A better find and replace function

Robert Sneidar bobs at twft.com
Thu Sep 28 02:12:05 EDT 2006


This may interest some of you. You may have stumbled across something  
with the replace command. Let's say I have a chunk that contains the  
word "blah". I want to replace all occurrences of "blah" with  
"noblah" which those of you who are astute see contains the word it  
replaces. If I call the replace command once, all my "blah"'s become  
"noblah"'s okay, but if I execute the command again, my noblahs  
become nonoblahs.

What I needed was something that would replace the WORD blah with the  
WORD noblah, ignoring words that CONTAIN blah, including subsequent  
search and replace executions. In other words I needed a whole word  
search and replace that only did it once per occurrence of a whole  
word (in the true sense of the "word" ) no matter how many times I  
ran it. I mainly need it for renaming function and command calls in a  
script, but it could be useful elsewhere.

Just do a wordoffset you say? Oh but nay, because wordoffset will hit  
on blah, noblah, blahse etc. I ONLY want the whole word blah, even if  
it is a function like say blah(). (Revolution thinks "blah 
(somearg,somearg2)" is all one word. :-) It's a lot harder than you  
might think, because Revolution's word offset is very liberal when it  
comes to what it thinks is a word.

One last note: I employ a technique that I think may also be of  
interest. When working with large chunks of text that I am executing  
offset functions on I progressively "chop" it down by working with a  
copy of the chunk and then on each loop I put the text after what I  
already processed into the chunk. So each iteration in the loop I am  
working with a progressively smaller chunk. I am not sure what speed  
gains I may get, but one added benifit is that when I have dealt with  
every occurrence of what I am looking for, the offset function in the  
last bit will return 0 and that is my cue to bail from the repeat  
loop. Otherwise if I replaced "blah" with "noblah" and did a  
wordoffset call on the whole chunk, it would find the "noblah" that I  
just created and never exit!!

So here it is, a whole word search and replace function. I am sure  
some will find ways to make it better, Have at it.

FUNCTION findreplace thetarget thesearchstring thereplacestring
     -- I need a search and replace feature that will not keep  
replacing a partial word
     -- repeatedly
     #breakpoint
     put 1 into mfirstline

     REPEAT
         put line mfirstline to -1 of thetarget into thechunk

         -- get the first line containing the search string
         put lineoffset(thesearchstring, thechunk) into fline
         IF fline is 0 THEN
             exit REPEAT
         END IF

         put line fline of thechunk into theline

         -- Check to see that the line is not a comment
         IF word 1 of theline is "--" THEN
             put mfirstline + fline into mfirstline
             next REPEAT
         END IF

         -- see if this has already been replaced

         put word wordoffset(thesearchstring, theline) of theline  
into thesearchword
         put word wordoffset(thereplacestring, theline) of theline  
into thereplaceword
         -- thesearchword will not equal thereplaceword IF  
thesearchstring IS IN thereplacestring
         -- because thereplacestring does not exist in the line  
containing thesearchstring
         -- BUT WILL if thesearchstring CONTAINS thereplacestring!  
It's a gotcha.

         IF thesearchword is not thereplaceword OR \
                 thereplacestring is in thesearchstring THEN
             #breakpoint
             replace thesearchstring with thereplacestring in theline
             put theline into line fline of thechunk
             put thechunk into line mfirstline to -1 of thetarget
         END IF

         put mfirstline + fline into mfirstline
     END REPEAT

     return thetarget
END findreplace

Bob Sneidar
IT Manager
Logos Management
Calvary Chapel CM





More information about the Use-livecode mailing list