fix language extensions and make schedule/paths customizable

Author: John Corser

.github/workflows/docker-publish.yml

@@ -79,10 +79,18 @@ jobs:
                 image: mrorbitman/subsyncarr:latest
                 container_name: subsyncarr
                 volumes:
-                  - ${MEDIA_PATH:-/path/to/your/media}:/scan_dir
+                  # Any path configured with SCAN_PATHS env var must be mounted
+                  # If no scan paths are specified, it defaults to scan `/scan_dir` like example below.
+                  # - ${MEDIA_PATH:-/path/to/your/media}:/scan_dir
+                  - /path/to/movies:/movies
+                  - /path/to/tv:/tv
+                  - /path/to/anime:/anime
                 restart: unless-stopped
                 environment:
                   - TZ=${TZ:-UTC}
+                  - CRON_SCHEDULE=0 0 * * *  # Runs every day at midnight by default
+                  - SCAN_PATHS=/movies,/tv,/anime # Remember to mount these as volumes. Must begin with /. Default valus is `/scan_dir`
+                  - EXCLUDE_PATHS=/movies/temp,/tv/downloads # Exclude certain sub-directories from the scan
             ```
 
             Docker Hub URL: https://hub.docker.com/r/mrorbitman/subsyncarr/tags

Dockerfile

@@ -38,9 +38,13 @@ COPY . .
 RUN npm run build
 
 # Create startup script
+# Set default cron schedule (if not provided by environment variable)
+ENV CRON_SCHEDULE="0 0 * * *"
+
+# Create startup script with environment variable
 RUN echo '#!/bin/bash\n\
 # Add cron job\n\
-echo "0 0 * * * cd /app && /usr/local/bin/node /app/dist/index.js >> /var/log/cron.log 2>&1" > /etc/cron.d/subsyncarr\n\
+echo "${CRON_SCHEDULE} cd /app && /usr/local/bin/node /app/dist/index.js >> /var/log/cron.log 2>&1" > /etc/cron.d/subsyncarr\n\
 chmod 0644 /etc/cron.d/subsyncarr\n\
 crontab /etc/cron.d/subsyncarr\n\
 \n\

docker-compose.yaml

@@ -5,7 +5,15 @@ services:
     image: mrorbitman/subsyncarr:latest
     container_name: subsyncarr
     volumes:
-      - ${MEDIA_PATH:-/path/to/your/media}:/scan_dir
+      # Any path configured with SCAN_PATHS env var must be mounted
+      # If no scan paths are specified, it defaults to scan `/scan_dir` like example below.
+      # - ${MEDIA_PATH:-/path/to/your/media}:/scan_dir
+      - /path/to/movies:/movies
+      - /path/to/tv:/tv
+      - /path/to/anime:/anime
     restart: unless-stopped
     environment:
       - TZ=${TZ:-UTC}
+      - CRON_SCHEDULE=0 0 * * * # Runs every day at midnight by default
+      - SCAN_PATHS=/movies,/tv,/anime # Remember to mount these as volumes. Must begin with /. Default valus is `/scan_dir`
+      - EXCLUDE_PATHS=/movies/temp,/tv/downloads # Exclude certain sub-directories from the scan

src/config.ts

@@ -0,0 +1,46 @@
+export interface ScanConfig {
+  includePaths: string[];
+  excludePaths: string[];
+}
+
+function validatePath(path: string): boolean {
+  // Add any path validation logic you need
+  return path.startsWith('/') && !path.includes('..');
+}
+
+export function getScanConfig(): ScanConfig {
+  const scanPaths = process.env.SCAN_PATHS?.split(',').filter(Boolean) || ['/scan_dir'];
+  const excludePaths = process.env.EXCLUDE_PATHS?.split(',').filter(Boolean) || [];
+
+  // Validate paths
+  const validIncludePaths = scanPaths.filter((path) => {
+    const isValid = validatePath(path);
+    if (!isValid) {
+      console.warn(`Invalid include path: ${path}`);
+    }
+    return isValid;
+  });
+
+  const validExcludePaths = excludePaths.filter((path) => {
+    const isValid = validatePath(path);
+    if (!isValid) {
+      console.warn(`Invalid exclude path: ${path}`);
+    }
+    return isValid;
+  });
+
+  if (validIncludePaths.length === 0) {
+    console.warn('No valid scan paths provided, defaulting to /scan_dir');
+    validIncludePaths.push('/scan_dir');
+  }
+
+  console.log('Scan configuration:', {
+    includePaths: validIncludePaths,
+    excludePaths: validExcludePaths,
+  });
+
+  return {
+    includePaths: validIncludePaths,
+    excludePaths: validExcludePaths,
+  };
+}

src/findAllSrtFiles.ts

@@ -1,10 +1,16 @@
 import { readdir } from 'fs/promises';
 import { extname, join } from 'path';
+import { ScanConfig } from './config';
 
-export async function findAllSrtFiles(dir: string): Promise<string[]> {
+export async function findAllSrtFiles(config: ScanConfig): Promise<string[]> {
   const files: string[] = [];
 
   async function scan(directory: string): Promise<void> {
+    // Check if this directory should be excluded
+    if (config.excludePaths.some((excludePath) => directory.startsWith(excludePath))) {
+      return;
+    }
+
     const entries = await readdir(directory, { withFileTypes: true });
 
     for (const entry of entries) {
@@ -18,6 +24,10 @@ export async function findAllSrtFiles(dir: string): Promise<string[]> {
     }
   }
 
-  await scan(dir);
+  // Scan all included paths
+  for (const includePath of config.includePaths) {
+    await scan(includePath);
+  }
+
   return files;
 }

src/findMatchingVideoFile.ts

@@ -8,13 +8,27 @@ export function findMatchingVideoFile(srtPath: string): string | null {
   const directory = dirname(srtPath);
   const srtBaseName = basename(srtPath, '.srt');
 
-  // Try to find a video file with the same name but different extension
+  // Try exact match first
   for (const ext of VIDEO_EXTENSIONS) {
     const possibleVideoPath = join(directory, `${srtBaseName}${ext}`);
     if (existsSync(possibleVideoPath)) {
       return possibleVideoPath;
     }
   }
 
+  // Progressive tag removal - split by dots and try removing one segment at a time
+  const segments = srtBaseName.split('.');
+  while (segments.length > 1) {
+    segments.pop(); // Remove the last segment
+    const baseNameToTry = segments.join('.');
+
+    for (const ext of VIDEO_EXTENSIONS) {
+      const possibleVideoPath = join(directory, `${baseNameToTry}${ext}`);
+      if (existsSync(possibleVideoPath)) {
+        return possibleVideoPath;
+      }
+    }
+  }
+
   return null;
 }

src/index.ts

@@ -3,6 +3,7 @@ import { findAllSrtFiles } from './findAllSrtFiles';
 import { findMatchingVideoFile } from './findMatchingVideoFile';
 import { generateFfsubsyncSubtitles } from './generateFfsubsyncSubtitles';
 import { generateAutosubsyncSubtitles } from './generateAutosubsyncSubtitles';
+import { getScanConfig } from './config';
 
 const SCAN_DIR = '/scan_dir';
 
@@ -12,7 +13,8 @@ async function main(): Promise<void> {
 
   try {
     // Find all .srt files
-    const srtFiles = await findAllSrtFiles(scanDir);
+    const scanConfig = getScanConfig();
+    const srtFiles = await findAllSrtFiles(scanConfig);
     console.log(`${new Date().toLocaleString()} Found ${srtFiles.length} SRT files`);
 
     // Process each SRT file