From 92bb3527dea5af48dee90df5b10fc5372195d4cc Mon Sep 17 00:00:00 2001
From: advplyr <advplyr@protonmail.com>
Date: Wed, 19 Mar 2025 17:39:23 -0500
Subject: [PATCH] Add logs when sanitizing filename and update podcast episode
 download to set targetFilename on init #4121

---
 server/managers/PodcastManager.js        |  2 +-
 server/objects/PodcastEpisodeDownload.js | 26 +++++++++++++++++++-----
 server/utils/fileUtils.js                | 10 +++++++++
 3 files changed, 32 insertions(+), 6 deletions(-)

diff --git a/server/managers/PodcastManager.js b/server/managers/PodcastManager.js
index 5ed0fb89..086238d6 100644
--- a/server/managers/PodcastManager.js
+++ b/server/managers/PodcastManager.js
@@ -108,7 +108,7 @@ class PodcastManager {
     //  e.g. "/tagesschau 20 Uhr.mp3" becomes "/tagesschau 20 Uhr (ep_asdfasdf).mp3"
     //  this handles podcasts where every title is the same (ref https://github.com/advplyr/audiobookshelf/issues/1802)
     if (await fs.pathExists(this.currentDownload.targetPath)) {
-      this.currentDownload.appendRandomId = true
+      this.currentDownload.setAppendRandomId(true)
     }
 
     // Ignores all added files to this dir
diff --git a/server/objects/PodcastEpisodeDownload.js b/server/objects/PodcastEpisodeDownload.js
index 7ff89395..3c1d82ac 100644
--- a/server/objects/PodcastEpisodeDownload.js
+++ b/server/objects/PodcastEpisodeDownload.js
@@ -20,6 +20,8 @@ class PodcastEpisodeDownload {
 
     this.appendRandomId = false
 
+    this.targetFilename = null
+
     this.startedAt = null
     this.createdAt = null
     this.finishedAt = null
@@ -74,11 +76,6 @@ class PodcastEpisodeDownload {
   get episodeTitle() {
     return this.rssPodcastEpisode.title
   }
-  get targetFilename() {
-    const appendage = this.appendRandomId ? ` (${this.id})` : ''
-    const filename = `${this.rssPodcastEpisode.title}${appendage}.${this.fileExtension}`
-    return sanitizeFilename(filename)
-  }
   get targetPath() {
     return filePathToPOSIX(Path.join(this.libraryItem.path, this.targetFilename))
   }
@@ -93,6 +90,23 @@ class PodcastEpisodeDownload {
     return new Date(this.rssPodcastEpisode.publishedAt).getFullYear()
   }
 
+  /**
+   * @param {string} title
+   */
+  getSanitizedFilename(title) {
+    const appendage = this.appendRandomId ? ` (${this.id})` : ''
+    const filename = `${title.trim()}${appendage}.${this.fileExtension}`
+    return sanitizeFilename(filename)
+  }
+
+  /**
+   * @param {boolean} appendRandomId
+   */
+  setAppendRandomId(appendRandomId) {
+    this.appendRandomId = appendRandomId
+    this.targetFilename = this.getSanitizedFilename(this.rssPodcastEpisode.title || '')
+  }
+
   /**
    *
    * @param {import('../utils/podcastUtils').RssPodcastEpisode} rssPodcastEpisode - from rss feed
@@ -112,6 +126,8 @@ class PodcastEpisodeDownload {
       this.url = encodeURI(url)
     }
 
+    this.targetFilename = this.getSanitizedFilename(this.rssPodcastEpisode.title || '')
+
     this.libraryItem = libraryItem
     this.isAutoDownload = isAutoDownload
     this.createdAt = Date.now()
diff --git a/server/utils/fileUtils.js b/server/utils/fileUtils.js
index 4b6915b7..2c935557 100644
--- a/server/utils/fileUtils.js
+++ b/server/utils/fileUtils.js
@@ -362,6 +362,9 @@ module.exports.sanitizeFilename = (filename, colonReplacement = ' - ') => {
     return false
   }
 
+  // Normalize the string first to ensure consistent byte calculations
+  filename = filename.normalize('NFC')
+
   // Most file systems use number of bytes for max filename
   //   to support most filesystems we will use max of 255 bytes in utf-16
   //   Ref: https://doc.owncloud.com/server/next/admin_manual/troubleshooting/path_filename_length.html
@@ -390,8 +393,11 @@ module.exports.sanitizeFilename = (filename, colonReplacement = ' - ') => {
   const ext = Path.extname(sanitized) // separate out file extension
   const basename = Path.basename(sanitized, ext)
   const extByteLength = Buffer.byteLength(ext, 'utf16le')
+
   const basenameByteLength = Buffer.byteLength(basename, 'utf16le')
   if (basenameByteLength + extByteLength > MAX_FILENAME_BYTES) {
+    Logger.debug(`[fileUtils] Filename "${filename}" is too long (${basenameByteLength + extByteLength} bytes), trimming basename to ${MAX_FILENAME_BYTES - extByteLength} bytes.`)
+
     const MaxBytesForBasename = MAX_FILENAME_BYTES - extByteLength
     let totalBytes = 0
     let trimmedBasename = ''
@@ -407,6 +413,10 @@ module.exports.sanitizeFilename = (filename, colonReplacement = ' - ') => {
     sanitized = trimmedBasename + ext
   }
 
+  if (filename !== sanitized) {
+    Logger.debug(`[fileUtils] Sanitized filename "${filename}" to "${sanitized}" (${Buffer.byteLength(sanitized, 'utf16le')} bytes)`)
+  }
+
   return sanitized
 }