@gobing-ai/knowledge-kit 0.0.17 → 0.0.18

package/plugins/generations/news-report-gen/dist/index.js

@@ -14642,72 +14642,6 @@ var init_zod = __esm(() => {
   init_external();
 });
 
-// ../../node_modules/.bun/@gobing-ai+ts-utils@0.4.39/node_modules/@gobing-ai/ts-utils/dist/errors.js
-var ErrorCode;
-var init_errors3 = __esm(() => {
-  ErrorCode = {
-    NotFound: "NOT_FOUND",
-    Validation: "VALIDATION",
-    Conflict: "CONFLICT",
-    Internal: "INTERNAL"
-  };
-});
-
-// ../../node_modules/.bun/@gobing-ai+ts-utils@0.4.39/node_modules/@gobing-ai/ts-utils/dist/api-response.js
-var API_ERROR_CODES, ERROR_CODE_TO_HTTP, CLIENT_SAFE_CODES;
-var init_api_response = __esm(() => {
-  init_errors3();
-  API_ERROR_CODES = {
-    SUCCESS: 0,
-    NOT_FOUND: 404,
-    VALIDATION_ERROR: 422,
-    BAD_REQUEST: 400,
-    UNAUTHORIZED: 401,
-    FORBIDDEN: 403,
-    CONFLICT: 409,
-    INTERNAL_ERROR: 500
-  };
-  ERROR_CODE_TO_HTTP = {
-    [ErrorCode.NotFound]: API_ERROR_CODES.NOT_FOUND,
-    [ErrorCode.Validation]: API_ERROR_CODES.VALIDATION_ERROR,
-    [ErrorCode.Conflict]: API_ERROR_CODES.CONFLICT,
-    [ErrorCode.Internal]: API_ERROR_CODES.INTERNAL_ERROR
-  };
-  CLIENT_SAFE_CODES = new Set([ErrorCode.NotFound, ErrorCode.Validation, ErrorCode.Conflict]);
-});
-
-// ../../node_modules/.bun/@gobing-ai+ts-utils@0.4.39/node_modules/@gobing-ai/ts-utils/dist/cursor.js
-var init_cursor = () => {};
-
-// ../../node_modules/.bun/@gobing-ai+ts-utils@0.4.39/node_modules/@gobing-ai/ts-utils/dist/event-severity.js
-var init_event_severity = () => {};
-
-// ../../node_modules/.bun/@gobing-ai+ts-utils@0.4.39/node_modules/@gobing-ai/ts-utils/dist/output.js
-function processStream(name) {
-  const proc = globalThis.process;
-  const stream = proc?.[name];
-  if (stream === undefined) {
-    throw new Error(`No ${name} target available: set one via setDefaultOutputTargets or pass an explicit target`);
-  }
-  return stream;
-}
-function writeLine(message, target) {
-  target.write(`${message}
-`);
-}
-function echoError(message, target = defaultStderrTarget ?? processStream("stderr")) {
-  writeLine(message, target);
-}
-var defaultStderrTarget;
-
-// ../../node_modules/.bun/@gobing-ai+ts-utils@0.4.39/node_modules/@gobing-ai/ts-utils/dist/index.js
-var init_dist = __esm(() => {
-  init_api_response();
-  init_cursor();
-  init_errors3();
-  init_event_severity();
-});
-
 // ../../node_modules/.bun/yaml@2.9.0/node_modules/yaml/dist/nodes/identity.js
 var require_identity = __commonJS((exports) => {
   var ALIAS = Symbol.for("yaml.alias");
@@ -17612,15 +17546,15 @@ var require_timestamp = __commonJS((exports) => {
         throw new Error("!!timestamp expects a date, starting with yyyy-mm-dd");
       const [, year, month, day, hour, minute, second] = match.map(Number);
       const millisec = match[7] ? Number((match[7] + "00").substr(1, 3)) : 0;
-      let date6 = Date.UTC(year, month - 1, day, hour || 0, minute || 0, second || 0, millisec);
+      let date5 = Date.UTC(year, month - 1, day, hour || 0, minute || 0, second || 0, millisec);
       const tz = match[8];
       if (tz && tz !== "Z") {
         let d = parseSexagesimal(tz, false);
         if (Math.abs(d) < 30)
           d *= 60;
-        date6 -= 60000 * d;
+        date5 -= 60000 * d;
       }
-      return new Date(date6);
+      return new Date(date5);
     },
     stringify: ({ value }) => value?.toISOString().replace(/(T00:00:00)?\.000Z$/, "") ?? ""
   };
@@ -19480,7 +19414,7 @@ var require_composer = __commonJS((exports) => {
   var node_process = __require("process");
   var directives = require_directives();
   var Document = require_Document();
-  var errors4 = require_errors();
+  var errors3 = require_errors();
   var identity = require_identity();
   var composeDoc = require_compose_doc();
   var resolveEnd = require_resolve_end();
@@ -19531,9 +19465,9 @@ var require_composer = __commonJS((exports) => {
       this.onError = (source, code, message, warning) => {
         const pos = getErrorPos(source);
         if (warning)
-          this.warnings.push(new errors4.YAMLWarning(pos, code, message));
+          this.warnings.push(new errors3.YAMLWarning(pos, code, message));
         else
-          this.errors.push(new errors4.YAMLParseError(pos, code, message));
