Use UNSTRING with one or more delimiters, or reference modification for fixed positions.

How do I normalize text?

Use intrinsic functions like UPPER-CASE/LOWER-CASE and replace sequences with STRING/UNSTRING patterns.

How to parse structured lines?

Combine UNSTRING, reference modification, and validation to extract fields step by step.

COBOL Text Processing Concepts

Text processing turns raw lines into structured fields you can validate and store. Focus on predictable delimiters and defensive parsing. This covers parsing, normalization, and validation of text data in COBOL programs.

Tokenize a Delimited Line

cobol

1
2
3
4
5
6
7
8
9
10
11
12
13
14
       01 IN-LINE     PIC X(120) VALUE 'ID=123|NAME=ALICE|CITY=AUSTIN'.
       01 ID-FLD       PIC 9(9).
       01 NAME-FLD     PIC X(30).
       01 CITY-FLD     PIC X(20).
       01 P            PIC 9(3) VALUE 1.
       UNSTRING IN-LINE
           DELIMITED BY '|' OR '='
           INTO FILLER
                ID-FLD
                FILLER
                NAME-FLD
                FILLER
                CITY-FLD
           WITH POINTER P.

This pattern alternates label and value; send labels to FILLER, real values into fields. UNSTRING splits on multiple delimiters ('|' separates fields, '=' separates labels from values).

Normalize Case and Trim

cobol

1
2
3
4
5
       MOVE FUNCTION UPPER-CASE(NAME-FLD) TO NAME-FLD
       MOVE FUNCTION LOWER-CASE(CITY-FLD) TO CITY-FLD
       *> Trim right spaces by moving into a smaller PIC
       MOVE NAME-FLD TO NAME-FLD-TRIMMED
       MOVE CITY-FLD TO CITY-FLD-TRIMMED

Use intrinsic functions for case conversion. To trim trailing spaces, move to a smaller PIC field. Always normalize data early in processing to ensure consistent comparisons.

Parse CSV Data

cobol

1
2
3
4
5
6
7
8
9
10
11
       01 CSV-LINE PIC X(200) VALUE '123,Alice Smith,Austin,TX'.
       01 FIELD-COUNT PIC 9 VALUE 0.
       01 FIELD-1 PIC X(20).
       01 FIELD-2 PIC X(30).
       01 FIELD-3 PIC X(20).
       01 FIELD-4 PIC X(10).
       UNSTRING CSV-LINE
           DELIMITED BY ','
           INTO FIELD-1 FIELD-2 FIELD-3 FIELD-4
           WITH POINTER P
           TALLYING FIELD-COUNT.

CSV parsing uses comma as delimiter. TALLYING counts how many fields were successfully parsed. Check FIELD-COUNT to ensure all expected fields were found.

Fixed-Position Records

cobol

1
2
3
4
       MOVE IN-LINE(1:2)   TO STATE
       MOVE IN-LINE(3:5)   TO ZIP
       MOVE IN-LINE(8:20)  TO STREET
       MOVE IN-LINE(21:50) TO CITY

Reference modification is best for fixed columns. Keep a spec documenting column ranges. Format: field(start:length) where start is 1-based position and length is number of characters.

Validate Parsed Data

cobol

1
2
3
4
5
6
7
8
9
10
11
       IF ID-FLD NOT NUMERIC
           DISPLAY 'Error: ID must be numeric'
       ELSE IF ID-FLD = 0
           DISPLAY 'Error: ID cannot be zero'
       ELSE IF NAME-FLD = SPACES
           DISPLAY 'Error: Name is required'
       ELSE IF FUNCTION LENGTH(FUNCTION TRIM(CITY-FLD)) < 2
           DISPLAY 'Error: City name too short'
       ELSE
           DISPLAY 'Data validation passed'
       END-IF.

Always validate parsed data: check for numeric fields, required values, reasonable lengths. Use FUNCTION TRIM to remove leading/trailing spaces before length checks.

Be explicit about delimiters and positions in documentation.
Validate each field (NUMERIC, ranges, required values).
Normalize case and whitespace early in processing.
Handle errors gracefully with meaningful messages.
Test with various input formats and edge cases.

String Processing

Strings, UNSTRING, reference modification

Character Handling

Character validation and sets