fix(tools): scale a digit followed by a spelled magnitude in parse_spoken_number

[jeffrey701]

Summary

parse_spoken_number has a digit fast-path that returns a leading digit token immediately, silently dropping a spelled magnitude that follows it:

if let Some(Ok(value)) = tokens.get(start).map(|token| token.parse::<u64>()) {
    return Some((value, start + 1));   // "5 hundred" -> (5, 1), "hundred" lost
}

So parse_spoken_number(&["5", "hundred"], 0) returns (5, 1) instead of (500, 2). This produces a confidently-wrong calculator answer: route("what is 20 percent of 5 hundred") builds calc_number_starting_at from parse_spoken_number, gets 5, and emits "5 * 20 / 100" → 1, instead of "500 * 20 / 100" → 100. The duration path is affected the same way ("5 hundred seconds" → a 100s timer, not 500).

Only the mixed digit + spelled-magnitude form was broken — the fully-spelled ("five hundred" → 500) and fully-digit ("500") forms already work, and "five hundred" is exactly what "5 hundred" means. This seeds the accumulator from the leading digit when a magnitude follows, so the existing hundred/thousand loop scales it; a bare digit with no magnitude still returns immediately, unchanged.

(Note: parse_amount has a separate first-digit shortcut with the same blind spot on the temperature-setpoint path; that is a distinct function and left for a follow-up so this change stays focused on parse_spoken_number.)

Closes #837

Real Behavior Proof

• I have built and run the affected code locally (or noted why I could not).
• I have verified the change end-to-end on Jetson hardware.
• I have NOT verified on Jetson hardware, and I explain the equivalent verification path or validation gap below.

Tested profile / hardware (check all that apply):

• jetson
• raspberry_pi
• portable_sbc
• laptop
• mac
• CI-only / docs-only
• Not run locally

genie-core is a Linux-only crate, so it doesn't compile on my Windows dev box — CI runs the real cargo test on Linux, exercising the four added tests. Locally I reproduced parse_spoken_number (old vs new) plus cardinal_word in a rustc harness.

What I ran

rustc --edition 2021 pr_nw_test.rs && ./pr_nw_test

What I observed

PR-A OK: OLD '5 hundred'->5 (dropped magnitude); NEW ->500; no regression.

The harness asserts OLD ["5","hundred"] → (5,1) (the dropped magnitude), NEW → (500,2), ["3","thousand"] → (3000,2), ["2","thousand","five","hundred"] → (2500,4), ["5","hundred","and","twenty"] → (520,4), and no change for the bare-digit (["5","minutes"]→(5,1), ["42"]→(42,1)) and fully-spelled (["five","hundred"]→(500,2)) forms.

Summary by CodeRabbit

• Bug Fixes

  • Improved spoken-number parsing for expressions such as “5 hundred” and “3 thousand.”
  • Supports continued parsing across mixed digits and spelled-out magnitudes, including optional “and.”
  • Preserves existing behavior for standalone digits and fully spelled-out magnitudes.

• Tests

  • Added coverage for digit-magnitude scaling and mixed-format number expressions.

[coderabbitai]

📝 Walkthrough

Walkthrough

parse_spoken_number now scales a leading digit followed by hundred or thousand, while preserving bare-digit behavior. Tests cover mixed digit and magnitude inputs, optional “and”, and existing spelled-number parsing.

Changes

Spoken number parsing

Layer / File(s)	Summary
Scale digit-magnitude inputs crates/genie-core/src/tools/number_words.rs	parse_spoken_number continues parsing when a leading digit is followed by hundred or thousand, and tests cover scaling, continuation, bare digits, and spelled magnitudes.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related PRs

• GeniePod/genie-claw#825: Updates the same parser and adds routed calculator coverage for the mixed digit-and-magnitude case.

Suggested labels: bug

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly states the main fix: scaling a digit before a spelled magnitude in parse_spoken_number.
Linked Issues check	✅ Passed	The change addresses the mixed digit+magnitude bug in #837 and preserves the listed existing behaviors.
Out of Scope Changes check	✅ Passed	No obvious unrelated changes are indicated; the fix and tests stay within the linked parsing bug scope.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

🧹 Nitpick comments (1)

crates/genie-core/src/tools/number_words.rs (1)

125-150: 🗄️ Data Integrity & Integration | 🔵 Trivial | ⚡ Quick win

Add calculator and duration regression tests.

These tests validate parse_spoken_number directly, but not its downstream consumers in crates/genie-core/src/tools/calc_input.rs and crates/genie-core/src/tools/quick.rs. Add one regression case through each path to ensure "5 hundred" remains scaled after integration.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@crates/genie-core/src/tools/number_words.rs` around lines 125 - 150, Add
integration regression tests for the calculator path in calc_input.rs and the
duration path in quick.rs, each exercising the input “5 hundred” and asserting
it produces the scaled value 500. Keep the existing parse_spoken_number unit
tests unchanged and follow the established test helpers and assertions in each
consumer.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Nitpick comments:
In `@crates/genie-core/src/tools/number_words.rs`:
- Around line 125-150: Add integration regression tests for the calculator path
in calc_input.rs and the duration path in quick.rs, each exercising the input “5
hundred” and asserting it produces the scaled value 500. Keep the existing
parse_spoken_number unit tests unchanged and follow the established test helpers
and assertions in each consumer.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 47a636d0-19c9-4354-8e69-c62f23345970

📥 Commits

Reviewing files that changed from the base of the PR and between 7c67906 and 5144a9b.

📒 Files selected for processing (1)

• crates/genie-core/src/tools/number_words.rs

[jeffrey701]

Closing in favor of #825, which was opened a day earlier and covers the same parse_spoken_number digit-plus-magnitude fix (including the quick.rs routing coverage). Deferring to it to avoid a duplicate.