GNU bug report logs - #8988
24.0.50; [PATCH] pop3-open-server fails to parse response to CAPA

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 8988 in the body.
You can then email your comments to 8988 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Wolfgang Jenkner <wjenkner <at> inode.at>
To: bug-gnu-emacs <at> gnu.org
Cc: larsi <at> gnus.org
Subject: 24.0.50; [PATCH] pop3-open-server fails to parse response to CAPA
Date: Sun, 03 Jul 2011 04:33:11 +0200

When I evaluate

(progn
  (load "pop3.el")
  (trace-function-background 'open-protocol-stream)
  (let ((pop3-stream-type 'starttls))
    (ignore-errors
      (pop3-open-server "pop.gmx.com" 110))))

buffer *trace-output* contains

======================================================================
1 -> open-protocol-stream: name="POP" buffer=#<buffer  trace of POP session to pop.gmx.com> host="pop.gmx.com" service=110 parameters=(:type starttls :capability-command "CAPA
" :end-of-command "^\\.?
\\|^\\(-ERR\\|+OK \\).*
" :success "^\\+OK.*
" :return-list t :starttls-function (lambda (capabilities) (and (string-match "\\bSTLS\\b" capabilities) "STLS
")))
1 <- open-protocol-stream: (nil :greeting "+OK POP server ready H migmx003
" :capabilities "+OK Capability list follows
" :type plain :error "Server does not support TLS")

However, in the shell

$ telnet pop.gmx.com 110
Trying 212.227.17.187...
Connected to pop.gmx.com.
Escape character is '^]'.
+OK POP server ready H migmx007
CAPA
+OK Capability list follows
TOP
USER
UIDL
STLS
SASL PLAIN
IMPLEMENTATION trinity
.
QUIT
+OK POP server signing off
Connection closed by foreign host.
$ 

So it seems that "+OK Capability list follows" arrived in a separate
chunk and was matched by the second alternative of the regexp specified
for :end-of-command in pop3-open-server before the actual capability
list followed and the regexp's first alternative had a chance to match.
(I haven't tested this theory, though, since it seems valid as
a theoretical possibility anyway :-)

I include ChangeLog entries and patches for lisp/gnus/pop3.el and
lisp/net/network-stream.el (I could also send a bzr bundle, but given
that most of lisp/gnus is merged from the main Gnus repository this
doesn't seem too practical).

Wolfgang

2011-07-03  Wolfgang Jenkner  <wjenkner <at> inode.at>

	* net/network-stream.el (open-network-stream)
	(network-stream-open-starttls, network-stream-open-tls)
	(network-stream-open-shell): New parameter :end-of-capability,
	like :end-of-command but matches the end of the response to
	a :capability-command query.

2011-07-03  Wolfgang Jenkner  <wjenkner <at> inode.at>

	* pop3.el (pop3-open-server): Use :end-of-capability.

=== modified file 'lisp/net/network-stream.el'
--- lisp/net/network-stream.el	2011-06-27 00:11:22 +0000
+++ lisp/net/network-stream.el	2011-07-02 23:38:45 +0000
@@ -98,6 +98,10 @@
 
 :end-of-command specifies a regexp matching the end of a command.
 
+:end-of-capability specifies a regexp matching the end of the
+  response to the command specified for :capability-command.
+  It defaults to the regexp specified for :end-of-command.
+
 :success specifies a regexp matching a message indicating a
   successful STARTTLS negotiation.  For instance, the default
   should be \"^3\" for an NNTP connection.
@@ -129,6 +133,10 @@
   asynchronously, if possible."
   (unless (featurep 'make-network-process)
     (error "Emacs was compiled without networking support"))
+  ;; (unless (plist-get parameters :end-of-capability)
+  ;;     (setq parameters (plist-put parameters
+  ;; 				  :end-of-capability
+  ;; 				  (plist-get parameters :end-of-command))))
   (let ((type (plist-get parameters :type))
 	(return-list (plist-get parameters :return-list)))
     (if (and (not return-list)
@@ -203,11 +211,12 @@
 	 (success-string     (plist-get parameters :success))
 	 (capability-command (plist-get parameters :capability-command))
 	 (eoc                (plist-get parameters :end-of-command))
+	 (eo-capa            (or (plist-get parameters :end-of-capability) eoc))
 	 ;; Return (STREAM GREETING CAPABILITIES RESULTING-TYPE)
 	 (stream (make-network-process :name name :buffer buffer
 				       :host host :service service))
 	 (greeting (network-stream-get-response stream start eoc))
-	 (capabilities (network-stream-command stream capability-command eoc))
+	 (capabilities (network-stream-command stream capability-command eo-capa))
 	 (resulting-type 'plain)
 	 (builtin-starttls (and (fboundp 'gnutls-available-p)
 				(gnutls-available-p)))
@@ -250,7 +259,7 @@
 	;; Requery capabilities for protocols that require it; i.e.,
 	;; EHLO for SMTP.
 	(when (plist-get parameters :always-query-capabilities)
-	  (network-stream-command stream capability-command eoc)))
+	  (network-stream-command stream capability-command eo-capa)))
       (when (string-match success-string
 			  (network-stream-command stream starttls-command eoc))
 	;; The server said it was OK to begin STARTTLS negotiations.
@@ -271,7 +280,7 @@
 	    (network-stream-get-response stream start eoc)))
 	;; Re-get the capabilities, which may have now changed.
 	(setq capabilities
-	      (network-stream-command stream capability-command eoc))))
+	      (network-stream-command stream capability-command eo-capa))))
 
