Skip to content

severity

Type: str Default: null (all violations reported) Accepted values: "low", "medium", "high"

Description

Filters violations by their static performance impact level. Only violations whose impact meets or exceeds the configured value are reported. Violations below the threshold produce no output and do not affect the exit code.

Each rule carries one of three impact levels:

Level Badge Meaning
"low" 🟢 LOW Minor impact — style, API correctness, small inefficiencies
"medium" 🟡 MEDIUM Moderate impact — avoidable overhead at scale
"high" 🔴 HIGH Major impact — OOM risk, full scans, shuffle explosion

The impact level is displayed as a colored badge in the terminal output next to the rule ID:

error[D001][HIGH]: Avoid using collect()
error[F005][LOW]: Avoid stacking multiple withColumn() calls

Example

[tool.pyspark-antipattern]
# Report only MEDIUM and HIGH violations
severity = "medium"

[tool.pyspark-antipattern]
# Report only HIGH violations — useful for blocking CI on critical issues only
severity = "high"

CLI equivalent

pyspark-antipattern check src/ --severity=medium
pyspark-antipattern check src/ --severity=high

When both pyproject.toml and --severity are set, the CLI flag takes precedence.

Notes

  • Setting severity = "low" is equivalent to the default (all violations reported).
  • severity filters by the rule's static impact level, which is independent of the warn / ignore / select options. A rule silenced via ignore is never reported regardless of severity.
  • To see the impact level of every rule, browse the rules reference — each rule page has a ## Severity section.