1
0
mirror of https://git.savannah.gnu.org/git/emacs.git synced 2024-12-13 09:32:47 +00:00
emacs/lisp/url/url-queue.el
Peder O. Klingenberg 9822a6a570 Change gnutls-verify-error to be first-match
* doc/misc/url.texi (Customization): Describe the new user
option url-lastloc-privacy-level.

* lisp/net/eww.el (eww-render): Set url-current-lastloc to the
url we are rendering, to get the referer header right on
subsequent requests.

* lisp/url/url-http.el (url-http--get-referer): New function
to determine which referer to send, if any, considering the
users privacy settings and the target url we are visiting.
(url-http-referer): New variable keeping track of the referer
computed by url-http--get-referer
(url-http-create-request): Use url-http-referer instead of the
optional argument to set up the referer header.  Leave
checking of privacy settings to url-http--get-referer.
(url-http): Set up url-http-referer by using
url-http--get-referer.

* lisp/url/url-queue.el (url-queue): New struct member
context-buffer for keeping track of the context a queued job
started from.
(url-queue-retrieve): Store the current buffer in the queue
object.
(url-queue-start-retrieve): Make sure url-retrieve is called
in the context of the original buffer, if available.

* lisp/url/url-util.el (url-domain): New function to determine
the domain of a given URL.

* lisp/url/url-vars.el (url-current-lastloc): New variable to
keep track of the desired "last location" (referer header).
(url-lastloc-privacy-level): New custom setting for more
fine-grained control over how lastloc (referer) is sent to
servers (Bug#27012).
2018-04-13 15:08:18 +02:00

