Re: [Erc-discuss] [PATCH] erc-dcc: allow SEND commands containing quoted filenames with spaces in them

[Michael Olson]

Do you also need to unquote backslashes before doing anything with the
filename?  E.g. // => /, /" => "

On Sat, Oct 29, 2011 at 10:31 PM, Mike Kazantsev <address@hidden> wrote:
>
> Made a lot of mistakes here, apologies.
> Guess my attention-span is even worse in the morning.
>
> Few comments and corrected PATCH follows.
>
>
> On Sun, 30 Oct 2011 10:50:20 +0600
> Mike Kazantsev <address@hidden> wrote:
>
>> On Sat, 29 Oct 2011 12:08:01 -0700
>> Michael Olson <address@hidden> wrote:
>>
>> > Be careful with that regexp.  It's not going to be able to handle
>> > escaped quotes (with backslash, presumably).  I'd rather have it be
>> > flawless before checking in, unless it's IRC convention to not parse
>> > that kind of thing.  Also, <<.*+?>> is not very meaningful -- perhaps
>> > you meant <<.*?>> ?
>> >
>>
>> Indeed I didn't think about escaped quotes inside and .*+? is a
>> mistake, but second bug seem to cancel the first on out - since regexp
>> there is greedy, all the quotes inside should be matched:
>>
>> (string-match
>>  "^DCC SEND \\(\\([^ ]+\\)\\|\"\\(.+?\\)\"\\) \\([0-9]+\\) \\([0-9]+\\) 
>> *\\([0-9]*\\)"
>>  "DCC SEND \"some \\\"stuff with quotes\" 12345 123 123")
>>
>> Will produce following groups:
>>
>>  "some \\\"stuff with quotes" "12345" "123" "123"
>>
>
> Bad example with non-greedy +, it works only because later fields
> match, correct one should be:
>
> (let ((query "DCC SEND \"some \\\" 1 1 1 stuff with quotes\" 12345 123 123"))
>  (string-match
>    "^DCC SEND \\(\\([^ ]+\\)\\|\"\\(.*\\)\"\\) \\([0-9]+\\) \\([0-9]+\\) 
> *\\([0-9]*\\)"
>    query)
>  (map 'list (lambda (n) (match-string n query)) '(3 4 5 6)))
>
> but I don't think it's really relevant.
>
>
>> Which should be ok, unless quotes in the later fields should be
>> supported.
>
>> I've never seen these, but it's not hard to make a cleaner version,
>> which follows.
>>
>
> And I've managed to omit corrections to the group numbers there, which
> is fixed in the version, attached here.
>
>
>> While testing it, I've stumbled upon another quirk which ERC (and this
>> regexp in particular) doesn't seem to support, which results in
>> "private" messages like this:
>>
>> [10:20:12]-seekbot- DCC Send SearchBot_results_for_Rule 34.txt.zip 
>> (XX.XX.XX.XX)
>>
>> Problem here is "Send" instead of "SEND", used by the IRC bot.
>> No specs I've been able to find mandate such usage, yet apparently it's
>> supported by xChat, at least.
>> It's trivial to make ERC support this one case (I've done it using
>> "(SEND|Send)" and incrementing the group numbers), and I can post it as
>> a separate patch, do you think it should be supported?
>>
>
> Wrong, problem was the same group numbers which I've messed up, as
> indicated above and actually fixed after needlessly modifying the
> regexp.
>
>
>
> From 970552d3e65b4d85d6602014c1d73d74693c7ed3 Mon Sep 17 00:00:00 2001
> From: Mike Kazantsev <address@hidden>
> Date: Sat, 29 Oct 2011 11:36:49 +0600
> Subject: [PATCH] erc-dcc: allow SEND commands containing quoted filenames
>  with spaces in them
>
> * erc-dcc.el (erc-dcc-ctcp-query-send-regexp): Updated regexp to match
>    quoted filenames with spaces inside.
>  (erc-dcc-handle-ctcp-send): Updated regexp match group numbers.
> ---
>  erc-dcc.el |   16 +++++++++++-----
>  1 files changed, 11 insertions(+), 5 deletions(-)
>
> diff --git a/erc-dcc.el b/erc-dcc.el
> index 9e53edc..acbdbdf 100644
> --- a/erc-dcc.el
> +++ b/erc-dcc.el
> @@ -646,7 +646,12 @@ that subcommand."
>        ?q query ?n nick ?u login ?h host))))
>
>  (defconst erc-dcc-ctcp-query-send-regexp
> -  "^DCC SEND \\([^ ]+\\) \\([0-9]+\\) \\([0-9]+\\) *\\([0-9]*\\)")
> +  (concat "^DCC SEND \\("
> +    ;; Following part matches either filename without spaces
> +    ;; or filename enclosed in double quotes with any number
> +    ;; of escaped double quotes inside.
> +    "\\([^ ]+\\)\\|\"\\(\\(.*?\\(\\\\\"\\)?\\)+?\\)\""
> +    "\\) \\([0-9]+\\) \\([0-9]+\\) *\\([0-9]*\\)"))
>
>  (defun erc-dcc-handle-ctcp-send (proc query nick login host to)
>   "This is called if a CTCP DCC SEND subcommand is sent to the client.
> @@ -661,10 +666,11 @@ It extracts the information about the dcc request and 
> adds it to
>        'dcc-request-bogus
>        ?r "SEND" ?n nick ?u login ?h host))
>      ((string-match erc-dcc-ctcp-query-send-regexp query)
> -      (let ((filename (match-string 1 query))
> -            (ip       (erc-decimal-to-ip (match-string 2 query)))
> -            (port     (match-string 3 query))
> -            (size     (match-string 4 query)))
> +      (let ((filename
> +              (or (match-string 2 query) (match-string 3 query)))
> +            (ip       (erc-decimal-to-ip (match-string 6 query)))
> +            (port     (match-string 7 query))
> +            (size     (match-string 8 query)))
>         ;; FIXME: a warning really should also be sent
>         ;; if the ip address != the host the dcc sender is on.
>         (erc-display-message
> --
> 1.7.7
>
>
>
> --
> Mike Kazantsev // fraggod.net
>
> _______________________________________________
> Erc-discuss mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/erc-discuss
>
>



-- 
Michael Olson  |  http://mwolson.org/