+          this.errors.push(new errors3.YAMLParseError(pos, code, message));
       };
       this.directives = new directives.Directives({ version: options.version || "1.2" });
       this.options = options;
@@ -19619,7 +19553,7 @@ ${cb}` : comment;
           break;
         case "error": {
           const msg = token.source ? `${token.message}: ${JSON.stringify(token.source)}` : token.message;
-          const error51 = new errors4.YAMLParseError(getErrorPos(token), "UNEXPECTED_TOKEN", msg);
+          const error51 = new errors3.YAMLParseError(getErrorPos(token), "UNEXPECTED_TOKEN", msg);
           if (this.atDirectives || !this.doc)
             this.errors.push(error51);
           else
@@ -19629,7 +19563,7 @@ ${cb}` : comment;
         case "doc-end": {
           if (!this.doc) {
             const msg = "Unexpected doc-end without preceding document";
-            this.errors.push(new errors4.YAMLParseError(getErrorPos(token), "UNEXPECTED_TOKEN", msg));
+            this.errors.push(new errors3.YAMLParseError(getErrorPos(token), "UNEXPECTED_TOKEN", msg));
             break;
           }
           this.doc.directives.docEnd = true;
@@ -19644,7 +19578,7 @@ ${end.comment}` : end.comment;
           break;
         }
         default:
-          this.errors.push(new errors4.YAMLParseError(getErrorPos(token), "UNEXPECTED_TOKEN", `Unsupported token ${token.type}`));
+          this.errors.push(new errors3.YAMLParseError(getErrorPos(token), "UNEXPECTED_TOKEN", `Unsupported token ${token.type}`));
       }
     }
     *end(forceDoc = false, endOffset = -1) {
@@ -19670,7 +19604,7 @@ ${end.comment}` : end.comment;
 var require_cst_scalar = __commonJS((exports) => {
   var resolveBlockScalar = require_resolve_block_scalar();
   var resolveFlowScalar = require_resolve_flow_scalar();
-  var errors4 = require_errors();
+  var errors3 = require_errors();
   var stringifyString = require_stringifyString();
   function resolveAsScalar(token, strict = true, onError) {
     if (token) {
@@ -19679,7 +19613,7 @@ var require_cst_scalar = __commonJS((exports) => {
         if (onError)
           onError(offset, code, message);
         else
-          throw new errors4.YAMLParseError([offset, offset + 1], code, message);
+          throw new errors3.YAMLParseError([offset, offset + 1], code, message);
       };
       switch (token.type) {
         case "scalar":
@@ -21559,7 +21493,7 @@ var require_parser = __commonJS((exports) => {
 var require_public_api = __commonJS((exports) => {
   var composer = require_composer();
   var Document = require_Document();
-  var errors4 = require_errors();
+  var errors3 = require_errors();
   var log = require_log();
   var identity = require_identity();
   var lineCounter = require_line_counter();
@@ -21576,8 +21510,8 @@ var require_public_api = __commonJS((exports) => {
     const docs = Array.from(composer$1.compose(parser$1.parse(source)));
     if (prettyErrors && lineCounter2)
       for (const doc2 of docs) {
-        doc2.errors.forEach(errors4.prettifyError(source, lineCounter2));
-        doc2.warnings.forEach(errors4.prettifyError(source, lineCounter2));
+        doc2.errors.forEach(errors3.prettifyError(source, lineCounter2));
+        doc2.warnings.forEach(errors3.prettifyError(source, lineCounter2));
       }
     if (docs.length > 0)
       return docs;
@@ -21592,13 +21526,13 @@ var require_public_api = __commonJS((exports) => {
       if (!doc2)
         doc2 = _doc;
       else if (doc2.options.logLevel !== "silent") {
-        doc2.errors.push(new errors4.YAMLParseError(_doc.range.slice(0, 2), "MULTIPLE_DOCS", "Source contains multiple documents; please use YAML.parseAllDocuments()"));
+        doc2.errors.push(new errors3.YAMLParseError(_doc.range.slice(0, 2), "MULTIPLE_DOCS", "Source contains multiple documents; please use YAML.parseAllDocuments()"));
         break;
       }
     }
     if (prettyErrors && lineCounter2) {
-      doc2.errors.forEach(errors4.prettifyError(source, lineCounter2));
-      doc2.warnings.forEach(errors4.prettifyError(source, lineCounter2));
+      doc2.errors.forEach(errors3.prettifyError(source, lineCounter2));
+      doc2.warnings.forEach(errors3.prettifyError(source, lineCounter2));
     }
     return doc2;
   }
@@ -21650,12 +21584,12 @@ var require_public_api = __commonJS((exports) => {
 });
 
