chat-agent-toolkit 1.2.379 → 1.2.380

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "chat-agent-toolkit",
-  "version": "1.2.379",
+  "version": "1.2.380",
   "description": "Multi-provider AI agent toolkit: generate language responses, search the web, extract content, and manage memory across 10+ LLM providers.",
   "author": "vtempest <grokthiscontact@gmail.com>",
   "license": "rights.institute/PROSPER",
@@ -98,7 +98,7 @@
     "html-entities": "^2.6.0",
     "manage-storage": "^0.0.47",
     "marked": "^18.0.12",
-    "qwksearch-api-client": "^1.0.51",
+    "qwksearch-api-client": "^1.0.52",
     "write-language": "^0.1.153",
     "zod": "^4.5.4"
   },

package/src/tools/search/doc-utils.d.ts

@@ -39,11 +39,25 @@ export type UploadFileLoader = (fileId: string) => Promise<LoadedUpload | null>;
  * the S3-credentials fallback.
  */
 export declare function registerUploadFileLoader(loader: UploadFileLoader): void;
+/**
+ * Resolves every uploaded fileId once, in order — sequentially, so one raw
+ * payload is live at a time, and once per request, with the result shared by
+ * the reranker and the image loader.
+ */
+export declare function loadUploads(fileIds: string[], r2Credentials?: R2CredentialsInput): Promise<LoadedUpload[]>;
+/**
+ * Picks the image attachments out of already-resolved uploads, ready to be
+ * passed to the LLM as image content parts.
+ */
+export declare function selectUploadImages(uploads: LoadedUpload[]): UploadImageAttachment[];
 /**
  * Resolves image attachments for the given uploaded fileIds so they can be
  * passed to the LLM as image content parts. Non-image uploads (documents) and
  * images stored without inline data are skipped.
  */
-export declare function loadUploadImages(fileIds: string[], r2Credentials?: R2CredentialsInput): Promise<UploadImageAttachment[]>;
-export declare function rerankDocs(query: string, docs: Document[], fileIds: string[], optimizationMode: "speed" | "balanced" | "quality", r2Credentials?: R2CredentialsInput): Promise<Document[]>;
+export declare function loadUploadImages(fileIds: string[], r2Credentials?: R2CredentialsInput, preloaded?: LoadedUpload[]): Promise<UploadImageAttachment[]>;
+/**
+ * Builds the answer context: uploaded file content first, then web results.
+ */
+export declare function rerankDocs(query: string, docs: Document[], fileIds: string[], optimizationMode: "speed" | "balanced" | "quality", r2Credentials?: R2CredentialsInput, preloaded?: LoadedUpload[]): Promise<Document[]>;
 export declare function processDocs(docs: Document[]): string;

