Release v0.8.0

[hitalin]

変更一覧（v0.7.0 → v0.8.0）

• notes キャッシュの実体/タイムライン所属を分離（feat(db): ノート実体とタイムライン所属の分離 (V6 note_timelines) #52, closes notes キャッシュの実体/タイムライン所属の分離（複数 TL 所属とキー不一致の解消） #30）— 同一ノートの複数 TL 所属と REST/streaming のキー不一致（孤児化）を解消するデータモデル再設計
• キャッシュ済みノートの述語注入スキャン API（feat(db): キャッシュ済みノートの述語注入スキャン API #51）
• はなみすきーの notes/hanamisearch-v1 に対応（feat: はなみすきーの notes/hanamisearch-v1 に対応 #50）
• ストリーミング接続ライフサイクルの状態遷移テスト（test: ストリーミング接続ライフサイクルの状態遷移を固める #49）
• エラー型の粒度を細分化（refactor: エラー型の粒度を細分化する #48）

破壊的変更

• DB スキーマ V6: notes_cache.timeline_type を除去し、所属を note_timelines テーブルへ分離。初回起動時に自動移行（既存のノート・FTS・アカウント・チャットは保持。streaming が壊れたキーで書いた孤児行のみ削除 — 元々どのタイムライン読み出しにも乗っていない行）
• TimelineType を廃止し TimelineKey に一本化（parse / as_canonical / api_endpoint / ws_channel）
• Database API: cache_notes / cache_note → ingest_notes、読み出し系は &TimelineKey を取る、delete_cached_note は account スコープ、get_cached_timeline_before に keyset cursor 引数、remove_membership / clear_timeline / sweep_orphan_notes 新設
• StreamingManager: subscribe_timeline / subscribe_antenna / subscribe_channel / subscribe_role → subscribe_notes(TimelineKey) に統合
• NoteDeckError::Api に api_code フィールド追加（refactor: エラー型の粒度を細分化する #48）
• daemon: GET /api/{host}/timeline/{tl_type} は Basic キーのみ受理（パラメータ付きキーは 400）、InvalidInput は 500 → 400

新機能・修正

• リスト TL の WS チャンネル名バグ修正（user-listTimeline → userList）、フォーク TL の kebab→lowerCamel 対応（vmimi-relay → vmimiRelayTimeline）
• polling モードで antenna / channel / role / user-list も更新されるように
• CLI: timeline antenna:{id} 等の prefix 付きキーに対応、notecli cache sweep サブコマンド新設
• eviction に per_timeline_limit（バケット単位のトリム。チャンク分割 tx で長時間 writer 占有を回避）

更新時の注意

• daemon を先に停止してから更新すること（稼働中の旧 daemon は実行時 "no such column" になる）
• V6 適用は 1 回だけ長い（1M 行で 1 分前後・warn ログあり）。一時的に DB サイズの最大 2 倍弱の空きディスクが必要。途中で中断しても DB は無傷（再起動で再試行）
• V6 適用済み DB は旧バージョンでは開けない。ダウングレードの可能性があるなら事前バックアップを推奨
• 同一 DB を共有する notetui / notebot は対応版（rev bump 済み）に更新すること

🤖 Generated with Claude Code

Summary by CodeRabbit

• New Features

  • Added support for expanded timeline types, including lists, antennas, channels, roles, clips, mentions, and favorites.
  • Added Hanami note search with date-based pagination.
  • Added a cache cleanup command for removing orphaned cached notes.
  • Added configurable per-timeline cache limits and improved timeline pagination.

• Bug Fixes

  • Improved authentication and API error reporting with stable error codes.
  • Invalid timeline requests now return clearer client errors.

• Improvements

  • Updated to version 0.8.0.

[coderabbitai]

📝 Walkthrough

Walkthrough

The package moves to version 0.8.0. It introduces typed timeline keys, separates cached note entities from timeline memberships, updates REST and streaming flows, adds structured errors, and provides cache orphan cleanup.

Changes

Timeline cache and API modernization

Layer / File(s)	Summary
Typed timeline contract and API integration src/models.rs, src/api.rs, src/http_server.rs, src/commands/notes.rs	TimelineKey replaces TimelineType. REST, WebSocket, CLI, Hanami search, and pagination logic use the typed key.
Separated cache entities and memberships migrations/V6__split_note_timeline_membership.sql, src/db.rs	The V6 migration adds note_timelines. Database ingestion, pagination, trimming, clearing, scanning, and orphan cleanup use separate entity and membership records.
Typed streaming subscriptions src/streaming.rs	Subscriptions store typed targets and derive channels, parameters, timeline keys, reconnect behavior, and note ingestion from those targets.
Structured authentication and API errors src/error.rs, src/lib.rs, src/api.rs	Authentication errors use typed variants and stable codes. API errors preserve optional server codes. Internal error details are sanitized before frontend output.
Cache sweep command and formatting updates src/cli.rs, src/commands/mod.rs, src/commands/auth.rs, src/commands/doctor.rs, src/main.rs, Cargo.toml	The CLI adds cache sweep for orphan-note cleanup. The package version changes to 0.8.0. Several existing expressions are reformatted without behavior changes.

