lmgrep 0.1.17 → 0.2.0

package/dist/infrastructure/process/ProcessRegistry.js

@@ -0,0 +1,109 @@
+import { readdirSync, readFileSync } from "node:fs";
+import { join } from "node:path";
+import { PidFileLock } from "../fs/PidFileLock.js";
+/**
+ * Discovers which lmgrep processes are alive and what they hold.
+ *
+ * Maintainer lock files double as the registry: each records its owner's pid,
+ * worktree and database, so scanning them and testing liveness gives an
+ * accurate picture without any separate bookkeeping that could drift from
+ * reality.
+ */
+export class ProcessRegistry {
+    state;
+    static LOCK_SUFFIX = ".lock";
+    constructor(state) {
+        this.state = state;
+    }
+    discoverRunning() {
+        const results = [];
+        // One process can hold several databases; report it once.
+        const seen = new Set();
+        for (const path of this.lockFiles()) {
+            const owner = new PidFileLock(path).read();
+            if (owner === undefined)
+                continue;
+            if (!PidFileLock.isAlive(owner.pid) || seen.has(owner.pid))
+                continue;
+            seen.add(owner.pid);
+            const info = this.inspect(owner.pid);
+            if (!info)
+                continue;
+            results.push({
+                pid: owner.pid,
+                processName: info.name,
+                cmdline: info.cmdline,
+                kind: this.classify(info),
+                // The lock records the worktree its owner actually watches,
+                // which is what the user wants to see — several worktrees can
+                // share one database, so the database's own recorded root
+                // would name the wrong tree.
+                projectRoot: owner.root,
+                databasePath: owner.database,
+                // Every lock in this scan is a maintainer lock, and only a
+                // watcher ever takes one — so holding it *is* watching. This
+                // used to be inferred from the command line, which cannot
+                // work: setting process.title overwrites that buffer, so the
+                // subcommand a process was launched with is no longer there
+                // to match against.
+                watching: true,
+            });
+        }
+        return results;
+    }
+    /**
+     * Lock files, from the locks directory and from the state root.
+     *
+     * The root is scanned too because a watcher started before locks moved is
+     * still running and still holds its database — reporting "no running
+     * processes" right after an upgrade would be wrong in exactly the moment
+     * someone is most likely to check.
+     */
+    lockFiles() {
+        const out = [];
+        for (const directory of [this.state.locksDirectory(), this.state.root()]) {
+            let entries;
+            try {
+                entries = readdirSync(directory);
+            }
+            catch {
+                continue;
+            }
+            for (const entry of entries) {
+                if (entry.endsWith(ProcessRegistry.LOCK_SUFFIX)) {
+                    out.push(join(directory, entry));
+                }
+            }
+        }
+        return out;
+    }
+    inspect(pid) {
+        try {
+            return {
+                name: readFileSync(`/proc/${pid}/comm`, "utf-8").trim(),
+                cmdline: readFileSync(`/proc/${pid}/cmdline`, "utf-8")
+                    .replace(/\0/g, " ")
+                    .trim(),
+            };
+        }
+        catch {
+            // No procfs (macOS) or the process exited mid-scan.
+            return undefined;
+        }
+    }
+    /**
+     * What kind of process this is, from its title.
+     *
+     * The title is the only reliable signal: `process.title` rewrites the argv
+     * buffer, so /proc/<pid>/cmdline reads back as the title alone and carries
+     * no subcommand. The MCP entry point titles itself distinctly; anything
+     * else holding a maintainer lock reached it through `lmgrep serve`.
+     */
+    classify(info) {
+        if (info.name === "lmgrep-mcp" || info.cmdline.includes("mcp")) {
+            return "mcp";
+        }
+        return "serve";
+    }
+}
+//# sourceMappingURL=ProcessRegistry.js.map

package/dist/infrastructure/process/ProcessRegistry.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ProcessRegistry.js","sourceRoot":"","sources":["../../../src/infrastructure/process/ProcessRegistry.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,YAAY,EAAE,MAAM,SAAS,CAAC;AACpD,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAC;AAEjC,OAAO,EAAE,WAAW,EAAE,MAAM,sBAAsB,CAAC;AAoBnD;;;;;;;GAOG;AACH,MAAM,OAAO,eAAe;IAGE;IAFrB,MAAM,CAAU,WAAW,GAAG,OAAO,CAAC;IAE9C,YAA6B,KAAyB;QAAzB,UAAK,GAAL,KAAK,CAAoB;IAAG,CAAC;IAE1D,eAAe;QACd,MAAM,OAAO,GAAqB,EAAE,CAAC;QACrC,0DAA0D;QAC1D,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAC;QAE/B,KAAK,MAAM,IAAI,IAAI,IAAI,CAAC,SAAS,EAAE,EAAE,CAAC;YACrC,MAAM,KAAK,GAAG,IAAI,WAAW,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;YAC3C,IAAI,KAAK,KAAK,SAAS;gBAAE,SAAS;YAClC,IAAI,CAAC,WAAW,CAAC,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC;gBAAE,SAAS;YACrE,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;YAEpB,MAAM,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;YACrC,IAAI,CAAC,IAAI;gBAAE,SAAS;YAEpB,OAAO,CAAC,IAAI,CAAC;gBACZ,GAAG,EAAE,KAAK,CAAC,GAAG;gBACd,WAAW,EAAE,IAAI,CAAC,IAAI;gBACtB,OAAO,EAAE,IAAI,CAAC,OAAO;gBACrB,IAAI,EAAE,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC;gBACzB,4DAA4D;gBAC5D,8DAA8D;gBAC9D,0DAA0D;gBAC1D,6BAA6B;gBAC7B,WAAW,EAAE,KAAK,CAAC,IAAI;gBACvB,YAAY,EAAE,KAAK,CAAC,QAAQ;gBAC5B,2DAA2D;gBAC3D,6DAA6D;gBAC7D,0DAA0D;gBAC1D,6DAA6D;gBAC7D,4DAA4D;gBAC5D,oBAAoB;gBACpB,QAAQ,EAAE,IAAI;aACd,CAAC,CAAC;QACJ,CAAC;QAED,OAAO,OAAO,CAAC;IAChB,CAAC;IAED;;;;;;;OAOG;IACK,SAAS;QAChB,MAAM,GAAG,GAAa,EAAE,CAAC;QACzB,KAAK,MAAM,SAAS,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,cAAc,EAAE,EAAE,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC;YAC1E,IAAI,OAAiB,CAAC;YACtB,IAAI,CAAC;gBACJ,OAAO,GAAG,WAAW,CAAC,SAAS,CAAC,CAAC;YAClC,CAAC;YAAC,MAAM,CAAC;gBACR,SAAS;YACV,CAAC;YACD,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;gBAC7B,IAAI,KAAK,CAAC,QAAQ,CAAC,eAAe,CAAC,WAAW,CAAC,EAAE,CAAC;oBACjD,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,KAAK,CAAC,CAAC,CAAC;gBAClC,CAAC;YACF,CAAC;QACF,CAAC;QACD,OAAO,GAAG,CAAC;IACZ,CAAC;IAEO,OAAO,CAAC,GAAW;QAC1B,IAAI,CAAC;YACJ,OAAO;gBACN,IAAI,EAAE,YAAY,CAAC,SAAS,GAAG,OAAO,EAAE,OAAO,CAAC,CAAC,IAAI,EAAE;gBACvD,OAAO,EAAE,YAAY,CAAC,SAAS,GAAG,UAAU,EAAE,OAAO,CAAC;qBACpD,OAAO,CAAC,KAAK,EAAE,GAAG,CAAC;qBACnB,IAAI,EAAE;aACR,CAAC;QACH,CAAC;QAAC,MAAM,CAAC;YACR,oDAAoD;YACpD,OAAO,SAAS,CAAC;QAClB,CAAC;IACF,CAAC;IAED;;;;;;;OAOG;IACK,QAAQ,CAAC,IAAuC;QACvD,IAAI,IAAI,CAAC,IAAI,KAAK,YAAY,IAAI,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;YAChE,OAAO,KAAK,CAAC;QACd,CAAC;QACD,OAAO,OAAO,CAAC;IAChB,CAAC"}