package/src/tools/search/doc-utils.ts

@@ -1,204 +1,246 @@
-/**
- * @module research/search/doc-utils
- * @description Document utilities: fallback docs, reranking, and formatting.
- */
-import type { Document } from "./document";
-
-export interface R2CredentialsInput {
-  accountId: string;
-  accessKeyId: string;
-  secretAccessKey: string;
-  bucket: string;
-}
-
-export function buildFallbackDocs(query: string): Document[] {
-  const trimmedQuery = (query || "").trim();
-  const searchQuery = trimmedQuery.length > 0 ? trimmedQuery : "web search";
-  const fallbackUrl = `https://www.google.com/search?q=${encodeURIComponent(searchQuery)}`;
-
-  return [
-    {
-      pageContent:
-        "No indexed sources were returned by the configured search providers for this query.",
-      metadata: {
-        title: `Search results for: ${searchQuery}`,
-        url: fallbackUrl,
-        source: "Google Search",
-      },
-    },
-  ];
-}
-
-export function normalizeSourcesOutput(output: unknown, query: string): Document[] {
-  if (Array.isArray(output) && output.length > 0) {
-    return output as Document[];
-  }
-  return buildFallbackDocs(query);
-}
-
-/** The payload resolved for an uploaded fileId. */
-export interface LoadedUpload {
-  title: string;
-  content: string;
-  /** MIME type for image uploads (e.g. `"image/png"`). */
-  mediaType?: string;
-  /** `data:`/`http(s)` URL of an image upload, passed to the LLM as image content. */
-  image?: string;
-}
-
-/** An image attachment resolved from an uploaded fileId. */
-export interface UploadImageAttachment {
-  mediaType: string;
-  /** `data:`/`http(s)` URL of the image, ready for an AI SDK image content part. */
-  image: string;
-}
-
-/**
- * Resolves an uploaded fileId to its extracted payload. Hosts register a loader
- * (e.g. reading the native R2 binding) so the search pipeline does not depend
- * on filesystem or credential-based access. Image uploads additionally carry
- * `mediaType` and `image` so they can be passed to the LLM directly.
- */
-export type UploadFileLoader = (fileId: string) => Promise<LoadedUpload | null>;
-
-let uploadFileLoader: UploadFileLoader | null = null;
-
-/**
- * Registers the loader used by {@link rerankDocs} to resolve uploaded
- * fileIds to extracted content. The registered loader takes precedence over
- * the S3-credentials fallback.
- */
-export function registerUploadFileLoader(loader: UploadFileLoader): void {
-  uploadFileLoader = loader;
-}
-
-async function downloadExtractedContent(fileId: string, r2Credentials: R2CredentialsInput): Promise<LoadedUpload | null> {
-  try {
-    const { manageStorage } = await import("manage-storage");
-    const config = {
-      provider: "cloudflare" as const,
-      BUCKET_NAME: r2Credentials.bucket,
-      ACCESS_KEY_ID: r2Credentials.accessKeyId,
-      SECRET_ACCESS_KEY: r2Credentials.secretAccessKey,
-      BUCKET_URL: `https://${r2Credentials.accountId}.r2.cloudflarestorage.com`,
-    };
-    const extractedKey = `${fileId}-extracted.json`;
-    const data = await manageStorage("download", { ...config, key: extractedKey });
-    const parsed = JSON.parse(data);
-    return {
-      title: parsed.title || "Uploaded Document",
-      content: parsed.content || "",
-      ...(parsed.mediaType ? { mediaType: parsed.mediaType } : {}),
-      ...(parsed.image ? { image: parsed.image } : {}),
-    };
-  } catch (error) {
-    console.error(`[rerankDocs] Failed to download extracted content for fileId ${fileId}:`, error);
-    return null;
-  }
-}
-
-/**
- * Resolves a single uploaded fileId via the registered loader (preferred) or
- * the R2 credentials fallback. Returns `null` when neither can resolve it.
- */
-async function resolveUpload(
-  fileId: string,
-  r2Credentials?: R2CredentialsInput,
-): Promise<LoadedUpload | null> {
-  if (uploadFileLoader) {
-    try {
