unsplashunsplash

Unicode displaying incorrectly

#39 · October 3, 2016 · Estimated 3 minute read

Problem

We made content changes on an MVC view in a .NET application. The Unicode characters were displaying fine until the change was made which resulted in gibberish on the screen where the Unicode characters used to be.

Investigation

  1. Using Git we looked at the diff to see if we could find anything strange that was introduced to break the display. We didn't see anything out of the ordinary.

  2. We confirm the <meta charset="utf-8" /> tag was present in the shared layout and that the content-type of the response was text/html; charset=utf-8.

  3. We replaced the Unicode character with its HTML entity and it worked. This is more explicit and doesn't rely on any behind-the-scenes black magic but unsustainable with the way we receive the content unless it was automated.

  4. We took to Google and started searching. We came across a StackOverflow question where there was mention of the BOM and something struck me.

Cause of failure

Hunch

We experimented with a Visual Studio add-on called Strip'em to automatically save line-endings in LF as Visual Studio doesn't have a global setting for this.

In retrospect this wasn't a great idea but we were hoping to avoid seeing:

  • the wall of pink in our Git diffs and be able to diff without having to ignore white spaces.

  • nasty dialogs every time we open a file that doesn't have consistent line-endings especially when working with files that were created on other operating systems.

Tip: In Visual Studio 2015 you can change the encoding and line ending for an individual file at File > Advanced Save Options.

This add-on had an unexpected "feature" that was not advertised in the dialog. When it removed the line-endings, the BOM at the start of the file was also removed.

There's two problems with it [Strip-em], firstly it kills the utf-8 magic bytes which windows likes, and also causes a change after file save so VS asks to reload changes, I know that the latter can be avoided but sometimes you don't want to reload changes automatically. ~ Brett Ryan commented on StackOverflow

Experiment

As we couldn't find any reason for the Unicode characters to be displaying differently now, we believed that it must be the add-on. We needed to test it.

  • We disabled the line-ending conversion in the add-on but had to add the BOM back.

  • We opened the file in Notepad++ (before we knew we could save it in Visual Studio) Encoding > Encode in UTF-8-BOM, saved the file, viewed it and it displayed correctly.

  • We changed the encoding back to UTF-8 without the BOM and saw that it broke.

  • This confirmed that the BOM character allowed the hardcoded Unicode characters to display correctly on Windows but we had to prove that the add-on was the culprit. (We hadn't seen the StackOverflow comment at that time)

  • We opened the BOM file in Visual Studio and saved it. It worked.

You can also see the BOM in a Hex editor.

Solution

Saving the UTF-8 file with the BOM signature on Windows solves the problem.

Microsoft compilers and interpreters, and many pieces of software on Microsoft Windows such as Notepad treat the BOM as a required magic number rather than use heuristics. These tools add a BOM when saving text as UTF-8, and cannot interpret UTF-8 unless the BOM is present, or the file contains only ASCII bytes. ~ Wikipedia

Note: As I was so curious about the byte order mark, I investigated and documented my findings.


References