yenc functions decode 1 part files correctly

package-lock.json

@@ -15,6 +15,7 @@
         "log4js": "^6.9.1",
         "nntp-js": "^1.0.4",
         "node-unrar-js": "^2.0.0",
+        "simple-yenc": "^1.0.4",
         "sqlite": "^5.1.1",
         "sqlite3": "^6.0.1",
         "xmlbuilder2": "^3.1.1",
@@ -1019,6 +1020,16 @@
         "simple-concat": "^1.0.0"
       }
     },
+    "node_modules/simple-yenc": {
+      "version": "1.0.4",
+      "resolved": "https://registry.npmjs.org/simple-yenc/-/simple-yenc-1.0.4.tgz",
+      "integrity": "sha512-5gvxpSd79e9a3V4QDYUqnqxeD4HGlhCakVpb6gMnDD7lexJggSBJRBO5h52y/iJrdXRilX9UCuDaIJhSWm5OWw==",
+      "license": "MIT",
+      "funding": {
+        "type": "individual",
+        "url": "https://github.com/sponsors/eshaz"
+      }
+    },
     "node_modules/sprintf-js": {
       "version": "1.0.3",
       "resolved": "https://registry.npmjs.org/sprintf-js/-/sprintf-js-1.0.3.tgz",

package.json

@@ -18,9 +18,10 @@
     "log4js": "^6.9.1",
     "nntp-js": "^1.0.4",
     "node-unrar-js": "^2.0.0",
+    "simple-yenc": "^1.0.4",
     "sqlite": "^5.1.1",
     "sqlite3": "^6.0.1",
     "xmlbuilder2": "^3.1.1",
     "yencode": "^1.0.1"
   }
-}
+}

src/body.worker.js

@@ -2,6 +2,7 @@ import { Queue, Worker } from 'bullmq';
 import log4js from './logger.js';
 import { headerQueue } from './header.worker.js';
 import { acquire, release } from './nntp.pool.js';
+import { decodeYenc } from './yenc.util.js';
 
 const bodyLogger = log4js.getLogger('body');
 
@@ -23,10 +24,8 @@ export const startBodyWorker = () => {
     try {
       conn = await acquire();
       const bodyBuffer = (await conn.body(header['message-id'])).data;
-
-      const firstNewlineIndex = bodyBuffer.indexOf('\\n');
-      const firstLineBuffer = (firstNewlineIndex !== -1) ? bodyBuffer.slice(0, firstNewlineIndex) : bodyBuffer;
-      const firstLine = firstLineBuffer.toString();
+      const decodedBuffer = decodeYenc(bodyBuffer);
+      const firstLine = decodedBuffer.toString().split('\\n')[0];
 
       const match = firstLine.match(YENC_REGEX);
 

src/collection.worker.js

@@ -3,7 +3,7 @@ import log4js from './logger.js';
 import { getDb } from './database.js';
 import { acquire, release } from './nntp.pool.js';
 import { createExtractorFromData } from 'node-unrar-js';
-import * as yEnc from 'simple-yenc';
+import { decodeYenc } from './yenc.util.js';
 
 const logger = log4js.getLogger('collection');
 
@@ -45,9 +45,8 @@ export const startCollectionWorker = () => {
         conn = await acquire();
         await conn.group('alt.binaries.test');
         const bodyBuffer = (await conn.body(`<${firstPart.id}>`)).data;
-        const decodedUint8Array = yEnc.decode(bodyBuffer.toString('latin1'));
-        const buffer = Buffer.from(decodedUint8Array);
-        const extractor = await createExtractorFromData({ data: buffer });
+        const decodedBuffer = decodeYenc(bodyBuffer);
+        const extractor = await createExtractorFromData({ data: decodedBuffer });
         const fileList = extractor.getFileList();
         logger.info(`Files in "${file.filename}":`, fileList);
       } catch (error) {

src/download.js

@@ -1,7 +1,7 @@
 import 'dotenv/config';
 import { getDb } from './database.js';
 import { acquire, release, shutdown } from './nntp.pool.js';
-import * as yEnc from 'simple-yenc';
+import { decodeYenc } from './yenc.util.js';
 import fs from 'fs/promises';
 import log4js from './logger.js';
 
@@ -35,9 +35,8 @@ async function downloadFile(fileId) {
       try {
         logger.debug(`Downloading part ${partNumber}/${file.parts} with message ID: ${segment.id}`);
         const bodyBuffer = (await conn.body(`<${segment.id}>`)).data;
-        const decodedUint8Array = yEnc.decode(bodyBuffer.toString('latin1'));
-        const buffer = Buffer.from(decodedUint8Array);
-        parts.push(buffer);
+        const decodedPart = decodeYenc(bodyBuffer);
+        parts.push(decodedPart);
       } catch (error) {
         if (error.code === 430) {
           logger.error(`Article not found for part ${partNumber} (Message ID: ${segment.id})`);

src/yenc.test.js

@@ -1,13 +1,12 @@
-import yencode from 'yencode';
+import { decodeYenc } from './yenc.util.js';
 import fs from 'fs/promises';
 import { Buffer } from 'buffer';
 
 async function runTest() {
-  const encodedData = await fs.readFile('HjVfQlWmHdUrQeQkRiLkTwEj-1779830864932@nyuu.bin');
-  const correctlyDecodedData = await fs.readFile('Dragon.Ball.S01E119.MULTI.BDRip.REMASTERED.1080p.x264.DTS-LILAS.par2');
+  const encodedData = await fs.readFile('files/HjVfQlWmHdUrQeQkRiLkTwEj-1779830864932@nyuu.bin');
+  const correctlyDecodedData = await fs.readFile('files/Dragon.Ball.S01E119.MULTI.BDRip.REMASTERED.1080p.x264.DTS-LILAS.par2-good');
 
