In other words rather than have the link name in the form of:
1984 decot bugs
it is now:
1984/decot in bugs.html
so that it matches how the entry itself looks (as in how it's
identified) and to say where (what file) it is (in). The reason it says
'in bugs.html' rather than 'bugs' (saying that the entry has bugs) is
because not all are bugs and it's saying where to see it anyway. But
some are features, others are missing links or files or other things
like that and to keep the same form it now says 'in bugs.html'.
This was done with my sgit(1) tool like:
sgit -e 's,\([0-9][0-9][0-9][0-9]\) \([-a-z\.0-9]*\) bugs,\1/\2 in bugs.html,g' '*/*md'
The link to bugs.html itself has the path '../../bugs.html'. This is
important as it's meant to be rendered in the browser. At first this was
not remembered and I did something like:
sgit -e 's,\[\([0-9][0-9][0-9][0-9]\) \([-a-z\.0-9]*\) bugs\](../../bugs.html,\[\1/\2 in bugs.html\](%%REPO_URL%%/bugs.html,g' '*/*md'
but this was changed to the simpler one due to the problem described
above. It's worth putting this note here as a reminder, though, so I
have included it here.
To build:
make
Bugs and (Mis)features:
The current status of this entry is:
STATUS: INABIAF - please **DO NOT** fix
For more detailed information see 2001/schweikh in bugs.html.
To use:
./schweikh string string2
This program will very likely crash or do something else if you do not give it two args.
Note also that the glob pattern must match the whole string. See the author's comments for details and a workaround.
There's also no way to escape meta characters.
This is supposed to happen. As is written in the The Jargon File:
That's not a bug, that's a feature.
Try:
./try.sh
Judges' remarks:
Do you get the pattern here?
Author's remarks:
This one-liner program is a glob pattern matcher. It understands
the glob characters * meaning 'zero or more characters' and
? meaning exactly one character, just like your unix shell.
Run it with two args, the string and the glob pattern. The exit status is 0
(shell true) when the pattern matches, 1 otherwise. The glob pattern must match
the whole string, so you may want to use * at the beginning and end of the
pattern if you are looking for something in the middle.
You can use it for your shell scripting needs similar to a silent grep
(and without stdout redirected to /dev/null):
if prog "${VARIABLE}" '<glob>'; then
echo VARIABLE matches
else
echo no match
fi
Bugs
There is no way to escape the meta characters. Dammit Jim, I'm a one-liner, not a regexp lib.
All obfuscation is obviously in the recursive m() function, an 86
character glob pattern evaluator, returning nonzero for a match:
- Just a single complex return expression.
- Nested ternary operator
?:to save onif/elseverbosity. - Short circuiting
&&and||to save even more onif/elseverbosity. - Subtraction instead of an equality operator in
*t - 42. - ASCII codes for
*and?. - Careful use of blanks even though this year's rules have extended that resource limit.
- The source is a complete preprocessed C program. Because it communicates
with the world out there by means of
argvand the exit status, there is no need forstdio.hbloat. An asm guru could surely squeeze this program in less than a screenful. indent(1)is probably not too helpful.
The recursion wonderfully and magically takes care of iteration and
backtracking. I could tell you how it works but then I would have to
kill request a NDA from you :-) Run it under a debugger with several
inputs, it will be quite helpful and instructive. Handling of ? is
straightforward; for * start out with *foo and foo* against foo.
How does it deal with sequences of adjacent *? How could this be
improved? If all else fails, recode the ?: operators with if/else and
try again. For extra credit implement character classes like [a-z].