feat: support big file downloads (read from WA cache + streaming)

.eslintignore

@@ -3,6 +3,7 @@ dist
 coverage
 docs
 *.min.js
+*.d.ts
 .wa-version
 .wwebjs_auth
 .wwebjs_cache

example.js

@@ -1,3 +1,4 @@
+const fs = require('fs');
 const { Client, Location, Poll, List, Buttons, LocalAuth } = require('./index');
 
 const client = new Client({
@@ -232,6 +233,18 @@ client.on('message', async (msg) => {
             Platform: ${info.platform}
         `,
         );
+    } else if (msg.body === '!streamdownload' && msg.hasMedia) {
+        const result = await msg.downloadMediaStream();
+        if (result) {
+            const filePath = `./${result.filename || 'download'}`;
+            const writeStream = fs.createWriteStream(filePath);
+            result.stream.pipe(writeStream);
+            writeStream.on('finish', () => {
+                msg.reply(
+                    `Media saved to ${filePath} (${result.mimetype}, ${result.filesize} bytes)`,
+                );
+            });
+        }
     } else if (msg.body === '!mediainfo' && msg.hasMedia) {
         const attachmentData = await msg.downloadMedia();
         msg.reply(`

index.d.ts

@@ -1,4 +1,5 @@
 import { EventEmitter } from 'events';
+import { Readable } from 'stream';
 import { RequestInit } from 'node-fetch';
 import * as puppeteer from 'puppeteer';
 import InterfaceController from './src/util/InterfaceController';
@@ -198,7 +199,7 @@ declare namespace WAWebJS {
         ): Promise<string>;
 
         /** Cancels an active pairing code session and returns to QR code mode */
-        cancelPairingCode(): Promise<void>
+        cancelPairingCode(): Promise<void>;
 
         /** Force reset of connection state for the client */
         resetState(): Promise<void>;
@@ -1308,7 +1309,11 @@ declare namespace WAWebJS {
         /** Deletes the message from the chat */
         delete: (everyone?: boolean, clearMedia?: boolean) => Promise<void>;
         /** Downloads and returns the attached message media */
-        downloadMedia: () => Promise<MessageMedia>;
+        downloadMedia: () => Promise<MessageMedia | undefined>;
+        /** Downloads the attached message media as a Node.js Readable stream */
+        downloadMediaStream: (
+            options?: MediaStreamOptions,
+        ) => Promise<MessageMediaStream | undefined>;
         /** Returns the Chat this message was sent in */
         getChat: () => Promise<Chat>;
         /** Returns the Contact this message was sent from */
@@ -1616,8 +1621,18 @@ declare namespace WAWebJS {
         reqOptions?: RequestInit;
     }
 
+    /** Common metadata for media attached to a message */
+    export interface MessageMediaMetadata {
+        /** MIME type of the attachment */
+        mimetype: string;
+        /** Document file name. Value can be null */
+        filename?: string | null;
+        /** Document file size in bytes. Value can be null. */
+        filesize?: number | null;
+    }
+
     /** Media attached to a message */
-    export class MessageMedia {
+    export class MessageMedia implements MessageMediaMetadata {
         /** MIME type of the attachment */
         mimetype: string;
         /** Base64-encoded data of the file */
@@ -1650,6 +1665,17 @@ declare namespace WAWebJS {
         ) => Promise<MessageMedia>;
     }
 
+    /** Options for downloadMediaStream */
+    export interface MediaStreamOptions {
+        /** Size in bytes of each chunk read from the browser (default 10MB) */
+        chunkSize?: number;
+    }
+
+    /** Result of downloadMediaStream: a Readable stream with media metadata */
+    export interface MessageMediaStream extends MessageMediaMetadata {
+        stream: Readable;
+    }
+
     export type MessageContent =
         | string
         | MessageMedia

src/structures/Message.js

@@ -1,5 +1,6 @@
 'use strict';
 
+const { Readable } = require('stream');
 const Base = require('./Base');
 const MessageMedia = require('./MessageMedia');
 const Location = require('./Location');
@@ -507,84 +508,25 @@ class Message extends Base {
     }
 
     /**
-     * Downloads and returns the attatched message media
-     * @returns {Promise<MessageMedia>}
+     * Downloads and returns the attached message media
+     * @returns {Promise<MessageMedia|undefined>}
      */
     async downloadMedia() {
-        if (!this.hasMedia) {
-            return undefined;
-        }
+        if (!this.hasMedia) return undefined;
 
         const result = await this.client.pupPage.evaluate(async (msgId) => {
-            const msg =
-                window.require('WAWebCollections').Msg.get(msgId) ||
-                (
-                    await window
-                        .require('WAWebCollections')
-                        .Msg.getMessagesById([msgId])
-                )?.messages?.[0];
-
-            // REUPLOADING mediaStage means the media is expired and the download button is spinning, cannot be downloaded now
-            if (
-                !msg ||
-                !msg.mediaData ||
-                msg.mediaData.mediaStage === 'REUPLOADING'
-            ) {
-                return null;
-            }
-            if (msg.mediaData.mediaStage != 'RESOLVED') {
-                // try to resolve media
-                await msg.downloadMedia({
-                    downloadEvenIfExpensive: true,
-                    rmrReason: 1,
-                });
-            }
-
-            if (
-                msg.mediaData.mediaStage.includes('ERROR') ||
-                msg.mediaData.mediaStage === 'FETCHING'
-            ) {
-                // media could not be downloaded
-                return undefined;
-            }
-
-            try {
-                const mockQpl = {
-                    addAnnotations: function () {
-                        return this;
-                    },
-                    addPoint: function () {
-                        return this;
-                    },
-                };
-                const decryptedMedia = await window
-                    .require('WAWebDownloadManager')
-                    .downloadManager.downloadAndMaybeDecrypt({
-                        directPath: msg.directPath,
-                        encFilehash: msg.encFilehash,
-                        filehash: msg.filehash,
-                        mediaKey: msg.mediaKey,
-                        mediaKeyTimestamp: msg.mediaKeyTimestamp,
-                        type: msg.type,
-                        signal: new AbortController().signal,
-                        downloadQpl: mockQpl,
-                    });
-
-                const data =
-                    await window.WWebJS.arrayBufferToBase64Async(
-                        decryptedMedia,
-                    );
+            const resolved = await window.WWebJS.resolveMediaBlob(msgId);
+            if (!resolved) return null;
 