-  const decodedBuffer = yencode.decode(encodedData);
-  await fs.writeFile('decoded.bin', decodedBuffer)
+  const decodedBuffer = decodeYenc(encodedData);
 
   if (Buffer.compare(decodedBuffer, correctlyDecodedData) === 0) {
     console.log('Test passed: Decoded data matches the correctly decoded file.');
@@ -15,7 +14,7 @@ async function runTest() {
     console.error('Test failed: Decoded data does not match the correctly decoded file.');
     console.error('Decoded buffer length:', decodedBuffer.length);
     console.error('Correct buffer length:', correctlyDecodedData.length);
-    await fs.writeFile('test-decoded-output.bin', decodedBuffer);
+    await fs.writeFile('files/test-decoded-output.bin', decodedBuffer);
   }
 }
 

src/yenc.util.js

@@ -0,0 +1,91 @@
+import yencode from 'yencode';
+
+/**
+ * Parses yEnc metadata from a header or footer line.
+ * @param {string} line The line to parse.
+ * @returns {object} A key-value map of the metadata.
+ */
+function parseMetaLine(line) {
+  const meta = {};
+  const parts = line.split(' ');
+  for (const part of parts) {
+    const eqIndex = part.indexOf('=');
+    if (eqIndex !== -1) {
+      meta[part.slice(0, eqIndex)] = part.slice(eqIndex + 1);
+    }
+  }
+  return meta;
+}
+
+/**
+ * Extracts metadata from a yEnc-encoded buffer.
+ * @param {Buffer} encodedBuffer The yEnc-encoded buffer.
+ * @returns {object} An object containing all metadata from the header and footer.
+ */
+export function parseYencMeta(encodedBuffer) {
+  const headerBegin = Buffer.from('=ybegin');
+  const headerPart = Buffer.from('=ypart');
+  const footer = Buffer.from('\n=yend');
+
+  const headerStartIndex = encodedBuffer.indexOf(headerBegin);
+  const partStartIndex = encodedBuffer.indexOf(headerPart);
+  const footerStartIndex = encodedBuffer.lastIndexOf(footer);
+
+  if ((headerStartIndex === -1 && partStartIndex === -1) || footerStartIndex === -1) {
+    throw new Error('Invalid yEnc data: missing header or footer.');
+  }
+
+  const headerIndex = headerStartIndex !== -1 ? headerStartIndex : partStartIndex;
+  const headerEndIndex = encodedBuffer.indexOf(Buffer.from('\\n'), headerIndex);
+  const headerLine = encodedBuffer.subarray(headerIndex, headerEndIndex).toString();
+
+  const footerEndIndex = encodedBuffer.indexOf(Buffer.from('\\n'), footerStartIndex);
+  const footerLine = encodedBuffer.subarray(footerStartIndex, footerEndIndex).toString();
+
+  const meta = {
+    ...parseMetaLine(headerLine),
+    ...parseMetaLine(footerLine),
+  };
+
+  // Convert numeric values
+  for (const key in meta) {
+    if (!isNaN(meta[key])) {
+      meta[key] = parseInt(meta[key], 10);
+    }
+  }
+
+  return meta;
+}
+
+/**
+ * Decodes a yEnc-encoded buffer and optionally writes it to a target buffer.
+ * @param {Buffer} encodedBuffer The yEnc-encoded buffer.
+ * @param {Buffer} [targetBuffer] An optional buffer to write the decoded data into.
+ * @returns {Buffer} The decoded data (or the target buffer if provided).
+ */
+export function decodeYenc(encodedBuffer, targetBuffer) {
+  const meta = parseYencMeta(encodedBuffer);
+
+  const header = meta.part ? Buffer.from(`=ypart`) : Buffer.from(`=ybegin`);
+  const footer = Buffer.from(`\n=yend`);
+
+  const contentStartIndex = encodedBuffer.indexOf('\n', encodedBuffer.indexOf(header)) + 1;
+  const contentEndIndex = encodedBuffer.lastIndexOf(footer);
+  const dataToDecode = encodedBuffer.subarray(contentStartIndex, contentEndIndex);
+
+  const decoded = yencode.decode(dataToDecode);
+
+  if (decoded.length !== meta.size) {
+    throw new Error(`Decoded size (${decoded.length}) does not match expected size (${meta.size}).`);
+  }
+
+  if (targetBuffer) {
+    if (meta.begin === undefined) {
+      throw new Error('Cannot write to target buffer: missing "begin" offset in yEnc metadata.');
+    }
+    decoded.copy(targetBuffer, meta.begin - 1);
+    return targetBuffer;
+  }
+
+  return decoded;
+}