Remove specific values out of string in cobol - cobol

Question

Would like to ask how to remove specific values in different lengths out of a string.
I have this:
'{4:72:SELLS¬#:73:ABC¬#:PPF:TESTPPF¬#:74:BLA¬#:PPF:ABC¬#:74:BLA¬#-}'
( want to remove all ¬#:PPF: Tags with its content. In my example this should be removed:
¬#:PPF:TESTPPF
¬#:PPF:ABC )
And would like to have this:
'{4:72:SELLS¬#:73:ABC¬#:74:BLA¬#:74:BLA¬#-}'
I have this code:
01 TINP.
05 TINPFIELD PIC X(2000) VALUE
'{4:72:SELLS¬#:73:ABC¬#:PPF:TESTPPF¬#:74:BLA¬#:PPF:TESTPPF¬#:74:BLA¬#-}'.
01 WA-OUTPUT PIC X(2000) value spaces.
01 WA-TEMP PIC X(2500).
01 WP-MSG PIC 9(8) BINARY value zero.
01 WN-ROWCNT PIC S9(8) BINARY.
01 WN-ROWCNT2 PIC S9(8) BINARY.
01 WP-BEG PIC 9(8) BINARY.
01 WP-END PIC 9(8) BINARY.
01 WN-OUT-LEN PIC 9(8) BINARY value zero.
0000-TESTPROCESSING SECTION.
display TINPFIELD.
INSPECT TINPFIELD
TALLYING WN-Rowcnt FOR ALL "¬#".
MOVE 1 TO WP-MSG
PERFORM UNTIL WN-ROWCNT2 >= WN-ROWCNT
MOVE WP-MSG TO WP-BEG
display 'WP-BEG' WP-BEG
UNSTRING TINPFIELD
DELIMITED BY "¬#"
INTO WA-TEMP
POINTER WP-MSG
END-UNSTRING
MOVE WP-MSG TO WP-END
display 'WP-END' WP-END
if WA-OUTPUT = space
subtract 1 from wp-end
STRING TINPFIELD(WP-BEG:WP-END)
delimited by SIZE
INTO WA-OUTPUT
END-STRING
move wp-end to WN-OUT-LEN
else
STRING WA-OUTPUT(1:WN-OUT-LEN)
delimited by SIZE
TINPFIELD(WP-BEG:WP-END)
delimited by SIZE
'¬#'
delimited by SIZE
INTO WA-OUTPUT
END-STRING
end-if
move WP-END TO WN-OUT-LEN
display 'WN-OUT-LEN' WN-OUT-LEN
ADD 1 TO WN-ROWCNT2
END-Perform
.
EXIT.
//edit: The Input data is always the same. In the output (after my code runs) I have sometimes a "Tag" twice, sometimes only once and so on. Its not consistent at all. I suppose this is a code issue on my side.

Answer 1

The code was changed to reflect a change in the question.
This code splits the input into delimited segments copying each segment directly to the output using reference modification. If there is no text to be removed (no delimiter) it will copy all the input directly to the output.
There are performance implications with this method. Specifically, the move of each segment will cause space-filling in the output. The larger the number of segments the worse the performance.
I changed WA-OUTPUT to X(2000), since the output can never be larger than the input.
01 TINP.
05 TINPFIELD PIC X(2000) VALUE
'{4:72:SELLS¬#:73:ABC¬#:PPF:TESTPPF¬#:74:BLA' &
'¬#:PPF:TESTPPF¬#:74:BLA¬#-}'.
01 WA-OUTPUT PIC X(2000) value spaces.
01 segment-length binary pic 9(4).
01 additional-characters binary pic 9(4).
01 input-pointer binary pic 9(4).
01 output-pointer binary pic 9(4).
01 input-length binary pic 9(4).
01 output-length binary pic 9(4).
procedure division.
begin.
move 1 to input-pointer output-pointer
input-length output-length
inspect TINPFIELD tallying
input-length for characters before "}".
display input-length
display TINPFIELD
perform until input-pointer > function length(TINPFIELD)
unstring TINPFIELD delimited "¬#:PPF:"
into WA-OUTPUT (output-pointer:)
count in segment-length
with pointer input-pointer
add segment-length to output-pointer
if input-pointer <= function length(TINPFIELD)
move 0 to additional-characters
inspect TINPFIELD (input-pointer:) tallying
additional-characters for characters before "¬"
add additional-characters to input-pointer
end-if
end-perform
inspect WA-OUTPUT tallying
output-length for characters before "}".
display wa-output
display output-length
goback
.
Output:
0070
{4:72:SELLS¬#:73:ABC¬#:PPF:TESTPPF¬#:74:BLA¬#:PPF:TESTPPF¬#:74:BLA¬#-}
{4:72:SELLS¬#:73:ABC¬#:74:BLA¬#:74:BLA¬#-}
0042
This is a modification of the above that, while slightly more complicated, eliminates the performance penalty for space-filling when there is a larger number of segments. It is unusual because, while it uses the UNSTRING statement, it does not actually "unstring" anything.
01 TINP.
05 TINPFIELD PIC X(2000) VALUE
'{4:72:SELLS¬#:73:ABC¬#:PPF:TESTPPF¬#:74:BLA' &
'¬#:PPF:TESTPPF¬#:74:BLA¬#-}'.
01 WA-OUTPUT PIC X(2000) value spaces.
01 segment-length binary pic 9(4).
01 segment-holder pic x.
01 additional-characters binary pic 9(4).
01 input-pointer binary pic 9(4).
01 backup-pointer binary pic 9(4).
01 output-pointer binary pic 9(4).
01 input-length binary pic 9(4).
01 output-length binary pic 9(4).
procedure division.
begin.
move 1 to input-pointer output-pointer
input-length output-length
inspect TINPFIELD tallying
input-length for characters before "}".
display input-length
display TINPFIELD
perform until input-pointer > function length(TINPFIELD)
move input-pointer to backup-pointer
unstring TINPFIELD delimited "¬#:PPF:"
into segment-holder
count in segment-length
with pointer input-pointer
move TINPFIELD (backup-pointer:segment-length)
to WA-OUTPUT (output-pointer:segment-length)
add segment-length to output-pointer
if input-pointer <= function length(TINPFIELD)
move 0 to additional-characters
inspect TINPFIELD (input-pointer:) tallying
additional-characters for characters before "¬"
add additional-characters to input-pointer
end-if
end-perform
if output-pointer < function length (WA-OUTPUT)
move space to WA-OUTPUT (output-pointer:)
inspect WA-OUTPUT tallying
output-length for characters before "}".
display wa-output
display output-length
goback
.
The output is the same.