Copybooks Explained

A copybook is a standalone member of source code, most often a set of DATA DIVISION field definitions, that is stored in a library and pulled into a COBOL program at compile time using the COPY statement, so that many programs can share one authoritative definition instead of retyping the same record layout repeatedly. A typical copybook might define a customer record with fields like CUST-ID, CUST-NAME, and CUST-BALANCE, and any program that needs to read or write that record simply writes COPY CUSTREC to bring in that exact layout. Because the copybook is expanded inline by the compiler's preprocessor before syntax checking, the resulting compiled program behaves as if every field had been typed directly into the source, giving zero runtime overhead compared to hand-copied code.

The basic form, COPY CUSTREC, tells the compiler to locate a member named CUSTREC in the library concatenation specified for the compile job, such as an MVS partitioned data set or a library path in a distributed COBOL environment, and insert its text at that point in the source before compilation proceeds. Copybooks can be placed in the DATA DIVISION for record layouts, but they are equally valid in the PROCEDURE DIVISION for shared paragraphs of logic, in the FILE-SECTION for FD entries, or even in the ENVIRONMENT DIVISION for SELECT clauses, making COPY a general-purpose textual-inclusion mechanism rather than something limited to data definitions. Multiple copybooks are frequently nested, where one copybook itself contains a COPY statement pulling in another, so a change to a deeply shared low-level field only needs to be made in a single place.

The REPLACING phrase lets a program customize a generic copybook without editing the shared source, using syntax like COPY GENERIC-REC REPLACING ==PREFIX== BY ==CUST==, which substitutes every occurrence of the pseudo-text PREFIX with CUST as the copybook is expanded. This is commonly used when the same record structure needs to be included twice in one program under different names, such as a WS-OLD-RECORD and a WS-NEW-RECORD built from the identical field layout for a before-and-after comparison, or when a shared copybook uses a placeholder like ==FILLER-NAME== that each calling program fills in with its own file's actual name. Because REPLACING operates on the raw text before compilation, it can substitute identifiers, literals, or even whole clauses, giving copybooks template-like flexibility while still guaranteeing every program derives from the same base definition.