package/dist/infrastructure/treesitter/ChunkContextBuilder.d.ts

@@ -0,0 +1,45 @@
+import type { Node } from "web-tree-sitter";
+import type { LanguageConfig } from "./LanguageCatalog.js";
+/** What kind of thing a chunk is, structurally. */
+export type StructuralRole = "definition" | "orchestration" | "implementation";
+/** One enclosing scope, e.g. `class Foo`. */
+export interface ScopeEntry {
+    kind: string;
+    name: string;
+}
+/**
+ * Builds the context header prepended to a chunk before embedding.
+ *
+ * The header is what makes an isolated function retrievable: on its own,
+ * `async function handle(req)` shares little vocabulary with "how are webhooks
+ * authenticated", but stamped with its file, enclosing class and doc comment
+ * it does. Retrieval quality depends on this at least as much as on the text.
+ */
+export declare class ChunkContextBuilder {
+    /** How far above a node to look for its doc comment. */
+    private static readonly MAX_COMMENT_LOOKBACK;
+    /** AST node type -> the word used for it in the header. */
+    private static readonly SCOPE_KINDS;
+    /** Node types that declare a shape rather than behaviour. */
+    private static readonly DEFINITION_TYPES;
+    /** Node types that wire things together rather than implement them. */
+    private static readonly ORCHESTRATION_TYPES;
+    /** The full header for one chunk. */
+    build(node: Node, filePath: string, source: string, language: LanguageConfig): string;
+    /** The minimal header for a file with no parsed structure. */
+    buildFileOnly(filePath: string): string;
+    /** Walk up from a node collecting the named scopes that enclose it. */
+    extractScope(node: Node, language: LanguageConfig): ScopeEntry[];
+    /**
+     * Collect the comment block or decorators immediately above a node.
+     *
+     * A blank line ends the block only once something has been collected, so a
+     * comment separated from its node by whitespace is still picked up, while
+     * unrelated code above it is not.
+     */
+    extractLeadingComment(node: Node, source: string): string | null;
+    classifyRole(node: Node): StructuralRole;
+    /** Comment syntax across every supported language, plus decorators. */
+    private looksLikeCommentOrDecorator;
+    private nodeName;
+}

package/dist/infrastructure/treesitter/ChunkContextBuilder.js

@@ -0,0 +1,159 @@
+/**
+ * Builds the context header prepended to a chunk before embedding.
+ *
+ * The header is what makes an isolated function retrievable: on its own,
+ * `async function handle(req)` shares little vocabulary with "how are webhooks
+ * authenticated", but stamped with its file, enclosing class and doc comment
+ * it does. Retrieval quality depends on this at least as much as on the text.
+ */
+export class ChunkContextBuilder {
+    /** How far above a node to look for its doc comment. */
+    static MAX_COMMENT_LOOKBACK = 10;
+    /** AST node type -> the word used for it in the header. */
+    static SCOPE_KINDS = {
+        // TypeScript / JavaScript
+        class_declaration: "class",
+        class_body: "class",
+        interface_declaration: "interface",
+        module: "module",
+        namespace_declaration: "namespace",
+        // Python
+        class_definition: "class",
+        // Rust
+        impl_item: "impl",
+        trait_item: "trait",
+        mod_item: "mod",
+        // Go
+        type_declaration: "type",
+        // Ruby
+        class: "class",
+        // C / C++
+        struct_specifier: "struct",
+        class_specifier: "class",
+        namespace_definition: "namespace",
+        // Swift
+        struct_declaration: "struct",
+        extension_declaration: "extension",
+        // Scala
+        object_definition: "object",
+        trait_definition: "trait",
+        // Generic
+        ContainerDecl: "container",
+    };
+    /** Node types that declare a shape rather than behaviour. */
+    static DEFINITION_TYPES = new Set([
+        "class_declaration",
+        "class_definition",
+        "interface_declaration",
+        "type_alias_declaration",
+        "enum_declaration",
+        "struct_item",
+        "enum_item",
+        "trait_item",
+        "type_item",
+        "type_declaration",
+        "struct_specifier",
+        "class_specifier",
+        "enum_specifier",
+        "type_definition",
+        "struct_declaration",
+        "protocol_declaration",
+        "trait_definition",
+        "ContainerDecl",
+    ]);
+    /** Node types that wire things together rather than implement them. */
+    static ORCHESTRATION_TYPES = new Set([
+        "export_statement",
+        "decorated_definition",
+    ]);
+    /** The full header for one chunk. */
+    build(node, filePath, source, language) {
+        const lines = [
+            `[file: ${filePath}]`,
+            `[role: ${this.classifyRole(node)}]`,
+        ];
+        const scope = this.extractScope(node, language);
+        if (scope.length > 0) {
+            lines.push(`[scope: ${scope.map((s) => `${s.kind} ${s.name}`).join(" > ")}]`);
+        }
+        const comment = this.extractLeadingComment(node, source);
+        if (comment)
+            lines.push(`[doc: ${comment}]`);
+        return lines.join("\n");
+    }
+    /** The minimal header for a file with no parsed structure. */
+    buildFileOnly(filePath) {
+        return `[file: ${filePath}]`;
+    }
+    /** Walk up from a node collecting the named scopes that enclose it. */
+    extractScope(node, language) {
+        const scopes = [];
+        let current = node.parent;
+        while (current) {
+            if (language.scopeTypes.includes(current.type)) {
+                const name = this.nodeName(current);
+                if (name) {
+                    scopes.unshift({
+                        kind: ChunkContextBuilder.SCOPE_KINDS[current.type] ?? current.type,
+                        name,
+                    });
+                }
+            }
+            current = current.parent;
+        }
+        return scopes;
+    }
+    /**
+     * Collect the comment block or decorators immediately above a node.
+     *
+     * A blank line ends the block only once something has been collected, so a
+     * comment separated from its node by whitespace is still picked up, while
+     * unrelated code above it is not.
+     */
+    extractLeadingComment(node, source) {
+        const lines = source.split("\n");
+        const start = node.startPosition.row;
+        const collected = [];
+        for (let i = start - 1; i >= 0 && i >= start - ChunkContextBuilder.MAX_COMMENT_LOOKBACK; i--) {
+            const line = lines[i].trim();
+            if (this.looksLikeCommentOrDecorator(line)) {
+                collected.unshift(lines[i]);
+            }
+            else if (line === "") {
+                if (collected.length > 0)
+                    break;
+            }
+            else {
+                break;
+            }
+        }
+        return collected.length === 0 ? null : collected.join("\n").trim();
+    }
+    classifyRole(node) {
+        if (ChunkContextBuilder.DEFINITION_TYPES.has(node.type)) {
+            return "definition";
+        }
+        if (ChunkContextBuilder.ORCHESTRATION_TYPES.has(node.type)) {
+            return "orchestration";
+        }
+        return "implementation";
+    }
+    /** Comment syntax across every supported language, plus decorators. */
+    looksLikeCommentOrDecorator(line) {
+        return (line.startsWith("//") ||
+            line.startsWith("#") ||
+            line.startsWith("*") ||
+            line.startsWith("/*") ||
+            line.startsWith("*/") ||
+            line.startsWith("///") ||
+            line.startsWith("--") ||
+            line.startsWith("@") ||
+            line.startsWith('"""') ||
+            line.startsWith("'''"));
+    }
+    nodeName(node) {
+        return (node.childForFieldName("name") ??
+            node.children.find((c) => c.type === "identifier" || c.type === "type_identifier"))?.text;
+    }
+}
+//# sourceMappingURL=ChunkContextBuilder.js.map

