** Per-file byte cap; files larger than this are skipped with a warning. */
+    readonly byteCapPerFile?: number;
+    /** Overall file cap; scan stops accepting files once this is reached. */
+    readonly maxTotalFiles?: number;
+    /**
+     * Populate the embeddings table. Requires `codehub setup --embeddings` to
+     * have installed weights on disk. When false (default) the embeddings
+     * phase returns an empty result without loading ONNX.
+     */
+    readonly embeddings?: boolean;
+    /**
+     * Which embedder variant to load when `embeddings=true`. Defaults to
+     * `fp32`. Ignored when `embeddings=false`.
+     */
+    readonly embeddingsVariant?: "fp32" | "int8";
+    /** Override the model directory used by the embedder. */
+    readonly embeddingsModelDir?: string;
+    /**
+     * Hierarchical tiers to emit when `embeddings=true` (P03). Defaults to
+     * `["symbol"]` so v1.0 behaviour is preserved. Pass
+     * `["symbol", "file", "community"]` to emit all three tiers; the phase
+     * de-dupes and normalizes the order. Unknown tier names are silently
+     * filtered at the TS type level.
+     */
+    readonly embeddingsGranularity?: readonly ("symbol" | "file" | "community")[];
+    /**
+     * Number of ONNX embedder workers to run in parallel. `undefined` or
+     * `<= 1` preserves the legacy in-process path (single main-thread
+     * embedder, no Piscina overhead). Values >= 2 spin up a worker pool of
+     * independent OnnxEmbedder instances. Each worker holds its own
+     * session (~300 MB RSS on fp32), so sizing above `os.cpus().length - 1`
+     * buys nothing and risks memory pressure. Ignored when the HTTP
+     * backend is selected via `CODEHUB_EMBEDDING_URL`.
+     */
+    readonly embeddingsWorkers?: number;
+    /**
+     * Batch size for cross-node inference. The embeddings phase groups
+     * chunks across symbols/files/communities into a single
+     * `embedder.embedBatch()` call; this knob controls that batch size.
+     * Defaults to 32 (see `DEFAULT_EMBEDDING_BATCH_SIZE` in the phase
+     * module). `1` restores the legacy one-node-per-call pattern.
+     */
+    readonly embeddingsBatchSize?: number;
+    /**
+     * When `true`, the SBOM phase emits `.codehub/sbom.cyclonedx.json` and
+     * `.codehub/sbom.spdx.json` from Dependency nodes. When `false` (the
+     * default), the phase is a no-op. Toggled via the `codehub analyze
+     * --sbom` flag.
+     */
+    readonly sbom?: boolean;
+    /**
+     * When `true` (default), SBOM emission uses a fixed epoch timestamp and
+     * a content-addressed document namespace so byte-identical SBOMs are
+     * produced for byte-identical Dependency sets. This keeps downstream
+     * attestations and diff-based review stable. Setting `false` swaps the
+     * epoch for `Date.now()` (floored to the second) and the namespace for
+     * a random UUID; reserved for ad-hoc local use.
+     */
+    readonly reproducibleSbom?: boolean;
+    /**
+     * When `true`, the coverage phase probes for lcov / cobertura / jacoco /
+     * coverage.py artifacts and annotates `File` nodes with
+     * `coveragePercent` + `coveredLines`. Default `false` so analyze stays
+     * quiet in repos without a coverage pipeline. Toggled via the
+     * `codehub analyze --coverage` flag.
+     */
+    readonly coverage?: boolean;
+    /**
+     * When `true`, the `summarize` phase walks callable symbols, computes a
+     * content hash from their source span, and issues a Bedrock summarize
+     * call for cache misses. Gated OFF by default because the phase spends
+     * real money. The `offline` flag always wins — when `offline === true`
+     * the phase is a no-op regardless of this flag.
+     */
+    readonly summaries?: boolean | undefined;
+    /**
+     * Upper bound on the number of Bedrock summarize calls per pipeline run.
+     * Defaults to `0`, which runs the phase in dry-run mode: it enumerates
+     * eligible symbols and reports `wouldHaveSummarized` without contacting
+     * Bedrock. Set to a positive integer (e.g. 10) to actually summarize a
+     * bounded subset. Ignored when `summaries !== true`. The CLI resolves
+     * `--max-summaries auto` to a concrete integer before calling into the
+     * pipeline, so this field is always numeric inside the pipeline.
+     */
+    readonly maxSummariesPerRun?: number | undefined;
+    /**
+     * Override the Bedrock model id used by the summarize phase. When
+     * undefined, the phase uses `DEFAULT_MODEL_ID` from
+     * `@opencodehub/summarizer`.
+     */
+    readonly summaryModel?: string | undefined;
+    /**
+     * When `true`, detectors that pattern-match on receiver identifiers
+     * skip heuristic-only matches entirely — edges are emitted only when a
+     * receiver's module origin was confirmed via the import graph or
+     * ts-morph. Exposed by the `codehub analyze --strict-detectors` flag
+     * (DET-O-001).
+     */
+    readonly strictDetectors?: boolean;
+}
+/** Lightweight progress event emitted during pipeline execution. */
+export interface ProgressEvent {
+    readonly phase: string;
+    readonly kind: "start" | "end" | "note" | "warn" | "error";
+    readonly pct?: number;
+    readonly message?: string;
+    readonly elapsedMs?: number;
+}
+/** Declarative phase description. */
+export interface PipelinePhase<TOutput = unknown> {
+    /** Unique phase identifier — must be distinct within a DAG. */
+    readonly name: string;
+    /** Other phase names that must complete before this one runs. */
+    readonly deps: readonly string[];
+    /**
+     * Execute the phase against the shared context. `deps` is pre-filtered to
+     * only include outputs of phases this phase declared as dependencies.
+     */
+    run(ctx: PipelineContext, deps: ReadonlyMap<string, unknown>): Promise<TOutput>;
+}
+/** Per-phase summary returned by {@link runPipeline}. */
+export interface PhaseResult {
+    readonly name: string;
+    readonly output: unknown;
+    readonly durationMs: number;
+    readonly warnings: readonly string[];
+}
+//# sourceMappingURL=types.d.ts.map