197 lines
6.7 KiB
EmacsLisp

;;; url-queue.el --- Fetching web pages in parallel -*- lexical-binding: t -*-
;; Copyright (C) 2011-2018 Free Software Foundation, Inc.
;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
;; Keywords: comm
;; This file is part of GNU Emacs.
;; GNU Emacs is free software: you can redistribute it and/or modify
;; it under the terms of the GNU General Public License as published by
;; the Free Software Foundation, either version 3 of the License, or
;; (at your option) any later version.
;; GNU Emacs is distributed in the hope that it will be useful,
;; but WITHOUT ANY WARRANTY; without even the implied warranty of
;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
;; GNU General Public License for more details.
;; You should have received a copy of the GNU General Public License
;; along with GNU Emacs. If not, see <https://www.gnu.org/licenses/>.
;;; Commentary:
;; The point of this package is to allow fetching web pages in
;; parallel -- but control the level of parallelism to avoid DoS-ing
;; web servers and Emacs.
;;; Code:
(eval-when-compile (require 'cl-lib))
(require 'browse-url)
(require 'url-parse)
(defcustom url-queue-parallel-processes 6
"The number of concurrent processes."
:version "24.1"
:type 'integer
:group 'url)
(defcustom url-queue-timeout 5
"How long to let a job live once it's started (in seconds)."
:version "24.1"
:type 'integer
:group 'url)
;;; Internal variables.
(defvar url-queue nil)
(defvar url-queue-progress-timer nil)
(cl-defstruct url-queue
url callback cbargs silentp
buffer start-time pre-triggered
inhibit-cookiesp context-buffer)
;;;###autoload
(defun url-queue-retrieve (url callback &optional cbargs silent inhibit-cookies)
"Retrieve URL asynchronously and call CALLBACK with CBARGS when finished.
This is like `url-retrieve' (which see for details of the arguments),
but with limits on the degree of parallelism. The variable
`url-queue-parallel-processes' sets the number of concurrent processes.
The variable `url-queue-timeout' sets a timeout."
(setq url-queue
(append url-queue
(list (make-url-queue :url url
:callback callback
:cbargs cbargs
:silentp silent
:inhibit-cookiesp inhibit-cookies
:context-buffer (current-buffer)))))
(url-queue-setup-runners))
;; To ensure asynch behavior, we start the required number of queue
;; runners from `run-with-idle-timer'. So we're basically going
;; through the queue in two ways: 1) synchronously when a program
;; calls `url-queue-retrieve' (which will then start the required
;; number of queue runners), and 2) at the exit of each job, which
;; will then not start any further threads, but just reuse the
;; previous "slot".
(defun url-queue-setup-runners ()
(let ((running 0)
waiting)
(dolist (entry url-queue)
(cond
((or (url-queue-start-time entry)
(url-queue-pre-triggered entry))
(cl-incf running))
((not waiting)
(setq waiting entry))))
(when (and waiting
(< running url-queue-parallel-processes))
(setf (url-queue-pre-triggered waiting) t)
;; We start fetching from this idle timer...
(run-with-idle-timer 0.01 nil #'url-queue-run-queue)
;; And then we set up a separate timer to ensure progress when a
;; web server is unresponsive.
(unless url-queue-progress-timer
(setq url-queue-progress-timer
(run-with-idle-timer 1 1 #'url-queue-check-progress))))))
(defun url-queue-run-queue ()
(url-queue-prune-old-entries)
(let ((running 0)
waiting)
(dolist (entry url-queue)
(cond
((url-queue-start-time entry)
(cl-incf running))
((not waiting)
(setq waiting entry))))
(when (and waiting
(< running url-queue-parallel-processes))
(setf (url-queue-start-time waiting) (float-time))
(url-queue-start-retrieve waiting))))
(defun url-queue-check-progress ()
(when url-queue-progress-timer
(if url-queue
(url-queue-run-queue)
(cancel-timer url-queue-progress-timer)
(setq url-queue-progress-timer nil))))
(defun url-queue-callback-function (status job)
(setq url-queue (delq job url-queue))
(when (and (eq (car status) :error)
(eq (cadr (cadr status)) 'connection-failed))
;; If we get a connection error, then flush all other jobs from
;; the host from the queue. This particularly makes sense if the
;; error really is a DNS resolver issue, which happens
;; synchronously and totally halts Emacs.
(url-queue-remove-jobs-from-host
(plist-get (nthcdr 3 (cadr status)) :host)))
(url-queue-run-queue)
(apply (url-queue-callback job) (cons status (url-queue-cbargs job))))
(defun url-queue-remove-jobs-from-host (host)
(let ((jobs nil))
(dolist (job url-queue)
(when (equal (url-host (url-generic-parse-url (url-queue-url job)))
host)
(push job jobs)))
(dolist (job jobs)
(url-queue-kill-job job)
(setq url-queue (delq job url-queue)))))
(defun url-queue-start-retrieve (job)
(setf (url-queue-buffer job)
(ignore-errors
(with-current-buffer (if (buffer-live-p (url-queue-context-buffer job))
(url-queue-context-buffer job)
(current-buffer))
(let ((url-request-noninteractive t))
(url-retrieve (url-queue-url job)
#'url-queue-callback-function (list job)
(url-queue-silentp job)
(url-queue-inhibit-cookiesp job)))))))
(defun url-queue-prune-old-entries ()
(let (dead-jobs)
(dolist (job url-queue)
;; Kill jobs that have lasted longer than the timeout.
(when (and (url-queue-start-time job)
(> (- (float-time) (url-queue-start-time job))
url-queue-timeout))
(push job dead-jobs)))
(dolist (job dead-jobs)
(url-queue-kill-job job)
(setq url-queue (delq job url-queue)))))
(defun url-queue-kill-job (job)
(when (bufferp (url-queue-buffer job))
(let (process)
(while (setq process (get-buffer-process (url-queue-buffer job)))
(set-process-sentinel process 'ignore)
(ignore-errors
(delete-process process)))))
;; Call the callback with an error message to ensure that the caller
;; is notified that the job has failed.
(with-current-buffer
(if (and (bufferp (url-queue-buffer job))
(buffer-live-p (url-queue-buffer job)))
;; Use the (partially filled) process buffer if it exists.
(url-queue-buffer job)
;; If not, just create a new buffer, which will probably be
;; killed again by the caller.
(generate-new-buffer " *temp*"))
(apply (url-queue-callback job)
(cons (list :error (list 'error 'url-queue-timeout
"Queue timeout exceeded"))
(url-queue-cbargs job)))))
(provide 'url-queue)
;;; url-queue.el ends here