regexps

regexps

[Scott Schwartz]

Hmm.  No syntax for matching word boundaries?  Sigh.  That comes up a lot.

Re: regexps

[Paul Haahr]

as i remember, the rules in sam for where the selection goes after an
x command that modified data were pretty vague.  the justification that
i came up with for this (which may or may not have anything to do with
the code) was the way in which all modifications which are supposed to
happen under an x// are batched and happen in a second pass.

Re: regexps

[Scott Schwartz]

Here's another question:

Given a file containing

foo dir foo
bar dir bar

,x/[^a-z]dir[^a-z]/s/dir/path/  yields

foo path foo
bar path bar

and leaves "r path" in the last line selected.  That's odd, because
running ,x/[^a-z]dir[^a-z]/p leaves " dir " selected, and then running
s/dir/path leaves " path " selected, as I would have expected.

Re: regexps

[Scott Schwartz]

| beginning of word:      (^|[^a-zA-Z0-9])
| end of word:            ($|[^a-zA-Z0-9])

I lust for \< and \>.  In vi, something like
	%s/\<foo\>/gnu/
is wonderfully concise.

| 'course, you still have to narrow dot by one then.  

Hey, no fair leaving parts out. :-)

Re: regexps

[Paul Haahr]

beginning of word:      (^|[^a-zA-Z0-9])
end of word:            ($|[^a-zA-Z0-9])

'course, you still have to narrow dot by one then.  your definition of 
word may be different from mine, of course, so, maybe
        (^|[ \t]) and ($|[ \t])
would be more appropriate.  (i don't think sam takes \t, but it's
much clearer for purposes of the example.)