package/dist/pipeline/types.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../../src/pipeline/types.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AAEH,OAAO,KAAK,EAAE,YAAY,EAAE,SAAS,EAAE,cAAc,EAAE,MAAM,yBAAyB,CAAC;AAEvF,8CAA8C;AAC9C,MAAM,WAAW,eAAe;IAC9B,qCAAqC;IACrC,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;IAC1B,mCAAmC;IACnC,QAAQ,CAAC,OAAO,EAAE,eAAe,CAAC;IAClC,uEAAuE;IACvE,QAAQ,CAAC,KAAK,EAAE,cAAc,CAAC;IAC/B,kEAAkE;IAClE,QAAQ,CAAC,YAAY,EAAE,WAAW,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;IACpD,6EAA6E;IAC7E,QAAQ,CAAC,UAAU,CAAC,EAAE,CAAC,EAAE,EAAE,aAAa,KAAK,IAAI,CAAC;CACnD;AAED;;;;;;;;;GASG;AACH,MAAM,WAAW,aAAa;IAC5B;;;;;;OAMG;IACH,QAAQ,CAAC,KAAK,EAAE,SAAS;QAAE,QAAQ,CAAC,OAAO,EAAE,MAAM,CAAC;QAAC,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAA;KAAE,EAAE,CAAC;IACrF;;;;OAIG;IACH,QAAQ,CAAC,WAAW,EAAE,SAAS;QAAE,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;QAAC,QAAQ,CAAC,MAAM,EAAE,MAAM,CAAA;KAAE,EAAE,CAAC;IACxF;;;;OAIG;IACH,QAAQ,CAAC,aAAa,EAAE,SAAS;QAC/B,QAAQ,CAAC,SAAS,EAAE,MAAM,CAAC;QAC3B,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;KAC7B,EAAE,CAAC;IACJ;;;;;;;OAOG;IACH,QAAQ,CAAC,KAAK,CAAC,EAAE,SAAS,SAAS,EAAE,CAAC;IACtC;;;;;;OAMG;IACH,QAAQ,CAAC,KAAK,CAAC,EAAE,SAAS,YAAY,EAAE,CAAC;CAC1C;AAED,qEAAqE;AACrE,MAAM,WAAW,eAAe;IAC9B,kEAAkE;IAClE,QAAQ,CAAC,KAAK,CAAC,EAAE,OAAO,CAAC;IACzB;;;;;;OAMG;IACH,QAAQ,CAAC,eAAe,CAAC,EAAE,aAAa,CAAC;IACzC;;;;OAIG;IACH,QAAQ,CAAC,OAAO,CAAC,EAAE,OAAO,CAAC;IAC3B,uEAAuE;IACvE,QAAQ,CAAC,OAAO,CAAC,EAAE,OAAO,CAAC;IAC3B,wEAAwE;IACxE,QAAQ,CAAC,OAAO,CAAC,EAAE,OAAO,CAAC;IAC3B,4EAA4E;IAC5E,QAAQ,CAAC,cAAc,CAAC,EAAE,MAAM,CAAC;IACjC,yEAAyE;IACzE,QAAQ,CAAC,aAAa,CAAC,EAAE,MAAM,CAAC;IAChC;;;;OAIG;IACH,QAAQ,CAAC,UAAU,CAAC,EAAE,OAAO,CAAC;IAC9B;;;OAGG;IACH,QAAQ,CAAC,iBAAiB,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC;IAC7C,yDAAyD;IACzD,QAAQ,CAAC,kBAAkB,CAAC,EAAE,MAAM,CAAC;IACrC;;;;;;OAMG;IACH,QAAQ,CAAC,qBAAqB,CAAC,EAAE,SAAS,CAAC,QAAQ,GAAG,MAAM,GAAG,WAAW,CAAC,EAAE,CAAC;IAC9E;;;;;;;;OAQG;IACH,QAAQ,CAAC,iBAAiB,CAAC,EAAE,MAAM,CAAC;IACpC;;;;;;OAMG;IACH,QAAQ,CAAC,mBAAmB,CAAC,EAAE,MAAM,CAAC;IACtC;;;;;OAKG;IACH,QAAQ,CAAC,IAAI,CAAC,EAAE,OAAO,CAAC;IACxB;;;;;;;OAOG;IACH,QAAQ,CAAC,gBAAgB,CAAC,EAAE,OAAO,CAAC;IACpC;;;;;;OAMG;IACH,QAAQ,CAAC,QAAQ,CAAC,EAAE,OAAO,CAAC;IAC5B;;;;;;OAMG;IACH,QAAQ,CAAC,SAAS,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;IACzC;;;;;;;;OAQG;IACH,QAAQ,CAAC,kBAAkB,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IACjD;;;;OAIG;IACH,QAAQ,CAAC,YAAY,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC3C;;;;;;OAMG;IACH,QAAQ,CAAC,eAAe,CAAC,EAAE,OAAO,CAAC;CACpC;AAED,oEAAoE;AACpE,MAAM,WAAW,aAAa;IAC5B,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAC;IACvB,QAAQ,CAAC,IAAI,EAAE,OAAO,GAAG,KAAK,GAAG,MAAM,GAAG,MAAM,GAAG,OAAO,CAAC;IAC3D,QAAQ,CAAC,GAAG,CAAC,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,OAAO,CAAC,EAAE,MAAM,CAAC;IAC1B,QAAQ,CAAC,SAAS,CAAC,EAAE,MAAM,CAAC;CAC7B;AAED,qCAAqC;AACrC,MAAM,WAAW,aAAa,CAAC,OAAO,GAAG,OAAO;IAC9C,+DAA+D;IAC/D,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,iEAAiE;IACjE,QAAQ,CAAC,IAAI,EAAE,SAAS,MAAM,EAAE,CAAC;IACjC;;;OAGG;IACH,GAAG,CAAC,GAAG,EAAE,eAAe,EAAE,IAAI,EAAE,WAAW,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,OAAO,CAAC,OAAO,CAAC,CAAC;CACjF;AAED,yDAAyD;AACzD,MAAM,WAAW,WAAW;IAC1B,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,MAAM,EAAE,OAAO,CAAC;IACzB,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;IAC5B,QAAQ,CAAC,QAAQ,EAAE,SAAS,MAAM,EAAE,CAAC;CACtC"}

