Regexp: Case-insensitive matching | N factorial

[gentsquash]

In a setting where I can specify only a JS regular
expression, but not the JS code that will use it, I seek
a regexp component that matches a string of letters,
ignoring case. E.g, for "cat" I'd like the effect of

([Cc][Aa][Tt])

but without having to have many occurrences of [Xx].
Secondly, what is an efficient regexp that matches a
string exactly when ALL words in a certain list occur in
the string. I'd like the effect of

(cat.*nip|nip.*cat)

except that there are N words rather than just the two
words "cat" and "nip". (I can assume that no word in the
list is a prefix of any other.) Naturally, I'm looking for
a regexp-solution that does not involve listing all
N factorial
many orderings.

--Jonathan LF King, Mathematics dept, Univ. of Florida

[Thomas 'PointedEars' Lahn]

RobG wrote:

If you want to match the word cat exactly, then:

var reA = /\bcat\b/i;

That depends on how you define a word. If you define a word as a sequence
of word characters as specified in the ECMAScript Language Specification,
Ed. 3 Final, section 15.10.2.6 (i.e. those matching /[0-9A-Za-z_]/), you are
right.

However, for example "Menü" is a word in German, and

var reA = /\bmen\b/i;

will (only) match the "Men" in "Menü" there. Because `ü' is not considered
a word character per the Specification, and so the empty word ε between "n"
and "ü" constitutes a word boundary matched by /\b/ (as e.g.

"Menü".match(/\bmen\b/i)

shows).

So for matching Unicode words in strings, you have to use

var reA = /(^|\s)cat(\s|$)/i;

instead; that is, a character sequence (here: without whitespace in-between)
bounded by whitespace, or one or two input boundaries.
PointedEars
--
Anyone who slaps a 'this page is best viewed with Browser X' label on
a Web page appears to be yearning for the bad old days, before the Web,
when you had very little chance of reading a document written on another
computer, another word processor, or another network. -- Tim Berners-Lee

[RobG]

On Jun 26, 4:17 pm, Thomas 'PointedEars' Lahn <[e-mail]>
wrote:

RobG wrote:

If you want to match the word cat exactly, then:

var reA = /\bcat\b/i;

That depends on how you define a word. If you define a word as a sequence
of word characters as specified in the ECMAScript Language Specification,
Ed. 3 Final, section 15.10.2.6 (i.e. those matching /[0-9A-Za-z_]/), you are
right.

However, for example "Men¨¹" is a word in German, and

var reA = /\bmen\b/i;

will (only) match the "Men" in "Men¨¹" there. Because `¨¹' is not considered
a word character per the Specification,

Hence I included the sentence "Also, the regular expression's idea of
a word
boundary might be different to what you expect."

and so the empty word ¦Å between "n"
and "¨¹" constitutes a word boundary matched by /\b/ (as e.g.

"Men¨¹".match(/\bmen\b/i)

shows).

So for matching Unicode words in strings, you have to use

var reA = /(^|\s)cat(\s|$)/i;

That expression is commonly used for matching values in the HTML class
attribute where the separator is specified as being whitespace. It is
not sufficient for matching words in general where they may be
followed by punctuation marks such as commas, semi-colons, colons,
dashes, periods and so on.
--
Rob

[Thomas 'PointedEars' Lahn]

RobG wrote:

Thomas 'PointedEars' Lahn wrote:

>RobG wrote:

>>If you want to match the word cat exactly, then:
var reA = /\bcat\b/i;

That depends on how you define a word. If you define a word as a sequence
of word characters as specified in the ECMAScript Language Specification,
Ed. 3 Final, section 15.10.2.6 (i.e. those matching /[0-9A-Za-z_]/), you are
right.

However, for example "Menü" is a word in German, and

var reA = /\bmen\b/i;

will (only) match the "Men" in "Menü" there. Because `ü' is not considered
a word character per the Specification,

Hence I included the sentence "Also, the regular expression's idea of
a word boundary might be different to what you expect."

It was easy to overlook and provides no explanation as to what should be
expected instead.

>and so the empty word ε between "n"
and "ü" constitutes a word boundary matched by /\b/ (as e.g.

"Menü".match(/\bmen\b/i)

shows).

So for matching Unicode words in strings, you have to use

var reA = /(^|\s)cat(\s|$)/i;

That expression is commonly used for matching values in the HTML class
attribute where the separator is specified as being whitespace. It is
not sufficient for matching words in general where they may be
followed by punctuation marks such as commas, semi-colons, colons,
dashes, periods and so on.

Good point. However, a character class can take care of that. For example,
in Unicode text that uses only ASCII and Latin-1 punctuation:

var reA = /(^|[\s,;:.-])cat([\s,;:.-]|$)/i;

But whether a punctuation mark really delimits a word is a matter of
language, interpretation, and personal taste. For example, the HYPHEN-MINUS
character ("-") may have been used as hyphen in compounds.

An alternative would be to use the \w escape sequence to build your own
character class:

var reA = /(^|[^\wäöü])cat([^\wäöü]|$)/i;
PointedEars
--
Anyone who slaps a 'this page is best viewed with Browser X' label on
a Web page appears to be yearning for the bad old days, before the Web,
when you had very little chance of reading a document written on another
computer, another word processor, or another network. -- Tim Berners-Lee

[Dr J R Stockton]

In comp.lang.javascript message <6aa0c1c4-b785-4da1-9107-b681df097261@c5
8g2000hsc.googlegroups.com>, Wed, 25 Jun 2008 15:31:37,
ge********@gmail.com posted:

>In a setting where I can specify only a JS regular
expression, but not the JS code that will use it, I seek
a regexp component that matches a string of letters,
ignoring case. E.g, for "cat" I'd like the effect of

([Cc][Aa][Tt])

but without having to have many occurrences of [Xx].

If all else fails, read the manual. There are links in <URL:http://www.
merlyn.demon.co.uk/js-valid.htm>.
Note that the average intellectual level of those who post with @gmail
addresses is so low that readers may kill-file it /in toto/.

Secondly, what is an efficient regexp that matches a
string exactly when ALL words in a certain list occur in
the string. I'd like the effect of

(cat.*nip|nip.*cat)

except that there are N words rather than just the two
words "cat" and "nip". (I can assume that no word in the
list is a prefix of any other.) Naturally, I'm looking for
a regexp-solution that does not involve listing all
N factorial
many orderings.

I doubt whether one exists to do a direct match, at least if it is to be
compatible with any user agent that knows RegExps.

But one could use S2 = S1.replace(/cat|nip/gi, "") and see whether the
difference of the lengths matches the total of the strings, provided
that no string can occur more than once and matchable strings cannot
overlap.

--Jonathan LF King, Mathematics dept, Univ. of Florida

DSS.

--
(c) John Stockton, nr London, UK. ?@merlyn.demon.co.uk Turnpike v6.05 MIME.
Web <URL:http://www.merlyn.demon.co.uk/- FAQish topics, acronyms, & links.
Proper <= 4-line sig. separator as above, a line exactly "-- " (SonOfRFC1036)
Do not Mail News to me. Before a reply, quote with ">" or "" (SonOfRFC1036)

[RobG]

On Jun 26, 10:52 pm, Dr J R Stockton <[e-mail]:
[...]

Note that the average intellectual level of those who post with @gmail
addresses is so low that readers may kill-file it /in toto/.

Bad day? My Google Groups profile has a non-gmail address that is
easily discovered by those who care to do so.

<URL: http://www.prejudicenoway.com.au/activities/2156.html >
--
Rob