-      const loaded = await uploadFileLoader(fileId);
-      if (loaded) return loaded;
-    } catch (error) {
-      console.error(
-        `[resolveUpload] Registered upload loader failed for fileId ${fileId}:`,
-        error,
-      );
-    }
-  }
-  if (r2Credentials) {
-    return downloadExtractedContent(fileId, r2Credentials);
-  }
-  return null;
-}
-
-/**
- * Resolves image attachments for the given uploaded fileIds so they can be
- * passed to the LLM as image content parts. Non-image uploads (documents) and
- * images stored without inline data are skipped.
- */
-export async function loadUploadImages(
-  fileIds: string[],
-  r2Credentials?: R2CredentialsInput,
-): Promise<UploadImageAttachment[]> {
-  if (!fileIds || fileIds.length === 0) return [];
-
-  const resolved = await Promise.all(
-    fileIds.map((fileId) => resolveUpload(fileId, r2Credentials)),
-  );
-
-  return resolved
-    .filter(
-      (r): r is LoadedUpload =>
-        r !== null && typeof r.image === "string" && r.image.length > 0,
-    )
-    .map((r) => ({
-      mediaType: r.mediaType || "image/png",
-      image: r.image as string,
-    }));
-}
-
-export async function rerankDocs(
-  query: string,
-  docs: Document[],
-  fileIds: string[],
-  optimizationMode: "speed" | "balanced" | "quality",
-  r2Credentials?: R2CredentialsInput,
-): Promise<Document[]> {
-  if (docs.length === 0 && fileIds.length === 0) {
-    return docs;
-  }
-
-  let filesData: LoadedUpload[] = [];
-
-  if (fileIds.length > 0) {
-    const results = await Promise.all(
-      fileIds.map((fileId) => resolveUpload(fileId, r2Credentials)),
-    );
-    filesData = results.filter((r): r is LoadedUpload => r !== null);
-  }
-
-  // Uploaded documents must always reach the LLM. Build their docs up front so
-  // every return path below keeps them in the answer context. Image uploads
-  // carry no text content — they are passed separately as image parts (see
-  // loadUploadImages) — so they are excluded here to avoid empty docs.
-  const fileDocs: Document[] = filesData
-    .filter((fileData) => fileData.content && fileData.content.length > 0)
-    .map((fileData) => ({
-      pageContent: fileData.content,
-      metadata: { title: fileData.title, url: "File" },
-    }));
-
-  if (query.toLocaleLowerCase() === "summarize") {
-    // Skip web-result reranking for an explicit "summarize" request, but keep
-    // the uploaded file content — otherwise attachments are silently dropped
-    // and never analysed.
-    return [...fileDocs, ...docs].slice(0, 15);
-  }
-
-  const docsWithContent = docs.filter(
-    (doc) => doc.pageContent && doc.pageContent.length > 0,
-  );
-
-  // Combine file docs with web results, cap at 15
-  return [...fileDocs, ...docsWithContent].slice(0, 15);
-}
-
-export function processDocs(docs: Document[]): string {
-  return docs
-    .map(
-      (_, index) =>
-        `${index + 1}. ${docs[index].metadata.title} ${docs[index].pageContent}`,
-    )
-    .join("\n");
-}
+/**
+ * @module research/search/doc-utils
+ * @description Document utilities: fallback docs, reranking, and formatting.
+ */
+import type { Document } from "./document";
+
+export interface R2CredentialsInput {
+  accountId: string;
+  accessKeyId: string;
+  secretAccessKey: string;
+  bucket: string;
+}
+
+export function buildFallbackDocs(query: string): Document[] {
+  const trimmedQuery = (query || "").trim();
+  const searchQuery = trimmedQuery.length > 0 ? trimmedQuery : "web search";
+  const fallbackUrl = `https://www.google.com/search?q=${encodeURIComponent(searchQuery)}`;
+
+  return [
+    {
+      pageContent:
+        "No indexed sources were returned by the configured search providers for this query.",
+      metadata: {
+        title: `Search results for: ${searchQuery}`,
+        url: fallbackUrl,
+        source: "Google Search",
+      },
