# Comparison rules for regression testing
# Defines how different collector outputs should be compared

# Global configuration (applies to both preflight and supportbundle)
global:
  # Files that should be compared exactly (byte-for-byte)
  exact_match:
    - "static-data.txt/static-data"
    # Note: version.yaml is NOT here - versionNumber varies between builds

  # Default behavior for unknown files
  non_empty_default: true

# Preflight-specific rules
preflight:
  # Files that should be compared exactly
  exact_match:
    - "static-data.txt/static-data"
    - "files/example.yaml"      # From data collector in v1beta2
    - "files/example.json"       # From data collector in v1beta2
    - "config/replicas.txt"      # From data collector in v1beta2
    # Note: version.yaml is NOT here - it uses non-empty check (versionNumber varies)

  # Files that need structural/field-specific comparison
  # Format: "pattern": "comparator_function_name"
  structural_compare:
    # NOTE: Most collectors are now non-empty checks only
    # We're testing that collectors RUN and PRODUCE OUTPUT, not that
    # specific environmental values (versions, connection status, image availability) match

    # Keep structural comparison only for truly deterministic collectors:
    # (Currently none - all moved to non-empty or exact match)

  # Everything else uses non-empty check by default
  # This tests that collectors RUN and PRODUCE OUTPUT, not environmental state
  # This includes:
  # - postgres/*.json, mysql/*.json, etc. (connection status varies)
  # - dns/debug.json (IPs, pod names vary)
  # - registry/*.json (image availability varies)
  # - http/*.json (endpoint status varies)
  # - cluster-info/*.json (k8s version varies)
  # - analysis.json (analyzer results vary with cluster state)
  # - cluster-resources/**/*.json (UIDs, timestamps, status vary)
  # - node-metrics/**/*.json (all metric values vary)
  # - goldpinger/**/*.json (latencies vary)
  # - certificates/**/*.json (validity time-based)
  # - sysctl/* (some values vary)
  # - And everything else...

# Support bundle-specific rules
supportbundle:
  # Files that should be compared exactly
  exact_match:
    - "static-data.txt/static-data"
    # Note: version.yaml is NOT here - it uses non-empty check (versionNumber varies)

  # Files that need structural comparison
  structural_compare:
    # NOTE: Like preflight, we only test that collectors produce output
    # Environmental state (DB connections, registry access, etc.) will vary
    # (Currently none - all moved to non-empty or exact match)

  # Everything else uses non-empty check (see list above in preflight section)

# Notes on comparison strategies:
#
# EXACT MATCH:
# - Use for static data that should never change between runs
# - Byte-for-byte comparison
# - Any difference is a regression
#
# STRUCTURAL COMPARISON:
# - Use for semi-deterministic output with consistent structure but variable values
# - Compare specific fields only (e.g., status codes, booleans)
# - Ignore timing-dependent or environment-specific values
#
# NON-EMPTY CHECK (default):
# - Use for highly variable output where exact comparison is impractical
# - Verifies file exists and is not empty
# - For JSON files, also validates JSON is parseable
# - Appropriate for:
#   * Kubernetes resources (UIDs, resourceVersions, timestamps)
#   * Metrics (all values constantly change)
#   * Logs (timestamps, dynamic content)
#   * Generated pod/resource names
#   * Runtime state (pod status, replica counts)
#
# This strategy catches major regressions (collectors breaking, files missing)
# while avoiding false positives from expected variability.