@itradingai/aiwiki 0.2.5

package/dist/src/ingest.js

@@ -0,0 +1,428 @@
+import { promises as fs } from "node:fs";
+import path from "node:path";
+import { randomBytes } from "node:crypto";
+import { normalizePayload } from "./payload.js";
+import { appendRunIdBeforeExt, relativePath, safeJoin, slugify } from "./paths.js";
+import { initWorkspace } from "./workspace.js";
+export async function ingestPayload(rootPath, rawPayload) {
+    await initWorkspace(rootPath);
+    const root = path.resolve(rootPath);
+    const runStartedAt = new Date().toISOString();
+    const runId = createRunId(runStartedAt);
+    const payload = normalizePayload(rawPayload, runStartedAt);
+    const runDirName = payload.source.fetch_status === "failed" ? `${runId}-fetch-failed` : runId;
+    const runDir = safeJoin(root, "09-runs", runDirName);
+    await fs.mkdir(runDir, { recursive: false });
+    const generatedFiles = [];
+    await writeFile(path.join(runDir, "payload.json"), `${JSON.stringify(payload, null, 2)}\n`, generatedFiles);
+    if (payload.source.fetch_status === "failed") {
+        await writeSummary(root, runDir, payload, generatedFiles, [
+            ...payload.warnings,
+            "宿主 Agent 未能提供正文，AIWiki CLI 没有自行抓取网页。"
+        ]);
+        return { runId: runDirName, runDir, generatedFiles, warnings: payload.warnings, agentReport: buildAgentReport(root, runDir, payload, generatedFiles) };
+    }
+    const slug = slugify(payload.source.title ?? payload.source.url);
+    const content = payload.source.content ?? "";
+    const collisionWarnings = [];
+    const longTermTargets = await chooseLongTermTargets(root, slug, runId, collisionWarnings);
+    const links = buildArtifactLinks(root, slug, runDirName, runStartedAt, longTermTargets);
+    await writeFile(path.join(runDir, "raw.md"), contentFile(payload, content, links), generatedFiles);
+    await writeFile(path.join(runDir, "source-card.md"), sourceCard(payload, runDirName, links), generatedFiles);
+    await writeFile(path.join(runDir, "creative-assets.md"), creativeAssets(payload, links), generatedFiles);
+    await writeFile(path.join(runDir, "topics.md"), topics(payload, links), generatedFiles);
+    await writeFile(path.join(runDir, "draft-outline.md"), outline(payload, links), generatedFiles);
+    await writeFile(longTermTargets.raw, contentFile(payload, content, links), generatedFiles);
+    await writeFile(longTermTargets.sourceCard, sourceCard(payload, runDirName, links), generatedFiles);
+    await writeFile(longTermTargets.claims, claims(payload, links), generatedFiles);
+    await writeFile(longTermTargets.assets, creativeAssets(payload, links), generatedFiles);
+    await writeFile(longTermTargets.topics, topics(payload, links), generatedFiles);
+    await writeFile(longTermTargets.outline, outline(payload, links), generatedFiles);
+    const warnings = [...payload.warnings, ...collisionWarnings];
+    await writeSummary(root, runDir, payload, generatedFiles, warnings, links);
+    return { runId, runDir, generatedFiles, warnings, agentReport: buildAgentReport(root, runDir, payload, generatedFiles) };
+}
+export async function ingestFile(rootPath, filePath) {
+    const content = await fs.readFile(filePath, "utf8");
+    if (!content.trim()) {
+        throw new Error("input file is empty");
+    }
+    return ingestPayload(rootPath, {
+        schema_version: "aiwiki.agent_payload.v1",
+        source: {
+            kind: "file",
