Metadata not recognized when file starts with BOM

[DanielSundberg]

When a content file starts with a UTF8 BOM I don't get any metadata which can result in hard-to-find bugs. We're using Wintersmith to generate a docs site and we allow multiple users to add content. It is very easy to unintentionally save a file with wrong BOM configuration.

To reproduce:

Create a Wintersmith site:

C:\Code\> wintersmith new .\test-site
initializing new wintersmith site in C:\Code\test-site using template blog
C:\Code\test-site
+-- [email protected]
+-- [email protected]
`-- [email protected]
done!
preinstall:. -> lifecycle / |#########################################################################--------|

Run wintersmith preview and navigate to the index page and check that the readme article is present:

Next open contents\articles\hello-world\index.md using an editor that can add a BOM char at the beginning of the file. In Notepad++ select Encoding/Convert to UTF-8 and save the file.

Refresh the preview and notice that the README link is missing:

The root cause is that the metadata section is not detected properly since you're reading the first 3 chars of the content looking for "---" (markdown.coffee:123) which in this case is the BOM char.

The naive approach would be to use something like stripBOM() in https://github.com/jonschlinkert/fs-utils, but I don't know if that's enough for all cases?

[jnordberg]

I don't know either, didn't know BOM was a thing even :) We could modify the regex to allow it, maybe \s matches those chars by default.