package/dist/pipeline/types.js

@@ -0,0 +1,15 @@
+/**
+ * Shared types for the pipeline runner and phase modules.
+ *
+ * The pipeline is a static DAG: each phase declares its dependencies by name,
+ * the runner validates (duplicate / missing / cycle) and sorts phases
+ * topologically, then executes in order. Phases communicate via a typed
+ * output map keyed by phase name; each phase only sees outputs of phases it
+ * declared as dependencies (dep isolation prevents hidden coupling).
+ *
+ * The runner plus scan/structure/parse are the core; markdown, routes,
+ * tools, orm, crossFile, mro, communities, and processes extend the set
+ * without touching this file.
+ */
+export {};
+//# sourceMappingURL=types.js.map

package/dist/pipeline/types.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"types.js","sourceRoot":"","sources":["../../src/pipeline/types.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG"}

package/dist/providers/c.d.ts

@@ -0,0 +1,3 @@
+import type { LanguageProvider } from "./types.js";
+export declare const cProvider: LanguageProvider;
+//# sourceMappingURL=c.d.ts.map

package/dist/providers/c.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"c.d.ts","sourceRoot":"","sources":["../../src/providers/c.ts"],"names":[],"mappings":"AAcA,OAAO,KAAK,EAKV,gBAAgB,EACjB,MAAM,YAAY,CAAC;AAqKpB,eAAO,MAAM,SAAS,EAAE,gBAcvB,CAAC"}

