@gobing-ai/knowledge-kit 0.0.16 → 0.0.18

package/plugins/generations/episode-plan-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/episode-plan-gen/src/index.ts
-init_dist();
 var SOURCE_BASE = {
   aihot: 30,
   rss: 12,
@@ -21968,6 +21919,7 @@ var SOURCE_BASE = {
 };
 var DEFAULT_SOURCE_BASE = 5;
 var TITLE_SIMILARITY_THRESHOLD = 0.6;
+var STUB_BODY_FLOOR = 300;
 var FRESHNESS_DAYS = 7;
 var PER_SOURCE_CAP = 5;
 var NON_NEWS_TAGS = new Set(["geopolitics", "military-conflict", "war"]);
@@ -21999,11 +21951,9 @@ var TAG_CATEGORY = {
   war: "non-ai"
 };
 var NON_AI_SOURCES = new Set(["polymarket"]);
+var AI_DEFAULT_SOURCES = new Set(["aihot"]);
+var JUNK_KEYWORDS = ["error:", "http error", "unable to download"];
 var CATEGORY_KEYWORDS = [
-  {
-    category: "non-ai",
-    words: ["error:", "http error", "unable to download"]
-  },
   {
     category: "ai-technical",
     words: ["api", "tutorial", "benchmark", "architecture", "sdk", "cli", "quantization", "inference", "fine-tun"]
@@ -22021,13 +21971,28 @@ var CATEGORY_KEYWORDS = [
       "\u4E0A\u7EBF",
       "\u63A8\u51FA",
       "\u5F00\u6E90",
+      "ai",
+      "\u4EBA\u5DE5\u667A\u80FD",
+      "\u5927\u6A21\u578B",
       "sora",
       "gpt",
+      "chatgpt",
       "claude",
       "gemini",
       "llama",
       "grok",
-      "copilot"
+      "copilot",
+      "openai",
+      "anthropic",
+      "deepseek",
+      "qwen",
+      "vllm",
+      "glm",
+      "mistral",
+      "nvidia",
+      "\u82F1\u4F1F\u8FBE",
+      "huawei",
+      "\u534E\u4E3A"
     ]
   },
   {
@@ -22051,6 +22016,9 @@ var CATEGORY_KEYWORDS = [
 ];
 var IMPORTANCE_BASE = { aihot: 4, rss: 3, hackernews: 3 };
 var DEFAULT_IMPORTANCE = 2;
+function planAuditOutPath(outputPath) {
+  return outputPath.endsWith(".json") ? `${outputPath.slice(0, -".json".length)}.audit.json` : `${outputPath}.audit.json`;
+}
 function docTimestamp(doc2) {
   const meta3 = asRecord(doc2.metadata);
   for (const key of ["publishedAt", "discoveredAt"]) {
@@ -22061,9 +22029,9 @@ function docTimestamp(doc2) {
         return t;
     }
   }
-  const date6 = meta3.date;
-  if (typeof date6 === "string" && /^\d{4}-\d{2}-\d{2}/.test(date6.trim())) {
-    return Date.parse(date6.trim().slice(0, 10));
+  const date5 = meta3.date;
+  if (typeof date5 === "string" && /^\d{4}-\d{2}-\d{2}/.test(date5.trim())) {
+    return Date.parse(date5.trim().slice(0, 10));
   }
   return null;
 }
@@ -22154,10 +22122,15 @@ function classifyCategory(doc2) {
     if (hit)
       return hit;
   }
-  if (NON_AI_SOURCES.has(typeof meta3.source === "string" ? meta3.source : ""))
+  const source = typeof meta3.source === "string" ? meta3.source : "";
+  if (NON_AI_SOURCES.has(source))
     return "non-ai";
   const title = (doc2.title ?? "").toLowerCase();
   const bodyHead = (doc2.body ?? "").slice(0, 200).toLowerCase();
+  if (JUNK_KEYWORDS.some((w) => matchesKeyword(title, w) || matchesKeyword(bodyHead, w)))
+    return "non-ai";
+  if (AI_DEFAULT_SOURCES.has(source))
+    return "ai-news";
   for (const { category, words } of CATEGORY_KEYWORDS) {
