fix: code-review findings (bugs, robustness, refactors)#1
Merged
Conversation
Two correctness bugs found in code review: - app.rs: navigating Down in an empty lesson menu (e.g. Custom with no markdown files) evaluated `filtered_count - 1` on a usize, panicking in debug and wrapping to usize::MAX in release. Extracted next_menu_index() returning Option to guard the empty case. - analytics.rs: per-key timings pushed the absolute cumulative timestamp instead of the per-keystroke delta, so average key time grew with position in the lesson and skewed slow-key detection. Now records the interval from the previous input. Also removed a no-op checked_sub(Duration::from_secs(0)) in the bigram path. Adds regression tests for both.
Generation loops used `result.len() < length` (byte length) but truncated with `chars().take(length)` (char count). French content is multibyte (é, è, à, ç...), so accented lessons came out shorter than requested. Standardized all generators on chars().count(). Also extracts the shared drill-assembly loop from finger_generator's base and shift variants into assemble_drill(), and updates stale byte-length test assertions to char counts. Adds an accented-content regression test.
- save()/save_config() now write to a temp file and rename over the target, so an interrupted write can't truncate the existing file. - load()/load_config() back up a corrupt JSON file to .bak and return defaults instead of propagating an error that made the app unlaunchable. - Error messages now include the file path. - config_path() falls back to "." instead of expect(); HOME-unset error now explains how to fix it. Adds tests for atomic write and corrupt-file recovery; replaces the stale "corrupt config returns error" test with the new graceful behavior.
- render.rs: extract render_lesson_item() (collapses 5 duplicated blocks); remove 3 dead VisibleWindow fields; wrap_text counts chars not bytes. - scoring.rs: calculate_results() now calls calculate_accuracy() instead of duplicating the formula; drops the dead_code annotation. - types.rs: is_complete() uses the cached content_buffer_size instead of rescanning chars() on every keystroke. - custom.rs: warn on stderr when a custom-lesson directory can't be read. - ui/analytics.rs: clamp progress bar fill so >100% can't underflow; use !is_empty() over len() > 0. - stats.rs: add update_analytics accumulation regression test.
Test count 145 -> 178; dead-code annotations 29 -> 22 after removing the scoring duplicate and the unused VisibleWindow fields.
- types/analytics: accented content (déjà, café, ré bigram) through add_input, is_complete, and SessionAnalyzer. - adaptive: pin the exact identify_slow_keys result and add a boundary-tie case so an off-by-one in the percentile index is caught. - custom: missing closing ---, CRLF front matter, and FileTooLarge guard.
bigram_generator and trigram_generator duplicated the drill/word/mixed-mode loops verbatim (differing only in item type). Introduces a DrillItem trait (pattern/examples) implemented by Bigram and Trigram, and generic generate_drill_mode/word_mode/mixed_mode in a new ngram_generator module. Both generators now delegate, removing ~150 lines of copy-paste. No behavior change.
add_input, is_complete, and the per-frame render lookups used content.chars().nth()/count(), an O(n) walk that made a session O(n^2) as append_content grew the buffer. TypingSession now keeps a `chars: Vec<char>` in sync with `content` and exposes char_at() for O(1) indexing.
The LessonMenu draw closure cloned every filtered Lesson each frame, deep- copying LessonType::Custom content (up to 1MB). render_menu now takes &[&Lesson] and receives filtered_lessons() references directly.
…tion render.rs (1862 lines) is now a render/ directory module split by screen: session, menu, results, statistics, analytics_screen. mod.rs is a thin aggregator (shared imports + re-exports); the public API is unchanged. Extracts keys_for_level() — the mastery grouping/sorting previously buried in render_mastery_breakdown — as a pure function with a unit test.
- Test count 178 -> 186. - ui/render is now a directory module; note ngram_generator.rs.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Addresses findings from a whole-codebase review, then implements the suggested follow-up improvements. 11 commits, grouped by concern.
Verified:
cargo fmt --checkclean · 186 tests pass ·cargo clippy --all-targets -- -D warningsclean.Bug fixes
app.rs): navigating Down in an empty lesson menu (Custom with no files) evaluatedfiltered_count - 1on ausize— panicked in debug, wrapped in release. Now guarded vianext_menu_index().engine/analytics.rs): recorded the absolute cumulative timestamp instead of the per-keystroke delta, skewing slow-key detection. Now records the interval; removed a no-opchecked_sub.content/*_generator.rs): loops used byte length but truncated by char count, so accented French lessons came out short. Standardized onchars().count().ui/render): column budgeting was byte-based; accented words wrapped too early.Robustness
data/storage.rs):save/save_configwrite to a temp file then rename; corruptstats.json/config.jsonis backed up to.bakand defaults returned so the app always launches. Error messages now include the file path;config_pathno longer usesexpect().custom.rswarns on unreadable lesson directories instead of silently returning empty.Quality / refactors
content/ngram_generator.rs): removed ~150 lines of duplicated bigram/trigram drill logic.engine/types.rs): parallelVec<char>removes the per-keystroke O(n) walk.ui/render_menutakes&[&Lesson]): stopped deep-copying lesson content every frame.render/module by screen; extracted a pure, unit-testedkeys_for_level()aggregation.calculate_accuracy, clamp progress bar against >100%,is_completeuses cached count, removed deadVisibleWindowfields and 5 duplicated menu-item blocks.Tests added
Multibyte input (déjà/café/ré),
identify_slow_keysexact-set + boundary tie, custom parser edges (missing---, CRLF, FileTooLarge),update_analyticsaccumulation, storage atomic write + corrupt recovery, progress-bar overflow.Not changed (decisions, not code)
ParseError::InvalidFrontMatter— kept as a documented strategic placeholder.🤖 Generated with Claude Code