+            title: path.basename(filePath),
+            content_format: "markdown",
+            content,
+            fetcher: "local-file",
+            fetch_status: "ok",
+            captured_at: new Date().toISOString()
+        },
+        request: {
+            mode: "ingest",
+            outputs: ["source_card", "creative_assets", "topics", "draft_outline", "processing_summary"],
+            language: "zh-CN"
+        }
+    });
+}
+async function chooseLongTermTargets(root, slug, runId, warnings) {
+    return {
+        raw: await chooseLongTermTarget(root, "02-raw/articles", `${slug}.md`, runId, warnings),
+        sourceCard: await chooseLongTermTarget(root, "03-sources/article-cards", `${slug}.md`, runId, warnings),
+        claims: await chooseLongTermTarget(root, "04-claims/_suggestions", `${slug}-claims.md`, runId, warnings),
+        assets: await chooseLongTermTarget(root, "06-assets/_suggestions", `${slug}-assets.md`, runId, warnings),
+        topics: await chooseLongTermTarget(root, "07-topics/ready", `${slug}-topics.md`, runId, warnings),
+        outline: await chooseLongTermTarget(root, "08-outputs/outlines", `${slug}-outline.md`, runId, warnings)
+    };
+}
+async function chooseLongTermTarget(root, dir, fileName, runId, warnings) {
+    const target = safeJoin(root, dir, fileName);
+    try {
+        await fs.access(target);
+    }
+    catch (error) {
+        if (error.code === "ENOENT") {
+            return target;
+        }
+        throw error;
+    }
+    const renamed = appendRunIdBeforeExt(fileName, runId);
+    const renamedTarget = safeJoin(root, dir, renamed);
+    warnings.push(`collision renamed: ${relativePath(root, target)} -> ${relativePath(root, renamedTarget)}`);
+    return renamedTarget;
+}
+async function writeFile(target, content, generatedFiles) {
+    try {
+        await fs.writeFile(target, content, { encoding: "utf8", flag: "wx" });
+    }
+    catch (error) {
+        if (error.code === "EEXIST") {
+            throw new Error(`target file already exists: ${target}`);
+        }
+        throw error;
+    }
+    generatedFiles.push(target);
+}
+async function writeSummary(root, runDir, payload, generatedFiles, warnings, links) {
+    const summaryPath = path.join(runDir, "processing-summary.md");
+    const files = [...generatedFiles, summaryPath];
+    const runId = path.basename(runDir);
+    const slug = links?.slug ?? slugify(payload.source.title ?? payload.source.url);
+    const createdAt = links?.createdAt ?? new Date().toISOString();
+    const lines = [
+        "---",
+        `aiwiki_id: "${escapeYaml(`${slug}:run:${runId}`)}"`,
+        `type: "processing_summary"`,
+        `status: "${payload.source.fetch_status === "failed" ? "fetch-failed" : "to-review"}"`,
+        `slug: "${escapeYaml(slug)}"`,
+        `title: "${escapeYaml(payload.source.title ?? "Untitled")}"`,
+        `source_url: "${escapeYaml(payload.source.url ?? "")}"`,
+        `source_type: "${escapeYaml(payload.source.kind)}"`,
+        `created_at: "${escapeYaml(createdAt)}"`,
+        `captured_at: "${escapeYaml(payload.source.captured_at)}"`,
+        `run_id: "${escapeYaml(runId)}"`,
+        ...(links ? relationshipFrontmatter(links) : []),
+        `tags: ["aiwiki/run"]`,
+        "---",
+        "",
+        "# 处理记录",
+        "",
+        `来源类型：${payload.source.kind}`,
+        `读取状态：${payload.source.fetch_status}`,
+        `宿主读取器：${payload.source.fetcher ?? "unknown"}`,
