Program Organization (with Examples in C)

Program Organization in Tidy, a Mid-Sized C Program

Introduction

In my book Why Learn C, Chapter 13, Program Organization, I discussed how most C programs consist of multiple source (.c) and header (.h) files. Typically, each .c file implements a specific functionality, while the corresponding .h file provides a public API for using that functionality. These pairs of files often work together to form a module or specialized aspect of the program. For example, a color.c file contains functions for printing text in color to a terminal, while color.h contains their declarations.

Program Structure in Tidy

Tidy, a mid-sized C program, demonstrates how to organize source files in a more complex project. Compared to the smaller ad program, Tidy is approximately three times larger. Here are some of its source files:

- array.c

- file_ext.c

- path_util.c

- symbol.h

- array.h

- file_ext.h

- path_util.h

- toml_lite.c

- bit_util.c

- fnv1a.c

- path_util_test.c

- toml_lite.h

- bit_util.h

- fnv1a.h

- pjl_config.h

- toml_test.c

- clang_util.c

- hash_table.c

- print.c

- trans_unit.c

- cli_options.c

- proxies.c

- type_traits.h

- cli_options.h

- include-tidy.c

- proxies.h

- typedef.c

- color.h

- include-tidy.h

- red_black.c

- typedef.h

- color.h

- include.c

- red_black.h

- unit_test.c

- config.h

- include.h

- red_black_test.c

- unit_test.h

- config_file.c

- ipath.c

- strbuf.c

- util.c

- config_file.h

- ipath.h

- strbuf.h

- util.h

- cxx.c

- options.c

- strbuf_test.c

- cxx.h

- symbol.c

Components of Tidy

Tidy can be broken down into several types of components:

1. Program-specific data structures and algorithms: Tidy includes a data structure to represent symbols referenced in a program, defined in symbol.h with implementations in symbol.c. This C++-specific code was later moved to cxx.c for better organization.

2. Generic data structures and algorithms: Tidy utilizes dynamic arrays, hash tables, red-black trees, and string buffers, which are necessary in C programs as they don't exist in the C standard library.

3. Program options: These allow the behavior of Tidy to be customized, such as comment styles, alignment columns, and colored output.

4. Command-line parsing: Tidy uses command-line options to set program options, often in a 1:1 correspondence with the program options.

5. Configuration file parsing: Tidy can also parse configuration files to set program options, which can be cumbersome to set from the command line.

6. Program-specific utilities: Functions that aren't entirely generic but are used in multiple places or improve the use of third-party libraries, like functions for dealing with Libclang in Tidy.

7. Generic utilities: Macros, error-handling functions, string-handling functions, and others that are useful across various parts of the program.

8. Test code: Tidy includes a small unit-test framework, test executables, and a functional-test framework.

Conclusion

Organizing source files effectively is crucial for maintaining and understanding larger programs. By separating program-specific and generic components, maximizing the reuse of generic components, and keeping program-specific code manageable, developers can create cleaner, more maintainable codebases. Tidy serves as an example of how to apply these principles in a mid-sized C program.