Questions Ab Initio

  Step-by-Step Process: Reads Grouped Data Takes in records that are already sorted by a specific key (like customer ID or date) Optional Filter Check If you have a filter expression: Filter returns FALSE (0) : Record is skipped Filter returns NULL : Record goes to reject port with error message Filter returns TRUE : Record continues processing If no filter: All records are processed Handles Record Groups Groups consecutive records with the same key value together For single-record groups : Record goes directly to output For multi-record groups (duplicates) : Uses your "keep" setting to decide which record to keep If keep = "first" or "last" : Keeps one record (sends to output), sends others to duplicate port If keep = "unique-only" : Sends ALL records from duplicate groups to duplicate port (output gets nothing) Key Points: Data MUST be sorted first for this to work correctly Works with consecutive duplicate records You control which duplicate t...

 Input: 1,a,b,c,2,d,e,f,g   Output: field1 field2 1       a 1       b 1       c 2       d 2       e 2       f 2       g     input---->normalize--->output   type temporary_type =  record   decimal("") v_col1;   decimal("") v_col2; end; out::length(in) = begin   out :: length_of(string_split(in.col, ",")); end; temp::normalize(temp, in, index) = begin   temp.v_col1 :: if (string_is_numeric(string_split(in.col, ",")[index])==1)                      string_split(in.col, ",")[index] else temp.v_col1;   temp.v_col2 :: string_split(in.col, ",")[index]; end; temp::initialize(in) = begin   temp.v_col1 :: 0;   temp.v_col2 :: 0; end; out::finalize(temp, in) = begin   out.col1 :: temp.v_col1;   out.col2 :: temp.v_col2; end; ...

 air object rm <EME object eme>

The Reformat component processes your data records through these steps: Basic Setup: Each output port is numbered (out0, out1, out2, etc.) Each output has matching reject and error ports (reject0/error0, reject1/error1, etc.) Step-by-Step Process: Read a Record Takes one record from the input port Check Selection Filter (if you have one) If the filter expression returns FALSE : Record is skipped, go to next record If the filter returns NULL : Graph stops with an error message If the filter returns TRUE (or no filter exists): Continue processing Determine Where to Send the Record If you have only one output port : Record goes to that port automatically If you have multiple output ports : Uses output-index to decide which port(s) get this record Can send to one port or multiple ports If there's an error choosing ports, record goes to reject0 Transform the Record For each designated output port: If no transform is defined: Uses automatic field mapping If transform is defined: Runs ...

  Single stage components process each record in a single step.Below are example of single stage component.  Filter by expression  Reformat  Dedup    Multi stage components process records in multiple stages allowing more complex operations like  initialisation,  finalization   Below are the example of the multi stage components    Normalize  Rollup  Scan  Scan with rollup etc.

      Input data       over run   1 6   2 5   3 13           Rollup    Use key change function   type temporary_type =   record   decimal("\n ") run _per_over ;   end;   out:: key_change (in 1,in 2)=   begin   out:: if(( decimal( "|" ))in2.ball %6==1)1 else 0;   end;   temp:: initialize(in)=   begin   temp.run_per_ over : :0;   end;   temp ::rollup ( temp,in )=   begin   temp.run_per_ over :: temp.run _per_over +( decimal( "|")) in.run ;   end;   out : : finalize( temp, in) =   begin   out.over ::( decimal("|")) next_in_sequence ();   out.run :: temp.run _per_over ;   end;