+        "",
+        "生成文件：",
+        ...files.map((file) => `- ${obsidianFileReference(root, file)}`),
+        "",
+        "告警：",
+        ...(warnings.length ? warnings.map((warning) => `- ${warning}`) : ["- none"]),
+        "",
+        "下一步审阅：",
+        "- 请在 Obsidian 中人工审阅资料卡、Claim 建议、素材建议、选题和大纲。",
+        "- AIWiki CLI 不负责网页抓取稳定性。"
+    ];
+    await fs.writeFile(summaryPath, `${lines.join("\n")}\n`, { encoding: "utf8", flag: "wx" });
+    generatedFiles.push(summaryPath);
+}
+function contentFile(payload, content, links) {
+    return [
+        "---",
+        `aiwiki_id: "${escapeYaml(`${links.slug}:raw`)}"`,
+        `title: "${escapeYaml(payload.source.title ?? "Untitled")}"`,
+        `type: "raw_article"`,
+        `status: "to-review"`,
+        `slug: "${escapeYaml(links.slug)}"`,
+        `source_type: "${escapeYaml(payload.source.kind)}"`,
+        `source_url: "${escapeYaml(payload.source.url ?? "")}"`,
+        `created_at: "${escapeYaml(links.createdAt)}"`,
+        `captured_at: "${escapeYaml(payload.source.captured_at)}"`,
+        `run_id: "${escapeYaml(links.runId)}"`,
+        ...relationshipFrontmatter(links),
+        `tags: ["aiwiki/raw"]`,
+        "---",
+        "",
+        `# ${payload.source.title ?? "Untitled"}`,
+        "",
+        "## AIWiki 链接",
+        "",
+        `- 资料卡：${obsidianLink(links.sourceCard, "资料卡")}`,
+        `- 处理记录：${obsidianLink(links.runSummary, "处理记录")}`,
+        "",
+        content,
+        ""
+    ].join("\n");
+}
+function sourceCard(payload, runId, links) {
+    return [
+        "---",
+        `aiwiki_id: "${escapeYaml(`${links.slug}:source-card`)}"`,
+        `title: "${escapeYaml(payload.source.title ?? "Untitled")}"`,
+        `type: "source_card"`,
+        `status: "to-review"`,
+        `slug: "${escapeYaml(links.slug)}"`,
+        `source_type: "${escapeYaml(payload.source.kind)}"`,
+        `source_url: "${escapeYaml(payload.source.url ?? "")}"`,
+        `url: "${escapeYaml(payload.source.url ?? "")}"`,
+        `fetcher: "${escapeYaml(payload.source.fetcher ?? "unknown")}"`,
+        `created_at: "${escapeYaml(links.createdAt)}"`,
+        `captured_at: "${escapeYaml(payload.source.captured_at)}"`,
+        `run_id: "${escapeYaml(runId)}"`,
+        ...relationshipFrontmatter(links),
+        `aliases: ["${escapeYaml(payload.source.title ?? "Untitled")}"]`,
+        `tags: ["aiwiki/source-card"]`,
+        "---",
+        "",
+        `# ${payload.source.title ?? "Untitled"}`,
+        "",
+        "## Obsidian 链接",
+        "",
+        `- 原文：${obsidianLink(links.raw, "原文")}`,
+        `- Claim 建议：${obsidianLink(links.claims, "Claim 建议")}`,
+        `- 素材建议：${obsidianLink(links.assets, "素材建议")}`,
+        `- 选题：${obsidianLink(links.topics, "选题")}`,
+        `- 大纲：${obsidianLink(links.outline, "大纲")}`,
+        `- 处理记录：${obsidianLink(links.runSummary, "处理记录")}`,
+        "",
+        "## 摘要",
+        "",
+        trimPreview(payload.source.content ?? payload.source.fetch_notes ?? ""),
+        ""
+    ].join("\n");
+}
+function claims(payload, links) {
+    return [
+        "---",
+        `aiwiki_id: "${escapeYaml(`${links.slug}:claims`)}"`,
+        `title: "${escapeYaml(payload.source.title ?? "Untitled")} Claims"`,
+        `type: "claim_suggestions"`,
+        `status: "to-review"`,
+        `slug: "${escapeYaml(links.slug)}"`,
+        `source_url: "${escapeYaml(payload.source.url ?? "")}"`,
