@functional-systems/srcx 0.1.0-dev.4 → 0.1.0-dev.40

package/dist/implementation.js

@@ -0,0 +1,191 @@
+import { CHUNKER, detectLanguage, tokens, withSyntaxTree } from "./chunk.js";
+import { invariant, lineAt } from "./common.js";
+const children = (n) => n.namedChildren.filter((c) => !!c);
+/** Only imported typing overload aliases used exclusively as decorators are safe here. */
+function overloadAliases(root) {
+    const aliases = new Set(), imports = new Set();
+    for (const n of children(root)) {
+        if (n.type !== "import_from_statement" ||
+            !["typing", "typing_extensions"].includes(n.childForFieldName("module_name")?.text ?? ""))
+            continue;
+        for (const name of n.childrenForFieldName("name")) {
+            if (!name)
+                continue;
+            if (name.type === "dotted_name" && name.text === "overload") {
+                aliases.add("overload");
+                imports.add(name.id);
+            }
+            else if (name.type === "aliased_import" &&
+                name.childForFieldName("name")?.text === "overload") {
+                const alias = name.childForFieldName("alias");
+                if (alias) {
+                    aliases.add(alias.text);
+                    imports.add(name.id);
+                }
+            }
+        }
+    }
+    function visit(n) {
+        if (imports.has(n.id))
+            return;
+        if (n.type === "wildcard_import")
+            aliases.clear();
+        if (n.type === "identifier" &&
+            aliases.has(n.text) &&
+            !(n.parent?.type === "decorator" && n.parent.text === `@${n.text}`))
+            aliases.delete(n.text);
+        for (const child of children(n))
+            visit(child);
+    }
+    visit(root);
+    return aliases;
+}
+function definition(n, python, aliases) {
+    const d = n.childForFieldName(python ? "definition" : "declaration") ?? n;
+    const name = d.childForFieldName("name");
+    if (!name ||
+        ![
+            "identifier",
+            "property_identifier",
+            "private_property_identifier",
+        ].includes(name.type))
+        return;
+    if (python) {
+        if (d.type !== "function_definition")
+            return;
+        const decorators = children(n).filter((c) => c.type === "decorator");
+        const declaration = decorators.some((c) => aliases.has(c.text.slice(1)));
+        const body = d.childForFieldName("body");
+        // Stubs without an overload marker do not supply an implementation.
+        const executable = body &&
+            children(body).some((c) => c.type !== "comment" &&
+                c.type !== "pass_statement" &&
+                !(c.type === "expression_statement" &&
+                    ["ellipsis", "string"].includes(c.firstNamedChild?.type ?? "")));
+        if (!declaration && !executable)
+            return;
+        // Other decorators may change dispatch (property, singledispatch, etc.).
+        if (decorators.some((c) => !aliases.has(c.text.slice(1)) &&
+            !["@staticmethod", "@classmethod"].includes(c.text)))
+            return;
+        const modifier = decorators
+            .filter((c) => ["@staticmethod", "@classmethod"].includes(c.text))
+            .map((c) => c.text)
+            .sort()
+            .join(":");
+        const async = d.children.some((c) => c?.type === "async");
+        return {
+            node: n,
+            name: name.text,
+            key: `${name.text}:${modifier}:${async}`,
+            declaration,
+        };
+    }
+    if (n !== d && n.type !== "export_statement")
+        return;
+    if (![
+        "function_signature",
+        "function_declaration",
+        "method_signature",
+        "method_definition",
+    ].includes(d.type))
+        return;
+    const prefix = d.text.slice(0, name.startIndex - d.startIndex);
+    if (/\b(abstract|get|set|declare)\b/.test(prefix))
+        return;
+    const method = d.type.startsWith("method");
+    const declaration = d.type.endsWith("signature");
+    if (!declaration && !d.childForFieldName("body"))
+        return;
+    return {
+        node: n,
+        name: name.text,
+        key: `${method}:${/\bstatic\b/.test(prefix)}:${name.text}`,
+        declaration,
+    };
+}
+/** Follow only a contiguous, unambiguous overload group in the same AST container. */
+export async function implementationSpan(source, path, span) {
+    const raw = Buffer.from(source), lang = detectLanguage(path);
+    invariant(Number.isInteger(span.startByte) &&
+        Number.isInteger(span.endByte) &&
+        span.startByte >= 0 &&
+        span.endByte > span.startByte &&
+        span.endByte <= raw.length, "Invalid implementation source range.");
+    if (!["python", "typescript", "tsx"].includes(lang) ||
+        /\.(pyi|d\.ts|d\.tsx)$/i.test(path))
+        return { status: "unsupported" };
+    const start = raw.subarray(0, span.startByte).toString("utf8").length;
+    const end = raw.subarray(0, span.endByte).toString("utf8").length;
+    return withSyntaxTree(source, lang, (root) => {
+        if (root.hasError)
+            return { status: "parse-error" };
+        const python = lang === "python", aliases = python ? overloadAliases(root) : new Set();
+        const targets = [];
+        function visit(container) {
+            const siblings = children(container).filter((n) => n.type !== "comment");
+            const definitions = siblings.map((n) => definition(n, python, aliases));
+            for (let i = 0; i < definitions.length; i++) {
+                const d = definitions[i];
+                // A class/container hit is not a declaration hit. Require the selected
+                // span to end before the implementation, and contain a whole declaration.
