[CMake] FIND sub-command to the STRING command

[Michael Wild]

Very cool.

I think creating a ticket and attaching the patches (or even better,
posting a URL to a public GIT branch) would be best. In order for this
feature to be included by Kitware, you'd also need to provide unit tests.


Michael

On 02/04/2011 04:31 PM, Tim Hütz wrote:
> Ok, I've implemented a forward search (finds the first occurrence of the substring) and a reverse search (which finds the last occurrence of a substring). Should I open a ticket (with adding the required paches) for suggesting this for one of the next CMake releases or how should I proceed? The idea of a STRING( FIND REGEX ...) is nice too. I'm working on it as soon as I have some free minutes :)
> 
> Am 04.02.2011 um 16:02 schrieb Michael Wild:
> 
>> Exactly. The single-character matching would then just be a special
>> case. If you want to get fancy, having a string(FIND REGEX) command
>> would also be nice ;-)
>>
>> Michael
>>
>> On 02/04/2011 03:31 PM, Tim Hütz wrote:
>>> Mhh, okay. With substring matching you mean that I can get the position of the first occurrence of a word or sub-sentence, am I right?
>>>
>>>
>>> Am 04.02.2011 um 13:32 schrieb Michael Wild:
>>>
>>>> On 02/04/2011 12:49 PM, Tim Hütz wrote:
>>>>> I don't think that the speed-factor was important in this case. A complete CMake run takes about 25 seconds to run (in my case) and I see no influence in the processing time with my suggested patch or the regular expression sample provided by Michael. I thought that my solution is a bit more intuitive, but I don't know if this is really important for CMake. If you use the provided patch, you can do something like
>>>>>
>>>>> PROJECT( "STRING FIND command" )
>>>>> SET( TESTSTRING "This is a test" )
>>>>> STRING( FIND ${TESTSTRING} "a" APOSITION )
>>>>> MESSAGE( STATUS "Position of character 'a' is ${APOSITION}" )
>>>>>
>>>>> Regards,
>>>>> Tim
>>>>>
>>>>> Am 04.02.2011 um 12:00 schrieb SF Markus Elfring:
>>>>>
>>>>>>> mhh, yeah, thats a good idea. Ok, I'll use this for doing a "STRING( FIND ...)" :) It was just an idea to share this patch with you, but you're right. Its easier doing this with regular expressions.
>>>>>>
>>>>>> Can it be that your implementation was faster than a complete regex application for your use case?
>>>>>>
>>>>>> Are there any remarkable differences in the corresponding execution speed and performance characteristics?
>>>>>>
>>>>>> Regards,
>>>>
>>>> I just wanted to point out that this particular use-case can be handled
>>>> by a regular expression. I still think that a string(FIND) command would
>>>> be useful, although it would be even more so if it did substring matching.
>>>>
>>>> BTW, I just had an idea how to find substrings with regexes, but it is
>>>> even more tedious:
>>>>
>>>> set(str "Hello Tim. How are you Tim?")
>>>> set(pos -1)
>>>> if(str MATCHES "(Tim.*)")
>>>> string(LENGTH "${str}" l1)
>>>> string(LENGTH "${CMAKE_MATCH_1}" l2)
>>>> math(EXPR pos "${l1}-${l2}")
>>>> endif()
>>>>
>>>> This finds the position of the first occurence of "Tim". Another
>>>> alternative, without math(EXPR) is:
>>>>
>>>> set(str "Hello Tim. How are you Tim?")
>>>> set(pos -1)
>>>> string(REGEX REPLACE "Ali.*" "" tmp "${str}")
>>>> if(NOT "${tmp}" STREQUAL "${str}")
>>>> string(LENGTH "${tmp}" pos)
>>>> endif()
>>>>
>>>> Michael
>>>>