+    },
+  ];
+}
+
+export function normalizeSourcesOutput(output: unknown, query: string): Document[] {
+  if (Array.isArray(output) && output.length > 0) {
+    return output as Document[];
+  }
+  return buildFallbackDocs(query);
+}
+
+/** The payload resolved for an uploaded fileId. */
+export interface LoadedUpload {
+  title: string;
+  content: string;
+  /** MIME type for image uploads (e.g. `"image/png"`). */
+  mediaType?: string;
+  /** `data:`/`http(s)` URL of an image upload, passed to the LLM as image content. */
+  image?: string;
+}
+
+/** An image attachment resolved from an uploaded fileId. */
+export interface UploadImageAttachment {
+  mediaType: string;
+  /** `data:`/`http(s)` URL of the image, ready for an AI SDK image content part. */
+  image: string;
+}
+
+/**
+ * Resolves an uploaded fileId to its extracted payload. Hosts register a loader
+ * (e.g. reading the native R2 binding) so the search pipeline does not depend
+ * on filesystem or credential-based access. Image uploads additionally carry
+ * `mediaType` and `image` so they can be passed to the LLM directly.
+ */
+export type UploadFileLoader = (fileId: string) => Promise<LoadedUpload | null>;
+
+let uploadFileLoader: UploadFileLoader | null = null;
+
+/**
+ * Registers the loader used by {@link rerankDocs} to resolve uploaded
+ * fileIds to extracted content. The registered loader takes precedence over
+ * the S3-credentials fallback.
+ */
+export function registerUploadFileLoader(loader: UploadFileLoader): void {
+  uploadFileLoader = loader;
+}
+
+async function downloadExtractedContent(fileId: string, r2Credentials: R2CredentialsInput): Promise<LoadedUpload | null> {
+  try {
+    const { manageStorage } = await import("manage-storage");
+    const config = {
+      provider: "cloudflare" as const,
+      BUCKET_NAME: r2Credentials.bucket,
+      ACCESS_KEY_ID: r2Credentials.accessKeyId,
+      SECRET_ACCESS_KEY: r2Credentials.secretAccessKey,
+      BUCKET_URL: `https://${r2Credentials.accountId}.r2.cloudflarestorage.com`,
+    };
+    const extractedKey = `${fileId}-extracted.json`;
+    const data = await manageStorage("download", { ...config, key: extractedKey });
+    const parsed = JSON.parse(data);
+    return {
+      title: parsed.title || "Uploaded Document",
+      content: parsed.content || "",
+      ...(parsed.mediaType ? { mediaType: parsed.mediaType } : {}),
+      ...(parsed.image ? { image: parsed.image } : {}),
+    };
+  } catch (error) {
+    console.error(`[rerankDocs] Failed to download extracted content for fileId ${fileId}:`, error);
+    return null;
+  }
+}
+
+/**
+ * Resolves a single uploaded fileId via the registered loader (preferred) or
+ * the R2 credentials fallback. Returns `null` when neither can resolve it.
+ */
+async function resolveUpload(
+  fileId: string,
+  r2Credentials?: R2CredentialsInput,
+): Promise<LoadedUpload | null> {
+  if (uploadFileLoader) {
+    try {
+      const loaded = await uploadFileLoader(fileId);
+      if (loaded) return loaded;
+    } catch (error) {
+      console.error(
+        `[resolveUpload] Registered upload loader failed for fileId ${fileId}:`,
+        error,
+      );
+    }
+  }
+  if (r2Credentials) {
+    return downloadExtractedContent(fileId, r2Credentials);
+  }
+  return null;
+}
+
+/**
+ * Resolves every uploaded fileId once, in order.
+ *
+ * Two things here are deliberate, and both are about the 128MB an edge isolate
+ * gets. An attachment arrives as a JSON object whose `image` field is a data
+ * URL, so an 8MB picture is ~11MB of base64 that exists twice while it is
+ * parsed — and a request may carry ten of them. So: resolved **sequentially**,
+ * to keep one raw payload live at a time rather than ten; and resolved **once**
+ * per request, with the result shared by the reranker and the image loader,