package/dist/providers/c.js

@@ -0,0 +1,162 @@
+import { getLine, innermostEnclosingDef, isInside, pairDefinitionsWithNames, } from "./extract-helpers.js";
+/**
+ * C provider.
+ *
+ * Definitions: functions, typedefs, structs, unions, enums, static variables,
+ * macro definitions. The grammar tags `struct_specifier`/`union_specifier`
+ * with `@definition.class` and uses `@definition.type` for `typedef` and
+ * `enum_specifier`.
+ *
+ * Calls: `call_expression` with a bare identifier callee. Qualified calls
+ * (e.g. `namespace::foo()`) are a C++ concept — not applicable here.
+ *
+ * Imports: `#include` directives — system headers (`<h>`) and user headers
+ * (`"h"`) are both treated as `package-wildcard` imports at MVP.
+ *
+ * Heritage: C has no inheritance, so this is always `[]`.
+ *
+ * Exports: at MVP we treat `static` identifiers (file-scoped) as not
+ * exported; everything else is. We use the header line of the definition
+ * to check for the `static` storage-class specifier.
+ */
+const C_DEF_KIND_MAP = {
+    "definition.function": "Function",
+    "definition.class": "Struct",
+    "definition.type": "Typedef",
+    "definition.enum": "Enum",
+    "definition.union": "Union",
+    "definition.constant": "Const",
+    "definition.macro": "Macro",
+};
+function extractCDefinitions(input) {
+    const { filePath, captures, sourceText } = input;
+    const paired = pairDefinitionsWithNames(captures);
+    const defCaptures = captures.filter((c) => c.tag.startsWith("definition."));
+    const out = [];
+    for (const { def, name } of paired) {
+        const kind = C_DEF_KIND_MAP[def.tag];
+        if (kind === undefined)
+            continue;
+        let owner;
+        const ownerDef = innermostEnclosingDef(def, defCaptures);
+        if (ownerDef !== undefined) {
+            const ownerPaired = paired.find((p) => p.def === ownerDef);
+            if (ownerPaired !== undefined)
+                owner = ownerPaired.name.text;
+        }
+        const qualifiedName = owner !== undefined ? `${owner}.${name.text}` : name.text;
+        const headerLine = getLine(sourceText, def.startLine);
+        const isStatic = /\bstatic\b/.test(headerLine);
+        // C convention: identifiers prefixed with `_` tend to be internal; we
+        // respect both `static` and leading-underscore as non-exported.
+        const isExported = !isStatic && !name.text.startsWith("_");
+        out.push({
+            kind,
+            name: name.text,
+            qualifiedName,
+            filePath,
+            startLine: def.startLine,
+            endLine: def.endLine,
+            isExported,
+            ...(owner !== undefined ? { owner } : {}),
+        });
+    }
+    return out;
+}
+function extractCCalls(input) {
+    const { filePath, captures, definitions } = input;
+    const defCaptures = captures.filter((c) => c.tag.startsWith("definition."));
+    const callRefs = captures.filter((c) => c.tag === "reference.call");
+    const out = [];
+    for (const ref of callRefs) {
+        const innerName = findNameInside(captures, ref);
+        const calleeName = innerName?.text ?? ref.text;