package/dist/infrastructure/treesitter/ChunkContextBuilder.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ChunkContextBuilder.js","sourceRoot":"","sources":["../../../src/infrastructure/treesitter/ChunkContextBuilder.ts"],"names":[],"mappings":"AAYA;;;;;;;GAOG;AACH,MAAM,OAAO,mBAAmB;IAC/B,wDAAwD;IAChD,MAAM,CAAU,oBAAoB,GAAG,EAAE,CAAC;IAElD,2DAA2D;IACnD,MAAM,CAAU,WAAW,GAA2B;QAC7D,0BAA0B;QAC1B,iBAAiB,EAAE,OAAO;QAC1B,UAAU,EAAE,OAAO;QACnB,qBAAqB,EAAE,WAAW;QAClC,MAAM,EAAE,QAAQ;QAChB,qBAAqB,EAAE,WAAW;QAClC,SAAS;QACT,gBAAgB,EAAE,OAAO;QACzB,OAAO;QACP,SAAS,EAAE,MAAM;QACjB,UAAU,EAAE,OAAO;QACnB,QAAQ,EAAE,KAAK;QACf,KAAK;QACL,gBAAgB,EAAE,MAAM;QACxB,OAAO;QACP,KAAK,EAAE,OAAO;QACd,UAAU;QACV,gBAAgB,EAAE,QAAQ;QAC1B,eAAe,EAAE,OAAO;QACxB,oBAAoB,EAAE,WAAW;QACjC,QAAQ;QACR,kBAAkB,EAAE,QAAQ;QAC5B,qBAAqB,EAAE,WAAW;QAClC,QAAQ;QACR,iBAAiB,EAAE,QAAQ;QAC3B,gBAAgB,EAAE,OAAO;QACzB,UAAU;QACV,aAAa,EAAE,WAAW;KAC1B,CAAC;IAEF,6DAA6D;IACrD,MAAM,CAAU,gBAAgB,GAAG,IAAI,GAAG,CAAC;QAClD,mBAAmB;QACnB,kBAAkB;QAClB,uBAAuB;QACvB,wBAAwB;QACxB,kBAAkB;QAClB,aAAa;QACb,WAAW;QACX,YAAY;QACZ,WAAW;QACX,kBAAkB;QAClB,kBAAkB;QAClB,iBAAiB;QACjB,gBAAgB;QAChB,iBAAiB;QACjB,oBAAoB;QACpB,sBAAsB;QACtB,kBAAkB;QAClB,eAAe;KACf,CAAC,CAAC;IAEH,uEAAuE;IAC/D,MAAM,CAAU,mBAAmB,GAAG,IAAI,GAAG,CAAC;QACrD,kBAAkB;QAClB,sBAAsB;KACtB,CAAC,CAAC;IAEH,qCAAqC;IACrC,KAAK,CACJ,IAAU,EACV,QAAgB,EAChB,MAAc,EACd,QAAwB;QAExB,MAAM,KAAK,GAAa;YACvB,UAAU,QAAQ,GAAG;YACrB,UAAU,IAAI,CAAC,YAAY,CAAC,IAAI,CAAC,GAAG;SACpC,CAAC;QAEF,MAAM,KAAK,GAAG,IAAI,CAAC,YAAY,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAC;QAChD,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACtB,KAAK,CAAC,IAAI,CACT,WAAW,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,CAAC,IAAI,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,CACjE,CAAC;QACH,CAAC;QAED,MAAM,OAAO,GAAG,IAAI,CAAC,qBAAqB,CAAC,IAAI,EAAE,MAAM,CAAC,CAAC;QACzD,IAAI,OAAO;YAAE,KAAK,CAAC,IAAI,CAAC,SAAS,OAAO,GAAG,CAAC,CAAC;QAE7C,OAAO,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IACzB,CAAC;IAED,8DAA8D;IAC9D,aAAa,CAAC,QAAgB;QAC7B,OAAO,UAAU,QAAQ,GAAG,CAAC;IAC9B,CAAC;IAED,uEAAuE;IACvE,YAAY,CAAC,IAAU,EAAE,QAAwB;QAChD,MAAM,MAAM,GAAiB,EAAE,CAAC;QAChC,IAAI,OAAO,GAAG,IAAI,CAAC,MAAM,CAAC;QAC1B,OAAO,OAAO,EAAE,CAAC;YAChB,IAAI,QAAQ,CAAC,UAAU,CAAC,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC;gBAChD,MAAM,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAC;gBACpC,IAAI,IAAI,EAAE,CAAC;oBACV,MAAM,CAAC,OAAO,CAAC;wBACd,IAAI,EAAE,mBAAmB,CAAC,WAAW,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,OAAO,CAAC,IAAI;wBACnE,IAAI;qBACJ,CAAC,CAAC;gBACJ,CAAC;YACF,CAAC;YACD,OAAO,GAAG,OAAO,CAAC,MAAM,CAAC;QAC1B,CAAC;QACD,OAAO,MAAM,CAAC;IACf,CAAC;IAED;;;;;;OAMG;IACH,qBAAqB,CAAC,IAAU,EAAE,MAAc;QAC/C,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACjC,MAAM,KAAK,GAAG,IAAI,CAAC,aAAa,CAAC,GAAG,CAAC;QACrC,MAAM,SAAS,GAAa,EAAE,CAAC;QAE/B,KACC,IAAI,CAAC,GAAG,KAAK,GAAG,CAAC,EACjB,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,KAAK,GAAG,mBAAmB,CAAC,oBAAoB,EAC/D,CAAC,EAAE,EACF,CAAC;YACF,MAAM,IAAI,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;YAC7B,IAAI,IAAI,CAAC,2BAA2B,CAAC,IAAI,CAAC,EAAE,CAAC;gBAC5C,SAAS,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;YAC7B,CAAC;iBAAM,IAAI,IAAI,KAAK,EAAE,EAAE,CAAC;gBACxB,IAAI,SAAS,CAAC,MAAM,GAAG,CAAC;oBAAE,MAAM;YACjC,CAAC;iBAAM,CAAC;gBACP,MAAM;YACP,CAAC;QACF,CAAC;QAED,OAAO,SAAS,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;IACpE,CAAC;IAED,YAAY,CAAC,IAAU;QACtB,IAAI,mBAAmB,CAAC,gBAAgB,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YACzD,OAAO,YAAY,CAAC;QACrB,CAAC;QACD,IAAI,mBAAmB,CAAC,mBAAmB,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAC5D,OAAO,eAAe,CAAC;QACxB,CAAC;QACD,OAAO,gBAAgB,CAAC;IACzB,CAAC;IAED,uEAAuE;IAC/D,2BAA2B,CAAC,IAAY;QAC/C,OAAO,CACN,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC;YACrB,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC;YACpB,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC;YACpB,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC;YACrB,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC;YACrB,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC;YACtB,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC;YACrB,IAAI,CAAC,UAAU,CAAC,GAAG,CAAC;YACpB,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC;YACtB,IAAI,CAAC,UAAU,CAAC,KAAK,CAAC,CACtB,CAAC;IACH,CAAC;IAEO,QAAQ,CAAC,IAAU;QAC1B,OAAO,CACN,IAAI,CAAC,iBAAiB,CAAC,MAAM,CAAC;YAC9B,IAAI,CAAC,QAAQ,CAAC,IAAI,CACjB,CAAC,CAAO,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,YAAY,IAAI,CAAC,CAAC,IAAI,KAAK,iBAAiB,CACpE,CACD,EAAE,IAAI,CAAC;IACT,CAAC"}

