sacha chua living an awesome life

Inspired by the Emacs Starter Kit and the literate programming features in org-babel, I reviewed and organized my Emacs configuration. I’m looking forward to adding more notes to my configuration as I explore!

Literate programming is the idea that you should write first and program second, and that you can interweave the program into your explanation. Nifty.

Here’s my Emacs configuration. Share yours! =)

This patch modifies punchagan’s org2blog to allow you to publish an Org subtree with M-x org2blog-post-subtree. It posts a draft by default, and publishes the post if you call it with C-u M-x org2blog-post-subtree. It gets the posting date from SCHEDULED, DEADLINE, active or inactive timestamps, or the Post Date property, and lets you use tags as categories or use a separate Categories property. It inherits tags from parent headings, too. It picks up the title from the subtree heading or uses the Title property.

diff --git a/org2blog.el b/org2blog.el
index dc88291..b95caba 100644
--- a/org2blog.el
+++ b/org2blog.el
@@ -77,6 +77,11 @@
   :group 'org2blog 
   :type 'string)
 
+(defcustom org2blog-use-tags-as-categories nil
+  "Non-nil means assign :tags: to WordPress categories instead."
+  :group 'org2blog
+  :type 'boolean)
+
 (defvar org2blog-categories-list nil 
   "List of weblog categories")
 
@@ -433,4 +438,143 @@
        (goto-char current-pos)
        (command-execute (lookup-key org-mode-map (kbd "C-c t")))))))
 
+(defun org2blog-create-categories (categories)
+  "Create unknown CATEGORIES."
+  (mapcar
+   (lambda (cat)
+     (if (and (not (member cat org2blog-categories-list))
+              (y-or-n-p (format "Create %s category? " cat)))
+         (wp-new-category org2blog-server-xmlrpc-url
+                          org2blog-server-userid
+                          (org2blog-password)
+                          org2blog-server-blogid
+                          cat)))
+   categories))
+
+(defun org2blog-password ()
+  "Get password or prompt if needed."
+  (or org2blog-server-pass
+      (setq org2blog-server-pass (read-passwd "Weblog password? "))))
+
+(defun org2blog-upload-images-insert-links (&optional beg end)
+  "Upload images and replace with links in the region specified by BEG to END."
+  (interactive "r")
+  (let ((re 
+        (concat "\\[\\[\\(.*\\)" 
+                (substring (org-image-file-name-regexp) 0 -2)
+                "\\]\\]"))
+       file-all-urls file-name file-web-url blog-pass)
+    (save-excursion
+      (save-restriction
+        (narrow-to-region (or beg (point-min))
+                          (or end (point-max)))
+        (goto-char (point-min))
+        (while (re-search-forward re nil t 1)
+          (setq file-name (concat 
+                           (match-string-no-properties 1)
+                           "."
+                           (match-string-no-properties 2)))
+          (unless (save-match-data (string-match org-link-types-re file-name))
+            (save-match-data 
+              (if (assoc file-name file-all-urls)
+                  (setq file-web-url (cdr (assoc file-name file-all-urls)))
+                (setq file-web-url
+                      (cdr (assoc "url" 
+                                  (metaweblog-upload-image org2blog-server-xmlrpc-url
+                                                           org2blog-server-userid
+                                                           (org2blog-password)
+                                                           org2blog-server-weblog-id
+                                                           (get-image-properties file-name))))
+                      file-all-urls (append file-all-urls (list (cons 
+                                                                 file-name file-web-url))))))
+            (replace-match (concat "[[" file-web-url "]]") t t nil 0)))))
+    file-all-urls))
+
+(defun org2blog-post-subtree (&optional publish)
+  "Post the current entry as a draft. Publish if PUBLISH is non-nil."
+  (interactive "P")
+  (let ((post (org2blog-parse-subtree))
+        post-id)
+    (org2blog-create-categories (cdr (assoc "categories" post)))
+    (setq post-id (cdr (assoc "post-id" post)))
+    (save-excursion 
+      (org2blog-upload-images-insert-links (org-back-to-heading) (org-end-of-subtree)))
+    (if post-id
+        (metaweblog-edit-post org2blog-server-xmlrpc-url
+                             org2blog-server-userid
+                              (org2blog-password)
+                             post-id
+                              post
+                             publish)
+      (setq post-id
+            (metaweblog-new-post
+             org2blog-server-xmlrpc-url
+             org2blog-server-userid
+             (org2blog-password)
+             org2blog-server-blogid
+             post
+             publish))
+      (org-entry-put (point) "Post ID" post-id)
+      (message (if publish
+                   "Published (%s): %s"
+                 "Draft (%s): %s")
+               post-id
+               (cdr (assoc "title" post))))))
+
+(defun org2blog-parse-subtree ()
+  "Parse the current subtree as a blog entry."
+  (let (html-text
+        (post-title (or (org-entry-get (point) "Title")
+                        (org-get-heading t)))
+        (post-id (org-entry-get (point) "Post ID"))
+        ;; Set post-date to the Post Date property or look for timestamp
+        (post-date (or (org-entry-get (point) "Post Date")
+                       (org-entry-get (point) "SCHEDULED")
+                       (org-entry-get (point) "DEADLINE")                       
+                       (org-entry-get (point) "TIMESTAMP_IA")
+                       (org-entry-get (point) "TIMESTAMP")))
+        (tags (org-get-tags-at (point) nil))
+        (categories (org-split-string (or (org-entry-get (point) "CATEGORIES") "") ":")))
+    ;; Convert post date to ISO timestamp
+    (setq post-date
+          (format-time-string "%Y%m%dT%T"
+                              (if post-date
+                                  (apply 'encode-time (org-parse-time-string post-date))
+                                (current-time))
+                              t))
+    (if org2blog-use-tags-as-categories
+        (setq categories tags
+              tags nil))
+    (save-excursion
+      (setq html-text
+            (org-export-region-as-html
+             (and (org-back-to-heading) (line-end-position))
+             (org-end-of-subtree)
+             t 'string))
+      (setq html-text
+            (with-temp-buffer
+              (insert html-text)
+              (goto-char (point-min))
+              ;; Fix newlines
+             (let (start-pos end-pos)
+                (setq start-pos (point-min))
+               (goto-char start-pos)
+                (while (re-search-forward "<\\(pre\\|blockquote\\).*?>" nil t 1)
+                  (setq end-pos (match-beginning 0))
+                  (replace-string "\n" " " nil start-pos end-pos)
+                  (re-search-forward (concat "</" (match-string-no-properties 1) ">") nil t 1)
+                  (setq start-pos (match-end 0))
+                  (goto-char start-pos))
+               (setq end-pos (point-max))
+               (replace-string "\n" " " nil start-pos end-pos))
+              ;; Copy the text
+              (buffer-substring-no-properties (point-min) (point-max)))))
+    (list
+     (cons "date" post-date)
+     (cons "title" post-title)
+     (cons "tags" tags)
+     (cons "categories" categories)
+     (cons "post-id" post-id)
+     (cons "description" html-text))))
+
 (provide 'org2blog)

