Topic: PennMUSH RegExp Features
Author: Raevnos
Summary: Raevnos presents an informational lecture as a followup to Javelin's
RegExp Lecture.

Raevnos cracks his knuckles and turns on the log. If you have questions,
please raise your hand and I'll call on you when I get a chance. If you think
I missed you, page.

Code Classroom(#1061RnJ)

A chalkboard fills one wall, with an old beat-up lectern in front of it.
The other three walls are painted off-white. A few long flourescent lights are

attached to the ceiling making the room bright. Several chairs with attached
desktop-like surfaces (standard classroom chairs) are scattered about the room

in no particular order.

Contents:
FettaS
William
Codex
Pip
China
Obvious exits:
Hallway <O>

Raevnos says, "To start off with a little bit of history, regular expression
support, in the form of regmatch() and the regexp attribute flag, were added
in 1.7.2. In 1.7.3, the regular expression (I'll be using regexp as an
abbreviation from now to save on typing), backend was changed to a more
powerful one, and various new functions were added."

Raevnos says, "I'll cover the new functions first, with a +help system as a
working example, and then touch on some of the new features of the regexp
backend."

Spiffy Help(#7100V)
Type: Thing Flags: VISUAL
Spiffy +help. See +help +help
Owner: Raevnos Zone: Raevnos' ZMO Ducats: 10
Parent: *NOTHING*
Basic Lock: =Raevnos
Powers:
Warnings checked: none
Created: Thu Sep 21 18:45:53 2000
Last Modification: Thu Sep 21 19:52:45 2000
DISPLAY_HELP [#1622]: table(edit(%0, %b, |, _, %b), 19, 78, |)
HELP_CMD [#1622]: $+help:@pemit %#=Help topics:%r[u(display_help,
v(toplevel))]
HELP_FIND_CMD [#1622]: $+help/find *:@pemit %#=Help topics matching
%0:%r[u(display_help, graball(lattr(#7030), *%0*))]
HELP_RFIND_CMD [#1622]: $+help/rfind *:@pemit %#=Help topics matching
%0:%r[u(display_help, regraballi(lattr(#7030), %0))]
HELP_RSEARCH_CMD [#1622]: $+help/rsearch *:@pemit %#=Help topics matching
%0:%r[u(display_help, regrepi(#7030, *, %0))]
HELP_SEARCH_CMD [#1622]: $+help/search *:@pemit %#=Help topics matching
%0:%r[u(display_help, grepi(#7030, *, %0))]
HELP_TOPIC2_CMD [#1622]: $+help2 *:@pemit %#=Help on
%0:%r[udefault(#7030/[regeditall(%0, \\s+, _)], No such topic.)]
HELP_TOPIC_CMD [#1622]: $+help *:@pemit %#=Help on
%0:%r[udefault(#7030/[edit(%0, %b, _)], No such topic.)]
TOPLEVEL [#1622]: +help +who
Home: Rec Room -- Raevnos' House
Location: Code Classroom(#1061RnJ)

Raevnos says, "This is a +help system that supports searching through help
entries and files. Useful if you're looking for something but don't know the
exact name."

Raevnos says, "We'll start with the regrab() and regrall() functions. They're
just like grab() and graball(), except they use a regular expression instead
of normal wildcards. To refresh, grab() returns the first element of a list
that matches a certain pattern, and graball() returns every item from a list
that matches. The re- versions do the same; they just treat the pattern
differently."

Raevnos says, "This help system has two commands to search through the names
of help topics. +help/find <wildcard pattern> and +help/rfind <regexp>. You
can examine Spiffy Help/help_find_cmd and help_rfind command to see the code.
They're identical except for the function used to do the match."

HELP_FIND_CMD [#1622]: $+help/find *:@pemit %#=Help topics matching
%0:%r[u(display_help, graball(lattr(#7030), *%0*))]

HELP_RFIND_CMD [#1622]: $+help/rfind *:@pemit %#=Help topics matching
%0:%r[u(display_help, regraballi(lattr(#7030), %0))]

Raevnos says, "Actually, it uses regraballi(), not regraball(). Regular
expression functions who's names end in 'i' are the same as the versions that
don't, except they are case insensitive. Normally, capitaliztion matters in
regexps, which can be an inconvience."