 // ../../node_modules/.bun/yaml@2.9.0/node_modules/yaml/dist/index.js
-var composer, Document, Schema, errors4, Alias, identity, Pair, Scalar, YAMLMap, YAMLSeq, cst, lexer, lineCounter, parser, publicApi, visit, $Composer, $Document, $Schema, $YAMLError, $YAMLParseError, $YAMLWarning, $Alias, $isAlias, $isCollection, $isDocument, $isMap, $isNode, $isPair, $isScalar, $isSeq, $Pair, $Scalar, $YAMLMap, $YAMLSeq, $Lexer, $LineCounter, $Parser, $parse, $parseAllDocuments, $parseDocument, $stringify, $visit, $visitAsync;
-var init_dist2 = __esm(() => {
+var composer, Document, Schema, errors3, Alias, identity, Pair, Scalar, YAMLMap, YAMLSeq, cst, lexer, lineCounter, parser, publicApi, visit, $Composer, $Document, $Schema, $YAMLError, $YAMLParseError, $YAMLWarning, $Alias, $isAlias, $isCollection, $isDocument, $isMap, $isNode, $isPair, $isScalar, $isSeq, $Pair, $Scalar, $YAMLMap, $YAMLSeq, $Lexer, $LineCounter, $Parser, $parse, $parseAllDocuments, $parseDocument, $stringify, $visit, $visitAsync;
+var init_dist = __esm(() => {
   composer = require_composer();
   Document = require_Document();
   Schema = require_Schema();
-  errors4 = require_errors();
+  errors3 = require_errors();
   Alias = require_Alias();
   identity = require_identity();
   Pair = require_Pair();
@@ -21671,9 +21605,9 @@ var init_dist2 = __esm(() => {
   $Composer = composer.Composer;
   $Document = Document.Document;
   $Schema = Schema.Schema;
-  $YAMLError = errors4.YAMLError;
-  $YAMLParseError = errors4.YAMLParseError;
-  $YAMLWarning = errors4.YAMLWarning;
+  $YAMLError = errors3.YAMLError;
+  $YAMLParseError = errors3.YAMLParseError;
+  $YAMLWarning = errors3.YAMLWarning;
   $Alias = Alias.Alias;
   $isAlias = identity.isAlias;
   $isCollection = identity.isCollection;
@@ -21701,7 +21635,7 @@ var init_dist2 = __esm(() => {
 // ../../node_modules/.bun/@gobing-ai+ts-runtime@0.4.39/node_modules/@gobing-ai/ts-runtime/dist/config.js
 var configSchema;
 var init_config = __esm(() => {
-  init_dist2();
+  init_dist();
   init_zod();
   configSchema = exports_external.object({
     app: exports_external.object({
@@ -21952,10 +21886,27 @@ function getProcessPid() {
 function uniqueToken() {
   return `${Date.now()}.${crypto.randomUUID()}`;
 }
+// ../../node_modules/.bun/@gobing-ai+ts-utils@0.4.67/node_modules/@gobing-ai/ts-utils/dist/output.js
+var defaultStderrTarget;
+function processStream(name) {
+  const proc = globalThis.process;
+  const stream = proc?.[name];
+  if (stream === undefined) {
+    throw new Error(`No ${name} target available: set one via setDefaultOutputTargets or pass an explicit target`);
+  }
+  return stream;
+}
+function writeLine(message, target) {
+  target.write(`${message}
+`);
+}
+function echoError(message, target = defaultStderrTarget ?? processStream("stderr")) {
+  writeLine(message, target);
+}
+
 // ../../packages/kk-core/src/render-md.ts
 if (false) {}
 // ../../plugins/generations/news-report-gen/src/index.ts
-init_dist();
 var QC_DIMENSIONS = ["quality", "importance", "urgency", "impact"];
 var QC_MIN_ENVS = {
   quality: "QC_MIN_QUALITY",
@@ -22094,8 +22045,8 @@ function durationMs(entry) {
     return null;
   return Math.max(0, end - start);
 }
-function renderReportMarkdown(docs, date6, options = {}) {
-  const lines = [`# \u6BCF\u65E5 AI \u8BED\u97F3\u8FD0\u884C\u62A5\u544A \u2014 ${date6}`, ""];
+function renderReportMarkdown(docs, date5, options = {}) {
+  const lines = [`# \u6BCF\u65E5 AI \u8BED\u97F3\u8FD0\u884C\u62A5\u544A \u2014 ${date5}`, ""];
   lines.push("## \u9009\u7A3F\u8BC4\u5206\uFF08\u4FDD\u7559\u5019\u9009\uFF09", "");
   if (docs.length === 0) {
     lines.push("\u672C\u671F\u65E0\u4FDD\u7559\u5019\u9009\u3002", "");
@@ -22132,6 +22083,30 @@ function renderReportMarkdown(docs, date6, options = {}) {
   }
   lines.push(`| category allowlist | \u2014 | ${counts.category} |`);
   lines.push("");
+  const blended = options.blended ?? null;
+  if (blended !== null && blended.length > 0) {
+    const comp = corpusComposition(blended);
+    lines.push(`## \u8BED\u6599\u6784\u6210\uFF08fan-in \u8F93\u5165\uFF0C\u5171 ${comp.total} \u6761\uFF09`, "");
+    lines.push("| \u6765\u6E90 | \u6761\u6570 |", "| --- | --- |");
