@hunterzhu/pulse-adapters 0.6.0 → 0.7.0

package/dist/documents/docx.d.ts

@@ -0,0 +1,2 @@
+/** Parse untrusted Office input off the host event loop, with bounded lifetime. */
+export declare function extractDocxText(data: Buffer, signal: AbortSignal, timeoutMs?: number): Promise<string>;

package/dist/documents/docx.js

@@ -0,0 +1,45 @@
+import { Worker } from 'node:worker_threads';
+import { createRequire } from 'node:module';
+import { ToolError } from '@hunterzhu/pulse-tool-sdk';
+import { validateOfficeArchive } from './tools.js';
+/** Parse untrusted Office input off the host event loop, with bounded lifetime. */
+export async function extractDocxText(data, signal, timeoutMs = 30_000) {
+    if (signal.aborted)
+        throw new ToolError('ABORTED', 'ABORTED');
+    if (data.length > 20 * 1024 * 1024)
+        throw new ToolError('DOCUMENT_TOO_LARGE', 'DOCUMENT_TOO_LARGE');
+    try {
+        validateOfficeArchive(data);
+    }
+    catch (error) {
+        const code = error instanceof ToolError ? error.code.replaceAll('XLSX', 'DOCX') : 'INVALID_DOCX';
+        throw new ToolError(code, code);
+    }
+    const modulePath = createRequire(import.meta.url).resolve('mammoth');
+    return new Promise((resolve, reject) => {
+        const worker = new Worker(`
+      const { parentPort, workerData } = require('node:worker_threads');
+      require(workerData.modulePath).extractRawText({ buffer: Buffer.from(workerData.data) })
+        .then(result => parentPort.postMessage({ text: result.value }))
+        .catch(() => parentPort.postMessage({ error: 'INVALID_DOCX' }));
+    `, { eval: true, workerData: { modulePath, data }, resourceLimits: { maxOldGenerationSizeMb: 128 } });
+        let finished = false;
+        const finish = (error, text) => {
+            if (finished)
+                return;
+            finished = true;
+            clearTimeout(timer);
+            signal.removeEventListener('abort', abort);
+            void worker.terminate().then(() => error ? reject(new ToolError(error, error)) : resolve(text ?? ''));
+        };
+        const abort = () => finish('ABORTED');
+        const timer = setTimeout(() => finish('DOCX_PARSE_TIMEOUT'), timeoutMs);
+        signal.addEventListener('abort', abort, { once: true });
+        if (signal.aborted)
+            abort();
+        worker.once('message', message => typeof message?.text === 'string' ? finish(undefined, message.text) : finish('INVALID_DOCX'));
+        worker.once('error', () => finish('INVALID_DOCX'));
+        worker.once('exit', () => { if (!finished)
+            finish('INVALID_DOCX'); });
+    });
+}

package/dist/documents/tools.d.ts

@@ -2,16 +2,19 @@ import { z } from 'zod';
 import { type ToolDefinition } from '@hunterzhu/pulse-tool-sdk';
 declare const pdfInput: z.ZodObject<{
     path: z.ZodString;
+    offset: z.ZodOptional<z.ZodNumber>;
     startPage: z.ZodOptional<z.ZodNumber>;
     endPage: z.ZodOptional<z.ZodNumber>;
     maxChars: z.ZodDefault<z.ZodNumber>;
 }, "strict", z.ZodTypeAny, {
     path: string;
     maxChars: number;
+    offset?: number | undefined;
     startPage?: number | undefined;
     endPage?: number | undefined;
 }, {
     path: string;
+    offset?: number | undefined;
     startPage?: number | undefined;
     endPage?: number | undefined;
     maxChars?: number | undefined;
@@ -19,6 +22,7 @@ declare const pdfInput: z.ZodObject<{
 declare const xlsxInput: z.ZodObject<{
     path: z.ZodString;
     sheet: z.ZodOptional<z.ZodString>;
+    startRow: z.ZodOptional<z.ZodNumber>;
     maxRows: z.ZodDefault<z.ZodNumber>;
     maxColumns: z.ZodDefault<z.ZodNumber>;
 }, "strict", z.ZodTypeAny, {
@@ -26,9 +30,11 @@ declare const xlsxInput: z.ZodObject<{
     maxRows: number;
     maxColumns: number;
     sheet?: string | undefined;
+    startRow?: number | undefined;
 }, {
     path: string;
     sheet?: string | undefined;
+    startRow?: number | undefined;
     maxRows?: number | undefined;
     maxColumns?: number | undefined;
 }>;
@@ -37,6 +43,8 @@ type XlsxInput = z.infer<typeof xlsxInput>;
 export interface PdfReadOutput {
     path: string;
     totalPages: number;
+    nextPage?: number | null;
+    nextOffset?: number | null;
     firstPage: number;
     lastPage: number;
     title?: string;
@@ -63,5 +71,9 @@ export interface XlsxInspectOutput {
     }>;
     truncated: boolean;
 }
+export declare function readPdf(root: string, input: PdfInput, signal: AbortSignal): Promise<PdfReadOutput>;
+export declare function inspectXlsx(root: string, input: XlsxInput, signal: AbortSignal): Promise<XlsxInspectOutput>;
 export declare function createDocumentTools(workspaceRoot: string): [ToolDefinition<PdfInput, PdfReadOutput>, ToolDefinition<XlsxInput, XlsxInspectOutput>];
+/** Apply the same bounded ZIP admission used by XLSX before parsing DOCX. */
+export declare function validateOfficeArchive(data: Buffer): void;
 export {};

