Something like charIndex, but in a variable...

[Ben Rubinstein]

On 30/04/2015 09:07, Malte Brill wrote:
> I need to find the start and end character of chuncks in a variable. If we are in a field we can use charIndex for that. However, in a variable I have no good idea on how to do something similar to
>
> get charIndex(token 7 of line 15 of field „myField“)

(unhelpful to Malte, sorry) - I hadn't noticed the introduction of charIndex. 
  What a useful property.  It would be great if it could be extended to 
arbitrary sources of value.

In the meantime, could one write an ugly function to calculate the start index 
of a chunk by first getting the length of the preceeding chunks, then using 
that value as the charsToSkip for a call to offset on the value of the chunk?

So a first approximation - it would need some better work parsing the chunk 
expression to be robust -

function chunkIndex tFullChunkExpr, tValue
    local tChunkExpr, tPriorChunksExpr, tChunkValue, iConsumed, n, i

    -- convert nested chunk expression into lines, minor to major
    replace " of " with return in tFullChunkExpr
    if last word of tFullChunkExpr = "of" then \
	delete last word of tFullChunkExpr

    -- accumulate start of each level of chunk
    put 0 into iConsumed

    -- find the start of each chunk in the expression, from major to minor
    repeat with i = number of lines in tFullChunkExpr down to 1
       put line i of tFullChunkExpr into tChunkExpr

       -- make an expression for the chunks prior to this one
       put tChunkExpr into tPriorChunksExpr
       put format("1 to %d", word 2 of tChunkExpr - 1) \
             into word 2 of tPriorChunksExpr

       -- get the length of the chunks prior to this one
       do format("put length(%s of tValue) into n", tPriorChunksExpr)

       -- get this chunk, and find the first instance
       -- of it after the length of the prior chunks
       do ("put" && tChunkExpr && "of tValue into tChunkValue")
       add offset(tChunkValue, tValue, n) + n to iConsumed

       -- repeat within this chunk
       put tChunkValue into tValue
    end repeat

    return iConsumed - 1
end chunkIndex


To be called as e.g.
    get chunkIndex("token 7 of line 15 of", myVar)

I suspect that this could be fooled if there was a lot of leading white space 
(and a repetition of the target chunk prior to it), but that could probably be 
dealt with by an additional check for offset of the 'priorChunkValue'.


Ben