;;; ewb.el --- Emacs Web Browser
; M
;E B
; W

;; Copyright (C) 2002  Patware Unc.

;; Author: Patrick Anderson <patware@freeshell.org>
;; Keywords: browser, web, net

;; This file 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 2, or (at your option)
;; any later version.

;; Commentary:
;; type  C-h m  for keystroke help
;; sites that kind of work: emacswiki.org, google.com, slashdot.org
;; sites that don't work: patware.freeshell.org, sf.net

;; latest at  http://patware.freeshell.org/emacs/ewb.el
;; Version: -.2a --- removed referer string.
;; Version: -.2 --- sending referer.  changed version numbering.  ewb-href now accepts <a name=>.  need to move to state-machine parser.
;; Version: -.3 --- prefix arg or file extensions suppress formatting.  write bookmarks to file, customizeable
;; Version: -.4 --- added ewb-revert [(g)], minibuffer state messages, display ewb-current-url above header, mouse-face, help-echo
;; Version: -.5 --- first alpha.

;; Installation:
;; (autoload 'ewb "ewb" "emacs web browser" t) ;put in your .emacs
;; type  M-x eval-buffer RET  then  M-x ewb RET  to try it now
;; M-x customize-variable RET browse-url-browser-function  to make ewb your default browser.

;; goals:
;; handle form data (rfc 2616)
;; asynchronous operation
;;   look at 
;; display bold, italics
;; turn <frame name="main" src="indext.htm"> into a link

(defgroup ewb nil "emacs web browser"
  :group 'hypermedia)

(defcustom ewb-suppressed-extensions "\.el$\\|\.txt$" "documents with these extensions won't be parsed.")

(defcustom ewb-bookmark-file
  "~/notebook/emacs/.ewb.bookmarks"
  "bookmark file"
  :group 'ewb)

(defcustom ewb-history-file
  "~/notebook/emacs/.ewb.history"
  "history file"
  :group 'ewb)

;................................................................................
(defvar ewb-href "<[aA][^>]*\\(?:[hH][rR][eE][fF]\\|[nN][aA][mM][eE]\\)=\"?\\([^\">]+\\)[^>]*>")
(defvar ewb-history nil)
(defvar ewb-current-host nil)
(defvar ewb-current-dir nil)
(defvar ewb-current-url nil)

(defvar ewb-mode-map (make-sparse-keymap))

(define-key ewb-mode-map [(b)] 'ewb-bookmarks)
(defun ewb-bookmarks ()
  (interactive)
  (find-file ewb-bookmark-file))
(define-key ewb-mode-map [(a)] '(lambda () (interactive) (write-region (concat ewb-current-url "\n") nil ewb-bookmark-file t)))

(define-key ewb-mode-map [(o)] 'ewb)

(define-key ewb-mode-map [(g)] 'ewb-revert)
(defun ewb-revert (&optional raw)
  (interactive "P")
  (ewb ewb-current-url raw)
  (setq ewb-history (cdr ewb-history)))

(define-key ewb-mode-map [(9)] 'ewb-next)
(defun ewb-next ()
  "jump to next link"
  (interactive)
  (re-search-forward ewb-href)
  (message (match-string 1)))

(define-key ewb-mode-map [(shift 9)] 'ewb-prev)
(defun ewb-prev ()
  "jump to prev link"
  (interactive)
  (re-search-backward ewb-href)
  (message (match-string 1)))

(define-key ewb-mode-map [(13)] 'ewb-follow)
(defun ewb-follow (&optional raw)
  "follow this link"
  (interactive "P")
  (re-search-backward ewb-href)
  (ewb-parse-url (match-string 1) 1 raw))

(define-key ewb-mode-map [(mouse-2)] 'ewb-follow-mouse)
(defun ewb-follow-mouse (event &optional raw)
  (interactive "e\nP")
  (goto-char (posn-point (event-end event)))
  (ewb-follow raw))

(define-key ewb-mode-map [(s)] 'ewb-search)
(defun ewb-search (str)
  "use google"
  (interactive "sFor: ")
  (ewb (concat "www.google.com/search?q=" (dired-replace-in-string " " "+" str))))

(define-key ewb-mode-map [(q)] 'bury-buffer)
(define-key ewb-mode-map " " 'scroll-up)
(define-key ewb-mode-map [(u)] 'scroll-down)
(define-key ewb-mode-map [(h)] '(lambda () (interactive) (mapcar 'print ewb-history)))
(define-key ewb-mode-map [(U)] 'ewb-back)
(defun ewb-back (&optional raw)
  "go back in history"
  (interactive "P")
  (if ewb-history
	  (progn
		(setq ewb-history (cdr ewb-history))
		(let ((cur (car ewb-history)))
		  (setq ewb-history (cdr ewb-history)) ;remove self, since ewb-get adds
		  (ewb cur raw)))))

(defun ewb-mode ()
  "emacs web browser mode. \\{ewb-mode-map}"
  (interactive)
  (kill-all-local-variables)
  (use-local-map ewb-mode-map)
  (setq mode-name "ewb")
  (setq major-mode 'ewb-mode)
  (add-to-invisibility-spec 'ewb)
;  (make-local-variable 'font-lock-defaults)
;  (setq font-lock-defaults '(ewb-mode-font-lock-keywords t nil nil nil))
  (run-hooks 'ewb-mode-hook))