+    for (const g of comp.sources)
+      lines.push(`| ${g.key} | ${g.count} |`);
+    lines.push("");
+    lines.push("| \u7C7B\u522B | \u6761\u6570 |", "| --- | --- |");
+    for (const g of comp.categories)
+      lines.push(`| ${g.key} | ${g.count} |`);
+    lines.push("");
+  }
+  const planAudit = options.planAudit ?? null;
+  if (planAudit !== null && planAudit.length > 0) {
+    lines.push(`## \u9009\u7A3F\u6F0F\u6597\uFF08\u5168\u91CF\u5BA1\u8BA1\uFF0C\u5171 ${planAudit.length} \u6761\uFF09`, "");
+    lines.push("| \u6765\u6E90 | \u6807\u9898 | \u7C7B\u522B | q | i | u | im | score | \u53BB\u5411 |", "| --- | --- | --- | --- | --- | --- | --- | --- | --- |");
+    for (const row of planAudit) {
+      const s = row.scores ?? {};
+      const title = (row.title || row.id).replace(/\|/g, "\\|").slice(0, 60);
+      lines.push(`| ${row.source} | ${title} | ${row.category} | ${s.quality ?? "\u2014"} | ${s.importance ?? "\u2014"} | ${s.urgency ?? "\u2014"} | ${s.impact ?? "\u2014"} | ${row.planScore ?? "\u2014"} | ${row.disposition} |`);
+    }
+    lines.push("");
+  }
   const qc = options.qc ?? null;
   if (qc !== null) {
     lines.push("## \u97F3\u9891 QC", "");
@@ -22176,12 +22151,50 @@ async function readRejectedAudit(path2) {
     return [];
   }
 }
-function docsToContent(docs, date6, options) {
+async function readPlanAudit(path2) {
+  if (!path2 || path2.trim() === "")
+    return null;
+  try {
+    const parsed = JSON.parse(await Bun.file(path2).text());
+    const rows = parsed?.rows;
+    return Array.isArray(rows) ? rows : null;
+  } catch {
+    return null;
+  }
+}
+async function readBlendedCorpus(path2) {
+  if (!path2 || path2.trim() === "")
+    return null;
+  try {
+    const parsed = JSON.parse(await Bun.file(path2).text());
+    const docs = Array.isArray(parsed) ? parsed : parsed?.docs;
+    return Array.isArray(docs) ? docs : null;
+  } catch {
+    return null;
+  }
+}
+function corpusComposition(docs) {
+  const tally = (pick2) => {
+    const counts = new Map;
+    for (const d of docs) {
+      const raw = pick2(d);
+      const key = typeof raw === "string" && raw.trim() !== "" ? raw : "\uFF08\u65E0\uFF09";
+      counts.set(key, (counts.get(key) ?? 0) + 1);
+    }
+    return [...counts.entries()].map(([key, count]) => ({ key, count })).sort((a, b) => b.count - a.count || (a.key < b.key ? -1 : 1));
+  };
+  return {
+    total: docs.length,
+    sources: tally((d) => d.metadata?.source ?? d.source),
+    categories: tally((d) => d.metadata?.category ?? d.category)
+  };
+}
+function docsToContent(docs, date5, options) {
   return ContentSchema.parse({
-    title: `\u6BCF\u65E5 AI \u8BED\u97F3\u8FD0\u884C\u62A5\u544A \u2014 ${date6}`,
-    body: renderReportMarkdown(docs, date6, options),
+    title: `\u6BCF\u65E5 AI \u8BED\u97F3\u8FD0\u884C\u62A5\u544A \u2014 ${date5}`,
+    body: renderReportMarkdown(docs, date5, options),
     format: "markdown",
-    metadata: { generator: "kk:news-report", date: date6, doc_count: docs.length }
+    metadata: { generator: "kk:news-report", date: date5, doc_count: docs.length }
   });
 }
 function dateFromEnv() {
@@ -22193,7 +22206,9 @@ async function processGeneratorIO(inputPath, outputPath) {
   const rejected = await readRejectedAudit(process.env.NEWS_REPORT_REJECTED_FILE);
   const timing = await readTiming(process.env.NEWS_REPORT_TIMING_FILE);
   const qc = await readQcSummary(process.env.NEWS_REPORT_QC_FILE);
-  const content = docsToContent(docs, dateFromEnv(), { rejected, timing, qc });
+  const planAudit = await readPlanAudit(process.env.NEWS_REPORT_PLAN_AUDIT_FILE);
+  const blended = await readBlendedCorpus(process.env.NEWS_REPORT_BLENDED_FILE);
+  const content = docsToContent(docs, dateFromEnv(), { rejected, timing, qc, planAudit, blended });
   const fs = createNodeFileSystem();
   const outDir = dirname2(outputPath);
   if (outDir && outDir !== ".")
@@ -22232,8 +22247,11 @@ export {
   readTiming,
   readRejectedAudit,
   readQcSummary,
+  readPlanAudit,
+  readBlendedCorpus,
   processGeneratorIO,
   minimumsFromEnv,
   main2 as main,
-  docsToContent
+  docsToContent,
+  corpusComposition
 };