Estimated code review effort: 5 (Critical) | ~120 minutes

Sequence Diagram(s)

sequenceDiagram
  participant CLI
  participant TimelineKey
  participant MisskeyClient
  participant StreamingManager
  participant Database
  CLI->>TimelineKey: parse timeline identifier
  TimelineKey->>MisskeyClient: provide endpoint and parameters
  MisskeyClient-->>Database: ingest REST notes with TimelineKey
  StreamingManager->>TimelineKey: derive typed subscription key
  StreamingManager-->>Database: ingest streaming notes with TimelineKey
  Database-->>CLI: return timeline data or sweep count

Loading

Possibly related PRs

• notedeck-dev/notecli#47: Both PRs update the package version as release bumps.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Out of Scope Changes check	⚠️ Warning	Hanami search, authentication error restructuring, and other release-wide features are not covered by issue #30.	Move unrelated features to separate issues or link additional issues that define their scope.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title identifies the v0.8.0 release, which matches the package version change and the stated release objective.
Linked Issues check	✅ Passed	The changes separate note entities from memberships, adopt TimelineKey, unify ingest, and add membership-level trimming and orphan cleanup required by issue #30.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch develop

---

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

[coderabbitai]

Actionable comments posted: 6

🧹 Nitpick comments (1)

src/http_server.rs (1)

143-158: 🔒 Security & Privacy | 🔵 Trivial | ⚡ Quick win

Use safe_message() for the outward error body.

The new status mapping documents why e.to_string() is safe for InvalidInput. The same message field still carries e.to_string() for every other variant, including Api and Keychain, whose Display output comes from upstream text. The coding guidelines require safe_message() for error messages so a token cannot reach a log or a response.

🛡️ Proposed change

         Self {
             status,
             code,
-            message: e.to_string(),
+            message: e.safe_message(),
         }

As per coding guidelines: 「API トークンをログやエラーメッセージに含めず、safe_message() を使用する。」

🤖 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 `@src/http_server.rs` around lines 143 - 158, Update the From<NoteDeckError>
for ApiError implementation to populate message using e.safe_message() instead
of e.to_string(), while preserving the existing status and code mapping.

Source: Coding guidelines

🤖 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.

Inline comments:
In `@src/api.rs`:
- Around line 909-945: Update the shared search dispatch used by
commands/notes.rs::run_search and http_server.rs::search_notes so
Hanami/Misskey-compatible servers call ApiClient::search_notes_hanami instead of
search_notes. Reuse the server-software detection mechanism already used by both
callers, preserving the existing search_notes path for other software and
ensuring the selection is centralized rather than duplicated.

In `@src/commands/mod.rs`:
- Around line 44-52: Update the cache sweep branch in run_cli to honor the fmt
OutputFormat parameter: retain the existing human-readable message for the
default format, and emit a structured deletion-count record for
OutputFormat::Json and OutputFormat::Jsonl. Add CLI coverage verifying both
default output and JSON output, including the deleted count.

In `@src/db.rs`:
- Around line 1041-1074: Refactor cleanup_with_eviction and
trim_timelines_chunked so the helper accepts &self rather than a held
&Connection, and acquire the writer lock separately inside each chunk iteration.
Remove the long-lived lock_write guard from cleanup_with_eviction while
preserving TTL handling and deletion totals, ensuring the mutex is released
after every chunk so ingest_notes and other writers can interleave.

In `@src/error.rs`:
- Around line 56-57: Rename the canonical application error enum in error.rs
from NoteDeckError to NotecliError, then migrate all callers, imports, return
types, and references to use NotecliError consistently as the public error
boundary.
- Around line 71-73: Update the HTTP error adapter in http_server.rs,
specifically the response construction around the message field, to use
NoteDeckError::safe_message() instead of e.to_string(). Preserve the existing
status and response handling while ensuring Internal errors return the sanitized
message.