package/dist/infrastructure/treesitter/LanguageCatalog.d.ts

@@ -0,0 +1,30 @@
+export type LanguageId = "javascript" | "typescript" | "tsx" | "python" | "rust" | "go" | "ruby" | "c" | "cpp" | "swift" | "json" | "yaml" | "toml" | "lua" | "scala" | "zig" | "bash" | "html" | "css" | "java" | "kotlin" | "php" | "c_sharp" | "vue" | "markdown" | "nix";
+export interface LanguageConfig {
+    id: LanguageId;
+    extensions: string[];
+    wasmFile: string;
+    /** AST node types that represent top-level or nestable definitions */
+    chunkTypes: string[];
+    /** AST node types that represent import/require statements */
+    importTypes: string[];
+    /** AST node types that represent class/struct/module wrappers */
+    scopeTypes: string[];
+}
+export declare const LANGUAGES: LanguageConfig[];
+/**
+ * The grammar table and the lookups over it.
+ *
+ * Grammars ship as separate `@lumis-sh/wasm-<slug>` packages built for the
+ * current web-tree-sitter ABI, so resolution is a runtime `require.resolve`
+ * that may legitimately fail — a missing grammar means "chunk this file by
+ * sliding window", not an error.
+ */
+export declare class LanguageCatalog {
+    private readonly languages;
+    private readonly byExtension;
+    constructor(languages?: readonly LanguageConfig[]);
+    /** Grammar for a path's extension, or undefined when none is known. */
+    forFile(filePath: string): LanguageConfig | undefined;
+    /** Absolute path to a grammar's wasm, or undefined when not installed. */
+    wasmPathFor(lang: LanguageConfig): string | undefined;
+}

package/dist/{lib/chunker/languages.js → infrastructure/treesitter/LanguageCatalog.js}

