[PATCH] Add parser for Netscape/Mozilla cookie file format

[PATCH] Add parser for Netscape/Mozilla cookie file format

[Aurelien Aptel]

* lisp/url/url-cookie.el (url-cookie-parse-file-netscape): New function.
---
 lisp/url/url-cookie.el | 49 +++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)

diff --git a/lisp/url/url-cookie.el b/lisp/url/url-cookie.el
index 453d4fe5b6..b0dc40475b 100644
--- a/lisp/url/url-cookie.el
+++ b/lisp/url/url-cookie.el
@@ -74,6 +74,55 @@ url-cookie-parse-file
   ;; It's completely normal for the cookies file not to exist yet.
   (load (or fname url-cookie-file) t t))
 
+(defun url-cookie-parse-file-netscape (filename &optional long-session)
+  "Load cookies from FILENAME in Netscape/Mozilla format.
+When LONG-SESSION is non-nil, session cookies (expiring at t=0
+i.e. 1970-1-1) are loaded as expiring one year from now instead."
+  (interactive "fLoad Netscape/Mozilla cookie file: ")
+  (let ((n 0))
+    (with-temp-buffer
+      (insert-file-contents-literally filename)
+      (goto-char (point-min))
+      (when (not (looking-at-p "# Netscape HTTP Cookie File\n"))
+	(error (format "File %s doesn't look like a netscape cookie file" filename)))
+      (while (not (eobp))
+	(when (not (looking-at-p (rx bol (* space) "#")))
+	  (let* ((line (buffer-substring (point) (save-excursion (end-of-line) (point))))
+		 (fields (split-string line "\t")))
+	    (cond
+	     ;;((>= 1 (length line) 0)
+	     ;; (message "skipping empty line"))
+	     ((= (length fields) 7)
+	      (let ((dom (nth 0 fields))
+		    (match (nth 1 fields))
+		    (path (nth 2 fields))
+		    (secure (string= (nth 3 fields) "TRUE"))
+		    ;; session cookies (expire time = 0) are supposed
+		    ;; to be removed when the browser is closed, but
+		    ;; the main point of loading external cookie is to
+		    ;; reuse a browser session, so to prevent the
+		    ;; cookie from being detected as expired straight
+		    ;; away, make it expire a year from now
+		    (expires (format-time-string
+			      "%d %b %Y %T [GMT]"
+			      (seconds-to-time
+			       (let ((s (string-to-number (nth 4 fields))))
+				 (if (and (= s 0) long-session)
+				     (seconds-to-time (+ (* 365 24 60 60) (float-time)))
+				   s)))))
+		    (key (nth 5 fields))
+		    (val (nth 6 fields)))
+		(incf n)
+		;;(message "adding <%s>=<%s> exp=<%s> dom=<%s> path=<%s> sec=%S" key val expires dom path secure)
+		(url-cookie-store key val expires dom path secure)
+		))
+	     (t
+	      (message "ignoring malformed cookie line <%s>" line)))))
+	(forward-line))
+      (when (< 0 n)
+	(setq url-cookies-changed-since-last-save t))
+      (message "added %d cookies from file %s" n filename))))
+
 (defun url-cookie-clean-up (&optional secure)
   (let ((var (if secure 'url-cookie-secure-storage 'url-cookie-storage))
 	new new-cookies)
-- 
2.12.3

Re: [PATCH] Add parser for Netscape/Mozilla cookie file format

[Aurélien Aptel]

will commit this tomorrow if no one sees anything wrong with it.

Re: [PATCH] Add parser for Netscape/Mozilla cookie file format

[Eli Zaretskii]

> From: Aurélien Aptel <aurelien.aptel+emacs@gmail.com>
> Date: Mon, 9 Oct 2017 10:50:47 +0200
> Cc: Emacs development discussions <emacs-devel@gnu.org>
> 
> will commit this tomorrow if no one sees anything wrong with it.

Is this format still used nowadays?  E.g., on my system the cookies
are stored in an SQLite db, not a text file.

Re: [PATCH] Add parser for Netscape/Mozilla cookie file format

[Aurélien Aptel]

On Mon, Oct 9, 2017 at 10:53 AM, Eli Zaretskii <eliz@gnu.org> wrote:
> Is this format still used nowadays?  E.g., on my system the cookies
> are stored in an SQLite db, not a text file.

It's the closest thing we have to a standard I think. It's accepted by
wget, curl, lynx, w3m, ... In fact it's the only format they know how
to parse & generate IIUC.
Also implemented by various programming language libs, including python [1].

1: https://docs.python.org/3.3/library/http.cookiejar.html?highlight=cookie#file-cookie-jar-classes

Re: [PATCH] Add parser for Netscape/Mozilla cookie file format

[Aurélien Aptel]

I've found a little shell script that opens the firefox sqlite cookie
db and dumps them in netscape format (using the sqlite cli program).
Perhaps I could add another function doing something similar if the
sqlite executable is available on the system Emacs is running.

https://gist.githubusercontent.com/hackerb9/d382e09683a52dcac492ebcdaf1b79af/raw/fced8cae8fd1f9573646336dcc8093a10e373e03/extract_cookies.sh

Re: [PATCH] Add parser for Netscape/Mozilla cookie file format

[Eli Zaretskii]

> From: Aurélien Aptel <aurelien.aptel+emacs@gmail.com>
> Date: Mon, 9 Oct 2017 11:12:53 +0200
> Cc: Aurelien Aptel <aaptel@suse.com>, Emacs development discussions <emacs-devel@gnu.org>
> 
> On Mon, Oct 9, 2017 at 10:53 AM, Eli Zaretskii <eliz@gnu.org> wrote:
> > Is this format still used nowadays?  E.g., on my system the cookies
> > are stored in an SQLite db, not a text file.
> 
> It's the closest thing we have to a standard I think. It's accepted by
> wget, curl, lynx, w3m, ... In fact it's the only format they know how
> to parse & generate IIUC.
> Also implemented by various programming language libs, including python [1].
> 
> 1: https://docs.python.org/3.3/library/http.cookiejar.html?highlight=cookie#file-cookie-jar-classes

Thanks.  If this is useful, I have no objections, but please push to
the master branch.