package/plugins/generations/news-report-gen/src/index.ts

@@ -18,6 +18,9 @@ import { echoError } from '@gobing-ai/ts-utils';
  *   `metadata.scores` (quality/importance/urgency/impact 0-5).
  * - `NEWS_REPORT_REJECTED_FILE` — the filter's rejected audit path
  *   (`2-plan/candidates.rejected.json`, RejectedDoc[] of {doc, reasons})
+ * - `NEWS_REPORT_BLENDED_FILE` — the collect stage's blended fan-in output
+ *   (`1-ingest/<date>_02_collect_blended.json`); absent or unreadable -> the
+ *   语料构成 section is omitted, never throw.
  * - `NEWS_REPORT_QC_FILE` — the generate stage's QC audit path
  *   (`3-audio/<date>_11_generate_content.json`); absent -> the 音频 QC block is omitted;
  *   absent or unreadable -> reject counts render as 0, never throw.
@@ -231,6 +234,8 @@ export function renderReportMarkdown(
         rejected?: { doc: Doc; reasons: string[] }[];
         timing?: TimingEntry[];
         qc?: QcSummary | null;
+        planAudit?: PlanAuditRow[] | null;
+        blended?: BlendedDoc[] | null;
     } = {},
 ): string {
     const lines: string[] = [`# 每日 AI 语音运行报告 — ${date}`, ''];
@@ -272,6 +277,40 @@ export function renderReportMarkdown(
     lines.push(`| category allowlist | — | ${counts.category} |`);
     lines.push('');
 