+        `source_type: "${escapeYaml(payload.source.kind)}"`,
+        `created_at: "${escapeYaml(links.createdAt)}"`,
+        `captured_at: "${escapeYaml(payload.source.captured_at)}"`,
+        `run_id: "${escapeYaml(links.runId)}"`,
+        ...relationshipFrontmatter(links),
+        `tags: ["aiwiki/claims"]`,
+        "---",
+        "",
+        "# Claim 建议",
+        "",
+        `- 资料卡：${obsidianLink(links.sourceCard, "资料卡")}`,
+        `- 原文：${obsidianLink(links.raw, "原文")}`,
+        `- 待人工审阅：${payload.source.title ?? "Untitled"}`,
+        ""
+    ].join("\n");
+}
+function creativeAssets(payload, links) {
+    return [
+        "---",
+        `aiwiki_id: "${escapeYaml(`${links.slug}:assets`)}"`,
+        `title: "${escapeYaml(payload.source.title ?? "Untitled")} Assets"`,
+        `type: "asset_suggestions"`,
+        `status: "to-review"`,
+        `slug: "${escapeYaml(links.slug)}"`,
+        `source_url: "${escapeYaml(payload.source.url ?? "")}"`,
+        `source_type: "${escapeYaml(payload.source.kind)}"`,
+        `created_at: "${escapeYaml(links.createdAt)}"`,
+        `captured_at: "${escapeYaml(payload.source.captured_at)}"`,
+        `run_id: "${escapeYaml(links.runId)}"`,
+        ...relationshipFrontmatter(links),
+        `tags: ["aiwiki/assets"]`,
+        "---",
+        "",
+        "# 素材建议",
+        "",
+        `- 资料卡：${obsidianLink(links.sourceCard, "资料卡")}`,
+        `- 原文：${obsidianLink(links.raw, "原文")}`,
+        `- 可复用素材：${payload.source.title ?? "Untitled"}`,
+        ""
+    ].join("\n");
+}
+function topics(payload, links) {
+    return [
+        "---",
+        `aiwiki_id: "${escapeYaml(`${links.slug}:topics`)}"`,
+        `title: "${escapeYaml(payload.source.title ?? "Untitled")} Topics"`,
+        `type: "topic_candidates"`,
+        `status: "ready"`,
+        `slug: "${escapeYaml(links.slug)}"`,
+        `source_url: "${escapeYaml(payload.source.url ?? "")}"`,
+        `source_type: "${escapeYaml(payload.source.kind)}"`,
+        `created_at: "${escapeYaml(links.createdAt)}"`,
+        `captured_at: "${escapeYaml(payload.source.captured_at)}"`,
+        `run_id: "${escapeYaml(links.runId)}"`,
+        ...relationshipFrontmatter(links),
+        `tags: ["aiwiki/topics"]`,
+        "---",
+        "",
+        "# 选题候选",
+        "",
+        `- 资料卡：${obsidianLink(links.sourceCard, "资料卡")}`,
+        `- 大纲：${obsidianLink(links.outline, "大纲")}`,
+        `- ${payload.source.title ?? "Untitled"}`,
+        ""
+    ].join("\n");
+}
+function outline(payload, links) {
+    return [
+        "---",
+        `aiwiki_id: "${escapeYaml(`${links.slug}:outline`)}"`,
+        `title: "${escapeYaml(payload.source.title ?? "Untitled")} Outline"`,
+        `type: "draft_outline"`,
+        `status: "draft"`,
+        `slug: "${escapeYaml(links.slug)}"`,
+        `source_url: "${escapeYaml(payload.source.url ?? "")}"`,
+        `source_type: "${escapeYaml(payload.source.kind)}"`,
+        `created_at: "${escapeYaml(links.createdAt)}"`,
+        `captured_at: "${escapeYaml(payload.source.captured_at)}"`,
+        `run_id: "${escapeYaml(links.runId)}"`,
+        ...relationshipFrontmatter(links),
+        `tags: ["aiwiki/outline"]`,
+        "---",
+        "",
+        "# 草稿大纲",
+        "",
+        `资料卡：${obsidianLink(links.sourceCard, "资料卡")}`,
+        `原文：${obsidianLink(links.raw, "原文")}`,
+        "",
+        "1. 背景",
+        "2. 关键观点",
+        "3. 可复用方法",