+        const enclosingDef = innermostEnclosingDef(ref, defCaptures);
+        const callerQualifiedName = enclosingDef
+            ? qualifiedForCapture(enclosingDef, definitions)
+            : "<module>";
+        out.push({
+            callerQualifiedName,
+            calleeName,
+            filePath,
+            startLine: ref.startLine,
+        });
+    }
+    return out;
+}
+function findNameInside(captures, outer) {
+    let best;
+    for (const c of captures) {
+        if (c.tag !== "name")
+            continue;
+        if (!isInside(c, outer))
+            continue;
+        if (best === undefined || c.startLine < best.startLine)
+            best = c;
+    }
+    return best;
+}
+function qualifiedForCapture(def, definitions) {
+    for (const d of definitions) {
+        if (d.startLine === def.startLine)
+            return d.qualifiedName;
+    }
+    return "<module>";
+}
+/**
+ * Parse `#include` directives. Covers both forms:
+ *   `#include <stdio.h>`   system headers
+ *   `#include "local.h"`   user headers
+ *
+ * The quoted form may appear in `localAlias` in future revisions; today we
+ * just record the raw path. `kind` is `package-wildcard` since an include
+ * pulls in every symbol exposed by the header.
+ */
+function extractCImports(input) {
+    const { filePath, sourceText } = input;
+    // We deliberately avoid the shared `stripComments` helper — it treats `#`
+    // as a line-comment starter (Python/Ruby/shell convention), which would
+    // eat `#include` directives. `//` and `/* */` are handled by the regex
+    // itself: an include line must start with `#include` before any comment.
+    const stripped = stripCStyleComments(sourceText);
+    const out = [];
+    // System includes: #include <path>
+    const systemRe = /^\s*#\s*include\s+<([^>]+)>/gm;
+    for (const m of stripped.matchAll(systemRe)) {
+        out.push({ filePath, source: m[1], kind: "package-wildcard" });
+    }
+    // User includes: #include "path"
+    const userRe = /^\s*#\s*include\s+"([^"]+)"/gm;
+    for (const m of stripped.matchAll(userRe)) {
+        out.push({ filePath, source: m[1], kind: "package-wildcard" });
+    }
+    return out;
+}
+/**
+ * Strip C-style line and block comments without touching `#` — which is
+ * a preprocessor sigil, not a comment.
+ */
+function stripCStyleComments(src) {
+    // Replace block comments with spaces to preserve line numbers.
+    let out = src.replace(/\/\*[\s\S]*?\*\//g, (s) => s.replace(/[^\n]/g, " "));
+    out = out.replace(/\/\/[^\n]*/g, "");
+    return out;
+}
+function extractCHeritage(_input) {
+    return [];
+}
+export const cProvider = {
+    id: "c",
+    extensions: [".c", ".h"],
+    importSemantics: "named",
+    mroStrategy: "none",
+    typeConfig: { structural: false, nominal: true, generics: false },
+    heritageEdge: null,
+    isExportedIdentifier: (name) => !name.startsWith("_"),
+    extractDefinitions: extractCDefinitions,
+    extractCalls: extractCCalls,
+    extractImports: extractCImports,
+    isExported: (def) => def.isExported,
+    extractHeritage: extractCHeritage,
+};
+//# sourceMappingURL=c.js.map