In `@src/models.rs`:
- Around line 513-559: Update TimelineKey::parse so the fallback Basic branch
accepts only non-empty names composed of lowercase ASCII letters, digits, and
hyphens. Reject any other character with NoteDeckError::InvalidInput before
constructing Basic, while preserving the existing prefixed-key parsing and
reserved-name handling.

---

Nitpick comments:
In `@src/http_server.rs`:
- Around line 143-158: Update the From<NoteDeckError> for ApiError
implementation to populate message using e.safe_message() instead of
e.to_string(), while preserving the existing status and code mapping.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: b5968b5e-29ee-4db2-a950-f6e26a6cdeca

📥 Commits

Reviewing files that changed from the base of the PR and between e235701 and 114850d.

⛔ Files ignored due to path filters (1)

• Cargo.lock is excluded by !**/*.lock

📒 Files selected for processing (15)

• Cargo.toml
• migrations/V6__split_note_timeline_membership.sql
• src/api.rs
• src/cli.rs
• src/commands/auth.rs
• src/commands/doctor.rs
• src/commands/mod.rs
• src/commands/notes.rs
• src/db.rs
• src/error.rs
• src/http_server.rs
• src/lib.rs
• src/main.rs
• src/models.rs
• src/streaming.rs

[coderabbitai]

📐 Maintainability & Code Quality | 🟠 Major | ⚡ Quick win

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# Description: Find callers of search_notes_hanami and how the endpoint is selected.
set -euo pipefail

rg -nP -C5 '\bsearch_notes_hanami\s*\(' --type=rust
# Check for fork/software detection near search dispatch
rg -nP -C4 'hanami' --type=rust

Repository: notedeck-dev/notecli

Length of output: 158

---

🏁 Script executed:

set -euo pipefail
printf '%s\n' '--- matching files ---'
git ls-files | rg '(^|/)(api|notes)\.rs$|src/commands'

printf '%s\n' '--- search_notes_hanami references ---'
rg -n -C3 'search_notes_hanami|search_notes\s*\(' --glob '*.rs' .

printf '%s\n' '--- endpoint and software detection references ---'
rg -n -C4 'notes/search|hanami|software|version|instance' --glob '*.rs' src

Repository: notedeck-dev/notecli

Length of output: 41397

---

🏁 Script executed:

set -euo pipefail
printf '%s\n' '--- search dispatch ---'
sed -n '1,105p' src/commands/notes.rs
sed -n '490,525p' src/http_server.rs

printf '%s\n' '--- detection consumers and software branches ---'
rg -n -C3 'ServerDetection|get_server_detection|load_server_detections|software_name|software_repository|account\.software|\.software' src --glob '*.rs'

printf '%s\n' '--- direct-call verifier ---'
python3 - <<'PY'
from pathlib import Path
import re

files = [p for p in Path("src").rglob("*.rs")]
hanami = []
normal = []
software_branches = []
for p in files:
    text = p.read_text()
    for m in re.finditer(r'\bsearch_notes_hanami\s*\(', text):
        line = text.count("\n", 0, m.start()) + 1
        hanami.append((str(p), line, text.splitlines()[line-1].strip()))
    for m in re.finditer(r'\.search_notes\s*\(', text):
        line = text.count("\n", 0, m.start()) + 1
        normal.append((str(p), line, text.splitlines()[line-1].strip()))
    for i, line in enumerate(text.splitlines(), 1):
        if re.search(r'(?i)hanami', line) or re.search(r'software_(name|repository)|account\.software', line):
            software_branches.append((str(p), i, line.strip()))

print("search_notes_hanami references:")
for row in hanami:
    print(row)
print("search_notes call sites:")
for row in normal:
    print(row)
print("Hanami/software-related lines:")
for row in software_branches:
    print(row)
PY

Repository: notedeck-dev/notecli

Length of output: 25051

---

Route Hanami searches to search_notes_hanami.

src/commands/notes.rs::run_search and src/http_server.rs::search_notes always call search_notes. No production caller selects search_notes_hanami; Hanami instances still use the disabled notes/search endpoint. Add shared server-software dispatch.

🤖 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 `@src/api.rs` around lines 909 - 945, Update the shared search dispatch used by
commands/notes.rs::run_search and http_server.rs::search_notes so
Hanami/Misskey-compatible servers call ApiClient::search_notes_hanami instead of
search_notes. Reuse the server-software detection mechanism already used by both
callers, preserving the existing search_notes path for other software and
ensuring the selection is centralized rather than duplicated.

[coderabbitai]

🎯 Functional Correctness | 🟠 Major | ⚡ Quick win

Honor OutputFormat for cache sweep.

run_cli receives fmt, but this branch always prints plain text. When the caller selects OutputFormat::Json or OutputFormat::Jsonl, the command returns invalid machine-readable output. Match fmt and emit a structured deletion count for JSON modes. Add a CLI test for default and JSON output.