+ * which between them used to fetch and parse every attachment twice over.
+ *
+ * Duplicate ids in the list collapse to one fetch. A file that cannot be
+ * resolved is dropped, exactly as before.
+ */
+export async function loadUploads(
+  fileIds: string[],
+  r2Credentials?: R2CredentialsInput,
+): Promise<LoadedUpload[]> {
+  if (!fileIds || fileIds.length === 0) return [];
+
+  const seen = new Set<string>();
+  const loaded: LoadedUpload[] = [];
+
+  for (const fileId of fileIds) {
+    if (seen.has(fileId)) continue;
+    seen.add(fileId);
+    const upload = await resolveUpload(fileId, r2Credentials);
+    if (upload) loaded.push(upload);
+  }
+
+  return loaded;
+}
+
+/**
+ * Picks the image attachments out of already-resolved uploads, ready to be
+ * passed to the LLM as image content parts. Documents and images stored
+ * without inline data carry no `image`, so they drop out here.
+ */
+export function selectUploadImages(uploads: LoadedUpload[]): UploadImageAttachment[] {
+  return uploads
+    .filter((u) => typeof u.image === "string" && u.image.length > 0)
+    .map((u) => ({
+      mediaType: u.mediaType || "image/png",
+      image: u.image as string,
+    }));
+}
+
+/**
+ * Resolves image attachments for the given uploaded fileIds so they can be
+ * passed to the LLM as image content parts. Non-image uploads (documents) and
+ * images stored without inline data are skipped.
+ *
+ * Pass `preloaded` when the uploads have already been resolved for this
+ * request — {@link loadUploads} explains why fetching them twice is worth
+ * avoiding.
+ */
+export async function loadUploadImages(
+  fileIds: string[],
+  r2Credentials?: R2CredentialsInput,
+  preloaded?: LoadedUpload[],
+): Promise<UploadImageAttachment[]> {
+  if (preloaded) return selectUploadImages(preloaded);
+  if (!fileIds || fileIds.length === 0) return [];
+  return selectUploadImages(await loadUploads(fileIds, r2Credentials));
+}
+
+/**
+ * Builds the answer context: uploaded file content first, then web results.
+ *
+ * Pass `preloaded` when the uploads have already been resolved for this
+ * request — see {@link loadUploads}.
+ */
+export async function rerankDocs(
+  query: string,
+  docs: Document[],
+  fileIds: string[],
+  optimizationMode: "speed" | "balanced" | "quality",
+  r2Credentials?: R2CredentialsInput,
+  preloaded?: LoadedUpload[],
+): Promise<Document[]> {
+  if (docs.length === 0 && fileIds.length === 0) {
+    return docs;
+  }
+
+  const filesData: LoadedUpload[] =
+    preloaded ??
+    (fileIds.length > 0 ? await loadUploads(fileIds, r2Credentials) : []);
+
+  // Uploaded documents must always reach the LLM. Build their docs up front so
+  // every return path below keeps them in the answer context. Image uploads
+  // carry no text content — they are passed separately as image parts (see
+  // loadUploadImages) — so they are excluded here to avoid empty docs.
+  const fileDocs: Document[] = filesData
+    .filter((fileData) => fileData.content && fileData.content.length > 0)
+    .map((fileData) => ({
+      pageContent: fileData.content,
+      metadata: { title: fileData.title, url: "File" },
+    }));
+
+  if (query.toLocaleLowerCase() === "summarize") {
+    // Skip web-result reranking for an explicit "summarize" request, but keep
+    // the uploaded file content — otherwise attachments are silently dropped
+    // and never analysed.
+    return [...fileDocs, ...docs].slice(0, 15);
+  }
+
+  const docsWithContent = docs.filter(
+    (doc) => doc.pageContent && doc.pageContent.length > 0,
+  );
+
+  // Combine file docs with web results, cap at 15
+  return [...fileDocs, ...docsWithContent].slice(0, 15);
+}
+
+export function processDocs(docs: Document[]): string {
+  return docs
+    .map(
+      (_, index) =>
+        `${index + 1}. ${docs[index].metadata.title} ${docs[index].pageContent}`,
+    )
+    .join("\n");
+}