package/dist/providers/c.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"c.js","sourceRoot":"","sources":["../../src/providers/c.ts"],"names":[],"mappings":"AAEA,OAAO,EACL,OAAO,EACP,qBAAqB,EACrB,QAAQ,EACR,wBAAwB,GACzB,MAAM,sBAAsB,CAAC;AAe9B;;;;;;;;;;;;;;;;;;;GAmBG;AAEH,MAAM,cAAc,GAAuC;IACzD,qBAAqB,EAAE,UAAU;IACjC,kBAAkB,EAAE,QAAQ;IAC5B,iBAAiB,EAAE,SAAS;IAC5B,iBAAiB,EAAE,MAAM;IACzB,kBAAkB,EAAE,OAAO;IAC3B,qBAAqB,EAAE,OAAO;IAC9B,kBAAkB,EAAE,OAAO;CAC5B,CAAC;AAEF,SAAS,mBAAmB,CAAC,KAA8B;IACzD,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,UAAU,EAAE,GAAG,KAAK,CAAC;IACjD,MAAM,MAAM,GAAG,wBAAwB,CAAC,QAAQ,CAAC,CAAC;IAClD,MAAM,WAAW,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,UAAU,CAAC,aAAa,CAAC,CAAC,CAAC;IAC5E,MAAM,GAAG,GAA0B,EAAE,CAAC;IAEtC,KAAK,MAAM,EAAE,GAAG,EAAE,IAAI,EAAE,IAAI,MAAM,EAAE,CAAC;QACnC,MAAM,IAAI,GAAG,cAAc,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QACrC,IAAI,IAAI,KAAK,SAAS;YAAE,SAAS;QAEjC,IAAI,KAAyB,CAAC;QAC9B,MAAM,QAAQ,GAAG,qBAAqB,CAAC,GAAG,EAAE,WAAW,CAAC,CAAC;QACzD,IAAI,QAAQ,KAAK,SAAS,EAAE,CAAC;YAC3B,MAAM,WAAW,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,GAAG,KAAK,QAAQ,CAAC,CAAC;YAC3D,IAAI,WAAW,KAAK,SAAS;gBAAE,KAAK,GAAG,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC;QAC/D,CAAC;QAED,MAAM,aAAa,GAAG,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,GAAG,KAAK,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC;QAChF,MAAM,UAAU,GAAG,OAAO,CAAC,UAAU,EAAE,GAAG,CAAC,SAAS,CAAC,CAAC;QACtD,MAAM,QAAQ,GAAG,YAAY,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;QAC/C,sEAAsE;QACtE,gEAAgE;QAChE,MAAM,UAAU,GAAG,CAAC,QAAQ,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC;QAE3D,GAAG,CAAC,IAAI,CAAC;YACP,IAAI;YACJ,IAAI,EAAE,IAAI,CAAC,IAAI;YACf,aAAa;YACb,QAAQ;YACR,SAAS,EAAE,GAAG,CAAC,SAAS;YACxB,OAAO,EAAE,GAAG,CAAC,OAAO;YACpB,UAAU;YACV,GAAG,CAAC,KAAK,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;SAC1C,CAAC,CAAC;IACL,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED,SAAS,aAAa,CAAC,KAAwB;IAC7C,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,WAAW,EAAE,GAAG,KAAK,CAAC;IAClD,MAAM,WAAW,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,UAAU,CAAC,aAAa,CAAC,CAAC,CAAC;IAC5E,MAAM,QAAQ,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,GAAG,KAAK,gBAAgB,CAAC,CAAC;IACpE,MAAM,GAAG,GAAoB,EAAE,CAAC;IAEhC,KAAK,MAAM,GAAG,IAAI,QAAQ,EAAE,CAAC;QAC3B,MAAM,SAAS,GAAG,cAAc,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC;QAChD,MAAM,UAAU,GAAG,SAAS,EAAE,IAAI,IAAI,GAAG,CAAC,