Hi Randy
This is what we are doing now and it works
well if the person writing the regular expression is the same person who wrote
the code that interprets the matches. However in our case we would like users to
supply a regular expression to parse a line of a bank statement. In this case
the developer is not sure how many sub-expressions there will be or which one
contains the value of interest. We get around this by insisting the
user’s expression contain 5 sub-expressions and the one of interest is
the third one. Uses then pad out their regular expression with empty
sub-expressions as required.
Allowing named sub-expressions would solve
this problem nicely. Boris pointed out that Python does this and he was good
enough to send me a suggested syntax based on the Python regular expression
syntax:
Indexed sub-expression access:
=======================
matcher := RxMatcher forString:
'(Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)[
]+(:isDigit::isDigit:?)[
]*,[ ]*(19|20)(:isDigit::isDigit:)'.
(matcher matches: 'Aug 6, 1996')
ifTrue:
[Array
with:
(matcher subexpression: 5)
with: (matcher subexpression: 2)
with: (matcher subexpression: 3)]
ifFalse: ['no match']
Named sub-expression access:
=======================
matcher := RxMatcher forString:
'(?P<month>Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)[
]+(?P<day>:isDigit::isDigit:?)[
]*,[
]*(?P<year1>19|20)(?P<year2>:isDigit::isDigit:)'.
(matcher matches: 'Aug 6, 1996')
ifTrue:
[Array
with: (matcher subexpression: #year2)
with: (matcher subexpression: #month)
with: (matcher subexpression: #day)]
ifFalse: ['no match']
From: Randy Coulman
[mailto:[hidden email]]
Sent: Friday, March 17, 2006 10:08
AM
To: Joerg Beekmann
Subject: Re: Named subexpressions
in RxParser
On 3/17/06, Joerg
Beekmann <[hidden email]>
wrote:
Is
it possible to tag a sub-expression with a name and then when there is a match
extract the value associated with the tagged sub-expression?
I keep thinking I should come up with a nicer API for this, but here's
what you need to do:
matcher := RxMatcher forString: 'a*([^a*])a*([^a*])a*'.
matcher matches: 'aaaabbbbaccaaa'. "print it yields true"
(1 to: matcher subexpressionCount) collect: [:index | matcher subexpression:
index]. "print it"
Subexpression 1 is the entire matched string. The nested subexpressions
(anything enclosed in $( and $)) start at index 2. So the above example
has 3 sub-expressions. The first is the entire string. The second
is 'bbb' and the third is 'cc'.
Randy
--
Randy Coulman
[hidden email]
Locked
