`headerFrom` does partial character decoding

[divipp]

When the email header From field has multiple encoded blocks, only the first block is decoded with headerFrom.

For example, given the following header fragment

From: =?utf-8?Q?V=C3=A1ros_=2D_Parkol=C3=A1si_=C3=BCgyint=C3=A9z?=
 =?utf-8?Q?=C3=A9s?= <x@y.z>

headerFrom returns

[Single (Mailbox (Just "V\225ros - Parkol\225si \252gyint\233z =?utf-8?Q?=C3=A9s?=") (AddrSpec "x" (DomainDotAtom ("y" :| ["z"]))))]

instead of

[Single (Mailbox (Just "V\225ros - Parkol\225si \252gyint\233z\233s") (AddrSpec "x" (DomainDotAtom ("y" :| ["z"]))))]

Notes

• headerTo and headerCC seems to have the same bug
• Maybe it is a separate bug that the space at the beginning of the extra lines of a multiline header field is wrongly inserted in the address. (Just "V\225ros - Parkol\225si \252gyint\233z... <-- the space before the three dots is unwanted)

[frasertweedale]

@divipp thank you for your detailed bug report. Would you be able to test the following branch and provide feedback? https://github.com/purebred-mua/purebred-email/tree/fix/73-decode-folded-encoded-word

[divipp]

Hi, Thanks for the fast fix!

The example in the bug report is OK now. The space went away and the second part is decoded too. However the bugfix doesn't work with the following header fragment:

From: "Aegon =?UTF-8?Q?Biztos=C3=ADt=C3=B3?=" <x@y.z>

I think the result should be [Single (Mailbox (Just "Aegon Biztosító") (AddrSpec "x" (DomainDotAtom ("y" :| ["z"]))))] but it is [Single (Mailbox (Just "Aegon =?UTF-8?Q?Biztos=C3=ADt=C3=B3?=") (AddrSpec "x" (DomainDotAtom ("y" :| ["z"]))))] so the decoding is not done after the ASCII Aegon+space characters.

[frasertweedale]

@divipp thanks for the feedback. I believe for:

From: "Aegon =?UTF-8?Q?Biztos=C3=ADt=C3=B3?=" <x@y.z>

is correct. Per https://datatracker.ietf.org/doc/html/rfc2047#section-5:

   + An 'encoded-word' MUST NOT appear within a 'quoted-string'.

Without the quotes, it decodes properly:

λ> parse (addressList defaultCharsets) ("\"Aegon =?UTF-8?Q?Biztos=C3=ADt=C3=B3?=\" <x@y.z>" :: B.ByteString)
Right [Single (Mailbox (Just "Aegon =?UTF-8?Q?Biztos=C3=ADt=C3=B3?=") (AddrSpec "x" (DomainDotAtom ("y" :| ["z"]))))]
λ> parse (addressList defaultCharsets) ("Aegon =?UTF-8?Q?Biztos=C3=ADt=C3=B3?= <x@y.z>" :: B.ByteString)
Right [Single (Mailbox (Just "Aegon Biztos\237t\243") (AddrSpec "x" (DomainDotAtom ("y" :| ["z"]))))]

I will implement a test for the feature and include the fix in the next release.

[divipp]

OK, thanks! Maybe I'll use a lenient headerFrom in my application, because in Hungary there are lots encoded words in quoted addresses. :(

[frasertweedale]

@divipp That's unfortunate. Consider reporting as bug(s) against whatever program(s) are producing those messages.

[frasertweedale]

Fix released in v0.5.1: https://hackage.haskell.org/package/purebred-email-0.5.1