package/dist/documents/tools.js

@@ -15,6 +15,7 @@ const MAX_XLSX_ENTRY_BYTES = 16 * 1024 * 1024;
 const MAX_XLSX_SHEETS = 100;
 const pdfInput = z.object({
     path: z.string().min(1).max(1_024),
+    offset: z.number().int().min(0).optional(),
     startPage: z.number().int().positive().optional(),
     endPage: z.number().int().positive().optional(),
     maxChars: z.number().int().positive().max(MAX_OUTPUT_CHARS).default(20_000),
@@ -22,6 +23,7 @@ const pdfInput = z.object({
 const xlsxInput = z.object({
     path: z.string().min(1).max(1_024),
     sheet: z.string().min(1).max(128).optional(),
+    startRow: z.number().int().positive().optional(),
     maxRows: z.number().int().positive().max(100).default(20),
     maxColumns: z.number().int().positive().max(50).default(20),
 }).strict();
@@ -72,7 +74,7 @@ async function readWorkspaceFile(rootInput, path, signal) {
 function pdfText(items) {
     return items.map((item) => item !== null && typeof item === 'object' && 'str' in item && typeof item.str === 'string' ? item.str : '').join(' ');
 }
-async function readPdf(root, input, signal) {
+export async function readPdf(root, input, signal) {
     const { absolutePath, data } = await readWorkspaceFile(root, input.path, signal);
     let destroyDocument;
     try {
@@ -92,23 +94,31 @@ async function readPdf(root, input, signal) {
         const chunks = [];
         let used = 0;
         let truncated = false;
+        let nextPage = lastPage < pdf.numPages ? lastPage + 1 : null;
+        let nextOffset = nextPage === null ? null : 0;
         for (let pageNumber = firstPage; pageNumber <= lastPage; pageNumber++) {
             if (signal.aborted)
                 throw docError('ABORTED');
             const page = await pdf.getPage(pageNumber);
             try {
                 const content = await page.getTextContent();
-                const text = pdfText(content.items);
+                const pageOffset = pageNumber === firstPage ? input.offset ?? 0 : 0;
+                const text = pdfText(content.items).slice(pageOffset);
                 const remaining = input.maxChars - used;
                 if (remaining <= 0) {
                     truncated = true;
+                    nextPage = pageNumber;
+                    nextOffset = pageOffset;
                     break;
                 }
                 const bounded = text.slice(0, remaining);
-                if (bounded.length < text.length)
+                if (bounded.length < text.length) {
                     truncated = true;
-                chunks.push(`--- Page ${pageNumber} ---\n${bounded}`);
-                used += bounded.length;
+                    nextPage = pageNumber;
+                    nextOffset = pageOffset + bounded.length;
+                }
+                chunks.push(input.offset === undefined ? `--- Page ${pageNumber} ---\n${bounded}` : bounded);
+                used += bounded.length + (input.offset === undefined ? 0 : 2);
             }
             finally {
                 page.cleanup();
@@ -119,6 +129,7 @@ async function readPdf(root, input, signal) {
         return {
             path: input.path,
             totalPages: pdf.numPages,
+            ...(input.offset === undefined ? {} : { nextPage, nextOffset }),
             firstPage,
             lastPage: Math.min(lastPage, firstPage + chunks.length - 1),
             ...(typeof info.Title === 'string' && info.Title ? { title: info.Title.slice(0, 500) } : {}),
@@ -232,7 +243,7 @@ function cellValue(value, maxChars) {
     }
     return String(value).slice(0, maxChars);
 }
-async function inspectXlsx(root, input, signal) {
+export async function inspectXlsx(root, input, signal) {
     const { data } = await readWorkspaceFile(root, input.path, signal);
     checkXlsxArchive(Buffer.from(data));
     const workbook = new ExcelJS.Workbook();
@@ -262,6 +273,8 @@ async function inspectXlsx(root, input, signal) {
                 const previewLimit = Symbol('preview-limit');
                 try {
                     sheet.eachRow({ includeEmpty: false }, (row, rowNumber) => {
+                        if (rowNumber < (input.startRow ?? 1))
+                            return;
                         if (preview.length >= input.maxRows || remaining <= 0) {
                             truncated = true;
                             throw previewLimit;
@@ -288,7 +301,7 @@ async function inspectXlsx(root, input, signal) {
                     if (cause !== previewLimit)
                         throw cause;
                 }
