[PUSHPAK-JAISWAL]

What & why

Enables CI/CD pipelines and automated tooling to consume analysis reports and enforce minimum migration confidence thresholds:

• --format <text|json>: Adds support for machine-readable JSON output (--format json) for tooling and dashboards, with text remaining the default.
• --fail-under <percent>: Adds a threshold check (0–100) that exits with code 1 if the estimated migration confidence score is strictly below the threshold. The report (text or JSON) is always emitted prior to exiting so that build logs remain actionable.
• Deterministic Exit Codes:
  • 0: Success (analysis succeeded and confidence meets --fail-under if set).
  • 1: Threshold failure (confidence strictly below --fail-under).
  • 2: Invocation / CLI usage error (unknown options, missing arguments, invalid threshold, missing/non-existent directory).
• Documentation: Documents CLI options, exit codes, example JSON, and a formal JSON Schema (Draft 2020-12) in README.md.

Type of change

• New rule (MIG-xxx)
• Bug fix
• Documentation
• Internal refactor / test coverage
• New feature (CLI tooling & CI support)

Checklist

• mvn test passes
• New behaviour is covered by a test
• Nothing here modifies the user's original project (read-only by default)
• Rules are AST-based and deterministic, not AI-generated transformations
• Any new mapping was checked against the Selenium Boot docs first
• If a new rule was added, the table in README.md was updated

Notes for reviewers

• Zero new dependencies: Built a dedicated JsonRenderer to keep the project lightweight and avoid adding Jackson / external libraries to the shaded JAR.
• Cross-platform path normalization: Relative file paths (unparsableFiles and findings[].file) normalize backslashes to / so JSON outputs are deterministic across Windows and Linux.
• Test coverage:
  • ReportJsonTest.java: Validates JSON structure, empty reports, finding details, escaping of quotes/newlines/control characters.
  • CliTest.java: Covers in-memory execution and actual OS subprocess execution (ProcessBuilder) verifying exit codes 0, 1, 2, argument order permutations, and edge-case validation.

closes #9

[shossain786]

Thanks for this, reviewed the code. Two things to fix before merge:

1. Exit code 1 means two different things. Right now, if analysis crashes (e.g. can't read a file), it exits with code 1 — same code as "quality gate failed". In CI this would look like a real gate failure instead of a crash. Can we use a different exit code for real errors?

2. JSON schema says line is always ≥ 1, but code can output line: 0. This happens when a node's position isn't found. If a CI tool validates against your schema, it'll reject a valid report.

Smaller nit: the byRule counting logic in JsonRenderer is copy-pasted from Report.render() — worth pulling into one shared method so they don't drift apart.

[PUSHPAK-JAISWAL]

@shossain786 on it

[PUSHPAK-JAISWAL]

@shossain786 i made some changes those are :-
Introduced Exit Code 3 for Runtime Errors
Added ruleCounts(): Extracted the byRule aggregation into public Map<String, Long> ruleCounts().
Refactored render(): Updated render() to consume ruleCounts() instead of computing the map inline.
Deduplicated byRule: Replaced the duplicated counting logic with report.ruleCounts().
Removed Unused Import: Removed java.util.TreeMap.
Updated Exit Codes Table: Documented exit code 3 as runtime/I/O error.
Fixed JSON Schema: Changed "line" from "minimum": 1 to "minimum": 0 and clarified that 0 denotes an unavailable AST position.
Added unit test rendersLineZeroWhenPositionUnavailable() verifying line: 0 serializes properly.
Added unit test ruleCountsProvidesSharedAggregation() verifying shared rule-counting behavior.
Added in-memory test returnsExitCodeThreeOnRuntimeError verifying exit code 3 on I/O failure.
Added a subprocess test in processExitCodes verifying OS exit code 3.

[shossain786]

Both fixes look right — exit code 3 cleanly separates runtime errors from quality-gate failures now, and the schema/line:0 handling checks out. Ran the full suite locally too (55/55 green) plus a manual smoke test of --format json --fail-under. Thanks for the thorough follow-through on the review, merging this.