Merge pull request #12 from Wal33D/claude-image-sanitization

Remove image standardization and implement consistent filename sanitization

Author: Wal33D

src/indexers/archive/InternetArchiveIndexer.ts

@@ -301,14 +301,14 @@ export class InternetArchiveIndexer extends EventEmitter {
       // Download best screenshot
       if (images.screenshots.length > 0) {
         const screenshot = images.screenshots[0];
-        const ext = path.extname(screenshot.name).toLowerCase();
-        const localPath = path.join(levelDir, `screenshot_original${ext}`);
+        const sanitizedName = FileUtils.sanitizeFilename(screenshot.name);
+        const localPath = path.join(levelDir, sanitizedName);
         const downloadUrl = `https://archive.org/download/${metadata.identifier}/${encodeURIComponent(screenshot.name)}`;
 
         const downloadPromise = this.downloadFile(downloadUrl, localPath)
           .then(() => {
             levelFiles.push({
-              filename: `screenshot_original${ext}`,
+              filename: sanitizedName,
               path: localPath,
               size: parseInt(screenshot.size || '0'),
               type: 'other',
@@ -325,14 +325,14 @@ export class InternetArchiveIndexer extends EventEmitter {
       // Download best thumbnail
       if (images.thumbnails.length > 0) {
         const thumbnail = images.thumbnails[0];
-        const ext = path.extname(thumbnail.name).toLowerCase();
-        const localPath = path.join(levelDir, `thumbnail_original${ext}`);
+        const sanitizedName = FileUtils.sanitizeFilename(thumbnail.name);
+        const localPath = path.join(levelDir, sanitizedName);
         const downloadUrl = `https://archive.org/download/${metadata.identifier}/${encodeURIComponent(thumbnail.name)}`;
 
         const downloadPromise = this.downloadFile(downloadUrl, localPath)
           .then(() => {
             levelFiles.push({
-              filename: `thumbnail_original${ext}`,
+              filename: sanitizedName,
               path: localPath,
               size: parseInt(thumbnail.size || '0'),
               type: 'other',

src/indexers/discord/discordBase.ts

@@ -410,13 +410,14 @@ export abstract class DiscordBaseIndexer {
       await fs.ensureDir(levelDir);
 
       // Download the .dat file
-      const datPath = path.join(levelDir, datAttachment.filename);
+      const sanitizedDatName = FileUtils.sanitizeFilename(datAttachment.filename);
+      const datPath = path.join(levelDir, sanitizedDatName);
       await this.downloadFile(datAttachment.url, datPath);
 
       // Create level files array
       const files: LevelFile[] = [
         {
-          filename: datAttachment.filename,
+          filename: sanitizedDatName,
           path: datPath,
           size: datAttachment.size,
           type: 'dat',
@@ -426,25 +427,27 @@ export abstract class DiscordBaseIndexer {
       // Download associated images
       for (const imageAtt of associatedImages) {
         try {
-          const imageName = imageAtt.filename;
-          logger.info(`Downloading image: ${imageName} for level ${datAttachment.filename}`);
-          const imagePath = path.join(levelDir, imageName);
+          const sanitizedImageName = FileUtils.sanitizeFilename(imageAtt.filename);
+          logger.info(
+            `Downloading image: ${imageAtt.filename} for level ${datAttachment.filename}`
+          );
+          const imagePath = path.join(levelDir, sanitizedImageName);
           await this.downloadFile(imageAtt.url, imagePath);
 
           // Determine image type
           let imageType: 'thumbnail' | 'image' = 'image';
-          if (imageName.toLowerCase().includes('thumb')) {
+          if (sanitizedImageName.toLowerCase().includes('thumb')) {
             imageType = 'thumbnail';
           }
 
           files.push({
-            filename: imageName,
+            filename: sanitizedImageName,
             path: imagePath,
             size: imageAtt.size,
             type: imageType,
           });
 
-          logger.info(`Downloaded image: ${imageName} (${imageType})`);
+          logger.info(`Downloaded image: ${sanitizedImageName} (${imageType})`);
         } catch (error) {
           logger.warn(`Failed to download image ${imageAtt.filename}:`, error);
         }
@@ -569,7 +572,8 @@ export abstract class DiscordBaseIndexer {
 
       // Download ZIP file
       logger.info(`Downloading zip: ${zipAttachment.filename}`);
-      const zipPath = path.join(tempDir, zipAttachment.filename);
+      const sanitizedZipName = FileUtils.sanitizeFilename(zipAttachment.filename);
+      const zipPath = path.join(tempDir, sanitizedZipName);
       await this.downloadFile(zipAttachment.url, zipPath);
 
       // Extract ZIP

src/indexers/hognoseIndexer.ts

@@ -431,11 +431,8 @@ export class HognoseIndexer {
   ): Promise<Level | null> {
     try {
       // First, stream to a temporary location to calculate hash if skipExisting is enabled
-      const tempPath = path.join(
-        this.outputDir,
-        '.temp',
-        `${Date.now()}_${path.basename(fileName)}`
-      );
+      const sanitizedFileName = FileUtils.sanitizeFilename(path.basename(fileName));
+      const tempPath = path.join(this.outputDir, '.temp', `${Date.now()}_${sanitizedFileName}`);
       await FileUtils.ensureDir(path.dirname(tempPath));
 
       // Stream the .dat file to temp location
@@ -462,7 +459,7 @@ export class HognoseIndexer {
       const levelDir = path.join(this.outputDir, getSourceLevelsDir(MapSource.HOGNOSE), levelId);
       await FileUtils.ensureDir(levelDir);
 
-      const datFileName = path.basename(fileName);
+      const datFileName = sanitizedFileName;
       const localDatPath = path.join(levelDir, datFileName);
 
       // Move from temp to final location