     if (words.some((w) => matchesKeyword(title, w) || matchesKeyword(bodyHead, w)))
       return category;
@@ -22166,8 +22139,9 @@ function classifyCategory(doc2) {
 }
 function scoreDimensions(doc2, planDate) {
   const meta3 = asRecord(doc2.metadata);
-  let quality = 1 + Math.min(4, Math.round((doc2.body ?? "").length / 1500));
-  if (doc2.sourceUri)
+  const bodyLen = (doc2.body ?? "").length;
+  let quality = bodyLen < STUB_BODY_FLOOR ? 1 : 1 + Math.min(4, Math.round(bodyLen / 1500));
+  if (doc2.sourceUri && bodyLen >= STUB_BODY_FLOOR)
     quality = Math.min(5, quality + 1);
   const source = typeof meta3.source === "string" ? meta3.source : "";
   let importance = IMPORTANCE_BASE[source] ?? DEFAULT_IMPORTANCE;
@@ -22200,6 +22174,7 @@ function planScore(doc2) {
 }
 function planEpisode(docs, options) {
   const planDate = options.date ?? new Date().toISOString().slice(0, 10);
+  const rows = options.audit ? new Map : null;
   const byUrl = new Map;
   const scored = [];
   for (const doc2 of docs) {
@@ -22212,9 +22187,25 @@ function planEpisode(docs, options) {
         scores: scoreDimensions(doc2, planDate)
       }
     };
+    const meta3 = asRecord(cleaned.metadata);
+    rows?.set(cleaned, {
+      id: doc2.id,
+      source: sourceOf(doc2),
+      title: doc2.title ?? "",
+      category: meta3.category,
+      scores: meta3.scores,
+      planScore: null,
+      disposition: "gate"
+    });
     if (!passesGates(cleaned, planDate))
       continue;
-    const record2 = { doc: cleaned, score: planScore(cleaned), mergedFrom: [] };
+    const score = planScore(cleaned);
+    const row = rows?.get(cleaned);
+    if (row) {
+      row.planScore = score;
+      row.disposition = "topN";
+    }
+    const record2 = { doc: cleaned, score, mergedFrom: [] };
     const urlKey = normalizeUrl(doc2.sourceUri);
     if (urlKey) {
       const existing = byUrl.get(urlKey);
@@ -22226,6 +22217,9 @@ function planEpisode(docs, options) {
           title: loser.doc.title ?? "",
           url: loser.doc.sourceUri ?? null
         });
+        const loserRow = rows?.get(loser.doc);
+        if (loserRow)
+          loserRow.disposition = "url-merge";
         if (winner !== existing) {
           existing.doc = winner.doc;
           existing.score = winner.score;
@@ -22257,6 +22251,9 @@ function planEpisode(docs, options) {
         title: loser.doc.title ?? "",
         url: loser.doc.sourceUri ?? null
       });
+      const loserRow = rows?.get(loser.doc);
+      if (loserRow)
+        loserRow.disposition = "title-merge";
       absorbed.add(loser);
     }
   }
@@ -22266,13 +22263,22 @@ function planEpisode(docs, options) {
   for (const record2 of candidates) {
     const source = sourceOf(record2.doc);
     const count = perSource.get(source) ?? 0;
-    if (count >= PER_SOURCE_CAP)
+    if (count >= PER_SOURCE_CAP) {
+      const row2 = rows?.get(record2.doc);
+      if (row2)
+        row2.disposition = "cap";
       continue;
+    }
     perSource.set(source, count + 1);
     selected.push(record2);
+    const row = rows?.get(record2.doc);
+    if (row)
