Note: this post relates to CloverDX 2.x version with CTL version 1. Starting with CloverDX 3.x , CTL version 2 was introduced which has a bit different syntax. However you can still use CTL 1 in transformations - see CTL1 vs CTL2 comparison.
Recently, I have been facing a very common problem. Imagine this scenario: I have two files – the first one with origin records and the second one with slightly modified new records. Each record had a unique key and approximately 50 fields. My task was to compare these two files and find out how many fields in every pair differ from each other in the corresponding records.
The simplified graph can be seen in following picture:
A comparison of two records can be processed by the CloverDX DataIntersection component which joins the records with the same keys. In the joined records, you can compare fields that are not part of the key. But remember, you have to write the comparison in the CTL transformation ;-) . Of course you could write the following block of code for each pair of compared fields:
int count = 0;
if(nvl($0.field_N, '') != nvl($1.field_N, '')) {
count++;
//but typically more actions :-(
}
//and imagine this block 50times :-(
//final mapping
$0.key := $0.key;
$0.count := count;
But this solution takes too much time when you have to repeat it for many fields (approx. 50 in my case). It is also very slow, uncomfortable and increases the probability of making a mistake in your code (e.g. omitting some fields). Fortunately, CloverDX allows you to iterate through the fields of processed records! :-) The code is then more briefer and more generic:
//declaration of variables for copies of input records
record(Metadata1) myrec1;
record(Metadata1) myrec2;
function transform() {
int i = 0;
int count = 0;
//asign value of input records to local variables
myrec1 = @0; //myrec1 is a copy of a current record on input port 0
myrec2 = @1; //myrec2 is a copy of a current record on input port 1
//iterate through fields, suppose that field with index 0 is the key
for(i = 1; i < length(myrec1); i++) {
if(nvl(myrec1[i], '') != nvl(myrec2[i], '')) {
count++;
}
}
//final mapping
$0.key := $0.key;
$0.count := count;
}
Someone could object to the necessity of making copies of records. In this case I have good news. CTL in CloverDX version 2.9 introduces the possibility of iterating directly through the fields of input records.
@0[i];//i-th field of the input record on port 0
Moreover new functions for getting field names and data types are introduced in version 2.9. Personally, I am looking forward to such features that will make CTL code simplier and clearer.
