Problems in parsing message's "From" header

Problems in parsing message's "From" header

[Teemu Likonen]

There's a small bug somewhere in parsing and handling messages' "From"
header. Here's some background information first. Function
mail-header-parse-address works like this:

    (mail-header-parse-address "Teemu Likonen <tlikonen@iki.fi>")
    => ("tlikonen@iki.fi" . "Teemu Likonen")

But if there are some non-latin letters it fails to parse the name part,
for example:

    (mail-header-parse-address "Γιώργος Σεφέρης <address@invalid>")
    => ("address@invalid")

If I put double quotes ("...") around the name it is parsed correctly:

    (mail-header-parse-address "\"Γιώργος Σεφέρης\" <address@invalid>")
    => ("address@invalid" . "Γιώργος Σεφέρης")

Maybe this is a feature but there is a little problem anyway: I use my
own function to insert citation line when replying to a message with
Gnus. The function is defined through variable
message-citation-line-function. To get person's name and address I have
this in my custom citation function:

    (mail-header-parse-address 
     (mail-header-from message-reply-headers))

Vector message-reply-headers has message's "From" field but person's
name is not enclosed in double quotes. Hence mail-header-parse-address
fails to parse the name.

So, is this bug or "non-ideal behaviour" (a) in the function
mail-header-parse-address (b) in the code which produces the vector
message-reply-headers or (c) in the code which puts the "From" field in
the buffer in the first place? About the item "c" I should add that in
Gnus's Article buffer there are no double quotes around person's name.

Thanks. I'm using up-to-date Emacs CVS/Git version:

    GNU Emacs 23.1.50.3 (i686-pc-linux-gnu, GTK+ Version 2.12.12) of
    2009-06-22 on mithlond

Re: Problems in parsing message's "From" header

[Teemu Likonen]

On 2009-06-22 10:56 (+0300), Teemu Likonen wrote:

> But if there are some non-latin letters it fails to parse the name
> part, for example:
>
>     (mail-header-parse-address "Γιώργος Σεφέρης <address@invalid>")
>     => ("address@invalid")

OK, I found another function, mail-extract-address-components, which can
parse the above address string:

    (mail-extract-address-components "Γιώργος Σεφέρης <address@invalid>")
    => ("Γιώργος Σεφέρης" "address@invalid")

So, my problem actually went away. It's probably time to say the usual
"sorry for the noise", from my part at least. :-)

The problem in mail-header-parse-address is probably valid anyway:

>     (mail-header-parse-address 
>      (mail-header-from message-reply-headers))
>
> Vector message-reply-headers has message's "From" field but person's
> name is not enclosed in double quotes. Hence mail-header-parse-address
> fails to parse the name.
>
> So, is this bug or "non-ideal behaviour" (a) in the function
> mail-header-parse-address (b) in the code which produces the vector
> message-reply-headers or (c) in the code which puts the "From" field
> in the buffer in the first place? About the item "c" I should add that
> in Gnus's Article buffer there are no double quotes around person's
> name.
>
> Thanks. I'm using up-to-date Emacs CVS/Git version:
>
>     GNU Emacs 23.1.50.3 (i686-pc-linux-gnu, GTK+ Version 2.12.12) of
>     2009-06-22 on mithlond