Commit 428bbfaa authored by Chris Lamb's avatar Chris Lamb 👀
Browse files

Fix a file(1)-related regression where .changes files that contained non-ASCII... 

Fix a file(1)-related regression where .changes files that contained non-ASCII text were not identified as such, therefore resulting in seemingly arbitrary packages on (for example) tests.reproducible-builds.org not actually comparing the files themselves. The non-ASCII parts could have been in the Maintainer or in the upload changelog. (Closes: #291)
parent b0fb9c56

diffoscope/comparators/debian.py

+7 −1
Original line number Diff line number Diff line
@@ -237,7 +237,13 @@ class DotChangesFile(DebControlFile): 


    # .changes files can be identified "data" if they contain non-printable

    # characters (Re: reproducible-builds/diffoscope#286)

    FILE_TYPE_RE = re.compile(r"^(ASCII text|UTF-8 Unicode text|data)")

    #

    # In addition, different versions of file(1) can emit "UTF-8 Unicode text"

    # or "UTF-8 Unicode text" for .changes files that use non-ASCII characters,

    # so we have to look for both. (Re: reproducible-builds/diffoscope#291)

    FILE_TYPE_RE = re.compile(

        r"^(ASCII text|UTF-8 Unicode text|data|Unicode text, UTF-8 text)"

    )



    @classmethod

    def recognizes(cls, file):