     ;; If TLS is mandatory, close the connection if it's unencrypted.
     (when (and (or require-tls
@@ -320,7 +329,8 @@
 			 'open-gnutls-stream
 		       'open-tls-stream)
 		     name buffer host service))
-	   (eoc (plist-get parameters :end-of-command)))
+	   (eoc (plist-get parameters :end-of-command))
+	   (eo-capa (or (plist-get parameters :end-of-capability) eoc)))
       (if (null stream)
 	  (list nil nil nil 'plain)
 	;; If we're using tls.el, we have to delete the output from
@@ -335,13 +345,14 @@
 	(let* ((capability-command (plist-get parameters :capability-command)))
 	  (list stream
 		(network-stream-get-response stream start eoc)
-		(network-stream-command stream capability-command eoc)
+		(network-stream-command stream capability-command eo-capa)
 		'tls))))))
 
 (defun network-stream-open-shell (name buffer host service parameters)
   (require 'format-spec)
   (let* ((capability-command (plist-get parameters :capability-command))
 	 (eoc 		     (plist-get parameters :end-of-command))
+	 (eo-capa 	     (or (plist-get parameters :end-of-capability) eoc))
 	 (start (with-current-buffer buffer (point)))
 	 (stream (let ((process-connection-type nil))
 		   (start-process name buffer shell-file-name
@@ -353,7 +364,7 @@
 				    ?p service))))))
     (list stream
 	  (network-stream-get-response stream start eoc)
-	  (network-stream-command stream capability-command eoc)
+	  (network-stream-command stream capability-command eo-capa)
 	  'plain)))
 
 (provide 'network-stream)

=== modified file 'lisp/gnus/pop3.el'
--- lisp/gnus/pop3.el	2011-05-30 22:11:52 +0000
+++ lisp/gnus/pop3.el	2011-07-02 23:38:45 +0000
@@ -306,7 +306,12 @@
 		    (t
 		     (or pop3-stream-type 'network)))
 	     :capability-command "CAPA\r\n"
-	     :end-of-command "^\\.\r?\n\\|^\\(-ERR\\|+OK \\).*\n"
+	     :end-of-command "^\\(-ERR\\|+OK \\).*\n"
+	     ;; As it happens, the regexp above also matches the first
+	     ;; line of the response to CAPA, which might arrive in
+	     ;; a separate chunk, with the actual capabilities
+	     ;; following later.
+	     :end-of-capability "^\\.\r?\n"
 	     :success "^\\+OK.*\n"
 	     :return-list t
 	     :starttls-function

Message #8 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Lars Magne Ingebrigtsen <larsi <at> gnus.org>
To: bug-gnu-emacs <at> gnu.org
Subject: Re: 24.0.50; [PATCH] pop3-open-server fails to parse response to CAPA
Date: Sun, 03 Jul 2011 15:51:55 +0200

Wolfgang Jenkner <wjenkner <at> inode.at> writes:

> So it seems that "+OK Capability list follows" arrived in a separate
> chunk and was matched by the second alternative of the regexp specified
> for :end-of-command in pop3-open-server before the actual capability
> list followed and the regexp's first alternative had a chance to match.
> (I haven't tested this theory, though, since it seems valid as
> a theoretical possibility anyway :-)

I think you're totally right.

I've applied parts of your patch to Emacs 24/No Gnus, and redid other
parts slight.

Thanks for the bug fix.

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.