package/src/tools/search/index.d.ts

@@ -3,11 +3,11 @@
  * @description Meta search agent tools for AI-powered web search and research
  */
 export { default as MetaSearchAgent } from "./metaSearchAgent";
-export type { MetaSearchAgentType, Config, ChatTurnMessage, SearchingEvent, FewShotExample } from "./meta-search-types";
+export type { MetaSearchAgentType, Config, ChatTurnMessage, DrainableEmitter, SearchingEvent, FewShotExample } from "./meta-search-types";
 export { searchHandlers, createSearchHandlers } from "./search-handlers";
 export { default as generateSuggestions } from "./suggestionGeneratorAgent";
 export { groupAndSummarizeDocs } from "./link-summarizer";
 export type { Document } from "./document";
 export { splitTextIntoChunks } from "./document";
-export { buildFallbackDocs, rerankDocs, processDocs, normalizeSourcesOutput, registerUploadFileLoader, loadUploadImages } from "./doc-utils";
+export { buildFallbackDocs, rerankDocs, processDocs, normalizeSourcesOutput, registerUploadFileLoader, loadUploads, loadUploadImages, selectUploadImages } from "./doc-utils";
 export type { UploadFileLoader, LoadedUpload, UploadImageAttachment } from "./doc-utils";

package/src/tools/search/index.ts

@@ -1,14 +1,14 @@
-/**
- * @module agent-toolkit/tools/search
- * @description Meta search agent tools for AI-powered web search and research
- */
-
-export { default as MetaSearchAgent } from "./metaSearchAgent";
-export type { MetaSearchAgentType, Config, ChatTurnMessage, SearchingEvent, FewShotExample } from "./meta-search-types";
-export { searchHandlers, createSearchHandlers } from "./search-handlers";
-export { default as generateSuggestions } from "./suggestionGeneratorAgent";
-export { groupAndSummarizeDocs } from "./link-summarizer";
-export type { Document } from "./document";
-export { splitTextIntoChunks } from "./document";
-export { buildFallbackDocs, rerankDocs, processDocs, normalizeSourcesOutput, registerUploadFileLoader, loadUploadImages } from "./doc-utils";
-export type { UploadFileLoader, LoadedUpload, UploadImageAttachment } from "./doc-utils";
+/**
+ * @module agent-toolkit/tools/search
+ * @description Meta search agent tools for AI-powered web search and research
+ */
+
+export { default as MetaSearchAgent } from "./metaSearchAgent";
+export type { MetaSearchAgentType, Config, ChatTurnMessage, DrainableEmitter, SearchingEvent, FewShotExample } from "./meta-search-types";
+export { searchHandlers, createSearchHandlers } from "./search-handlers";
+export { default as generateSuggestions } from "./suggestionGeneratorAgent";
+export { groupAndSummarizeDocs } from "./link-summarizer";
+export type { Document } from "./document";
+export { splitTextIntoChunks } from "./document";
+export { buildFallbackDocs, rerankDocs, processDocs, normalizeSourcesOutput, registerUploadFileLoader, loadUploads, loadUploadImages, selectUploadImages } from "./doc-utils";
+export type { UploadFileLoader, LoadedUpload, UploadImageAttachment } from "./doc-utils";

package/src/tools/search/meta-search-types.d.ts

@@ -11,10 +11,16 @@ export interface ChatTurnMessage {
 }
 /** A `[role, content]` tuple used for few-shot prompt examples. */
 export type FewShotExample = [role: "user" | "assistant", content: string];
+/**
+ * The pipeline's event emitter, plus the hook a consumer uses to push back.
+ */
+export interface DrainableEmitter extends EventEmitter {
+    waitForDrain?: () => Promise<void>;
+}
 export interface MetaSearchAgentType {
     searchAndAnswer: (message: string, history: ChatTurnMessage[], llm: LanguageModel, optimizationMode: "speed" | "balanced" | "quality", fileIds: string[], systemInstructions: string, category?: string, sourceExtractionEnabled?: boolean, thinkingTimeLimit?: number, 
     /** User-authored replacement for the focus mode's query-expansion prompt. */
-    queryExpansionPrompt?: string) => Promise<EventEmitter>;
+    queryExpansionPrompt?: string) => Promise<DrainableEmitter>;
 }
 /** Emitted on the EventEmitter data channel to report live search progress. */
 export interface SearchingEvent {