+    const blended = options.blended ?? null;
+    if (blended !== null && blended.length > 0) {
+        // 20260917 funnel: selection can only be judged against what arrived — the blended
+        // input's source/category composition exposes the uncategorized mass that plan
+        // under-ranks (e.g. 47 of 52 docs with no category -> 2 selected).
+        const comp = corpusComposition(blended);
+        lines.push(`## 语料构成（fan-in 输入，共 ${comp.total} 条）`, '');
+        lines.push('| 来源 | 条数 |', '| --- | --- |');
+        for (const g of comp.sources) lines.push(`| ${g.key} | ${g.count} |`);
+        lines.push('');
+        lines.push('| 类别 | 条数 |', '| --- | --- |');
+        for (const g of comp.categories) lines.push(`| ${g.key} | ${g.count} |`);
+        lines.push('');
+    }
+
+    const planAudit = options.planAudit ?? null;
+    if (planAudit !== null && planAudit.length > 0) {
+        // 20260917 funnel: the report only showed the 2 survivors — every fetched title, its
+        // scores and its elimination reason now render here (selection is auditable end to end).
+        lines.push(`## 选稿漏斗（全量审计，共 ${planAudit.length} 条）`, '');
+        lines.push(
+            '| 来源 | 标题 | 类别 | q | i | u | im | score | 去向 |',
+            '| --- | --- | --- | --- | --- | --- | --- | --- | --- |',
+        );
+        for (const row of planAudit) {
+            const s = row.scores ?? {};
+            const title = (row.title || row.id).replace(/\|/g, '\\|').slice(0, 60);
+            lines.push(
+                `| ${row.source} | ${title} | ${row.category} | ${s.quality ?? '—'} | ${s.importance ?? '—'} | ${s.urgency ?? '—'} | ${s.impact ?? '—'} | ${row.planScore ?? '—'} | ${row.disposition} |`,
+            );
+        }
+        lines.push('');
+    }
+
     const qc = options.qc ?? null;
     if (qc !== null) {
         // 0139 R5b: the audio QC verdict is otherwise invisible in the report — this run published
@@ -322,6 +361,79 @@ export async function readRejectedAudit(path: string | undefined): Promise<Rejec
     }
 }
 