(add-to-list 'load-path "~/elisp/org2blog")
(require 'org2blog)
(setq org2blog-server-url "http://sachachua.com/blog/xmlrpc.php"
      org2blog-server-user "admin"
      org2blog-server-weblog-id ""
      org2blog-use-tags-as-categories t)
(org2blog-login)

As a developer, I often find myself working with multiple trees of source code. Sometimes, I’m comparing the trunk and the branches. Sometimes, I’m copying ideas from one place to another. Sometimes, I’m working on one project and an urgent defect comes in for a different project. With good development environments such as Eclipse, I’d still need to click on the project or directory in order to change my context, and then use a keyboard shortcut to find the resource in that tree.

With an awesome development environment like my customized Emacs, however, I can easily juggle multiple source trees. Here’s the macro I wrote to make setting this up much easier:

(defmacro sacha/file-cache-setup-tree (prefix shortcut directories)
  "Set up the file-cache tree for PREFIX using the keyboard SHORTCUT.
DIRECTORIES should be a list of directory names."
  `(let ((file-cache-alist nil)
	 (directories ,directories))
     (file-cache-clear-cache)
     (while directories
       (file-cache-add-directory-using-find (car directories))
       (setq directories (cdr directories)))
     (setq ,(intern (concat "sacha/file-cache-" prefix "-alist")) file-cache-alist)
     (defun ,(intern (concat "sacha/file-cache-ido-find-" prefix)) ()
       (interactive)
       (let ((file-cache-alist ,(intern (concat "sacha/file-cache-" prefix "-alist"))))
	 (call-interactively 'file-cache-ido-find-file)))
     (global-set-key (kbd ,shortcut)
		     (quote ,(intern (concat "sacha/file-cache-ido-find-" prefix))))))

With that, I can use the following to map C-c p to finding files in my personal directories:

(sacha/file-cache-setup-tree
 "personal"
 "C-c p"
 '("/var/www/html/drupal"
   "~/elisp"
   "~/personal"))

I’ve also set up keyboard shortcuts for the other trees I’m working on.