@@ -141,30 +141,21 @@ export const LANGUAGES = [
             "enum_specifier",
         ],
         importTypes: ["preproc_include", "using_declaration"],
-        scopeTypes: [
-            "class_specifier",
-            "struct_specifier",
-            "namespace_definition",
-        ],
+        scopeTypes: ["class_specifier", "struct_specifier", "namespace_definition"],
     },
     {
         id: "swift",
         extensions: [".swift"],
         wasmFile: "tree-sitter-swift.wasm",
+        // Modern tree-sitter-swift represents struct/enum/extension/actor all as
+        // class_declaration (distinguished by keyword), so it covers them all.
         chunkTypes: [
             "function_declaration",
             "class_declaration",
-            "struct_declaration",
-            "enum_declaration",
             "protocol_declaration",
-            "extension_declaration",
         ],
         importTypes: ["import_declaration"],
-        scopeTypes: [
-            "class_declaration",
-            "struct_declaration",
-            "extension_declaration",
-        ],
+        scopeTypes: ["class_declaration"],
     },
     {
         id: "json",
@@ -194,11 +185,8 @@ export const LANGUAGES = [
         id: "lua",
         extensions: [".lua"],
         wasmFile: "tree-sitter-lua.wasm",
-        chunkTypes: [
-            "function_declaration",
-            "local_function",
-            "function_definition",
-        ],
+        // Modern tree-sitter-lua folds `local function` into function_declaration.
+        chunkTypes: ["function_declaration", "function_definition"],
         importTypes: ["call"],
         scopeTypes: [],
     },
@@ -215,19 +203,19 @@ export const LANGUAGES = [
             "trait_definition",
         ],
         importTypes: ["import_declaration"],
-        scopeTypes: [
-            "class_definition",
-            "object_definition",
-            "trait_definition",
-        ],
+        scopeTypes: ["class_definition", "object_definition", "trait_definition"],
     },
     {
         id: "zig",
         extensions: [".zig"],
         wasmFile: "tree-sitter-zig.wasm",
-        chunkTypes: ["FnProto", "VarDecl", "ContainerDecl", "TestDecl"],
+        chunkTypes: [
+            "function_declaration",
+            "variable_declaration",
+            "test_declaration",
+        ],
         importTypes: [],
-        scopeTypes: ["ContainerDecl"],
+        scopeTypes: ["struct_declaration", "enum_declaration", "union_declaration"],
     },
     {
         id: "bash",
@@ -303,7 +291,7 @@ export const LANGUAGES = [
     {
         id: "c_sharp",
         extensions: [".cs"],
-        wasmFile: "tree-sitter-c_sharp.wasm",
+        wasmFile: "tree-sitter-csharp.wasm",
         chunkTypes: [
             "class_declaration",
             "interface_declaration",
@@ -330,23 +318,59 @@ export const LANGUAGES = [
         importTypes: [],
         scopeTypes: [],
     },
+    {
+        id: "markdown",
+        extensions: [".md", ".markdown", ".mdx"],
+        wasmFile: "tree-sitter-markdown.wasm",
+        // Markdown is chunked by a heading-aware splitter (see chunker), not by
+        // collectChunks — its `section` nodes nest, so emitting them whole would
+        // collapse a doc into one chunk. chunkTypes/scopeTypes are unused here.
+        chunkTypes: ["section"],
+        importTypes: [],
+        scopeTypes: ["section"],
+    },
+    {
+        id: "nix",
+        extensions: [".nix"],
+        wasmFile: "tree-sitter-nix.wasm",
+        chunkTypes: ["binding"],
+        importTypes: [],
+        scopeTypes: ["attrset_expression", "binding_set"],
+    },
 ];
-const extToLang = new Map();
-for (const lang of LANGUAGES) {
-    for (const ext of lang.extensions) {
-        extToLang.set(ext, lang);
+/**
+ * The grammar table and the lookups over it.
+ *
+ * Grammars ship as separate `@lumis-sh/wasm-<slug>` packages built for the
+ * current web-tree-sitter ABI, so resolution is a runtime `require.resolve`
+ * that may legitimately fail — a missing grammar means "chunk this file by
+ * sliding window", not an error.
+ */
+export class LanguageCatalog {
+    languages;
+    byExtension = new Map();
+    constructor(languages = LANGUAGES) {
+        this.languages = languages;
+        for (const lang of this.languages) {
+            for (const ext of lang.extensions)
+                this.byExtension.set(ext, lang);
+        }
     }
-}
-export function getLanguageForFile(filePath) {
-    const ext = filePath.slice(filePath.lastIndexOf("."));
-    return extToLang.get(ext);
-}
-export function getWasmPath(lang) {
-    try {
-        return require.resolve(`tree-sitter-wasms/out/${lang.wasmFile}`);
+    /** Grammar for a path's extension, or undefined when none is known. */
+    forFile(filePath) {
+        return this.byExtension.get(filePath.slice(filePath.lastIndexOf(".")));
     }
-    catch {
-        return undefined;
+    /** Absolute path to a grammar's wasm, or undefined when not installed. */
+    wasmPathFor(lang) {
+        const slug = lang.wasmFile
+            .replace(/^tree-sitter-/, "")
+            .replace(/\.wasm$/, "");
+        try {
+            return require.resolve(`@lumis-sh/wasm-${slug}/${lang.wasmFile}`);
+        }
+        catch {
+            return undefined;
+        }
     }
 }
-//# sourceMappingURL=languages.js.map
+//# sourceMappingURL=LanguageCatalog.js.map