Raevnos says, "Any questions?"

Raevnos says, "Okay. The next function is regrep(), which is a version of
grep() that uses a regexp. grep() will search through attributes matching a
pattern, and return the names of those who's text contains a third argument.
It's case-sensitive. grepi() is the case-insenstive version, and there's a
regrepi() also. regrep() treats the attribute-matching pattern the same was as

grep() - a normal wildmatch pattern. It differs in looking for attribute text
that matches a regular expression, instead of containing a string."

Raevnos says, "This lets the +help system support searching through help files

for something. +help/search <string> returns all help entries that contain
<string>, and +help/rsearch <regexp> those that match the regexp. They're on
Spiffy Help/help_search_cmd and help_rsearch_cmd. Once again, they're the same

except for the function that odes the actual search."

HELP_RSEARCH_CMD [#1622]: $+help/rsearch *:@pemit %#=Help topics matching
%0:%r[u(display_help, regrepi(#7030, *, %0))]

HELP_SEARCH_CMD [#1622]: $+help/search *:@pemit %#=Help topics matching
%0:%r[u(display_help, grepi(#7030, *, %0))]

Raevnos says, "The last, and probably most useful of the new functions is the
regedit() family. There's four of them - regedit(), regeditall(), and the
case-insenstive versions that end in i. If you're familar with perl, they do
the same as s///, and are similar to edit() in mushcode. They search a string
for a section matching a regexp, and replace it with a third argument.
regedit() only replaces the first match, while regeditall() replaces every
match."

Raevnos shows an example.

Raevnos types --> say regedit(this is a silly example, si.*y, good)
Raevnos says, "this is a good example"

Raevnos says, "Another example is in the +help code.
Spiffy Help/help_topic_cmd has problems with help entries that are more than
one word, if too many spaces are present."

HELP_TOPIC2_CMD [#1622]: $+help2 *:@pemit %#=Help on
%0:%r[udefault(#7030/[regeditall(%0, \\s+, _)], No such topic.)]

HELP_TOPIC_CMD [#1622]: $+help *:@pemit %#=Help on
%0:%r[udefault(#7030/[edit(%0, %b, _)], No such topic.)]

Raevnos says, "Try this:"

Raevnos types --> +help +who%b%b2

Help on +who 2:
No such topic.

You say, "And then this:"

Raevnos types --> +help +who 2

Help on +who 2:
+3who: List connected players in a three-column format.
+idle: List connected players in a five-column format.

Raevnos says, "Now, suppose you want the first example to work also.
Spiffy Help/help_topic2_cmd uses regedit() to convert any number of spaces
(And tabs, and newlines) into the single _ used in the actual help topic name
(Because you can't have spaces in attribute names)."

Raevnos types --> +help2 +who%b%b2
Help on +who 2:
+3who: List connected players in a three-column format.
+idle: List connected players in a five-column format.

Raevnos says, "The actual regexp it uses is \s+. \s is a special pair of
characters that stands for 'Any whitespace characters, like space and tab',
and is one of the new features of the regexp backend. The + just means 'One or

more of the previous thing', so it matches one or more spaces. In the above
examples, two spaces. But it also matches one, or three, or a %t."

Raevnos says, "Lots of characters that have special meanings in regexps, like
\ and [ also have special meanings to the mush parser, so they need to be
escaped in code so the regexp backend sees them properly."

Raevnos pauses a moment for questions, if there are any?

FettaS shakes his head and has none

Raevnos says, "Alrighty then. The regular expression backend the mush uses is
very similar to the one used by the perl language, if you're familar with
that. It has a bunch of neat features. I'll mention some of the more useful
ones."

Pip has disconnected.

Raevnos mentioned \s already. There's a few other sequences: \d matches any
number (0-9), and \w matches any 'word' character - Letters, numbers, and _.
You can reverse what these match by capitalizing the letter. \D matches
anything /but/ numbers, for example.