+        `4. 来源：${payload.source.title ?? "Untitled"}`,
+        ""
+    ].join("\n");
+}
+function trimPreview(value) {
+    return value.trim().slice(0, 500) || "待人工补充。";
+}
+function buildAgentReport(root, runDir, payload, generatedFiles) {
+    const content = payload.source.content ?? "";
+    const fetchFailed = payload.source.fetch_status === "failed";
+    const fitScore = fetchFailed ? 0 : estimateFitScore(payload, content);
+    return {
+        ingested: !fetchFailed,
+        recorded: true,
+        fetchStatus: payload.source.fetch_status,
+        sourceTitle: payload.source.title ?? "Untitled",
+        sourceUrl: payload.source.url,
+        fitScore,
+        fitLevel: fitLevel(fitScore, fetchFailed),
+        summary: fetchFailed ? (payload.source.fetch_notes ?? "宿主 Agent 没有提供可读正文。") : summarizeContent(content),
+        keyFiles: {
+            processingSummary: relativePath(root, path.join(runDir, "processing-summary.md")),
+            sourceCard: findGeneratedFileInDir(root, generatedFiles, "03-sources/article-cards"),
+            draftOutline: findGeneratedFile(root, generatedFiles, "draft-outline.md"),
+            dashboard: "dashboards/AIWiki Home.md",
+            reviewQueue: "dashboards/Review Queue.md"
+        }
+    };
+}
+function estimateFitScore(payload, content) {
+    let score = 45;
+    if (payload.source.kind === "url" || payload.source.kind === "file") {
+        score += 10;
+    }
+    if ((payload.source.title ?? "").trim()) {
+        score += 10;
+    }
+    if (content.trim().length >= 500) {
+        score += 20;
+    }
+    else if (content.trim().length >= 150) {
+        score += 10;
+    }
+    if (payload.source.url) {
+        score += 5;
+    }
+    return Math.min(95, score);
+}
+function fitLevel(score, fetchFailed) {
+    if (fetchFailed) {
+        return "fetch_failed";
+    }
+    if (score >= 80) {
+        return "high";
+    }
+    if (score >= 60) {
+        return "medium";
+    }
+    return "low";
+}
+function summarizeContent(content) {
+    const compact = content.replace(/\s+/g, " ").trim();
+    if (!compact) {
+        return "没有提供可读正文。";
+    }
+    return compact.length > 180 ? `${compact.slice(0, 180)}...` : compact;
+}
+function findGeneratedFile(root, files, basename) {
+    const match = files.find((file) => path.basename(file) === basename);
+    return match ? relativePath(root, match) : undefined;
+}
+function findGeneratedFileInDir(root, files, dir) {
+    const match = files.find((file) => relativePath(root, file).startsWith(`${dir}/`));
+    return match ? relativePath(root, match) : undefined;
+}
+function buildArtifactLinks(root, slug, runDirName, createdAt, longTermTargets) {
+    return {
+        slug,
+        runId: runDirName,
+        createdAt,
+        raw: relativePath(root, longTermTargets.raw),
+        sourceCard: relativePath(root, longTermTargets.sourceCard),
+        claims: relativePath(root, longTermTargets.claims),
+        assets: relativePath(root, longTermTargets.assets),
+        topics: relativePath(root, longTermTargets.topics),
+        outline: relativePath(root, longTermTargets.outline),
+        runSummary: `09-runs/${runDirName}/processing-summary.md`
+    };
+}
+function relationshipFrontmatter(links) {
+    return [
+        `source_card: "${escapeYaml(obsidianLink(links.sourceCard, "资料卡"))}"`,
+        `raw_note: "${escapeYaml(obsidianLink(links.raw, "原文"))}"`,