Suggested fix

-                    println!("Removed {deleted} orphan note(s) from cache");
+                    match fmt {
+                        OutputFormat::Json | OutputFormat::Jsonl => {
+                            println!(r#"{{"deleted":{deleted}}}"#);
+                        }
+                        _ => println!("Removed {deleted} orphan note(s) from cache"),
+                    }

🤖 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 `@src/commands/mod.rs` around lines 44 - 52, Update the cache sweep branch in
run_cli to honor the fmt OutputFormat parameter: retain the existing
human-readable message for the default format, and emit a structured
deletion-count record for OutputFormat::Json and OutputFormat::Jsonl. Add CLI
coverage verifying both default output and JSON output, including the deleted
count.

[coderabbitai]

🚀 Performance & Scalability | 🟠 Major | 🏗️ Heavy lift

Chunked trimming still blocks every in-process writer for the whole run.

cleanup_with_eviction acquires the writer MutexGuard at line 1052 and holds it until the function returns. trim_timelines_chunked receives that &Connection and only splits the SQLite transaction. The Mutex is never released between chunks, so ingest_notes and every other lock_write caller in this process waits for the complete trim, which the comments describe as taking minutes on a first run with 1M rows.

Pass &self into the trim helper and take the writer lock once per chunk. Each chunk then yields the mutex and lets streaming ingestion interleave.

♻️ Proposed restructure

-        let conn = self.lock_write()?;
         let mut total_deleted: u64 = 0;
 
         // ① TTL (単一 tx)
         if let Some(ttl_days) = config.ttl_days {
+            let conn = self.lock_write()?;
             ...
         }
 
         // ② per-timeline トリム (チャンク分割 tx)
         if let Some(per_timeline_limit) = config.per_timeline_limit {
-            total_deleted += self.trim_timelines_chunked(&conn, per_timeline_limit)?;
+            total_deleted += self.trim_timelines_chunked(per_timeline_limit)?;
         }
 
         // ③ per-account hard cap (単一 tx)
         if let Some(per_account_limit) = config.per_account_limit {
+            let conn = self.lock_write()?;
             let tx = conn.unchecked_transaction()?;

-    fn trim_timelines_chunked(
-        &self,
-        conn: &Connection,
-        per_timeline_limit: i64,
-    ) -> Result<u64, NoteDeckError> {
+    fn trim_timelines_chunked(&self, per_timeline_limit: i64) -> Result<u64, NoteDeckError> {
         let mut total: u64 = 0;
         loop {
+            // チャンクごとに writer lock を取り直し、WS ingest を割り込ませる
+            let conn = self.lock_write()?;
             let tx = conn.unchecked_transaction()?;

📝 Committable suggestion

‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.

Suggested change

	pub fn cleanup_with_eviction(&self, config: &EvictionConfig) -> Result<u64, NoteDeckError> {
	// どちらも無効なら早期 return (lock も取らない)。
	if config.per_account_limit.is_none() && config.ttl_days.is_none() {
	// 全フィールド無効なら早期 return (lock も取らない)。
	// per_timeline_limit を含む 3 フィールド判定であること (2 フィールド判定だと
	// per-timeline のみ設定時にトリムが走らない)。
	if config.per_account_limit.is_none()
	&& config.ttl_days.is_none()
	&& config.per_timeline_limit.is_none()
	{
	return Ok(0);
	}
	let conn = self.lock_write()?;
	let tx = conn.unchecked_transaction()?;
	let mut total_deleted: u64 = 0;
	// ① TTL (単一 tx)
	if let Some(ttl_days) = config.ttl_days {
	let now = std::time::SystemTime::now()
	.duration_since(std::time::UNIX_EPOCH)
	.unwrap_or_default()
	.as_secs() as i64;
	let ttl_cutoff = now - ttl_days * 86_400;
	let tx = conn.unchecked_transaction()?;
	let n = tx.execute(
	"DELETE FROM notes_cache WHERE cached_at < ?1",
	params![ttl_cutoff],
	)?;
	tx.commit()?;
	total_deleted += n as u64;
	}
	// ② per-timeline トリム (チャンク分割 tx)
	if let Some(per_timeline_limit) = config.per_timeline_limit {
	total_deleted += self.trim_timelines_chunked(&conn, per_timeline_limit)?;
	}
	pub fn cleanup_with_eviction(&self, config: &EvictionConfig) -> Result<u64, NoteDeckError> {
	// 全フィールド無効なら早期 return (lock も取らない)。
	// per_timeline_limit を含む 3 フィールド判定であること (2 フィールド判定だと
	// per-timeline のみ設定時にトリムが走らない)。
	if config.per_account_limit.is_none()
	&& config.ttl_days.is_none()
	&& config.per_timeline_limit.is_none()
	{
	return Ok(0);
	}
	let mut total_deleted: u64 = 0;
	// ① TTL (単一 tx)
	if let Some(ttl_days) = config.ttl_days {
	let conn = self.lock_write()?;
	let now = std::time::SystemTime::now()
	.duration_since(std::time::UNIX_EPOCH)
	.unwrap_or_default()
	.as_secs() as i64;
	let ttl_cutoff = now - ttl_days * 86_400;
	let tx = conn.unchecked_transaction()?;
	let n = tx.execute(
	"DELETE FROM notes_cache WHERE cached_at < ?1",
	params![ttl_cutoff],
	)?;
	tx.commit()?;
	total_deleted += n as u64;
	}
	// ② per-timeline トリム (チャンク分割 tx)
	if let Some(per_timeline_limit) = config.per_timeline_limit {
	total_deleted += self.trim_timelines_chunked(per_timeline_limit)?;
	}