package/dist/infrastructure/treesitter/LanguageCatalog.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"LanguageCatalog.js","sourceRoot":"","sources":["../../../src/infrastructure/treesitter/LanguageCatalog.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AAE5C,MAAM,OAAO,GAAG,aAAa,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;AA0C/C,MAAM,CAAC,MAAM,SAAS,GAAqB;IAC1C;QACC,EAAE,EAAE,YAAY;QAChB,UAAU,EAAE,CAAC,KAAK,EAAE,MAAM,CAAC;QAC3B,QAAQ,EAAE,6BAA6B;QACvC,UAAU,EAAE;YACX,sBAAsB;YACtB,gBAAgB;YAChB,mBAAmB;YACnB,mBAAmB;YACnB,kBAAkB;YAClB,qBAAqB;SACrB;QACD,WAAW,EAAE,CAAC,kBAAkB,EAAE,iBAAiB,CAAC;QACpD,UAAU,EAAE,CAAC,mBAAmB,EAAE,YAAY,CAAC;KAC/C;IACD;QACC,EAAE,EAAE,YAAY;QAChB,UAAU,EAAE,CAAC,KAAK,CAAC;QACnB,QAAQ,EAAE,6BAA6B;QACvC,UAAU,EAAE;YACX,sBAAsB;YACtB,gBAAgB;YAChB,mBAAmB;YACnB,mBAAmB;YACnB,uBAAuB;YACvB,wBAAwB;YACxB,kBAAkB;YAClB,kBAAkB;YAClB,qBAAqB;SACrB;QACD,WAAW,EAAE,CAAC,kBAAkB,CAAC;QACjC,UAAU,EAAE;YACX,mBAAmB;YACnB,uBAAuB;YACvB,QAAQ;YACR,uBAAuB;SACvB;KACD;IACD;QACC,EAAE,EAAE,KAAK;QACT,UAAU,EAAE,CAAC,MAAM,CAAC;QACpB,QAAQ,EAAE,sBAAsB;QAChC,UAAU,EAAE;YACX,sBAAsB;YACtB,gBAAgB;YAChB,mBAAmB;YACnB,mBAAmB;YACnB,uBAAuB;YACvB,wBAAwB;YACxB,kBAAkB;YAClB,kBAAkB;YAClB,qBAAqB;SACrB;QACD,WAAW,EAAE,CAAC,kBAAkB,CAAC;QACjC,UAAU,EAAE;YACX,mBAAmB;YACnB,uBAAuB;YACvB,QAAQ;YACR,uBAAuB;SACvB;KACD;IACD;QACC,EAAE,EAAE,QAAQ;QACZ,UAAU,EAAE,CAAC,KAAK,CAAC;QACnB,QAAQ,EAAE,yBAAyB;QACnC,UAAU,EAAE;YACX,qBAAqB;YACrB,kBAAkB;YAClB,sBAAsB;SACtB;QACD,WAAW,EAAE,CAAC,kBAAkB,EAAE,uBAAuB,CAAC;QAC1D,UAAU,EAAE,CAAC,kBAAkB,EAAE,QAAQ,CAAC;KAC1C;IACD;QACC,EAAE,EAAE,MAAM;QACV,UAAU,EAAE,CAAC,KAAK,CAAC;QACnB,QAAQ,EAAE,uBAAuB;QACjC,UAAU,EAAE;YACX,eAAe;YACf,WAAW;YACX,aAAa;YACb,WAAW;YACX,YAAY;YACZ,WAAW;YACX,YAAY;YACZ,aAAa;YACb,kBAAkB;SAClB;QACD,WAAW,EAAE,CAAC,iBAAiB,CAAC;QAChC,UAAU,EAAE,CAAC,WAAW,EAAE,YAAY,EAAE,UAAU,CAAC;KACnD;IACD;QACC,EAAE,EAAE,IAAI;QACR,UAAU,EAAE,CAAC,KAAK,CAAC;QACnB,QAAQ,EAAE,qBAAqB;QAC/B,UAAU,EAAE;YACX,sBAAsB;YACtB,oBAAoB;YACpB,kBAAkB;YAClB,mBAAmB;YACnB,iBAAiB;SACjB;QACD,WAAW,EAAE,CAAC,oBAAoB,CAAC;QACnC,UAAU,EAAE,CAAC,kBAAkB,CAAC;KAChC;IACD;QACC,EAAE,EAAE,MAAM;QACV,UAAU,EAAE,CAAC,KAAK,CAAC;QACnB,QAAQ,EAAE,uBAAuB;QACjC,UAAU,EAAE,CAAC,QAAQ,EAAE,kBAAkB,EAAE,OAAO,EAAE,QAAQ,CAAC;QAC7D,WAAW,EAAE,CAAC,MAAM,CAAC;QACrB,UAAU,EAAE,CAAC,OAAO,EAAE,QAAQ,CAAC;KAC/B;IACD;QACC,EAAE,EAAE,GAAG;QACP,UAAU,EAAE,CAAC,IAAI,EAAE,IAAI,CAAC;QACxB,QAAQ,EAAE,oBAAoB;QAC9B,UAAU,EAAE;YACX,qBAAqB;YACrB,kBAAkB;YAClB,gBAAgB;YAChB,iBAAiB;YACjB,aAAa;SACb;QACD,WAAW,EAAE,CAAC,iBAAiB,CAAC;QAChC,UAAU,EAAE,CAAC,kBAAkB,CAAC;KAChC;IACD;QACC,EAAE,EAAE,KAAK;QACT,UAAU,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,KAAK,CAAC;QACnC,QAAQ,EAAE,sBAAsB;QAChC,UAAU,EAAE;YACX,qBAAqB;YACrB,iBAAiB;YACjB,kBAAkB;YAClB,sBAAsB;YACtB,sBAAsB;YACtB,gBAAgB;SAChB;QACD,WAAW,EAAE,CAAC,iBAAiB,EAAE,mBAAmB,CAAC;QACrD,UAAU,EAAE,CAAC,iBAAiB,EAAE,kBAAkB,EAAE,sBAAsB,CAAC;KAC3E;IACD;QACC,EAAE,EAAE,OAAO;QACX,UAAU,EAAE,CAAC,QAAQ,CAAC;QACtB,QAAQ,EAAE,wBAAwB;QAClC,yEAAyE;QACzE,uEAAuE;QACvE,UAAU,EAAE;YACX,sBAAsB;YACtB,mBAAmB;YACnB,sBAAsB;SACtB;QACD,WAAW,EAAE,CAAC,oBAAoB,CAAC;QACnC,UAAU,EAAE,CAAC,mBAAmB,CAAC;KACjC;IACD;QACC,EAAE,EAAE,MAAM;QACV,UAAU,EAAE,CAAC,OAAO,CAAC;QACrB,QAAQ,EAAE,uBAAuB;QACjC,UAAU,EAAE,CAAC,MAAM,CAAC;QACpB,WAAW,EAAE,EAAE;QACf,UAAU,EAAE,CAAC,QAAQ,CAAC;KACtB;IACD;QACC,EAAE,EAAE,MAAM;QACV,UAAU,EAAE,CAAC,OAAO,EAAE,MAAM,CAAC;QAC7B,QAAQ,EAAE,uBAAuB;QACjC,UAAU,EAAE,CAAC,oBAAoB,CAAC;QAClC,WAAW,EAAE,EAAE;QACf,UAAU,EAAE,CAAC,eAAe,CAAC;KAC7B;IACD;QACC,EAAE,EAAE,MAAM;QACV,UAAU,EAAE,CAAC,OAAO,CAAC;QACrB,QAAQ,EAAE,uBAAuB;QACjC,UAAU,EAAE,CAAC,OAAO,EAAE,MAAM,CAAC;QAC7B,WAAW,EAAE,EAAE;QACf,UAAU,EAAE,CAAC,OAAO,CAAC;KACrB;IACD;QACC,EAAE,EAAE,KAAK;QACT,UAAU,EAAE,CAAC,MAAM,CAAC;QACpB,QAAQ,EAAE,sBAAsB;QAChC,2EAA2E;QAC3E,UAAU,EAAE,CAAC,sBAAsB,EAAE,qBAAqB,CAAC;QAC3D,WAAW,EAAE,CAAC,MAAM,CAAC;QACrB,UAAU,EAAE,EAAE;KACd;IACD;QACC,EAAE,EAAE,OAAO;QACX,UAAU,EAAE,CAAC,QAAQ,CAAC;QACtB,QAAQ,EAAE,wBAAwB;QAClC,UAAU,EAAE;YACX,qBAAqB;YACrB,gBAAgB;YAChB,gBAAgB;YAChB,kBAAkB;YAClB,mBAAmB;YACnB,kBAAkB;SAClB;QACD,WAAW,EAAE,CAAC,oBAAoB,CAAC;QACnC,UAAU,EAAE,CAAC,kBAAkB,EAAE,mBAAmB,EAAE,kBAAkB,CAAC;KACzE;IACD;QACC,EAAE,EAAE,KAAK;QACT,UAAU,EAAE,CAAC,MAAM,CAAC;QACpB,QAAQ,EAAE,sBAAsB;QAChC,UAAU,EAAE;YACX,sBAAsB;YACtB,sBAAsB;YACtB,kBAAkB;SAClB;QACD,WAAW,EAAE,EAAE;QACf,UAAU,EAAE,CAAC,oBAAoB,EAAE,kBAAkB,EAAE,mBAAmB,CAAC;KAC3E;IACD;QACC,EAAE,EAAE,MAAM;QACV,UAAU,EAAE,CAAC,KAAK,EAAE,OAAO,CAAC;QAC5B,QAAQ,EAAE,uBAAuB;QACjC,UAAU,EAAE,CAAC,qBAAqB,CAAC;QACnC,WAAW,EAAE,CAAC,SAAS,CAAC;QACxB,UAAU,EAAE,EAAE;KACd;IACD;QACC,EAAE,EAAE,MAAM;QACV,UAAU,EAAE,CAAC,OAAO,CAAC;QACrB,QAAQ,EAAE,uBAAuB;QACjC,UAAU,EAAE,CAAC,SAAS,EAAE,gBAAgB,EAAE,eAAe,CAAC;QAC1D,WAAW,EAAE,EAAE;QACf,UAAU,EAAE,CAAC,SAAS,CAAC;KACvB;IACD;QACC,EAAE,EAAE,KAAK;QACT,UAAU,EAAE,CAAC,MAAM,EAAE,OAAO,CAAC;QAC7B,QAAQ,EAAE,sBAAsB;QAChC,UAAU,EAAE,CAAC,UAAU,EAAE,iBAAiB,EAAE,qBAAqB,CAAC;QAClE,WAAW,EAAE,CAAC,kBAAkB,CAAC;QACjC,UAAU,EAAE,EAAE;KACd;IACD;QACC,EAAE,EAAE,MAAM;QACV,UAAU,EAAE,CAAC,OAAO,CAAC;QACrB,QAAQ,EAAE,uBAAuB;QACjC,UAAU,EAAE;YACX,mBAAmB;YACnB,uBAAuB;YACvB,kBAAkB;YAClB,oBAAoB;YACpB,yBAAyB;YACzB,oBAAoB;SACpB;QACD,WAAW,EAAE,CAAC,oBAAoB,CAAC;QACnC,UAAU,EAAE,CAAC,mBAAmB,EAAE,uBAAuB,CAAC;KAC1D;IACD;QACC,EAAE,EAAE,QAAQ;QACZ,UAAU,EAAE,CAAC,KAAK,CAAC;QACnB,QAAQ,EAAE,yBAAyB;QACnC,UAAU,EAAE;YACX,mBAAmB;YACnB,oBAAoB;YACpB,sBAAsB;YACtB,sBAAsB;SACtB;QACD,WAAW,EAAE,CAAC,eAAe,CAAC;QAC9B,UAAU,EAAE,CAAC,mBAAmB,EAAE,oBAAoB,CAAC;KACvD;IACD;QACC,EAAE,EAAE,KAAK;QACT,UAAU,EAAE,CAAC,MAAM,CAAC;QACpB,QAAQ,EAAE,sBAAsB;QAChC,UAAU,EAAE;YACX,mBAAmB;YACnB,uBAAuB;YACvB,mBAAmB;YACnB,kBAAkB;YAClB,qBAAqB;YACrB,oBAAoB;SACpB;QACD,WAAW,EAAE,CAAC,2BAA2B,CAAC;QAC1C,UAAU,EAAE;YACX,mBAAmB;YACnB,uBAAuB;YACvB,mBAAmB;SACnB;KACD;IACD;QACC,EAAE,EAAE,SAAS;QACb,UAAU,EAAE,CAAC,KAAK,CAAC;QACnB,QAAQ,EAAE,yBAAyB;QACnC,UAAU,EAAE;YACX,mBAAmB;YACnB,uBAAuB;YACvB,oBAAoB;YACpB,kBAAkB;YAClB,oBAAoB;YACpB,oBAAoB;YACpB,yBAAyB;YACzB,sBAAsB;SACtB;QACD,WAAW,EAAE,CAAC,iBAAiB,CAAC;QAChC,UAAU,EAAE;YACX,mBAAmB;YACnB,uBAAuB;YACvB,oBAAoB;YACpB,uBAAuB;SACvB;KACD;IACD;QACC,EAAE,EAAE,KAAK;QACT,UAAU,EAAE,CAAC,MAAM,CAAC;QACpB,QAAQ,EAAE,sBAAsB;QAChC,UAAU,EAAE,CAAC,gBAAgB,EAAE,kBAAkB,EAAE,eAAe,CAAC;QACnE,WAAW,EAAE,EAAE;QACf,UAAU,EAAE,EAAE;KACd;IACD;QACC,EAAE,EAAE,UAAU;QACd,UAAU,EAAE,CAAC,KAAK,EAAE,WAAW,EAAE,MAAM,CAAC;QACxC,QAAQ,EAAE,2BAA2B;QACrC,wEAAwE;QACxE,yEAAyE;QACzE,wEAAwE;QACxE,UAAU,EAAE,CAAC,SAAS,CAAC;QACvB,WAAW,EAAE,EAAE;QACf,UAAU,EAAE,CAAC,SAAS,CAAC;KACvB;IACD;QACC,EAAE,EAAE,KAAK;QACT,UAAU,EAAE,CAAC,MAAM,CAAC;QACpB,QAAQ,EAAE,sBAAsB;QAChC,UAAU,EAAE,CAAC,SAAS,CAAC;QACvB,WAAW,EAAE,EAAE;QACf,UAAU,EAAE,CAAC,oBAAoB,EAAE,aAAa,CAAC;KACjD;CACD,CAAC;AAEF;;;;;;;GAOG;AACH,MAAM,OAAO,eAAe;IAIT;IAHD,WAAW,GAAG,IAAI,GAAG,EAA0B,CAAC;IAEjE,YACkB,YAAuC,SAAS;QAAhD,cAAS,GAAT,SAAS,CAAuC;QAEjE,KAAK,MAAM,IAAI,IAAI,IAAI,CAAC,SAAS,EAAE,CAAC;YACnC,KAAK,MAAM,GAAG,IAAI,IAAI,CAAC,UAAU;gBAAE,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC;QACpE,CAAC;IACF,CAAC;IAED,uEAAuE;IACvE,OAAO,CAAC,QAAgB;QACvB,OAAO,IAAI,CAAC,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,QAAQ,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;IACxE,CAAC;IAED,0EAA0E;IAC1E,WAAW,CAAC,IAAoB;QAC/B,MAAM,IAAI,GAAG,IAAI,CAAC,QAAQ;aACxB,OAAO,CAAC,eAAe,EAAE,EAAE,CAAC;aAC5B,OAAO,CAAC,SAAS,EAAE,EAAE,CAAC,CAAC;QACzB,IAAI,CAAC;YACJ,OAAO,OAAO,CAAC,OAAO,CAAC,kBAAkB,IAAI,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC,CAAC;QACnE,CAAC;QAAC,MAAM,CAAC;YACR,OAAO,SAAS,CAAC;QAClB,CAAC;IACF,CAAC;CACD"}