You’ll need file-cache, ido, and probably something like this:

(require 'filecache)
(require 'ido)
(defun file-cache-ido-find-file (file)
  "Using ido, interactively open file from file cache'.
First select a file, matched using ido-switch-buffer against the contents
in `file-cache-alist'. If the file exist in more than one
directory, select directory. Lastly the file is opened."
  (interactive (list (file-cache-ido-read "File: "
                                          (mapcar
                                           (lambda (x)
                                             (car x))
                                           file-cache-alist))))
  (let* ((record (assoc file file-cache-alist)))
    (find-file
     (expand-file-name
      file
      (if (= (length record) 2)
          (car (cdr record))
        (file-cache-ido-read
         (format "Find %s in dir: " file) (cdr record)))))))

(defun file-cache-ido-read (prompt choices)
  (let ((ido-make-buffer-list-hook
	 (lambda ()
	   (setq ido-temp-list choices))))
    (ido-read-buffer prompt)))
(add-to-list 'file-cache-filter-regexps "docs/html")
(add-to-list 'file-cache-filter-regexps "\\.svn-base$")
(add-to-list 'file-cache-filter-regexps "\\.dump$")

Sometimes w3m is not enough. To make it easier to open the current page in a browser such as Mozilla Firefox, add the following to your ~/.emacs:

(defun wicked/w3m-open-current-page-in-firefox ()
  "Open the current URL in Mozilla Firefox."
  (interactive)
  (browse-url-firefox w3m-current-url)) ;; (1)

(defun wicked/w3m-open-link-or-image-in-firefox ()
  "Open the current link or image in Firefox."
  (interactive)
  (browse-url-firefox (or (w3m-anchor) ;; (2)
                          (w3m-image)))) ;; (3)

This defines a function that uses the current URL being browsed(1) and another function that takes the URL of the link at point(2). If no link is found, it takes the URL of the image at point(3).

You can use other browse-url functions instead of browse-url-firefox. For example, replacing browse-url-firefox with browse-url-kde will open the page, link, or image in Konqueror, KDE’s web browser.

I like binding f to the function that opens the current URL in Mozilla Firefox and F to the function that opens the current link or image in Mozilla Firefox. To do the same, add the following to your ~/.emacs:

(eval-after-load 'w3m
  (progn 
    (define-key w3m-mode-map "f" 'wicked/w3m-open-current-page-in-firefox)
    (define-key w3m-mode-map "F" 'wicked/w3m-open-link-or-image-in-firefox)))

—
This is part of the book that I’m writing about Emacs, which will be published by No Starch Press if I manage to get it together in time.

If you browse with a lot of open tabs, like I do, w3m will be much easier to use once you remap w3m-next-buffer and w3m-previous-buffer onto single-key shortcuts, allowing you to press a key to quickly flip between tabs.

By default, w3m-previous-buffer is mapped to C-c C-p and w3m-next-buffer is mapped to C-c C-n. On a QWERTY keyboard, you may want to remap w3m-previous-buffer to q and w3m-next-buffer to w. You’ll probably also want to remap w3m-close-window (which had been bound to q), and x is a good keybinding for that. To make all these changes, add the following to your ~/.emacs:

(eval-after-load 'w3m
  '(progn
     (define-key w3m-mode-map "q" 'w3m-previous-buffer)
     (define-key w3m-mode-map "w" 'w3m-next-buffer)
     (define-key w3m-mode-map "x" 'w3m-close-window)))

If you use a Dvorak keyboard layout, you can bind . to w3m-previous-buffer and , to w3m-next-buffer instead. Just add the following code to your ~/.emacs:

(eval-after-load 'w3m
  '(progn
     (define-key w3m-mode-map "." 'w3m-previous-buffer)
     (define-key w3m-mode-map "," 'w3m-next-buffer)))

(This is part of the draft for my book on Emacs, to be published by No Starch Press if I’m not too late.)