IAAI,CAAC;QAE/C,MAAM,YAAY,GAAG,qBAAqB,CAAC,GAAG,EAAE,WAAW,CAAC,CAAC;QAC7D,MAAM,mBAAmB,GAAG,YAAY;YACtC,CAAC,CAAC,mBAAmB,CAAC,YAAY,EAAE,WAAW,CAAC;YAChD,CAAC,CAAC,UAAU,CAAC;QAEf,GAAG,CAAC,IAAI,CAAC;YACP,mBAAmB;YACnB,UAAU;YACV,QAAQ;YACR,SAAS,EAAE,GAAG,CAAC,SAAS;SACzB,CAAC,CAAC;IACL,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED,SAAS,cAAc,CACrB,QAAiC,EACjC,KAAmB;IAEnB,IAAI,IAA8B,CAAC;IACnC,KAAK,MAAM,CAAC,IAAI,QAAQ,EAAE,CAAC;QACzB,IAAI,CAAC,CAAC,GAAG,KAAK,MAAM;YAAE,SAAS;QAC/B,IAAI,CAAC,QAAQ,CAAC,CAAC,EAAE,KAAK,CAAC;YAAE,SAAS;QAClC,IAAI,IAAI,KAAK,SAAS,IAAI,CAAC,CAAC,SAAS,GAAG,IAAI,CAAC,SAAS;YAAE,IAAI,GAAG,CAAC,CAAC;IACnE,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AAED,SAAS,mBAAmB,CAC1B,GAAiB,EACjB,WAA2C;IAE3C,KAAK,MAAM,CAAC,IAAI,WAAW,EAAE,CAAC;QAC5B,IAAI,CAAC,CAAC,SAAS,KAAK,GAAG,CAAC,SAAS;YAAE,OAAO,CAAC,CAAC,aAAa,CAAC;IAC5D,CAAC;IACD,OAAO,UAAU,CAAC;AACpB,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,eAAe,CAAC,KAA0B;IACjD,MAAM,EAAE,QAAQ,EAAE,UAAU,EAAE,GAAG,KAAK,CAAC;IACvC,0EAA0E;IAC1E,wEAAwE;IACxE,uEAAuE;IACvE,yEAAyE;IACzE,MAAM,QAAQ,GAAG,mBAAmB,CAAC,UAAU,CAAC,CAAC;IACjD,MAAM,GAAG,GAAsB,EAAE,CAAC;IAElC,mCAAmC;IACnC,MAAM,QAAQ,GAAG,+BAA+B,CAAC;IACjD,KAAK,MAAM,CAAC,IAAI,QAAQ,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC5C,GAAG,CAAC,IAAI,CAAC,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAC,CAAC,CAAW,EAAE,IAAI,EAAE,kBAAkB,EAAE,CAAC,CAAC;IAC3E,CAAC;IACD,iCAAiC;IACjC,MAAM,MAAM,GAAG,+BAA+B,CAAC;IAC/C,KAAK,MAAM,CAAC,IAAI,QAAQ,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QAC1C,GAAG,CAAC,IAAI,CAAC,EAAE,QAAQ,EAAE,MAAM,EAAE,CAAC,CAAC,CAAC,CAAW,EAAE,IAAI,EAAE,kBAAkB,EAAE,CAAC,CAAC;IAC3E,CAAC;IACD,OAAO,GAAG,CAAC;AACb,CAAC;AAED;;;GAGG;AACH,SAAS,mBAAmB,CAAC,GAAW;IACtC,+DAA+D;IAC/D,IAAI,GAAG,GAAG,GAAG,CAAC,OAAO,CAAC,mBAAmB,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC,CAAC;IAC5E,GAAG,GAAG,GAAG,CAAC,OAAO,CAAC,aAAa,EAAE,EAAE,CAAC,CAAC;IACrC,OAAO,GAAG,CAAC;AACb,CAAC;AAED,SAAS,gBAAgB,CAAC,MAA4B;IACpD,OAAO,EAAE,CAAC;AACZ,CAAC;AAED,MAAM,CAAC,MAAM,SAAS,GAAqB;IACzC,EAAE,EAAE,GAAG;IACP,UAAU,EAAE,CAAC,IAAI,EAAE,IAAI,CAAC;IACxB,eAAe,EAAE,OAAO;IACxB,WAAW,EAAE,MAAM;IACnB,UAAU,EAAE,EAAE,UAAU,EAAE,KAAK,EAAE,OAAO,EAAE,IAAI,EAAE,QAAQ,EAAE,KAAK,EAAE;IACjE,YAAY,EAAE,IAAI;IAClB,oBAAoB,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC;IAErD,kBAAkB,EAAE,mBAAmB;IACvC,YAAY,EAAE,aAAa;IAC3B,cAAc,EAAE,eAAe;IAC/B,UAAU,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,UAAU;IACnC,eAAe,EAAE,gBAAgB;CAClC,CAAC"}