+                if (!d?.declaration ||
+                    d.node.startIndex < start ||
+                    d.node.endIndex > end)
+                    continue;
+                let first = i;
+                while (first > 0 &&
+                    definitions[first - 1]?.declaration &&
+                    definitions[first - 1]?.key === d.key)
+                    first--;
+                let next = first;
+                while (definitions[next]?.declaration &&
+                    definitions[next]?.key === d.key)
+                    next++;
+                const target = definitions[next];
+                if (!target ||
+                    target.declaration ||
+                    target.key !== d.key ||
+                    target.node.startIndex < end)
+                    continue;
+                const matches = siblings.filter((n) => {
+                    const node = n.childForFieldName(python ? "definition" : "declaration") ?? n;
+                    // TypeScript namespaces augment a function without adding another body.
+                    // Keep other same-name declarations in the ambiguity check.
+                    if (!python && node.type === "internal_module")
+                        return false;
+                    return node.childForFieldName("name")?.text === d.name;
+                });
+                if (matches.length !== next - first + 1)
+                    continue;
+                // Reject unrelated code included in the selected chunk, but allow trivia
+                // between declarations (the chunker can attach comments/whitespace).
+                const selected = siblings.filter((n) => n.startIndex < end && n.endIndex > start);
+                if (selected.some((n) => !definitions.slice(first, next).some((x) => x?.node.id === n.id)))
+                    continue;
+                if (!targets.some((t) => t.node.id === target.node.id))
+                    targets.push(target);
+            }
+            for (const child of siblings)
+                visit(child);
+        }
+        visit(root);
+        if (targets.length !== 1)
+            return { status: "not-linked" };
+        const target = targets[0];
+        const startLine = lineAt(raw, Buffer.byteLength(source.slice(0, target.node.startIndex)));
+        const endLine = lineAt(raw, Buffer.byteLength(source.slice(0, target.node.endIndex)) - 1);
+        const text = (source.match(/[^\n]*\n|[^\n]+$/g) ?? [])
+            .slice(startLine - 1, endLine)
+            .join("");
+        if (tokens(text) > CHUNKER.maxTokens)
+            return { status: "too-large" };
+        return {
+            status: "resolved",
+            startLine,
+            endLine,
+            symbol: target.name,
+        };
+    });
+}
+//# sourceMappingURL=implementation.js.map

package/dist/implementation.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"implementation.js","sourceRoot":"","sources":["../src/implementation.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,OAAO,EAAE,cAAc,EAAE,MAAM,EAAE,cAAc,EAAE,MAAM,YAAY,CAAC;AAC7E,OAAO,EAAE,SAAS,EAAE,MAAM,EAAE,MAAM,aAAa,CAAC;AAehD,MAAM,QAAQ,GAAG,CAAC,CAAO,EAAU,EAAE,CACnC,CAAC,CAAC,aAAa,CAAC,MAAM,CAAC,CAAC,CAAC,EAAa,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;AAEhD,0FAA0F;AAC1F,SAAS,eAAe,CAAC,IAAU;IACjC,MAAM,OAAO,GAAG,IAAI,GAAG,EAAU,EAC/B,OAAO,GAAG,IAAI,GAAG,EAAU,CAAC;IAC9B,KAAK,MAAM,CAAC,IAAI,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;QAC/B,IACE,CAAC,CAAC,IAAI,KAAK,uBAAuB;YAClC,CAAC,CAAC,QAAQ,EAAE,mBAAmB,CAAC,CAAC,QAAQ,CACvC,CAAC,CAAC,iBAAiB,CAAC,aAAa,CAAC,EAAE,IAAI,IAAI,EAAE,CAC/C;YAED,SAAS;QACX,KAAK,MAAM,IAAI,IAAI,CAAC,CAAC,oBAAoB,CAAC,MAAM,CAAC,EAAE,CAAC;YAClD,IAAI,CAAC,IAAI;gBAAE,SAAS;YACpB,IAAI,IAAI,CAAC,IAAI,KAAK,aAAa,IAAI,IAAI,CAAC,IAAI,KAAK,UAAU,EAAE,CAAC;gBAC5D,OAAO,CAAC,GAAG,CAAC,UAAU,CAAC,CAAC;gBACxB,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;YACvB,CAAC;iBAAM,IACL,IAAI,CAAC,IAAI,KAAK,gBAAgB;gBAC9B,IAAI,CAAC,iBAAiB,CAAC,MAAM,CAAC,EAAE,IAAI,KAAK,UAAU,EACnD,CAAC;gBACD,MAAM,KAAK,GAAG,IAAI,CAAC,iBAAiB,CAAC,OAAO,CAAC,CAAC;gBAC9C,IAAI,KAAK,EAAE,CAAC;oBACV,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;oBACxB,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;gBACvB,CAAC;YACH,CAAC;QACH,CAAC;IACH,CAAC;IACD,SAAS,KAAK,CAAC,CAAO;QACpB,IAAI,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;YAAE,OAAO;QAC9B,IAAI,CAAC,CAAC,IAAI,KAAK,iBAAiB;YAAE,OAAO,CAAC,KAAK,EAAE,CAAC;QAClD,IACE,CAAC,CAAC,IAAI,KAAK,YAAY;YACvB,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC;YACnB,CAAC,CAAC,CAAC,CAAC,MAAM,EAAE,IAAI,KAAK,WAAW,IAAI,CAAC,CAAC,MAAM,CAAC,IAAI,KAAK,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC;YAEnE,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;QACzB,KAAK,MAAM,KAAK,IAAI,QAAQ,CAAC,CAAC,CAAC;YAAE,KAAK,CAAC,KAAK,CAAC,CAAC;IAChD,CAAC;IACD,KAAK,CAAC,IAAI,CAAC,CAAC;IACZ,OAAO,OAAO,CAAC;AACjB,CAAC;AACD,SAAS,UAAU,CACjB,CAAO,EACP,MAAe,EACf,OAAoB;IAEpB,MAAM,CAAC,GAAG,CAAC,CAAC,iBAAiB,CAAC,MAAM,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC;IAC1E,MAAM,IAAI,GAAG,CAAC,CAAC,iBAAiB,CAAC,MAAM,CAAC,CAAC;IACzC,IACE,CAAC,IAAI;QACL,CAAC;YACC,YAAY;YACZ,qBAAqB;YACrB,6BAA6B;SAC9B,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC;QAErB,OAAO;IACT,IAAI,MAAM,EAAE,CAAC;QACX,IAAI,CAAC,CAAC,IAAI,KAAK,qBAAqB;YAAE,OAAO;QAC7C,MAAM,UAAU,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,WAAW,CAAC,CAAC;QACrE,MAAM,WAAW,GAAG,UAAU,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QACzE,MAAM,IAAI,GAAG,CAAC,CAAC,iBAAiB,CAAC,MAAM,CAAC,CAAC;QACzC,oEAAoE;QACpE,MAAM,UAAU,GACd,IAAI;YACJ,QAAQ,CAAC,IAAI,CAAC,CAAC,IAAI,CACjB,CAAC,CAAC,EAAE,EAAE,CACJ,CAAC,CAAC,IAAI,KAAK,SAAS;gBACpB,CAAC,CAAC,IAAI,KAAK,gBAAgB;gBAC3B,CAAC,CACC,CAAC,CAAC,IAAI,KAAK,sBAAsB;oBACjC,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,eAAe,EAAE,IAAI,IAAI,EAAE,CAAC,CAC/D,CACJ,CAAC;QACJ,IAAI,CAAC,WAAW,IAAI,CAAC,UAAU;YAAE,OAAO;QACxC,yEAAyE;QACzE,IACE,UAAU,CAAC,IAAI,CACb,CAAC,CAAC,EAAE,EAAE,CACJ,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;YAC7B,CAAC,CAAC,eAAe,EAAE,cAAc,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,CACtD;YAED,OAAO;QACT,MAAM,QAAQ,GAAG,UAAU;aACxB,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,eAAe,EAAE,cAAc,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;aACjE,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC;aAClB,IAAI,EAAE;aACN,IAAI,CAAC,GAAG,CAAC,CAAC;QACb,MAAM,KAAK,GAAG,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,IAAI,KAAK,OAAO,CAAC,CAAC;QAC1D,OAAO;YACL,IAAI,EAAE,CAAC;YACP,IAAI,EAAE,IAAI,CAAC,IAAI;YACf,GAAG,EAAE,GAAG,IAAI,CAAC,IAAI,IAAI,QAAQ,IAAI,KAAK,EAAE;YACxC,WAAW;SACZ,CAAC;IACJ,CAAC;IACD,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,IAAI,KAAK,kBAAkB;QAAE,OAAO;IACrD,IACE,CAAC;QACC,oBAAoB;QACpB,sBAAsB;QACtB,kBAAkB;QAClB,mBAAmB;KACpB,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC;QAElB,OAAO;IACT,MAAM,MAAM,GAAG,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,UAAU,GAAG,CAAC,CAAC,UAAU,CAAC,CAAC;IAC/D,IAAI,gCAAgC,CAAC,IAAI,CAAC,MAAM,CAAC;QAAE,OAAO;IAC1D,MAAM,MAAM,GAAG,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAC;IAC3C,MAAM,WAAW,GAAG,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,WAAW,CAAC,CAAC;IACjD,IAAI,CAAC,WAAW,IAAI,CAAC,CAAC,CAAC,iBAAiB,CAAC,MAAM,CAAC;QAAE,OAAO;IACzD,OAAO;QACL,IAAI,EAAE,CAAC;QACP,IAAI,EAAE,IAAI,CAAC,IAAI;QACf,GAAG,EAAE,GAAG,MAAM,IAAI,YAAY,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC,IAAI,EAAE;QAC1D,WAAW;KACZ,CAAC;AACJ,CAAC;AAED,sFAAsF;AACtF,MAAM,CAAC,KAAK,UAAU,kBAAkB,CACtC,MAAc,EACd,IAAY,EACZ,IAA4C;IAE5C,MAAM,GAAG,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,EAC7B,IAAI,GAAG,cAAc,CAAC,IAAI,CAAC,CAAC;IAC9B,SAAS,CACP,MAAM,CAAC,SAAS,CAAC,IAAI,CAAC,SAAS,CAAC;QAC9B,MAAM,CAAC,SAAS,CAAC,IAAI,CAAC,OAAO,CAAC;QAC9B,IAAI,CAAC,SAAS,IAAI,CAAC;QACnB,IAAI,CAAC,OAAO,GAAG,IAAI,CAAC,SAAS;QAC7B,IAAI,CAAC,OAAO,IAAI,GAAG,CAAC,MAAM,EAC5B,sCAAsC,CACvC,CAAC;IACF,IACE,CAAC,CAAC,QAAQ,EAAE,YAAY,EAAE,KAAK,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC;QAC/C,wBAAwB,CAAC,IAAI,CAAC,IAAI,CAAC;QAEnC,OAAO,EAAE,MAAM,EAAE,aAAa,EAAE,CAAC;IACnC,MAAM,KAAK,GAAG,GAAG,CAAC,QAAQ,CAAC,CAAC,EAAE,IAAI,CAAC,SAAS,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC;IACtE,MAAM,GAAG,GAAG,GAAG,CAAC,QAAQ,CAAC,CAAC,EAAE,IAAI,CAAC,OAAO,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,MAAM,CAAC;IAClE,OAAO,cAAc,CAAC,MAAM,EAAE,IAAI,EAAE,CAAC,IAAI,EAAE,EAAE;QAC3C,IAAI,IAAI,CAAC,QAAQ;YAAE,OAAO,EAAE,MAAM,EAAE,aAAa,EAAE,CAAC;QACpD,MAAM,MAAM,GAAG,IAAI,KAAK,QAAQ,EAC9B,OAAO,GAAG,MAAM,CAAC,CAAC,CAAC,eAAe,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,GAAG,EAAU,CAAC;QAC/D,MAAM,OAAO,GAAiB,EAAE,CAAC;QACjC,SAAS,KAAK,CAAC,SAAe;YAC5B,MAAM,QAAQ,GAAG,QAAQ,CAAC,SAAS,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,SAAS,CAAC,CAAC;YACzE,MAAM,WAAW,GAAG,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,UAAU,CAAC,CAAC,EAAE,MAAM,EAAE,OAAO,CAAC,CAAC,CAAC;YACxE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,WAAW,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;gBAC5C,MAAM,CAAC,GAAG,WAAW,CAAC,CAAC,CAAC,CAAC;gBACzB,uEAAuE;gBACvE,0EAA0E;gBAC1E,IACE,CAAC,CAAC,EAAE,WAAW;oBACf,CAAC,CAAC,IAAI,CAAC,UAAU,GAAG,KAAK;oBACzB,CAAC,CAAC,IAAI,CAAC,QAAQ,GAAG,GAAG;oBAErB,SAAS;gBACX,IAAI,KAAK,GAAG,CAAC,CAAC;gBACd,OACE,KAAK,GAAG,CAAC;oBACT,WAAW,CAAC,KAAK,GAAG,CAAC,CAAC,EAAE,WAAW;oBACnC,WAAW,CAAC,KAAK,GAAG,CAAC,CAAC,EAAE,GAAG,KAAK,CAAC,CAAC,GAAG;oBAErC,KAAK,EAAE,CAAC;gBACV,IAAI,IAAI,GAAG,KAAK,CAAC;gBACjB,OACE,WAAW,CAAC,IAAI,CAAC,EAAE,WAAW;oBAC9B,WAAW,CAAC,IAAI,CAAC,EAAE,GAAG,KAAK,CAAC,CAAC,GAAG;oBAEhC,IAAI,EAAE,CAAC;gBACT,MAAM,MAAM,GAAG,WAAW,CAAC,IAAI,CAAC,CAAC;gBACjC,IACE,CAAC,MAAM;oBACP,MAAM,CAAC,WAAW;oBAClB,MAAM,CAAC,GAAG,KAAK,CAAC,CAAC,GAAG;oBACpB,MAAM,CAAC,IAAI,CAAC,UAAU,GAAG,GAAG;oBAE5B,SAAS;gBACX,MAAM,OAAO,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE;oBACpC,MAAM,IAAI,GACR,CAAC,CAAC,iBAAiB,CAAC,MAAM,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC;oBAClE,wEAAwE;oBACxE,4DAA4D;oBAC5D,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,IAAI,KAAK,iBAAiB;wBAAE,OAAO,KAAK,CAAC;oBAC7D,OAAO,IAAI,CAAC,iBAAiB,CAAC,MAAM,CAAC,EAAE,IAAI,KAAK,CAAC,CAAC,IAAI,CAAC;gBACzD,CAAC,CAAC,CAAC;gBACH,IAAI,OAAO,CAAC,MAAM,KAAK,IAAI,GAAG,KAAK,GAAG,CAAC;oBAAE,SAAS;gBAClD,yEAAyE;gBACzE,qEAAqE;gBACrE,MAAM,QAAQ,GAAG,QAAQ,CAAC,MAAM,CAC9B,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,UAAU,GAAG,GAAG,IAAI,CAAC,CAAC,QAAQ,GAAG,KAAK,CAChD,CAAC;gBACF,IACE,QAAQ,CAAC,IAAI,CACX,CAAC,CAAC,EAAE,EAAE,CACJ,CAAC,WAAW,CAAC,KAAK,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,EAAE,IAAI,CAAC,EAAE,KAAK,CAAC,CAAC,EAAE,CAAC,CACnE;oBAED,SAAS;gBACX,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,KAAK,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC;oBACpD,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;YACzB,CAAC;YACD,KAAK,MAAM,KAAK,IAAI,QAAQ;gBAAE,KAAK,CAAC,KAAK,CAAC,CAAC;QAC7C,CAAC;QACD,KAAK,CAAC,IAAI,CAAC,CAAC;QACZ,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC;YAAE,OAAO,EAAE,MAAM,EAAE,YAAY,EAAE,CAAC;QAC1D,MAAM,MAAM,GAAG,OAAO,CAAC,CAAC,CAAE,CAAC;QAC3B,MAAM,SAAS,GAAG,MAAM,CACtB,GAAG,EACH,MAAM,CAAC,UAAU,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC,CAC3D,CAAC;QACF,MAAM,OAAO,GAAG,MAAM,CACpB,GAAG,EACH,MAAM,CAAC,UAAU,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,GAAG,CAAC,CAC7D,CAAC;QACF,MAAM,IAAI,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,mBAAmB,CAAC,IAAI,EAAE,CAAC;aACnD,KAAK,CAAC,SAAS,GAAG,CAAC,EAAE,OAAO,CAAC;aAC7B,IAAI,CAAC,EAAE,CAAC,CAAC;QACZ,IAAI,MAAM,CAAC,IAAI,CAAC,GAAG,OAAO,CAAC,SAAS;YAAE,OAAO,EAAE,MAAM,EAAE,WAAW,EAAE,CAAC;QACrE,OAAO;YACL,MAAM,EAAE,UAAU;YAClB,SAAS;YACT,OAAO;YACP,MAAM,EAAE,MAAM,CAAC,IAAI;SACpB,CAAC;IACJ,CAAC,CAAC,CAAC;AACL,CAAC"}