(defun ewb-filter (proc string)
  "string is typically 1k"
  (with-current-buffer (process-buffer proc)
	  (save-excursion
		(goto-char (point-max))
		(insert string))))

(defun ewb-sentinel (process event)
  (message "formatting...")
  (save-excursion
 	(goto-char (point-min))
	(let ((href-end))
	  (while (setq href-end (re-search-forward ewb-href nil t))
 		(let* ((href-overlay (make-overlay (match-beginning 0) href-end))
			   (url (match-string 1))
			   (link-end (re-search-forward "</[aA]>" nil t))
			   (link-overlay (make-overlay href-end (- link-end 4))))
		  (overlay-put href-overlay 'invisible 'ewb)
		  (overlay-put href-overlay 'intangible 'ewb)
		  (overlay-put link-overlay 'face 'font-lock-keyword-face)
		  (overlay-put link-overlay 'mouse-face 'highlight)
		  (overlay-put link-overlay 'help-echo url))
		(replace-match "" t t)))

;; 	(goto-char (point-min))
;; 	(while (setq end (re-search-forward (concat
;; 					  "\\(<[bB][rR]>\\|<[pP]>\\|<[hH][rR]>\\|</[hH][1-9]>\\)"
;; 					   "\\|\\(&[nN][bB][sS][pP];\\|[ 	]+\\)"
;; 					   "\\|\\(</?[^aA][^>]*>\\|&[a-zA-Z0-9#]+;\\|\r\\)")
;; 					   nil t))
;; 	  (replace-match "\n" t t nil 1)
;; 	  (replace-match " " t t nil 2)
;; 	  (replace-match "" t t nil 3))

 	(goto-char (point-min))
 	(while (setq end (re-search-forward "<[bB][rR]>\\|<[pP]>\\|<[hH][rR]>\\|</[hH][1-9]>" nil t))
 	  (replace-match "\n" t t))

 	(goto-char (point-min))
 	(while (setq end (re-search-forward "&[nN][bB][sS][pP];\\|[ 	]+" nil t))
 	  (replace-match " " t t))

 	(goto-char (point-min))
 	(while (setq end (re-search-forward "</?[^aA][^>]*>\\|&[a-zA-Z0-9#]+;\\|\r" nil t))
 	  (replace-match "" t t))
 	)
  (message "done."))

(defun ewb (url &optional raw)
  "emacs web browser.
\\{ewb-mode-map}"
  (interactive "sURL: \nP")
  (ewb-parse-url url nil raw))

(defun ewb-parse-url (url followed &optional raw)
  "  a 'followed' link has a looser file spec, tighter host spec:
a host must begin with (http:)?//
a link such as \"a.b\" specifies file or directory.

  an 'interactive' url has a tighter 'file' spec, looser host spec:
a host might not begin with (http:)?//
a link such as \"a.b\" specifies a host."
  (if followed
	  (string-match "\\(http:\\)?\\(//[^/\n]+\\)?\\(/\\)?\\(\\(?:[^/\n]+/\\)*\\)\\(.*\\)" url)
;                  1=protocol?,   2=//host?       3=/?,  4=dir/*         5=file?
      (string-match "\\(http://\\)?\\([^/\n]+\\)\\(/\\)?\\(\\(?:[^/\n]+/\\)*\\)\\(.*\\)" url))
;                  1=protocol//?,   2=host        3=/?,  4=dir/*         5=file?
  (let* ((host (match-string 2 url))
		 (dir (match-string 4 url))
		 (file (match-string 5 url)))
	(if host
		(progn
		  (string-match "\\(//\\)?\\(.*\\)" host)
		  (setq host (match-string 2 host)))) ;just match the 2nd chunk
	(ewb-get host dir file raw)))

(defun ewb-get (host dir file &optional raw) ;port
  (let* ((port 80)
		 (buf (get-buffer-create "*Emacs Web Browser*"))
		 (ewb-connection)
		 (referer ewb-current-url))
	(if host (progn (setq ewb-current-host host)
					(setq ewb-current-dir dir))) ;force this (generally to clear)
	(if dir
		(if (not (= 0 (length dir)))
			(setq ewb-current-dir dir)))
	(setq ewb-current-url (concat "http://" ewb-current-host "/" ewb-current-dir file))
	(setq ewb-history (cons ewb-current-url ewb-history))
	(message (concat "downloading  " ewb-current-url "..."))
	(setq ewb-connection (open-network-stream "ewb-connection" buf ewb-current-host port))
	(switch-to-buffer buf)
	(erase-buffer)
	(ewb-mode)
	(insert (concat ewb-current-url "\n"))
	(overlay-put (make-overlay 1 (+ (length ewb-current-url) 1)) 'face 'font-lock-constant-face)
	(set-process-filter ewb-connection 'ewb-filter)
	(if (and (null raw) (not (string-match ewb-suppressed-extensions file)))
		(set-process-sentinel ewb-connection 'ewb-sentinel))
;	(process-send-string ewb-connection (concat "GET /" ewb-current-dir file " HTTP/1.0 " referer "\n\n"))))
	(process-send-string ewb-connection (concat "GET /" ewb-current-dir file " HTTP/1.0\n\n"))))
;	(process-send-string ewb-connection (concat "GET /" ewb-current-dir file " HTTP/1.1\r\n\r\n"))

(provide 'ewb)
;;; ewb.el ends here