Raevnos says, "You can get similar effects with character classes. [A-Za-z]
will match any letter from A to Z or a to z. However, Penn can run in
environments where other characters are valid letters. \w includes these
automatically. Inside a character class that includes other things, you can
use
the \sequences, or more flexible ranges, with the [:NAME:] notation. For
example, [[:lower:]] will match any lower-case letter, no matter the language.

Note the doubled brackets - the outer ones are for the character class, the
inner ones and :'s for a range. More than one of these can be in the same
class, and they can be mixed with other things. See HELP REGEXP CLASSES for
information on these."

Raevnos says, "The next thing is non-capturing parenthesis. Normally, in a
regexp, ()'s serve two purposes. 1, they 'capture' whatever matches their
contents so it can be referred to later. 2, they 'group' the sub-regexp they
contain so that it appears as one thing to modifiers like + and ?. 'foo?' will

match 'fo' or 'foo', but '(foo)?' will match 'foo' or nothing."

Raevnos says, "If you don't care what a parenthesis matches, but just want it
for the second, grouping behaivor, you can use (?:) instead of plain ().
With (?:), the match isn't kept for later use. This is most useful when you
have a regexp command pattern, where every capturing () is copied to a
%-variable. If you're never going to use what it matches, (?:) will prevent
that useless. copying."

Regexp Commands
Type: Thing Flags: VISUAL
Owner: Raevnos Zone: Raevnos' ZMO Ducats: 10
Parent: *NOTHING*
Basic Lock: =Raevnos
Powers:
Warnings checked: none
Created: Thu Sep 21 19:06:06 2000
Last Modification: Thu Sep 21 19:07:47 2000
BAD [#1622R]: $^\+t(est)?$:@pemit %#=The bad test. %%0 is %0. %%1 is %1.
GOOD [#1622R]: $^\+t(?\:est)?$:@pemit %#=The good test. %%0 is %0. %%1 is
%1.
Home: Code Classroom(#1061RnJ)
Location: Code Classroom(#1061RnJ)

Raevnos says, "The object I just dropped has two attributes that match the
same thing, one with (), one with (?:). Try them:"

Raevnos types --> +t

The bad test. %0 is +t. %1 is .
The good test. %0 is +t. %1 is .

Raevnos types --> +test

The bad test. %0 is +test. %1 is est.
The good test. %0 is +test. %1 is .

Raevnos says, "However, since : normally ends the pattern part of a command,
it needs to be escaped with \ so it's taken as part of the pattern, not the
end. You have to do the same thing with : in $-commands or ^-patterns that use

wildcards instead of regexps."

Raevnos says, "Examine the object if you haven't already to see how it looks."

Raevnos says, "There are some special regexp characters that match a position,

and not an actual character. ^ for the beginning of a string and $ for the end

are two examples. Penn also has, among others, \b, which matches at a word
boundry -- In between a character that matches \w and one that matches \W.
Remember that \Capital reverses \lower, so that \b matches at the start or end

of a normal word, but not in the middle."

Raevnos says, "This applies to \b as well. \B matches at places that aren't at

'word boundries'. Between two \w's or two \W's."

Raevnos says, "The last nifty thing I'm going to cover tonight since this is
turning out to be not-so-mini, is (?i). When this is found in a regexp,
everything from then on is case-insensitive. For example, regmatchi(string,
pattern) and regmatch(string, (?i)pattern) will both match the same thing.
This is useful if you don't know until you actually match if the pattern
should be case-sensitive or not."

Raevnos says, "There's a lot more that I haven't covered. Some of it might be
in HELP REGEXP along with the basics. Otherwise, find a copy of the perl
language manual online, or read one of the books on perl or regular
expressions in general. 'Mastering Regular Expressions', by Jeffrey Friedl,
is a pretty good one."

Raevnos asks for the last time, "Any questions?"

FettaS shakes his head and has none.

Codex shakes his head. "Thanks for the information"

FettaS needs to tinker with Regexp before he gets some.
China writes down the name of the book.

Raevnos will turn off the log now, then. Thanks for coming!