package/dist/publish.d.ts

@@ -25,6 +25,8 @@ export type Journal = {
     attemptId: string;
     writer: string;
     source: string;
+    gitRef?: string;
+    reuseWriter?: boolean;
     baseline?: Published;
     phase: "branch" | "writing" | "waiting" | "publishing";
     batches: {
@@ -37,6 +39,22 @@ export type Journal = {
         snapshotId: string;
     };
 };
+/** The applied snapshot may differ from the shared canonical commit Tag. */
+export type TrackedBranch = Doc & {
+    gitRef: string;
+    writer: string;
+    applied?: Published;
+    publishedTagName?: string;
+    pending?: {
+        attemptId: string;
+        buildId: string;
+        commitOid: string;
+    };
+};
+export declare function gitBranchName(ref: string): string;
+export declare const gitBranchId: (ref: string) => string;
+export declare const lastBuildPath: (state: string, ref?: string) => string;
+export declare function trackedBranch(store: CollectionStore, binding: Binding, ref: string): Promise<TrackedBranch | undefined>;
 export declare function versionTag(oid: string): string;
 export declare function versionId(oid: string): string;
 export declare function exclusive<T>(directory: string, work: () => Promise<T>): Promise<T>;

package/dist/publish.js

@@ -2,8 +2,27 @@ import { mkdir, rm, readFile } from "node:fs/promises";
 import { join } from "node:path";
 import { randomUUID } from "node:crypto";
 import { atomic, canonical, hash, invariant, optionalJson, sleep, } from "./common.js";
-import { corpusMetadata, records, validateBuild } from "./build.js";
+import { corpusMetadata, loadBuild, records, recordHash, supportedPreset, validateBuild, } from "./build.js";
 import { branch, tag, one } from "./remote.js";
+export function gitBranchName(ref) {
+    invariant(ref.startsWith("refs/heads/") && ref.length > 11, "Expected a full Git branch ref.");
+    return `git-${hash(ref).slice(0, 40)}`;
+}
+export const gitBranchId = (ref) => `branch-${hash(ref)}`;
+export const lastBuildPath = (state, ref) => join(state, ref ? `last-build-${hash(ref)}.json` : "last-build.json");
+export async function trackedBranch(store, binding, ref) {
+    const doc = await one(store, branch("main"), gitBranchId(ref), true);
+    if (!doc)
+        return undefined;
+    invariant(doc.kind === "manifest" &&
+        doc.role === "git-branch" &&
+        doc.gitRef === ref &&
+        doc.writer === gitBranchName(ref) &&
+        doc.repoId === binding.repoId &&
+        doc.indexId === binding.indexId &&
+        doc.configHash === binding.configHash, "Tracked branch identity mismatch.");
+    return doc;
+}
 export function versionTag(oid) {
     return `ver-${hash(oid).slice(0, 40)}`;
 }
@@ -60,8 +79,13 @@ export async function validateCandidate(store, name, b, binding, attemptId) {
     for (const d of corpusMetadata(b, binding.repoId, binding.indexId, attemptId))
         expected.set(d.id, hash(d));
     for await (const d of store.list(tag(name))) {
-        if (expected.get(d.id) !== hash(d))
+        try {
+            if (expected.get(d.id) !== recordHash(d, b.preset))
+                return false;
+        }
+        catch {
             return false;
+        }
         expected.delete(d.id);
     }
     if (expected.size)
@@ -70,14 +94,11 @@ export async function validateCandidate(store, name, b, binding, attemptId) {
         let query = {
             queryString: { query: "kind:chunk" },
         };
-        for await (const d of records(b.directory)) {
+        for await (const d of records(b)) {
             if (d.kind !== "chunk")
                 continue;
-            // Keep a complete surface token: the standard analyzer may retain dots
-            // (e.g. code.ts), so extracting an alphanumeric substring is not a valid probe.
