From 2f02ae096183bc4241ddf12e6ed367a27c0fb550 Mon Sep 17 00:00:00 2001
From: 24c02 <163450896+24c02@users.noreply.github.com>
Date: Fri, 20 Mar 2026 15:45:58 -0400
Subject: [PATCH] charset on content-type if not provided

---
 app/controllers/api/v4/uploads_controller.rb |  1 +
 app/controllers/uploads_controller.rb        |  1 +
 app/models/upload.rb                         | 11 +++++++++++
 3 files changed, 13 insertions(+)

diff --git a/app/controllers/api/v4/uploads_controller.rb b/app/controllers/api/v4/uploads_controller.rb
index e496705..df5bc02 100644
--- a/app/controllers/api/v4/uploads_controller.rb
+++ b/app/controllers/api/v4/uploads_controller.rb
@@ -15,6 +15,7 @@ module API
         end
 
         content_type = Marcel::MimeType.for(file.tempfile, name: file.original_filename) || file.content_type || "application/octet-stream"
+        content_type = Upload.normalize_content_type(content_type)
 
         # Pre-gen upload ID for predictable storage path
         upload_id = SecureRandom.uuid_v7
diff --git a/app/controllers/uploads_controller.rb b/app/controllers/uploads_controller.rb
index c284262..3de3df4 100644
--- a/app/controllers/uploads_controller.rb
+++ b/app/controllers/uploads_controller.rb
@@ -23,6 +23,7 @@ class UploadsController < ApplicationController
     end
 
     content_type = Marcel::MimeType.for(uploaded_file.tempfile, name: uploaded_file.original_filename) || uploaded_file.content_type || "application/octet-stream"
+    content_type = Upload.normalize_content_type(content_type)
 
     # pre-gen upload ID for predictable storage path
     upload_id = SecureRandom.uuid_v7
diff --git a/app/models/upload.rb b/app/models/upload.rb
index 1e3e0fc..1ff4aeb 100644
--- a/app/models/upload.rb
+++ b/app/models/upload.rb
@@ -36,6 +36,16 @@ class Upload < ApplicationRecord
   alias_method :file_size, :byte_size
   alias_method :mime_type, :content_type
 
+  # Ensure text content types include charset=utf-8 so browsers
+  # don't fall back to ISO-8859-1 and mangle unicode
+  def self.normalize_content_type(content_type)
+    if content_type&.start_with?("text/") && !content_type.include?("charset")
+      "#{content_type}; charset=utf-8"
+    else
+      content_type
+    end
+  end
+
   # Provenance enum
   enum :provenance, {
     slack: "slack",
@@ -93,6 +103,7 @@ class Upload < ApplicationRecord
     filename ||= File.basename(URI.parse(url).path)
     body = response.body
     content_type = Marcel::MimeType.for(StringIO.new(body), name: filename) || response.headers["content-type"] || "application/octet-stream"
+    content_type = normalize_content_type(content_type)
 
     # Pre-generate upload ID for predictable storage path
     upload_id = SecureRandom.uuid_v7