feat(search): add --snippet-max-length flag

[IlyaGusev]

Summary

• Add --snippet-max-length to keenable search. It maps to the API's snippet_max_length body param (accepted range 180-10000; the API validates and returns a structured error with exit 1 on out-of-range values).
• The param rides in the JSON body, so both the daemon and direct HTTP paths get it with no daemon changes.

Tests

• Verified live: cap 180 gives ~186-char average snippets; cap 5000 gives ~4850. Out-of-range 50 exits 1 with Invalid parameter.
• New e2e tests: test_snippet_max_length (loose comparative bounds, since caps overshoot ~10% per-fragment) and test_snippet_max_length_out_of_range.
• Full test_search.py e2e file passes against the live API (25 passed).
• cargo clippy and release build pass.

🤖 Generated with Claude Code

[qodo-code-review]

PR Summary by Qodo

Add --snippet-max-length flag to search CLI and pass through to API

✨ Enhancement 🧪 Tests 📝 Documentation 🕐 10-20 Minutes

AI Description

• Add --snippet-max-length option to keenable search CLI.
• Include snippet_max_length in the search request JSON body when provided.
• Add e2e coverage for in-range behavior and out-of-range API error handling.

Diagram

graph TD
  A["keenable CLI"] --> B["src/main.rs (clap)"] --> C["src/commands/search.rs"] --> D{{"Search API"}}
  E["README.md"] --> B
  F["tests/e2e/test_search.py"] --> A

Loading

High-Level Assessment

The following are alternative approaches to this PR:

1. Client-side range validation (clap value parser)

• ➕ Fail fast without an API call for clearly invalid values
• ➕ More precise, immediate CLI error messaging
• ➕ Reduces noise in API error logs for obvious mistakes
• ➖ Duplicates validation rules already enforced server-side
• ➖ Needs careful alignment if the API range changes

2. Accept human-friendly units (e.g., 2k/5k) and normalize

• ➕ Improves UX for common values
• ➕ Still sends the same numeric API parameter
• ➖ Adds parsing complexity and more edge cases
• ➖ Potential ambiguity/locale concerns; not requested by API

Recommendation: The current pass-through approach is appropriate because the API already validates and returns a structured error. Consider adding optional client-side range validation via clap’s value parser as a follow-up to improve UX and avoid unnecessary requests, but it’s not required for correctness.

Files changed (4) +42 / -3

Enhancement (2) +19 / -3

search.rsForward snippet_max_length into search JSON body +4/-0

Forward snippet_max_length into search JSON body

• Extends the search command handler to accept an optional 'snippet_max_length' parameter and conditionally include it as 'snippet_max_length' in the request JSON body.

src/commands/search.rs

main.rsAdd --snippet-max-length flag to CLI and wire into search() +15/-3

Add --snippet-max-length flag to CLI and wire into search()

• Introduces a new '--snippet-max-length' clap argument on the Search subcommand, updates help examples, and passes the parsed value through to 'commands::search::search'.

src/main.rs

Tests (1) +22 / -0

test_search.pyAdd e2e tests for snippet length cap and out-of-range error +22/-0

Add e2e tests for snippet length cap and out-of-range error

• Adds an e2e test that compares average snippet length for low vs high caps, and a second test that asserts the API returns exit code 1 with a structured 'Invalid parameter' error for out-of-range values.

tests/e2e/test_search.py

Documentation (1) +1 / -0

README.mdDocument --snippet-max-length in search examples +1/-0

Document --snippet-max-length in search examples

• Adds a CLI usage example showing '--snippet-max-length 2000' and notes the accepted range (180-10000).

README.md

[qodo-code-review]

Code Review by Qodo

🐞 Bugs (1) 📘 Rule violations (0) 📜 Skill insights (0)

1. Snippet test divide-by-zero 🐞 Bug ☼ Reliability

Description

test_snippet_max_length computes an average as sum(lens)/len(lens) without guarding against an empty
results list, which can raise ZeroDivisionError and fail the e2e suite with an unhandled exception.

Code

tests/e2e/test_search.py[R50-53]

+def test_snippet_max_length(kn):
+    def avg_snippet(data):
+        lens = [len(r.get("snippet") or "") for r in results_of(data)]
+        return sum(lens) / len(lens)

Evidence

The new test divides by the number of results without checking for zero. Separately, results_of()
only asserts that results is a list (it may be empty), and elsewhere in the same test file the
suite explicitly skips tests when results are empty—showing empty results are a handled/expected
possibility in this e2e setup.

tests/e2e/test_search.py[50-53]
tests/e2e/conftest.py[155-158]
tests/e2e/test_search.py[108-112]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`test_snippet_max_length` calculates an average snippet length by dividing by `len(lens)`. If the API returns zero results (an empty `results` list), this becomes a division by zero and the test errors out (not a clean assertion failure / skip).

## Issue Context
The e2e suite already has a pattern for handling vacuous assertions when search returns no results (`_non_empty_results` uses `pytest.skip`). The new snippet-length test should follow a similar approach or at least assert non-empty results before averaging.

## Fix Focus Areas
- tests/e2e/test_search.py[50-61]

### Suggested change
Update `avg_snippet()` to either:
- `assert results_of(data), "no results returned"` before computing the average, or
- `pytest.skip(...)` when results are empty (preferred if empty results are acceptable for live API variability).

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

Tip of the day

💡 Did you know, you can enable the Remediation agent and Qodo fixes findings in a dedicated fix PR

More tips ↗ | Customize Qodo ↗ | Qodo docs ↗

[qodo-code-review]

1. Snippet test divide-by-zero 🐞 Bug ☼ Reliability

test_snippet_max_length computes an average as sum(lens)/len(lens) without guarding against an empty
results list, which can raise ZeroDivisionError and fail the e2e suite with an unhandled exception.

Agent Prompt

## Issue description
`test_snippet_max_length` calculates an average snippet length by dividing by `len(lens)`. If the API returns zero results (an empty `results` list), this becomes a division by zero and the test errors out (not a clean assertion failure / skip).

## Issue Context
The e2e suite already has a pattern for handling vacuous assertions when search returns no results (`_non_empty_results` uses `pytest.skip`). The new snippet-length test should follow a similar approach or at least assert non-empty results before averaging.

## Fix Focus Areas
- tests/e2e/test_search.py[50-61]

### Suggested change
Update `avg_snippet()` to either:
- `assert results_of(data), "no results returned"` before computing the average, or
- `pytest.skip(...)` when results are empty (preferred if empty results are acceptable for live API variability).

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools