Implementation:Rapidsai Cuml Include Checker

Knowledge Sources	Rapidsai_Cuml
Domains	Machine_Learning, Code_Quality
Last Updated	2026-02-08 12:00 GMT

Overview

A Python linting script that validates C/C++/CUDA include syntax consistency across the cuML codebase, enforcing correct use of angle-bracket versus quoted includes.

Description

include_checker.py scans C++ source files and validates that #include directives use the correct syntax: quoted includes (#include "...") for files that exist as relative paths, and angle-bracket includes (#include <...>) for system or installed headers.

The script performs the following checks:

1. Include Style Validation -- If a quoted include references a file that does not exist relative to the source file, it flags an error recommending #include <...>. Conversely, if an angle-bracket include references a file that does exist as a relative path, it flags an error recommending #include "...".
2. Pragma Once Ordering -- Verifies that #pragma once appears before any #include directives.
3. Relative Include Warnings -- Generates warnings for questionable relative include patterns:

• • References to the current directory (./)
  • Relative paths that cross top-level include directories (src, src_prims)
  • Complex relative paths with more than two folders mixing parent references (..) with named directories

1. Inplace Fix -- With the --inplace flag, automatically rewrites include directives to use the correct syntax.

The script uses an Issue dataclass to track errors and warnings with file location, message, and fix information. It processes files matching a configurable regex pattern (defaulting to [.](cu|cuh|h|hpp|hxx|cpp)$).

Usage

Run this script as a code quality check during development or as part of CI to ensure consistent include syntax across the C++ codebase. Use --inplace to automatically fix detected issues.

Code Reference

Source Location

• Repository: Rapidsai_Cuml
• File: cpp/scripts/include_checker.py

Signature

def main():
def parse_args():
def check_includes_in(src, inplace, top_inc_dirs) -> typing.List[Issue]:
def list_all_source_file(file_regex, srcdirs):
def rel_include_warnings(dir, src, line_num, inc_file, top_inc_dirs) -> typing.List[Issue]:

@dataclasses.dataclass()
class Issue:
    is_error: bool
    msg: str
    file: str
    line: int
    fixed_str: str = None
    was_fixed: bool = False

Import

# Run from the cpp directory:
python scripts/include_checker.py [--regex PATTERN] [--inplace] [--top_include_dirs DIRS] [dirs...]

I/O Contract

Inputs

Name	Type	Required	Description
dirs	string(s)	Yes	Directories to scan for source files
--regex	string	No	cuh|h|hpp|hxx|cpp)$)
--inplace	flag	No	If set, automatically fix detected include issues in-place
--top_include_dirs	string	No	Comma-separated list of top-level include dirs (default: src,src_prims)

Outputs

Name	Type	Description
PASSED/FAILED	stdout	Overall check result
Error messages	stdout	Formatted error messages with file path and line number
Fixed files	Modified source files	Source files with corrected includes (when --inplace is used)

Usage Examples

# Check all source files in the src directory
python cpp/scripts/include_checker.py cpp/src cpp/src_prims

# Check and automatically fix issues
python cpp/scripts/include_checker.py --inplace cpp/src

# Check only header files
python cpp/scripts/include_checker.py --regex "[.](h|hpp)$" cpp/src

Related Pages

• Environment:Rapidsai_Cuml_CUDA_GPU