-            const term = String(d.symbol ?? d.searchText).match(/\S+/u)?.[0];
-            if (!term)
-                continue;
+            // Probe an indexed chunk ID: a valid source may contain only stop words
+            // or punctuation, so text-query matches cannot establish readiness.
             query = {
                 bool: [
                     {
@@ -90,16 +111,8 @@ export async function validateCandidate(store, name, b, binding, attemptId) {
                     },
                     {
                         queryString: {
-                            query: d.fileId,
-                            defaultField: "fileId",
-                            skipSyntax: true,
-                        },
-                        occur: "filter",
-                    },
-                    {
-                        queryString: {
-                            query: term,
-                            defaultField: "searchText",
+                            query: d.id,
+                            defaultField: "id",
                             skipSyntax: true,
                         },
                         occur: "must",
@@ -110,7 +123,8 @@ export async function validateCandidate(store, name, b, binding, attemptId) {
         }
         const found = await store.query(name, query, 1);
         if (!found.length ||
-            found.some((h) => h.doc.kind !== "chunk" || b.recordHashes[h.doc.id] !== hash(h.doc)))
+            found.some((h) => h.doc.kind !== "chunk" ||
+                b.recordHashes[h.doc.id] !== recordHash(h.doc, b.preset)))
             return false;
     }
     return true;
@@ -156,35 +170,133 @@ export async function published(store, binding) {
     return result.sort((a, b) => String(a.commitTime).localeCompare(String(b.commitTime)) ||
         a.commitOid.localeCompare(b.commitOid));
 }
+/** A local artifact is only a cache for this exact, previously published baseline. */
+async function baselineHashes(store, binding, preset, baseline, state, gitRef, local) {
+    const pin = (await store.tags()).find((t) => t.name === baseline.tagName);
+    const root = await one(store, tag(baseline.tagName), "__manifest__");
+    invariant(pin?.snapshotId === baseline.snapshotId &&
+        root?.repoId === binding.repoId &&
+        root.indexId === binding.indexId &&
+        root.configHash === binding.configHash &&
+        root.buildId === baseline.buildId &&
+        root.commitOid === baseline.commitOid &&
+        root.attemptId === baseline.attemptId &&
+        root.recordsHash === baseline.recordsHash &&
+        root.inventoryHash === baseline.inventoryHash, "Baseline Tag is missing or changed.");
+    if (!local) {
+        const last = await optionalJson(lastBuildPath(state, gitRef));
+        if (last) {
+            try {
+                local = await loadBuild(last.artifact);
+            }
+            catch (error) {
+                if (error.code !== "ENOENT")
+                    throw error;
+            }
+        }
+    }
+    if (local &&
+        local.repoKey === binding.repoKey &&
+        local.configHash === binding.configHash &&
+        local.buildId === baseline.buildId &&
+        local.commitOid === baseline.commitOid &&
+        local.recordsHash === baseline.recordsHash &&
+        local.inventoryHash === baseline.inventoryHash) {
+        try {
+            await validateBuild(local);
+            const hashes = new Map(Object.entries(local.recordHashes));
+            for (const doc of corpusMetadata(local, binding.repoId, binding.indexId, baseline.attemptId))
+                hashes.set(doc.id, hash(doc));
+            return hashes;
+        }
+        catch (error) {
+            if (error.code !== "ENOENT")
+                throw error;
+        }
+    }
+    // Retain recovery without local files, including explicit same-artifact retries.
+    const hashes = new Map();
+    for await (const doc of store.list(tag(baseline.tagName)))
+        hashes.set(doc.id, recordHash(doc, preset));
+    return hashes;
+}
 /** Caller holds the per-collection lock for the complete import. */
 export async function publish(args) {
     const { store, binding, build: b, state } = args;
     invariant(b.repoKey === binding.repoKey && b.configHash === binding.configHash, "Build does not belong to this index.");
-    invariant(!b.preset.embedding, "CLI publication currently supports only the embedding=none preset.");
+    invariant(!b.preset.embedding || supportedPreset(b.preset), "Publication requires a supported lexical or managed embedding preset.");
     await validateBuild(b);
     const journalPath = join(state, "pending.json");
     let j = await optionalJson(journalPath);
     if (j) {
         invariant(j.buildId === b.buildId &&
             j.recordsHash === b.recordsHash &&
-            j.inventoryHash === b.inventoryHash, "Another build is pending. Resume the artifact recorded in pending.json before importing a new commit.");
+            j.inventoryHash === b.inventoryHash &&
+            (j.gitRef === b.branchRef ||
+                (j.reuseWriter === undefined && j.writer.startsWith("work-"))), "Another build is pending. Resume the artifact recorded in pending.json before importing a new commit.");
         invariant(args.resume, "Pending import found; use --resume with the same artifact.");
     }
     const existing = (await published(store, binding)).find((v) => v.commitOid === b.commitOid);
+    const tracked = b.branchRef
+        ? await trackedBranch(store, binding, b.branchRef)
+        : undefined;
+    if (tracked?.pending)
+        invariant(j?.attemptId === tracked.pending.attemptId, "Git branch has a pending import; resume with its original retry journal.");
+    if (j?.gitRef && j.phase !== "branch")
+        invariant(tracked?.pending?.attemptId === j.attemptId ||
+            (j.phase === "publishing" &&
+                !tracked?.pending &&
+                tracked?.applied?.attemptId === j.attemptId), "Tracked branch import ownership changed; refusing to replay this journal.");
     if (existing) {
         invariant(await validateCandidate(store, existing.tagName, b, binding, existing.attemptId), "Existing published version differs from this build.");
-        await atomic(join(state, "last-build.json"), {
-            artifact: b.directory,
-            version: existing,
-        });
-        if (j)
+        // A shared commit Tag does not mean this Git branch has applied that commit.