+      row.disposition = "selected";
     if (selected.length >= Math.max(1, options.maxItems))
       break;
   }
+  if (rows)
+    options.audit?.push(...rows.values());
   return selected.map((r) => {
     const meta3 = asRecord(r.doc.metadata);
     const ascii = asciiRatio(r.doc.title ?? "");
@@ -22295,7 +22301,7 @@ function sourceOf(doc2) {
   const meta3 = asRecord(doc2.metadata);
   return typeof meta3.source === "string" ? meta3.source : "unknown";
 }
-function renderPlanMarkdown(docs, date6) {
+function renderPlanMarkdown(docs, date5) {
   const items = docs.map((doc2, i) => {
     const meta3 = asRecord(doc2.metadata);
     const parts = [
@@ -22309,25 +22315,25 @@ function renderPlanMarkdown(docs, date6) {
   }).join(`
 
 `);
-  return `# \u6BCF\u65E5 AI \u64AD\u5BA2\u9009\u7A3F\u8BA1\u5212 \u2014 ${date6}
+  return `# \u6BCF\u65E5 AI \u64AD\u5BA2\u9009\u7A3F\u8BA1\u5212 \u2014 ${date5}
 
 \u672C\u671F\u5171 ${docs.length} \u6761\uFF08\u8BC4\u5206/\u53BB\u91CD/\u6807\u8BB0\u540E\u7CBE\u9009\uFF09\u3002
 
 ${items}
 `;
 }
-function docsToContent(docs, date6) {
-  const planned = planEpisode(docs, { maxItems: maxItemsFromEnv(), date: date6 });
-  return planContent(planned, date6);
+function docsToContent(docs, date5, audit) {
+  const planned = planEpisode(docs, { maxItems: maxItemsFromEnv(), date: date5, audit });
+  return planContent(planned, date5);
 }
-function planContent(docs, date6) {
+function planContent(docs, date5) {
   return ContentSchema.parse({
-    title: `\u6BCF\u65E5 AI \u64AD\u5BA2\u9009\u7A3F\u8BA1\u5212 \u2014 ${date6}`,
-    body: renderPlanMarkdown(docs, date6),
+    title: `\u6BCF\u65E5 AI \u64AD\u5BA2\u9009\u7A3F\u8BA1\u5212 \u2014 ${date5}`,
+    body: renderPlanMarkdown(docs, date5),
     format: "markdown",
     metadata: {
       generator: "kk:episode-plan",
-      date: date6,
+      date: date5,
       doc_count: docs.length,
       docs
     }
@@ -22347,7 +22353,7 @@ function resolvePlanDate() {
 var MECE_CATEGORIES = ["ai-news", "ai-technical", "ai-research", "ai-industry", "non-ai"];
 var QC_DIMENSIONS = ["quality", "importance", "urgency", "impact"];
 var DEFAULT_FILTER_CONFIG = {
-  categories: ["ai-news", "ai-industry"],
+  categories: ["ai-news", "ai-technical", "ai-research", "ai-industry"],
   minQuality: 2,
   minImportance: 3,
   minUrgency: 2,
@@ -22441,8 +22447,10 @@ async function processGeneratorIO(inputPath, outputPath, mode = "plan") {
 }
 async function processPlanIO(inputPath, outputPath) {
   const docs = DocListSchema.parse(await readJsonFile(inputPath));
-  const content = docsToContent(docs, resolvePlanDate());
+  const audit = [];
+  const content = docsToContent(docs, resolvePlanDate(), audit);
   await atomicWrite(outputPath, content);
+  await atomicWrite(planAuditOutPath(outputPath), { date: resolvePlanDate(), total: docs.length, rows: audit });
   return content;
 }
 async function processFilterIO(inputPath, outputPath) {
@@ -22492,6 +22500,7 @@ export {
   processGeneratorIO,
   planScore,
   planEpisode,
+  planAuditOutPath,
   normalizeUrl,
   main2 as main,
   filterConfigFromEnv,