+        `claims_note: "${escapeYaml(obsidianLink(links.claims, "Claim 建议"))}"`,
+        `assets_note: "${escapeYaml(obsidianLink(links.assets, "素材建议"))}"`,
+        `topics_note: "${escapeYaml(obsidianLink(links.topics, "选题"))}"`,
+        `outline_note: "${escapeYaml(obsidianLink(links.outline, "大纲"))}"`,
+        `run_summary: "${escapeYaml(obsidianLink(links.runSummary, "处理记录"))}"`
+    ];
+}
+function obsidianFileReference(root, file) {
+    const vaultPath = relativePath(root, file);
+    if (!vaultPath.toLowerCase().endsWith(".md")) {
+        return vaultPath;
+    }
+    return `${obsidianLink(vaultPath, path.basename(vaultPath, ".md"))} (${vaultPath})`;
+}
+function obsidianLink(vaultPath, label) {
+    return `[[${vaultPath.replace(/\\/g, "/").replace(/\.md$/i, "")}|${label}]]`;
+}
+function escapeYaml(value) {
+    return value.replace(/\\/g, "\\\\").replace(/"/g, '\\"');
+}
+function createRunId(now) {
+    const stamp = now.replace(/[-:.]/g, "").replace("T", "-").replace("Z", "");
+    return `${stamp}-${randomBytes(3).toString("hex")}`;
+}

package/dist/src/output.js

@@ -0,0 +1,10 @@
+export function writeLine(stream, line = "") {
+    stream.write(`${line}\n`);
+}
+export class CliError extends Error {
+    exitCode;
+    constructor(message, exitCode = 1) {
+        super(message);
+        this.exitCode = exitCode;
+    }
+}

package/dist/src/paths.js

@@ -0,0 +1,32 @@
+import path from "node:path";
+export function safeJoin(root, ...parts) {
+    const resolvedRoot = path.resolve(root);
+    const target = path.resolve(resolvedRoot, ...parts);
+    const relative = path.relative(resolvedRoot, target);
+    if (relative.startsWith("..") || path.isAbsolute(relative)) {
+        throw new Error(`Refusing to write outside workspace: ${target}`);
+    }
+    return target;
+}
+export function toPosixPath(filePath) {
+    return filePath.split(path.sep).join("/");
+}
+export function relativePath(root, target) {
+    return toPosixPath(path.relative(root, target));
+}
+export function slugify(value) {
+    const source = value?.trim() || "item";
+    const ascii = source
+        .normalize("NFKD")
+        .replace(/[^\w\s-]/g, "")
+        .trim()
+        .toLowerCase()
+        .replace(/[-\s_]+/g, "-")
+        .replace(/^-+|-+$/g, "");
+    return ascii || "item";
+}
+export function appendRunIdBeforeExt(fileName, runId) {
+    const ext = path.extname(fileName);
+    const base = fileName.slice(0, fileName.length - ext.length);
+    return `${base}-${runId}${ext}`;
+}

package/dist/src/payload.js

@@ -0,0 +1,95 @@
+export function normalizePayload(raw, runStartedAt) {
+    if (!isRecord(raw)) {
+        throw new Error("payload must be a JSON object");
+    }
+    if (raw.schema_version !== "aiwiki.agent_payload.v1") {
+        throw new Error("schema_version must be aiwiki.agent_payload.v1");
+    }
+    rejectWriteControlFields(raw);
+    const sourceRaw = raw.source;
+    if (!isRecord(sourceRaw)) {
+        throw new Error("source is required");
+    }
+    const legacyContent = isRecord(raw.content) ? stringValue(raw.content.text) : undefined;
+    const legacyMetadata = isRecord(raw.metadata) ? raw.metadata : undefined;
+    const warnings = [];
+    const content = stringValue(sourceRaw.content) ?? legacyContent;
+    const fetcher = stringValue(sourceRaw.fetcher) ?? (legacyMetadata ? stringValue(legacyMetadata.fetcher) : undefined);
+    const legacyCapturedAt = legacyMetadata ? stringValue(legacyMetadata.captured_at) : undefined;