+        if (!j && (!b.branchRef || tracked?.applied?.commitOid === b.commitOid)) {
+            if (tracked?.applied) {
+                const writer = (await store.branches()).find((x) => x.name === tracked.writer);
+                invariant(writer?.snapshotId === tracked.applied.snapshotId, "Writer baseline has changed; refusing to update tracked branch.");
+                invariant(await validateCandidate(store, tracked.applied.tagName, b, binding, tracked.applied.attemptId), "Tracked branch baseline validation failed.");
+            }
+            await atomic(lastBuildPath(state, b.branchRef), {
+                artifact: b.directory,
+                version: existing,
+            });
+            return existing;
+        }
+        // Legacy manual imports can finish bookkeeping after publication on retry.
+        if (j && !j.gitRef) {
+            await atomic(lastBuildPath(state), {
+                artifact: b.directory,
+                version: existing,
+            });
             await rm(journalPath);
-        return existing;
+            return existing;
+        }
     }
     if (!j) {
         let source = "checkpoint-empty";
         let baseline;
-        if (args.baseline) {
+        if (b.branchRef) {
+            const writer = (await store.branches()).find((x) => x.name === gitBranchName(b.branchRef));
+            if (tracked?.applied) {
+                baseline = tracked.applied;
+                invariant(writer?.snapshotId === baseline.snapshotId, "Writer baseline has changed; refusing to update tracked branch.");
+                const pin = (await store.tags()).find((t) => t.name === baseline.tagName);
+                const root = await one(store, tag(baseline.tagName), "__manifest__");
+                invariant(pin?.snapshotId === baseline.snapshotId &&
+                    root?.repoId === binding.repoId &&
+                    root.indexId === binding.indexId &&
+                    root.configHash === binding.configHash &&
+                    root.buildId === baseline.buildId &&
+                    root.attemptId === baseline.attemptId &&
+                    root.recordsHash === baseline.recordsHash &&
+                    root.inventoryHash === baseline.inventoryHash &&
+                    root.commitOid === baseline.commitOid, "Tracked branch baseline Tag is missing or changed.");
+            }
+            else {
+                invariant(!writer && !tracked, "Unowned or unfinished tracked branch; retain its retry journal.");
+            }
+        }
+        else if (args.baseline?.version.writer.startsWith("work-")) {
             const prev = args.baseline;
             await validateBuild(prev.build);
             invariant(prev.build.repoKey === b.repoKey &&
@@ -202,7 +314,9 @@ export async function publish(args) {
             inventoryHash: b.inventoryHash,
             artifact: b.directory,
             attemptId,
-            writer: `work-${attemptId}`,
+            writer: b.branchRef ? gitBranchName(b.branchRef) : `work-${attemptId}`,
+            gitRef: b.branchRef,
+            reuseWriter: !!tracked?.applied,
             source,
             baseline,
             phase: "branch",
@@ -212,12 +326,39 @@ export async function publish(args) {
     }
     const save = () => atomic(journalPath, j);
     if (j.phase === "branch") {
+        if (j.gitRef) {
+            // Record intent before any corpus mutation, including before branch creation.
+            // A fresh local state must not treat a lagging, partially written head as clean.
+            const current = await trackedBranch(store, binding, j.gitRef);
+            invariant((!current?.pending || current.pending.attemptId === j.attemptId) &&
+                current?.applied?.snapshotId === j.baseline?.snapshotId, "Git branch ownership or baseline changed before import.");
+            await store.upsert("main", [
+                {
+                    ...current,
+                    id: gitBranchId(j.gitRef),
+                    kind: "manifest",
+                    role: "git-branch",
+                    schemaVersion: 1,
+                    repoId: binding.repoId,
+                    indexId: binding.indexId,
+                    configHash: binding.configHash,
+                    gitRef: j.gitRef,
+                    writer: j.writer,
+                    pending: {
+                        attemptId: j.attemptId,
+                        buildId: j.buildId,
+                        commitOid: b.commitOid,
+                    },
+                },
+            ]);
+        }
         const branches = await store.branches();
         const found = branches.find((x) => x.name === j.writer);
         if (found)
-            invariant(found.parent === j.source &&
+            invariant((j.reuseWriter || found.parent === j.source) &&
                 found.snapshotId === (j.baseline?.snapshotId ?? null), "Partially created writer has an unexpected baseline.");
         else {
+            invariant(!j.reuseWriter, "Tracked writer is missing; refusing to recreate a nonempty baseline.");
             const source = branches.find((x) => x.name === j.source);
             invariant(source &&
                 (j.baseline
@@ -232,18 +373,14 @@ export async function publish(args) {
     const marker = metadata.at(-1);
     if (j.phase === "writing") {
         // Reconcile from the immutable old Tag, never from a half-written writer.