-                if (rowCount > input.maxRows)
+                if (sheet.rowCount >= (input.startRow ?? 1) + input.maxRows)
                     truncated = true;
             }
             worksheets.push({ id: sheet.id, name: sheet.name.slice(0, 128), state: String(sheet.state), rowCount, columnCount, ...(preview === undefined ? {} : { preview }) });
@@ -318,7 +331,7 @@ function manifest(name, description, inputSchema, outputSchema) {
     return { name, version: '1.0.0', description, tags: ['documents', 'read'], inputSchema, outputSchema, concurrencyClass: 'tool', locks: [], supportsAbortSignal: true, sideEffectPolicy: 'read', retrySafety: 'read_only', defaultTimeoutMs: 60_000, maxResultSummaryBytes: 4_096 };
 }
 const pdfManifest = manifest('document.pdf.read', 'Extract bounded text and metadata from a PDF in the workspace.', { type: 'object', required: ['path'], properties: { path: { type: 'string' }, startPage: { type: 'integer', minimum: 1 }, endPage: { type: 'integer', minimum: 1 }, maxChars: { type: 'integer', minimum: 1, maximum: MAX_OUTPUT_CHARS, default: 20_000 } }, additionalProperties: false }, { type: 'object', required: ['path', 'totalPages', 'firstPage', 'lastPage', 'text', 'truncated'], properties: { path: { type: 'string' }, totalPages: { type: 'integer' }, firstPage: { type: 'integer' }, lastPage: { type: 'integer' }, title: { type: 'string' }, author: { type: 'string' }, text: { type: 'string' }, truncated: { type: 'boolean' } }, additionalProperties: false });
-const xlsxManifest = manifest('document.xlsx.inspect', 'List workbook metadata and worksheet dimensions; optionally return a bounded cell preview.', { type: 'object', required: ['path'], properties: { path: { type: 'string' }, sheet: { type: 'string' }, maxRows: { type: 'integer', minimum: 1, maximum: 100, default: 20 }, maxColumns: { type: 'integer', minimum: 1, maximum: 50, default: 20 } }, additionalProperties: false }, { type: 'object', required: ['path', 'sheetCount', 'worksheets', 'truncated'], properties: { path: { type: 'string' }, creator: { type: 'string' }, title: { type: 'string' }, subject: { type: 'string' }, sheetCount: { type: 'integer' }, worksheets: { type: 'array', items: { type: 'object', required: ['id', 'name', 'state', 'rowCount', 'columnCount'], properties: { id: { type: 'integer' }, name: { type: 'string' }, state: { type: 'string' }, rowCount: { type: 'integer' }, columnCount: { type: 'integer' }, preview: { type: 'array', items: { type: 'object', required: ['row', 'cells'], properties: { row: { type: 'integer' }, cells: { type: 'array', items: { anyOf: [{ type: 'string' }, { type: 'number' }, { type: 'boolean' }, { type: 'null' }] } } }, additionalProperties: false } } }, additionalProperties: false } }, truncated: { type: 'boolean' } }, additionalProperties: false });
+const xlsxManifest = manifest('document.xlsx.inspect', 'List workbook metadata and worksheet dimensions; optionally return a bounded cell preview.', { type: 'object', required: ['path'], properties: { path: { type: 'string' }, sheet: { type: 'string' }, startRow: { type: 'integer', minimum: 1 }, maxRows: { type: 'integer', minimum: 1, maximum: 100, default: 20 }, maxColumns: { type: 'integer', minimum: 1, maximum: 50, default: 20 } }, additionalProperties: false }, { type: 'object', required: ['path', 'sheetCount', 'worksheets', 'truncated'], properties: { path: { type: 'string' }, creator: { type: 'string' }, title: { type: 'string' }, subject: { type: 'string' }, sheetCount: { type: 'integer' }, worksheets: { type: 'array', items: { type: 'object', required: ['id', 'name', 'state', 'rowCount', 'columnCount'], properties: { id: { type: 'integer' }, name: { type: 'string' }, state: { type: 'string' }, rowCount: { type: 'integer' }, columnCount: { type: 'integer' }, preview: { type: 'array', items: { type: 'object', required: ['row', 'cells'], properties: { row: { type: 'integer' }, cells: { type: 'array', items: { anyOf: [{ type: 'string' }, { type: 'number' }, { type: 'boolean' }, { type: 'null' }] } } }, additionalProperties: false } } }, additionalProperties: false } }, truncated: { type: 'boolean' } }, additionalProperties: false });
 export function createDocumentTools(workspaceRoot) {
     const root = resolve(workspaceRoot);
     return [
@@ -336,3 +349,5 @@ export function createDocumentTools(workspaceRoot) {
         },
     ];
 }
+/** Apply the same bounded ZIP admission used by XLSX before parsing DOCX. */
+export function validateOfficeArchive(data) { checkXlsxArchive(data); }

