6.3. A More Complicated Macro Example

Sometimes you may want to find all the references to a particular topic in a file. Table 6-2 lists steps for creating a macro that takes takes every sentence in the buffer that contains the word Emacs and copies it to another buffer. If you try this macro, you'll need to type some text about Emacs into a buffer. You can also get a test file to work with by opening the Emacs NEWS file (using C-h n), then writing it to a file (C-x C-w NEWS). This buffer is in view mode by default; change to text mode by typing M-x text-mode Enter.

Table 6-2. Steps for macro that creates a buffer of Emacs references

Keystrokes	Action
F3 or C-x (	Start macro definition; Def appears on the mode line.
C-s emacs	Find the word Emacs.
Enter	Stop the search after it is successful; if the search is unsuccessful, it rings the bell and stops the macro.
M-a	Move to the beginning of the sentence.[3]
C-Space	Set the mark.
M-e	Move to the end of the sentence.
M-w	Copy the sentence to the kill ring.
C-x b emacsrefs Enter	Move to a buffer called emacsrefs.
C-y	Insert the sentence.
Enter	Start the next sentence on a new line.
C-x b Enter	Move back to the original buffer.
F4 or C-x )	End the macro definition; Def is removed from the mode line.

^[3] M-a's definition of a "sentence" is controlled by the variable sentence-end, which is a fairly complex regular expression. By default, a sentence ends with a period, question mark, or exclamation mark, optionally followed by a quotation mark or parenthesis (including brackets or braces), and followed by two or more spaces or a newline.

Now, assume that you've already constructed the macro outlined in Table 6-2 and that you can invoke it with F4. The following screen shows what happens when you run it five times and then display the emacsrefs buffer.

Type: M-5 F4 or M-5 C-x e, followed by C-x b Enter
By executing the macro repeatedly, we've created a buffer that contains references to the Emacs editor.

As in the previous example, you can jump back and forth between an unlimited number of buffers while defining a macro. Macros don't need to be confined to one buffer. Macros that work with several buffers are more difficult to debug; when several buffers are involved, it becomes harder for you to keep track of where the cursor and the mark are. It is also easy to make mistaken assumptions about what buffer you're visiting; hence, it's a good idea to specify the buffer name explicitly. However, after you get accustomed to working with macros and multiple buffers, you'll be amazed at how much work you can do with almost no effort.

Windows are sometimes useful in macros, but, again, you have to watch out. It's better to start a macro with one window on the screen, have the macro open other windows, and finally close all but one window (C-x 1). If you write a macro with two windows on the screen and later try to execute it with four windows on the screen, the results will be unpredictable at best! In general, moving to a named buffer, C-x b buffername, is preferable to moving to the "other" window using C-x o (too vague to be generally useful). The other window could be anything—a *Help* buffer, *Completion* buffer, *shell* buffer, and so on. Moving to a named buffer always gets you to the right place, no matter how (or whether) the buffer is displayed.