-        const oldIds = new Set();
-        const oldHashes = new Map();
-        if (j.baseline)
-            for await (const d of store.list(tag(j.baseline.tagName))) {
-                oldIds.add(d.id);
-                oldHashes.set(d.id, hash(d));
-            }
+        const oldHashes = j.baseline
+            ? await baselineHashes(store, binding, b.preset, j.baseline, state, j.gitRef, args.baseline?.build)
+            : new Map();
         const newIds = new Set([
             ...Object.keys(b.recordHashes),
             ...metadata.map((d) => d.id),
         ]);
-        const obsolete = [...oldIds].filter((id) => !newIds.has(id));
+        const obsolete = [...oldHashes.keys()].filter((id) => !newIds.has(id));
         const mutate = async (operation, payload, send) => {
             const receipt = {
                 operation,
@@ -261,11 +398,12 @@ export async function publish(args) {
             await mutate("delete", ids, () => store.delete(j.writer, ids));
         }
         async function* changed() {
-            for await (const d of records(b.directory))
+            for await (const d of records(b))
                 if (oldHashes.get(d.id) !== hash(d))
                     yield d;
             for (const d of metadata.slice(0, -1))
-                yield d;
+                if (oldHashes.get(d.id) !== hash(d))
+                    yield d;
         }
         for await (const docs of batches(changed()))
             await mutate("upsert", docs, () => store.upsert(j.writer, docs));
@@ -280,6 +418,7 @@ export async function publish(args) {
     const deadline = Date.now() + (args.timeoutMs ?? 120_000);
     const pause = args.pollMs ?? 1000;
     let rejectedSnapshot;
+    let validatedHere = false;
     while (j.phase === "waiting") {
         invariant(Date.now() < deadline, "Index validation timed out; build remains unpublished. Resume this artifact with --resume.");
         const visible = await one(store, branch(j.writer), "__manifest__", false);
@@ -301,16 +440,24 @@ export async function publish(args) {
             continue;
         }
         j.candidate = candidate;
+        validatedHere = true;
         j.phase = "publishing";
         await save();
     }
     invariant(j.candidate, "Missing validated candidate.");
-    invariant(await validateCandidate(store, j.candidate.name, b, binding, j.attemptId), "Validated candidate is missing or changed.");
+    // Never persist this shortcut: a resumed journal must revalidate the full corpus.
+    if (!validatedHere)
+        invariant(await validateCandidate(store, j.candidate.name, b, binding, j.attemptId), "Validated candidate is missing or changed.");
+    const pins = await store.tags();
+    invariant(pins.find((t) => t.name === j.candidate.name)?.snapshotId ===
+        j.candidate.snapshotId, "Validated candidate Tag is missing or changed.");
     const name = versionTag(b.commitOid);
-    const found = (await store.tags()).find((t) => t.name === name);
+    const found = pins.find((t) => t.name === name);
     const final = found ?? (await store.tag(name, tag(j.candidate.name)));
-    invariant(final.snapshotId === j.candidate.snapshotId, "Published Tag name already points to a different snapshot.");
-    const result = {
+    invariant(existing
+        ? final.snapshotId === existing.snapshotId
+        : final.snapshotId === j.candidate.snapshotId, "Published Tag name already points to a different snapshot.");
+    const result = existing ?? {
         id: versionId(b.commitOid),
         kind: "manifest",
         role: "published",
@@ -330,8 +477,39 @@ export async function publish(args) {
         counts: b.counts,
         validation: { inventory: true, content: true, query: true },
     };
-    await store.upsert("main", [result]);
-    await atomic(join(state, "last-build.json"), {
+    if (!existing)
+        await store.upsert("main", [result]);
+    if (j.gitRef) {
+        const current = await trackedBranch(store, binding, j.gitRef);
+        invariant(current?.pending?.attemptId === j.attemptId ||
+            (!current?.pending && current?.applied?.attemptId === j.attemptId), "Tracked branch publication ownership changed.");
+        // Pin the exact writer baseline even when another branch published this commit first.
+        const applied = {
+            ...result,
+            writer: j.writer,
+            snapshotId: j.candidate.snapshotId,
+            tagName: final.snapshotId === j.candidate.snapshotId
+                ? final.name
+                : j.candidate.name,
+            attemptId: j.attemptId,
+        };
+        await store.upsert("main", [
+            {
+                id: gitBranchId(j.gitRef),
+                kind: "manifest",
+                role: "git-branch",
+                schemaVersion: 1,
+                repoId: binding.repoId,
+                indexId: binding.indexId,
+                configHash: binding.configHash,
+                gitRef: j.gitRef,
+                writer: j.writer,
+                applied,
+                publishedTagName: result.tagName,
+            },
+        ]);
+    }
+    await atomic(lastBuildPath(state, j.gitRef), {
         artifact: b.directory,
         version: result,
     });