argparser.py: HTTPieArgumentParser has high cyclomatic complexity and mixed responsibilities #1828

Overview

Automated complexity analysis of the httpie/ package ranked httpie/cli/argparser.py as the highest-complexity module. The findings below expand on what the existing developer TODO comments already acknowledge.

Findings

_process_auth — cyclomatic complexity ~20, max nesting 5 levels

# TODO: refactor & simplify this method.   (line 283)

The method (74 lines) handles six distinct concerns in sequence:

1. Extracting embedded URL credentials (user:pass@host)
2. Looking up .netrc credentials
3. Resolving the auth plugin
4. Validating --auth required
5. Parsing / prompting for credentials
6. Calling plugin.get_auth()

Three nearly-identical AuthCredentials(key=…, value=…, sep=…, orig=…) blocks are copy-pasted rather than shared.

_process_output_options — nested closure

A check_options function is defined inside the method body and called twice immediately after — a straightforward candidate for extraction to class level.

Class-level TODO (line 79)

# TODO: refactor and design type-annotated data structures
#       for raw args + parsed args and keep things immutable.

HTTPieArgumentParser has no return type annotations on its 14 methods.

Metrics (automated scan)

Suggestion

A PR (#1827) is open that addresses _process_auth specifically — decomposing it into four focused helpers and adding return-type annotations. Happy to discuss scope or approach if this direction is welcome.

Found via automated codebase complexity scan.