package/dist/index.d.ts

@@ -12,3 +12,4 @@ export * from './documents/tools.js';
 export * from './tools/registry.js';
 export * from './workers/http.js';
 export * from './mcp/index.js';
+export * from './documents/docx.js';

package/dist/index.js

@@ -12,3 +12,4 @@ export * from './documents/tools.js';
 export * from './tools/registry.js';
 export * from './workers/http.js';
 export * from './mcp/index.js';
+export * from './documents/docx.js';

package/dist/native/srt-win.json

@@ -7,5 +7,5 @@
   "patchUrl": "https://github.com/anthropics/sandbox-runtime/pull/523",
   "patchSha256": "3cf1d16ef273c9d066ff0479b1cd28178a7453a41dbdf80cbc3239dd831ca067",
   "testAdaptation": "Parent SDDL round-trip permits only D:( to D:AI( canonicalization; all owner/group/ACE/protection comparisons remain exact.",
-  "sha256": "17d3dd079dbfc8ec42d6e623f612d31d8ca3a2633627910d8bcc4ae5eecfec06"
+  "sha256": "c091963f50f93cd5b3b14bad6b6d7dcbbe00f0836dc822e75ea9c4b93fcaabc5"
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@hunterzhu/pulse-adapters",
-  "version": "0.6.0",
+  "version": "0.7.0",
   "license": "PolyForm-Noncommercial-1.0.0",
   "repository": {
     "type": "git",
@@ -18,9 +18,10 @@
   },
   "dependencies": {
     "@anthropic-ai/sandbox-runtime": "0.0.77",
-    "@hunterzhu/pulse-runtime": "0.6.0",
-    "@hunterzhu/pulse-tool-sdk": "0.6.0",
+    "@hunterzhu/pulse-runtime": "0.7.0",
+    "@hunterzhu/pulse-tool-sdk": "0.7.0",
     "exceljs": "4.4.0",
+    "mammoth": "1.11.0",
     "pdfjs-dist": "4.10.38"
   }
 }