+/** One row of episode-plan-gen's `<out>.audit.json` selection trail (plan mode). */
+export interface PlanAuditRow {
+    id: string;
+    source: string;
+    title: string;
+    category: string;
+    scores: Record<string, number>;
+    planScore: number | null;
+    disposition: 'gate' | 'url-merge' | 'title-merge' | 'cap' | 'topN' | 'selected';
+}
+
+/** Read the plan-mode audit envelope `{ total, rows }`; absent/unreadable/malformed -> null (never throw). */
+export async function readPlanAudit(path: string | undefined): Promise<PlanAuditRow[] | null> {
+    if (!path || path.trim() === '') return null;
+    try {
+        const parsed: unknown = JSON.parse(await Bun.file(path).text());
+        const rows = (parsed as { rows?: unknown } | null)?.rows;
+        return Array.isArray(rows) ? (rows as PlanAuditRow[]) : null;
+    } catch {
+        return null;
+    }
+}
+
+/** Minimal shape the 语料构成 section needs from a blended fan-in doc (tolerant on purpose). */
+export interface BlendedDoc {
+    title?: unknown;
+    id?: unknown;
+    source?: unknown;
+    category?: unknown;
+    metadata?: { source?: unknown; category?: unknown };
+}
+
+export interface CorpusCount {
+    key: string;
+    count: number;
+}
+
+/** Read the blended fan-in input; absent/unreadable/malformed/not-an-array -> null (never throw). */
+export async function readBlendedCorpus(path: string | undefined): Promise<BlendedDoc[] | null> {
+    if (!path || path.trim() === '') return null;
+    try {
+        const parsed: unknown = JSON.parse(await Bun.file(path).text());
+        const docs = Array.isArray(parsed) ? parsed : (parsed as { docs?: unknown } | null)?.docs;
+        return Array.isArray(docs) ? (docs as BlendedDoc[]) : null;
+    } catch {
+        return null;
+    }
+}
+
+/** Tally blended docs by source and by category (missing fields bucket as （无）), most frequent first. */
+export function corpusComposition(docs: BlendedDoc[]): {
+    total: number;
+    sources: CorpusCount[];
+    categories: CorpusCount[];
+} {
+    const tally = (pick: (d: BlendedDoc) => unknown): CorpusCount[] => {
+        const counts = new Map<string, number>();
+        for (const d of docs) {
+            const raw = pick(d);
+            const key = typeof raw === 'string' && raw.trim() !== '' ? raw : '（无）';
+            counts.set(key, (counts.get(key) ?? 0) + 1);
+        }
+        return [...counts.entries()]
+            .map(([key, count]) => ({ key, count }))
+            .sort((a, b) => b.count - a.count || (a.key < b.key ? -1 : 1));
+    };
+    return {
+        total: docs.length,
+        sources: tally((d) => d.metadata?.source ?? d.source),
+        categories: tally((d) => d.metadata?.category ?? d.category),
+    };
+}
+
 /** Build the report Content from a kept-candidate Doc[]. */
 export function docsToContent(
     docs: Doc[],
@@ -330,6 +442,8 @@ export function docsToContent(
         rejected?: { doc: Doc; reasons: string[] }[];
         timing?: TimingEntry[];
         qc?: QcSummary | null;
+        planAudit?: PlanAuditRow[] | null;
+        blended?: BlendedDoc[] | null;
     },
 ): Content {
     return ContentSchema.parse({
@@ -350,7 +464,9 @@ export async function processGeneratorIO(inputPath: string, outputPath: string):
     const rejected = await readRejectedAudit(process.env.NEWS_REPORT_REJECTED_FILE);
     const timing = await readTiming(process.env.NEWS_REPORT_TIMING_FILE);
     const qc = await readQcSummary(process.env.NEWS_REPORT_QC_FILE);
-    const content = docsToContent(docs, dateFromEnv(), { rejected, timing, qc });
+    const planAudit = await readPlanAudit(process.env.NEWS_REPORT_PLAN_AUDIT_FILE);
+    const blended = await readBlendedCorpus(process.env.NEWS_REPORT_BLENDED_FILE);
+    const content = docsToContent(docs, dateFromEnv(), { rejected, timing, qc, planAudit, blended });
     const fs = createNodeFileSystem();
     const outDir = dirname(outputPath);
     if (outDir && outDir !== '.') await fs.ensureDir(outDir);

package/plugins/generations/omni-voice-gen/src/omni_voice_gen/pipeline.py

@@ -210,7 +210,7 @@ def _render(
             "profile": resolve_profile(profile, registry),
             "instruct": segment_instruct,
             "seed": None if segment.seed is None else int(segment.seed),
-            "speed": None,
+            "speed": None if segment.speed is None else float(segment.speed),
         }
         wav, duration, retries, transcription, dip = _render_verified(backend, spec)
 

package/plugins/generations/omni-voice-gen/src/omni_voice_gen/voicescript.py

@@ -70,6 +70,7 @@ class VoiceSegment:
     engine: Any = None
     language: Any = None
     gap_ms: Any = None
+    speed: Any = None
     max_chunk_chars: Any = None
     crossfade_ms: Any = None
     personality: Any = None
@@ -334,6 +335,13 @@ def validate_voice_script(script: VoiceScript, sink: WarnSink) -> None:
         ):
             raise ValueError("gap_ms must be non-negative")
 
+        if segment.speed is not None and (
+            isinstance(segment.speed, bool)
+            or not isinstance(segment.speed, (int, float))
+            or not 0.5 <= float(segment.speed) <= 2.0
+        ):
+            raise ValueError("speed must be between 0.5 and 2.0")
+
         if segment.seed is not None and (not _is_int(segment.seed) or segment.seed < 0):
             raise ValueError("seed must be a non-negative integer")