package/dist/infrastructure/treesitter/SlidingWindowChunker.d.ts

@@ -0,0 +1,15 @@
+import { Chunk } from "../../domain/corpus/Chunk.js";
+import type { ChunkerPort } from "../../domain/ports/ChunkerPort.js";
+/**
+ * Fixed-size windows over a file, for sources with no grammar or no chunkable
+ * structure (a linear shell script, say).
+ *
+ * Windows overlap slightly so a passage straddling a boundary is still
+ * retrievable. The overlap is kept small because search deduplicates
+ * overlapping hits anyway, making a larger one pure index bloat.
+ */
+export declare class SlidingWindowChunker implements ChunkerPort {
+    private static readonly WINDOW_LINES;
+    private static readonly STRIDE_LINES;
+    chunk(filePath: string, cwd: string): Promise<Chunk[]>;
+}

package/dist/infrastructure/treesitter/SlidingWindowChunker.js

@@ -0,0 +1,42 @@
+import { readFileSync } from "node:fs";
+import { join } from "node:path";
+import { Chunk } from "../../domain/corpus/Chunk.js";
+import { CodeLocation } from "../../domain/corpus/CodeLocation.js";
+import { ContentHash } from "../../domain/corpus/ContentHash.js";
+/**
+ * Fixed-size windows over a file, for sources with no grammar or no chunkable
+ * structure (a linear shell script, say).
+ *
+ * Windows overlap slightly so a passage straddling a boundary is still
+ * retrievable. The overlap is kept small because search deduplicates
+ * overlapping hits anyway, making a larger one pure index bloat.
+ */
+export class SlidingWindowChunker {
+    static WINDOW_LINES = 50;
+    static STRIDE_LINES = 40;
+    async chunk(filePath, cwd) {
+        const source = readFileSync(join(cwd, filePath), "utf-8");
+        const lines = source.split("\n");
+        if (lines.length === 0)
+            return [];
+        const chunks = [];
+        for (let i = 0; i < lines.length; i += SlidingWindowChunker.STRIDE_LINES) {
+            const content = lines
+                .slice(i, i + SlidingWindowChunker.WINDOW_LINES)
+                .join("\n");
+            if (content.trim().length === 0)
+                continue;
+            const endLine = Math.min(i + SlidingWindowChunker.WINDOW_LINES, lines.length);
+            chunks.push(new Chunk({
+                location: new CodeLocation(filePath, i + 1, endLine),
+                type: "block",
+                name: `lines_${i + 1}_${endLine}`,
+                content,
+                context: `[file: ${filePath}]`,
+                hash: ContentHash.of(content),
+            }));
+        }
+        return chunks;
+    }
+}
+//# sourceMappingURL=SlidingWindowChunker.js.map

package/dist/infrastructure/treesitter/SlidingWindowChunker.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"SlidingWindowChunker.js","sourceRoot":"","sources":["../../../src/infrastructure/treesitter/SlidingWindowChunker.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,MAAM,SAAS,CAAC;AACvC,OAAO,EAAE,IAAI,EAAE,MAAM,WAAW,CAAC;AACjC,OAAO,EAAE,KAAK,EAAE,MAAM,8BAA8B,CAAC;AACrD,OAAO,EAAE,YAAY,EAAE,MAAM,qCAAqC,CAAC;AACnE,OAAO,EAAE,WAAW,EAAE,MAAM,oCAAoC,CAAC;AAGjE;;;;;;;GAOG;AACH,MAAM,OAAO,oBAAoB;IACxB,MAAM,CAAU,YAAY,GAAG,EAAE,CAAC;IAClC,MAAM,CAAU,YAAY,GAAG,EAAE,CAAC;IAE1C,KAAK,CAAC,KAAK,CAAC,QAAgB,EAAE,GAAW;QACxC,MAAM,MAAM,GAAG,YAAY,CAAC,IAAI,CAAC,GAAG,EAAE,QAAQ,CAAC,EAAE,OAAO,CAAC,CAAC;QAC1D,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;QACjC,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,EAAE,CAAC;QAElC,MAAM,MAAM,GAAY,EAAE,CAAC;QAC3B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,IAAI,oBAAoB,CAAC,YAAY,EAAE,CAAC;YAC1E,MAAM,OAAO,GAAG,KAAK;iBACnB,KAAK,CAAC,CAAC,EAAE,CAAC,GAAG,oBAAoB,CAAC,YAAY,CAAC;iBAC/C,IAAI,CAAC,IAAI,CAAC,CAAC;YACb,IAAI,OAAO,CAAC,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC;gBAAE,SAAS;YAE1C,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,CACvB,CAAC,GAAG,oBAAoB,CAAC,YAAY,EACrC,KAAK,CAAC,MAAM,CACZ,CAAC;YACF,MAAM,CAAC,IAAI,CACV,IAAI,KAAK,CAAC;gBACT,QAAQ,EAAE,IAAI,YAAY,CAAC,QAAQ,EAAE,CAAC,GAAG,CAAC,EAAE,OAAO,CAAC;gBACpD,IAAI,EAAE,OAAO;gBACb,IAAI,EAAE,SAAS,CAAC,GAAG,CAAC,IAAI,OAAO,EAAE;gBACjC,OAAO;gBACP,OAAO,EAAE,UAAU,QAAQ,GAAG;gBAC9B,IAAI,EAAE,WAAW,CAAC,EAAE,CAAC,OAAO,CAAC;aAC7B,CAAC,CACF,CAAC;QACH,CAAC;QAED,OAAO,MAAM,CAAC;IACf,CAAC"}

package/dist/infrastructure/treesitter/TreeSitterChunker.d.ts

@@ -0,0 +1,42 @@
+import { Chunk } from "../../domain/corpus/Chunk.js";
+import type { ChunkerPort } from "../../domain/ports/ChunkerPort.js";
+import { ChunkContextBuilder } from "./ChunkContextBuilder.js";
+import { LanguageCatalog } from "./LanguageCatalog.js";
+import { SlidingWindowChunker } from "./SlidingWindowChunker.js";
+/**
+ * Splits source into chunks along AST boundaries.
+ *
+ * Node-bounded chunks are what make results readable: a hit is a whole
+ * function or class, not an arbitrary window through the middle of one. Files
+ * with no grammar, or no chunkable structure, fall back to a sliding window.
+ */
+export declare class TreeSitterChunker implements ChunkerPort {
+    private readonly catalog;
+    private readonly context;
+    private readonly fallback;
+    /** Chunks larger than this are split by recursing into their children. */
+    private static readonly MAX_CHUNK_TOKENS;
+    /** A one-line fragment shorter than this is noise, not a unit of code. */
+    private static readonly MIN_CHUNK_CHARS;
+    private parser;
+    private readonly loaded;
+    constructor(catalog?: LanguageCatalog, context?: ChunkContextBuilder, fallback?: SlidingWindowChunker);
+    chunk(filePath: string, cwd: string): Promise<Chunk[]>;
+    /**
+     * Markdown needs its own pass: the block grammar nests `section` nodes by
+     * heading level, so emitting them whole would collapse a document into one
+     * chunk. Splitting at every heading the grammar reports (so a `#` inside a
+     * fenced code block is not mistaken for one) gives complete,
+     * non-overlapping, heading-granular chunks instead.
+     */
+    private chunkMarkdown;
+    private collect;
+    private hasChunkableDescendants;
+    private nodeName;
+    private getParser;
+    /**
+     * Grammars are cached for the process lifetime. They are wasm modules whose
+     * linear memory only ever grows, so loading one repeatedly would leak.
+     */
+    private loadGrammar;
+}