Parse Error Detection & Type Safety

[jamescook]

Parse Error Detection & Type Safety

This PR adds comprehensive parse error detection to Cataract with minimal performance impact, along with type safety validation for the C extension.

Parse Error Detection

Adds optional raise_parse_errors option to Stylesheet.parse that validates CSS and raises ParseError exceptions for malformed input:

# Enable all error checking
Cataract::Stylesheet.parse(css, raise_parse_errors: true)

# Granular control over specific checks
Cataract::Stylesheet.parse(css, raise_parse_errors: {
  empty_values: true,               # color: ;
  malformed_declarations: true,     # color red (missing colon)
  invalid_selectors: true,          # empty or missing selectors
  invalid_selector_syntax: true,    # ..invalid, ##bad, ???
  malformed_at_rules: true,         # @media {}, @supports {}
  unclosed_blocks: true             # missing closing braces
})

ParseError Object

Errors include position tracking:

begin
  Cataract::Stylesheet.parse(css, raise_parse_errors: true)
rescue Cataract::ParseError => e
  e.message      # => "Empty value for property 'color' at line 3, column 12"
  e.line         # => 3
  e.column       # => 12
  e.error_type   # => :empty_value
end

Type Safety

Added comprehensive type validation for C extension constructor arguments:

• Stylesheet.parse validates string input
• Stylesheet.new validates options hash structure
• Parser options validated (import_fetcher must be Proc, base_uri must be String, etc.)
• Raises clear TypeError instead of segfaulting on invalid input

Selector Syntax Validation

The invalid_selector_syntax check uses a whitelist approach to validate CSS selector characters:

• Valid: a-z A-Z 0-9 - _ . # [ ] : * > + ~ ( ) ' " = ^ $ | \ & % / ! and whitespace
• Invalid sequences: .. (double dot), ## (double hash)
• Invalid characters: ???, $$$, @@@, etc.
• Supports: Pseudo-classes (:hover), pseudo-elements (::before), functions (:not())

Fuzzer Improvements

• Added 22 invalid CSS patterns to fuzzer corpus for crash testing
• Removed parse error tracking from fuzzer (now raises exceptions instead)

Implementation Details

• Early exit optimization: checks only run when enabled
• Both C and Pure Ruby implementations maintain parity
• Error position calculated from original CSS string for accurate line/column numbers