Text manipulation

"Ron Rosenfeld" <ronrosenfeld@nospam.org> wrote...
....
>How about a negative look-behind assertion?
>
>=REGEX.MID($A$1,".*?(?<!FP)(K|$)",ROWS($B$2:B2))

I keep forgetting Longre uses PCRE rather than VBScript regexp syntax. Too
many regexp varieties.

That should work for one literal substring in the assertion, but multiple
options, e.g., ".+?(?<!(FP|XYZ|D))K" [note: it's more efficient to append a
K to A1 and use a simple K at the end of the regexp than to use (K|$) - to
me at least that's a clearer indicator that trailing symbols after the last
K should be included] could cause backtracking problems. Assertions with
alternation subexpressions with closures (not the case here, but generally)
can become a big PITA. If there were multiple prohibited sequences, it may
be more efficient to mask immediately following cut characters and replace
them in the results. This is the regexp analog to using ancillary cells for
intermediate calculations rather than single huge formulas in spreadsheets.
Just as 2 or 3 formulas may recalc more quickly than a single equivalent
formula, 2 or 3 simpler regexp operations may recalc more quickly than a
single, more complex one. Assertions may work well in this case, but I tend
to avoid them when possible myself.

On Mon, 2 Jan 2006 14:26:16 -0800, "Harlan Grove" <hrlngrv@aol.com> wrote:

>"Ron Rosenfeld" <ronrosenfeld@nospam.org> wrote...
>...
>>How about a negative look-behind assertion?
>>
>>=REGEX.MID($A$1,".*?(?<!FP)(K|$)",ROWS($B$2:B2))
>
>I keep forgetting Longre uses PCRE rather than VBScript regexp syntax. Too
>many regexp varieties.
>
>That should work for one literal substring in the assertion, but multiple
>options, e.g., ".+?(?<!(FP|XYZ|D))K" [note: it's more efficient to append a
>K to A1 and use a simple K at the end of the regexp than to use (K|$) - to
>me at least that's a clearer indicator that trailing symbols after the last
>K should be included] could cause backtracking problems. Assertions with
>alternation subexpressions with closures (not the case here, but generally)
>can become a big PITA. If there were multiple prohibited sequences, it may
>be more efficient to mask immediately following cut characters and replace
>them in the results. This is the regexp analog to using ancillary cells for
>intermediate calculations rather than single huge formulas in spreadsheets.
>Just as 2 or 3 formulas may recalc more quickly than a single equivalent
>formula, 2 or 3 simpler regexp operations may recalc more quickly than a
>single, more complex one. Assertions may work well in this case, but I tend
>to avoid them when possible myself.
>

I will need to keep that in mind. Clearly I have insufficient experience with
regular expressions to have run into some of the issues you have mentioned.

I will need to think more about the backtracking issues in order to better
understand what you write. As well as the pros and cons of trying to avoid
assertions.

I have been trying to work out some of the questions posed on the NG's you
referred me to. Educational. Thanks.

Best wishes and Happy New Year,

--ron
--ron

Ron Rosenfeld wrote...
....
>I will need to keep that in mind. Clearly I have insufficient experience with
>regular expressions to have run into some of the issues you have mentioned.
>
>I will need to think more about the backtracking issues in order to better
>understand what you write. As well as the pros and cons of trying to avoid
>assertions.
....

If you want to get a real grounding in regular expressions, read
Jeffrey Friedl's book 'Mastering Regular Expressions, 2nd Ed.', ISBN
0-596-00289-0. Lots of discussion about backtracking and optimizing
regular expessions.

On 2 Jan 2006 19:05:16 -0800, "Harlan Grove" <hrlngrv@aol.com> wrote:

>Ron Rosenfeld wrote...
>...
>>I will need to keep that in mind. Clearly I have insufficient experience with
>>regular expressions to have run into some of the issues you have mentioned.
>>
>>I will need to think more about the backtracking issues in order to better
>>understand what you write. As well as the pros and cons of trying to avoid
>>assertions.
>...
>
>If you want to get a real grounding in regular expressions, read
>Jeffrey Friedl's book 'Mastering Regular Expressions, 2nd Ed.', ISBN
>0-596-00289-0. Lots of discussion about backtracking and optimizing
>regular expessions.

Thanks. Just ordered a copy from Amazon.


--ron