Here’s a handy shortcut that toggles between the W3M web browser and other buffers you’re working on. I use it to quickly switch between code and documentation (or your favorite timewasting site, as it also makes a handy boss key).

Define the function by adding the following code to your ~/.emacs:

(defun wicked/toggle-w3m ()
  "Switch to a w3m buffer or return to the previous buffer."
  (interactive)
  (if (derived-mode-p 'w3m-mode)
      ;; Currently in a w3m buffer
      ;; Bury buffers until you reach a non-w3m one
      (while (derived-mode-p 'w3m-mode)
	(bury-buffer))
    ;; Not in w3m
    ;; Find the first w3m buffer
    (let ((list (buffer-list)))
      (while list
	(if (with-current-buffer (car list)
	      (derived-mode-p 'w3m-mode))
	    (progn
	      (switch-to-buffer (car list))
	      (setq list nil))
	  (setq list (cdr list))))
      (unless (derived-mode-p 'w3m-mode)
	(call-interactively 'w3m)))))

Then bind it to a shortcut key sequence (F7 F7 in this example) by adding the following code to your ~/.emacs:

(global-set-key (kbd " ") 'wicked/toggle-w3m)

You can then use F7 F7 to switch back and forth between your web browser and whatever else you’re working on.

“Are you browsing Slashdot in Emacs?”, W- asked me after he glanced at my screen.

With Emacs’ reputation for including everything _and_ the kitchen sink, you probably won’t be surprised to hear that there’s more than one way to surf the Internet using your text editor. With today’s Javascript- and image-heavy websites, it can be hard to believe that anyone would use a text-based browser with limited support for many of the things we take for granted. Still, a Web browser in your text editor can be surprisingly useful. Here are some of the reasons why you might like it:

• Browsing is faster and less distracting. Forget flashing ads, garish colors, and large images. When you surf the Web in Emacs, you can focus on reading, and you can use all the typical Emacs shortcuts for navigating around. You can view images when you want to.
  If you need to see something that Emacs doesn’t support, you can easily open the current page in an external Web browser.
• You can integrate it into your work. With a little bit of Emacs Lisp, you can qucikly look up information on the Web based on what you’re currently working on. For example, PHP mode comes with a shortcut that lets you look up the current function’s documentation in the PHP manual. You can look up bug report details, dictionary definitions, and Wikipedia pages with minimal typing, too. If you use Emacspeak, you can set up the web browser to speech-synthesize more than what’s displayed on screen. The more you use Emacs, the more benefits you get from the integration.
• You can customize everything. You can customize your Emacs experience quickly and easily, and if you spend a lot of time on the Net, you’ll appreciate having your own shortcuts and functions. For example, I’ve completely remapped my keyboard shortcuts to support tabbed browsing on a Dvorak keyboard, and I’ve defined a few functions to make frequently-used commands much easier. You can even use functions to process Web pages and either summarize the information you’re interested in or make pages more navigable. It’s all just Emacs Lisp.
• You’re safe from browser exploits. No Javascript pop-ups, no image bugs, no browser-based malware that can take over your comuter or steal data. Just content.
• You need less memory. Why open up a memory-intensive graphical Web-based browser when you’ve got Emacs open anyway?

There’s more than one way to browse the Web in Emacs, of course. Browse-url is a package that makes it easy to open URLs in your preferred browser or browsers. For example, you can use it to browse the Web in Mozilla Firefox, and (of course) you can use it to browse the Web within Emacs itself. For browsing within Emacs, you can use w3m.el, an interface to the external W3M browser, or w3, a Web browser written entirely in Emacs Lisp. Of the two, I prefer w3m.el, which is much faster and more featureful than w3. Both can display graphics, tables, and frames, and w3 supports stylesheets.

More about Emacs and browsing the Web soon! Planned projects for this chapter of Wicked Cool Emacs:

*** Project XXX: Browse the Web
*** Project XXX: Open the current webpage in an external browser
*** Project XXX: Different browsers for different pages
*** Project XXX: Toggle between Web and work
*** Project XXX: Quick search
*** Project XXX: Customize your keymap
*** Project XXX: Download files
*** Project XXX: Add access keys
*** Project XXX: Use social bookmarking
*** Project XXX: Typeahead
*** Project XXX: Preview HTML
*** Project XXX: Read Web pages as news