-                return {
-                    data,
-                    mimetype: msg.mimetype,
-                    filename: msg.filename,
-                    filesize: msg.size,
-                };
-            } catch (e) {
-                if (e.status && e.status === 404) return undefined;
-                throw e;
-            }
+            const data = await window.WWebJS.arrayBufferToBase64Async(
+                await resolved.blob.arrayBuffer(),
+            );
+            return {
+                data,
+                mimetype: resolved.mimetype,
+                filename: resolved.filename,
+                filesize: resolved.filesize,
+            };
         }, this.id._serialized);
 
         if (!result) return undefined;
@@ -596,6 +538,67 @@ class Message extends Base {
         );
     }
 
+    /**
+     * Like downloadMedia(), but returns a Readable stream instead of loading the entire file into memory.
+     * @param {Object} [options]
+     * @param {number} [options.chunkSize=10485760] Size in bytes of each chunk read from the browser (default 10MB)
+     * @returns {Promise<MessageMediaStream|undefined>} undefined if media is unavailable
+     */
+    async downloadMediaStream({ chunkSize = 10 * 1024 * 1024 } = {}) {
+        if (!this.hasMedia) return undefined;
+
+        const blobHandle = await this.client.pupPage.evaluateHandle(
+            async (msgId) => {
+                const result = await window.WWebJS.resolveMediaBlob(msgId);
+                return result?.blob ?? null;
+            },
+            this.id._serialized,
+        );
+
+        let metadata;
+        try {
+            metadata = await blobHandle.evaluate((blob, msgId) => {
+                if (!blob) return null;
+                const msg = window.require('WAWebCollections').Msg.get(msgId);
+                return {
+                    blobSize: blob.size,
+                    mimetype: msg?.mimetype,
+                    filename: msg?.filename,
+                    filesize: msg?.size,
+                };
+            }, this.id._serialized);
+        } catch (err) {
+            await blobHandle.dispose().catch(() => {});
+            throw err;
+        }
+        if (!metadata) {
+            await blobHandle.dispose().catch(() => {});
+            return undefined;
+        }
+
+        const { blobSize, ...rest } = metadata;
+
+        async function* readChunks() {
+            try {
+                for (let offset = 0; offset < blobSize; offset += chunkSize) {
+                    const base64 = await blobHandle.evaluate(
+                        async (blob, s, e) =>
+                            window.WWebJS.arrayBufferToBase64Async(
+                                await blob.slice(s, e).arrayBuffer(),
+                            ),
+                        offset,
+                        offset + chunkSize,
+                    );
+                    yield Buffer.from(base64, 'base64');
+                }
+            } finally {
+                await blobHandle.dispose().catch(() => {});
+            }
+        }
+
+        return { stream: Readable.from(readChunks()), ...rest };
+    }
+
     /**
      * Deletes a message from the chat
      * @param {?boolean} everyone If true and the message is sent by the current user or the user is an admin, will delete it for everyone in the chat.

src/util/Injected/Utils.js

@@ -1105,6 +1105,63 @@ exports.LoadUtils = () => {
         });
     };
 
+    /**
+     * Resolves the media blob and metadata for a message.
+     * Shared by downloadMedia and downloadMediaStream.
+     * @param {string} msgId
+     * @returns {Promise<{blob: Blob, mimetype: string, filename: string, filesize: number}|null>}
+     */
+    window.WWebJS.resolveMediaBlob = async (msgId) => {
+        const { Msg } = window.require('WAWebCollections');
+        const msg =
+            Msg.get(msgId) ||
+            (await Msg.getMessagesById([msgId]))?.messages?.[0];
+
+        if (
+            !msg ||
+            !msg.mediaData ||
+            msg.mediaData.mediaStage === 'REUPLOADING'
+        ) {
+            return null;
+        }
+
+        // Always call internal downloadMedia - never skip based on
+        // mediaStage, because cache eviction can leave stage=RESOLVED
+        // with empty InMemoryMediaBlobCache.
+        await msg.downloadMedia({
+            downloadEvenIfExpensive: true,
+            rmrReason: 1,
+            isUserInitiated: true,
+        });
+
+        if (
+            msg.mediaData.mediaStage.includes('ERROR') ||
+            msg.mediaData.mediaStage === 'FETCHING'
+        ) {
+            return null;
+        }
+
+        const cached = window
+            .require('WAWebMediaInMemoryBlobCache')
+            .InMemoryMediaBlobCache.get(msg.mediaObject?.filehash);
+
+        let blob;
+        if (cached) {
+            blob = cached;
+        } else if (msg.mediaObject?.mediaBlob) {
+            blob = msg.mediaObject.mediaBlob.forceToBlob();
+        }
+
+        if (!blob) return null;
+
+        return {
+            blob,
+            mimetype: msg.mimetype,
+            filename: msg.filename,
+            filesize: msg.size,
+        };
+    };
+
     window.WWebJS.arrayBufferToBase64 = (arrayBuffer) => {
         let binary = '';
         const bytes = new Uint8Array(arrayBuffer);