+    const capturedAt = stringValue(sourceRaw.captured_at) ?? legacyCapturedAt ?? runStartedAt;
+    if (!stringValue(sourceRaw.captured_at) && legacyCapturedAt) {
+        warnings.push("metadata.captured_at 已规范化为 source.captured_at。");
+    }
+    if (!stringValue(sourceRaw.captured_at) && !legacyCapturedAt) {
+        warnings.push("缺少 captured_at，已使用 run_started_at 补齐。");
+    }
+    const fetchStatus = normalizeFetchStatus(stringValue(sourceRaw.fetch_status), content);
+    if (fetchStatus !== "failed" && !content?.trim()) {
+        throw new Error("source.content is required when source.fetch_status is ok");
+    }
+    const kind = stringValue(sourceRaw.kind);
+    if (!kind) {
+        throw new Error("source.kind is required");
+    }
+    const requestRaw = isRecord(raw.request) ? raw.request : {};
+    const requestedOutputs = Array.isArray(requestRaw.outputs)
+        ? requestRaw.outputs.filter((item) => typeof item === "string")
+        : ["source_card", "creative_assets", "topics", "draft_outline", "processing_summary"];
+    const outputs = ["source_card", "creative_assets", "topics", "draft_outline", "processing_summary"];
+    if (fetchStatus !== "failed" && requestedOutputs.length && requestedOutputs.length !== outputs.length) {
+        warnings.push("基础版会生成完整单条资料产物，request.outputs 已按全量输出处理。");
+    }
+    if (typeof raw.target_kb === "string" && raw.target_kb.trim()) {
+        warnings.push(`target_kb=${raw.target_kb} 已被单知识库流程忽略。`);
+    }
+    if (fetchStatus === "failed" && content?.trim()) {
+        throw new Error("source.content must be empty when source.fetch_status is failed");
+    }
+    return {
+        schema_version: "aiwiki.agent_payload.v1",
+        target_kb: stringValue(raw.target_kb),
+        source: {
+            kind,
+            url: stringValue(sourceRaw.url),
+            title: stringValue(sourceRaw.title),
+            author: stringValue(sourceRaw.author),
+            platform: stringValue(sourceRaw.platform),
+            content_format: stringValue(sourceRaw.content_format),
+            content,
+            fetcher,
+            fetch_status: fetchStatus,
+            fetch_notes: stringValue(sourceRaw.fetch_notes),
+            captured_at: capturedAt,
+            language: stringValue(sourceRaw.language) ?? (legacyMetadata ? stringValue(legacyMetadata.language) : undefined)
+        },
+        request: {
+            mode: stringValue(requestRaw.mode) ?? (fetchStatus === "failed" ? "record_fetch_failure" : "ingest"),
+            outputs,
+            language: stringValue(requestRaw.language) ?? stringValue(sourceRaw.language)
+        },
+        warnings
+    };
+}
+function normalizeFetchStatus(value, content) {
+    if (value === "failed") {
+        return "failed";
+    }
+    if (value === "ok") {
+        return "ok";
+    }
+    return content ? "ok" : "failed";
+}
+function rejectWriteControlFields(raw) {
+    if (isRecord(raw.output) && (typeof raw.output.path === "string" || typeof raw.output.dir === "string")) {
+        throw new Error("payload must not control output paths");
+    }
+    if (typeof raw.output_file === "string") {
+        throw new Error("payload must not control output paths");
+    }
+}
+function stringValue(value) {
+    return typeof value === "string" ? value : undefined;
+}
+function isRecord(value) {
+    return typeof value === "object" && value !== null && !Array.isArray(value);
+}