package/dist/providers/cobol.d.ts

@@ -0,0 +1,19 @@
+/**
+ * COBOL language provider — stub.
+ *
+ * COBOL has no tree-sitter grammar, so the parse pipeline does NOT route
+ * `.cbl` / `.cob` / `.cpy` files through the worker pool or this provider's
+ * extract methods. Instead, `packages/ingestion/src/parse/cobol-regex.ts`
+ * emits `CodeElement` graph nodes directly from a regex pass.
+ *
+ * This stub exists solely to satisfy the compile-time
+ * `satisfies Record<LanguageId, LanguageProvider>` constraint in
+ * `providers/registry.ts`. Every extract method returns an empty array; the
+ * receiver-inference and heritage hooks follow the "no inheritance" defaults.
+ * Calling any of these methods indicates the parse phase failed to route
+ * COBOL files correctly — the resulting empty output is preferable to a
+ * crash, but upstream callers should treat it as a bug signal.
+ */
+import type { LanguageProvider } from "./types.js";
+export declare const cobolProvider: LanguageProvider;
+//# sourceMappingURL=cobol.d.ts.map

package/dist/providers/cobol.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"cobol.d.ts","sourceRoot":"","sources":["../../src/providers/cobol.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAQH,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,YAAY,CAAC;AAEnD,eAAO,MAAM,aAAa,EAAE,gBA2B3B,CAAC"}

package/dist/providers/cobol.js

@@ -0,0 +1,44 @@
+/**
+ * COBOL language provider — stub.
+ *
+ * COBOL has no tree-sitter grammar, so the parse pipeline does NOT route
+ * `.cbl` / `.cob` / `.cpy` files through the worker pool or this provider's
+ * extract methods. Instead, `packages/ingestion/src/parse/cobol-regex.ts`
+ * emits `CodeElement` graph nodes directly from a regex pass.
+ *
+ * This stub exists solely to satisfy the compile-time
+ * `satisfies Record<LanguageId, LanguageProvider>` constraint in
+ * `providers/registry.ts`. Every extract method returns an empty array; the
+ * receiver-inference and heritage hooks follow the "no inheritance" defaults.
+ * Calling any of these methods indicates the parse phase failed to route
+ * COBOL files correctly — the resulting empty output is preferable to a
+ * crash, but upstream callers should treat it as a bug signal.
+ */
+export const cobolProvider = {
+    id: "cobol",
+    extensions: [".cbl", ".cob", ".cpy"],
+    importSemantics: "named",
+    mroStrategy: "none",
+    typeConfig: {
+        structural: false,
+        nominal: false,
+        generics: false,
+    },
+    heritageEdge: null,
+    extractDefinitions() {
+        return [];
+    },
+    extractCalls() {
+        return [];
+    },
+    extractImports() {
+        return [];
+    },
+    isExported() {
+        return false;
+    },
+    extractHeritage() {
+        return [];
+    },
+};
+//# sourceMappingURL=cobol.js.map

package/dist/providers/cobol.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"cobol.js","sourceRoot":"","sources":["../../src/providers/cobol.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAUH,MAAM,CAAC,MAAM,aAAa,GAAqB;IAC7C,EAAE,EAAE,OAAO;IACX,UAAU,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC;IACpC,eAAe,EAAE,OAAO;IACxB,WAAW,EAAE,MAAM;IACnB,UAAU,EAAE;QACV,UAAU,EAAE,KAAK;QACjB,OAAO,EAAE,KAAK;QACd,QAAQ,EAAE,KAAK;KAChB;IACD,YAAY,EAAE,IAAI;IAElB,kBAAkB;QAChB,OAAO,EAAE,CAAC;IACZ,CAAC;IACD,YAAY;QACV,OAAO,EAAE,CAAC;IACZ,CAAC;IACD,cAAc;QACZ,OAAO,EAAE,CAAC;IACZ,CAAC;IACD,UAAU;QACR,OAAO,KAAK,CAAC;IACf,CAAC;IACD,eAAe;QACb,OAAO,EAAE,CAAC;IACZ,CAAC;CACF,CAAC"}

package/dist/providers/cpp.d.ts

@@ -0,0 +1,3 @@
+import type { LanguageProvider } from "./types.js";
+export declare const cppProvider: LanguageProvider;
+//# sourceMappingURL=cpp.d.ts.map

package/dist/providers/cpp.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"cpp.d.ts","sourceRoot":"","sources":["../../src/providers/cpp.ts"],"names":[],"mappings":"AAeA,OAAO,KAAK,EAKV,gBAAgB,EACjB,MAAM,YAAY,CAAC;AAiNpB,eAAO,MAAM,WAAW,EAAE,gBAczB,CAAC"}