feat(localnet): add --json flag to logs command

[teyrebaz33]

Closes #38

Summary

localnet logs was the only diagnostic command without --json support. This PR adds it for consistency with localnet status --json and doctor --json.

Changes

• src/cli.rs: Added json: bool field to LocalnetLogsArgs
• src/commands/localnet.rs: cmd_localnet_logs now accepts json param and emits structured output
• tests/cli.rs: 3 new tests covering help text, empty log file, and tail output

Output format

{
  "tail": 50,
  "lines": [
    "2026-04-06T10:00:00Z sequencer started",
    "2026-04-06T10:00:01Z listening on 127.0.0.1:3040"
  ]
}

[weboko]

@teyrebaz33 please, sign commits so that we can merge :)

[weboko]

lgtm

[Copilot]

Pull request overview

Adds --json support to localnet logs to align diagnostic commands (localnet status, doctor) with machine-readable output expectations.

Changes:

• Added --json flag plumbing for localnet logs from CLI parsing through to command dispatch.
• Implemented structured JSON output for localnet logs, including tail and lines.
• Added CLI tests covering help output and JSON behavior for missing/tailed logs.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
src/cli.rs	Adds json: bool to LocalnetLogsArgs and forwards it into LocalnetAction::Logs.
src/commands/localnet.rs	Extends LocalnetAction::Logs with json and emits JSON output for logs.
tests/cli.rs	Adds tests validating --json appears in help and JSON output shape for missing log file and tailing.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

cmd_localnet_logs no longer distinguishes a missing log file from an empty one in text mode: when the file doesn’t exist it now flows through print_log_lines and prints “log file is empty”. This is misleading (and differs from read_log_tail, which reports missing files explicitly). Consider keeping the old “does not exist yet” message for non---json output, while still emitting an empty JSON payload for --json.

Suggested change

	return print_log_lines(tail, &[], log_path, json);
	if json {
	return print_log_lines(tail, &[], log_path, true);
	}
	println!("log file does not exist yet: {}", log_path.display());
	return Ok(());

[teyrebaz33]

Fixed in the latest commit: when the log file does not exist and --json is off, we now print the original 'log file does not exist yet' message. Only the JSON path goes through print_log_lines with an empty slice.

[Copilot]

The previous content.trim().is_empty() check was removed. As a result, files containing only whitespace/newlines will be treated as having log lines (possibly empty strings) and will be printed/emitted in JSON rather than reported as empty. If the intent is “no meaningful log output yet”, consider restoring a trim/blank check (and/or filtering empty lines) before building the lines array.

[teyrebaz33]

Fixed: added a blank-line filter before slicing the tail, so files containing only whitespace/newlines are treated as empty and reported accordingly.

[teyrebaz33]

All commits have been re-signed with GPG — they now show as Verified on GitHub.

[fryorcraken]

Thanks @teyrebaz33.

Please revise against the points below, then ping me for another pass. The repo's CONTRIBUTING.md is the contract these come from.

• Keep the PR to exactly the scope the linked issue asked for
• Treat the PR template and CONTRIBUTING.md as the bar, not boilerplate
• Read your own diff and run the repo's local checks before pushing
• Walk the DOGFOODING flow for diagnostics changes before asking for review

[fryorcraken]

First — apologies for the three-week silence on our side. That's on us, not you. Thanks for sticking with this and for already turning around two rounds of review feedback; the --json shape and CLI plumbing are both right where we want them.

A few small things to clean up before we land it:

Local checks before the next push:

cargo fmt              # currently fails: 3 spots in src/commands/localnet.rs and tests/cli.rs
cargo test --all-targets
git rebase origin/master   # branch picked up a conflict during the wait

The tail-handling refactor went a touch further than it needed to. Issue #38 only asked for --json, and the blank-line filter at src/commands/localnet.rs:354-358 answers the Copilot nit about whitespace-only files by also dropping blank lines mid-content. Two side effects:

1. --tail N now means "last N non-blank lines" instead of "last N lines of the file" — a quiet semantics change on the human path.
2. localnet logs --tail 0 on a non-empty file prints "log file is empty: …" because print_log_lines decides emptiness from the post-tail slice.

Easiest fix is the smaller one: restore the original if content.trim().is_empty() early-return that handles whitespace-only files, and keep the raw lines for slicing. The --tail 0 bug then falls out for free.

A few more tests would lock the edges down:

• whitespace-only log file → lines: [] / "log file is empty"
• non-empty file + --tail 0 → empty output, no "empty" message
• log file with embedded blank lines → preserved (or dropped — whichever you pick, assert it)

One note on goalposts that moved while you waited. Some of the contributor expectations (filling in Your Project or User Demand / User Story / Verification in the PR body, and updating DOGFOODING.md when user-facing behavior changes) were added to CONTRIBUTING.md after you opened this PR — so it's fair that you didn't follow them the first time. Since you'll be touching the PR anyway:

• Refill the PR body using .github/PULL_REQUEST_TEMPLATE.md.
• Add a --json line to DOGFOODING.md D2 with the expected JSON shape, mirroring how localnet status --json is treated in the same scenario.

Totally optional, not blocking: the new JSON payload uses serde_json::json!{...}; elsewhere in the file (e.g. cmd_localnet_status at localnet.rs:301-313) we emit via #[derive(Serialize)] structs from model.rs. A LocalnetLogsReport { tail, lines } type would match house style and give future fields a clean home — but happy to land it as-is and refactor later.

Thanks again for your patience — once the above is in we should be able to merge quickly this time.