🤖 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 `@src/db.rs` around lines 1041 - 1074, Refactor cleanup_with_eviction and
trim_timelines_chunked so the helper accepts &self rather than a held
&Connection, and acquire the writer lock separately inside each chunk iteration.
Remove the long-lived lock_write guard from cleanup_with_eviction while
preserving TTL handling and deletion totals, ensuring the mutex is released
after every chunk so ingest_notes and other writers can interleave.

[coderabbitai]

📐 Maintainability & Code Quality | 🟠 Major | 🏗️ Heavy lift

Use NotecliError as the canonical public error type.

Lines 56-57 continue to expose NoteDeckError as the application error boundary. Rename the canonical enum in src/error.rs to NotecliError and migrate its callers in this release.

As per coding guidelines, error types must be unified in error.rs as NotecliError.

🤖 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 `@src/error.rs` around lines 56 - 57, Rename the canonical application error
enum in error.rs from NoteDeckError to NotecliError, then migrate all callers,
imports, return types, and references to use NotecliError consistently as the
public error boundary.

Source: Coding guidelines

[coderabbitai]

🔒 Security & Privacy | 🟠 Major | ⚡ Quick win

Use safe_message() in the HTTP error adapter.

src/http_server.rs Lines 143-159 assigns message: e.to_string(). If NoteDeckError::Internal reaches an HTTP route, the response exposes its internal detail despite Lines 137-140 defining a sanitized message. Replace that conversion with e.safe_message().

Proposed fix

-            message: e.to_string(),
+            message: e.safe_message(),

As per coding guidelines, API error messages must use safe_message() and must not expose API tokens.

🤖 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 `@src/error.rs` around lines 71 - 73, Update the HTTP error adapter in
http_server.rs, specifically the response construction around the message field,
to use NoteDeckError::safe_message() instead of e.to_string(). Preserve the
existing status and response handling while ensuring Internal errors return the
sanitized message.

Source: Coding guidelines

[coderabbitai]

🔒 Security & Privacy | 🟠 Major | 🏗️ Heavy lift

Restrict the allowed characters for Basic timeline names.

parse rejects only empty input, over-length input, control characters, and unknown prefixes. Every other byte sequence becomes Basic(String). api_endpoint then interpolates that string into an API path: Cow::Owned(format!("notes/{other}-timeline")), and MisskeyClient::api_url inserts the result into {scheme}://{host}/api/{endpoint}.

A caller-supplied name that contains /, ., ?, or # therefore changes the effective request target. For example, notes/../../admin/x? resolves to /api/admin/x with -timeline pushed into the query string. src/http_server.rs get_timeline accepts the path segment and only checks matches!(key, TimelineKey::Basic(_)), so this input passes the allowlist and reaches the endpoint mapping with the account token attached.

Add a charset check in parse so Basic accepts only the shape the endpoint and channel mapping assume (lowercase letters, digits, and -).

🛡️ Proposed validation for `Basic` names

             match s {
                 "mentions" => Ok(Self::Mentions),
                 "specified" => Ok(Self::Specified),
                 "favorites" => Ok(Self::Favorites),
                 _ if RESERVED_PREFIXES.contains(&s) => Err(NoteDeckError::InvalidInput(format!(
                     "bare reserved timeline key '{s}' (id required)"
                 ))),
-                _ => Ok(Self::Basic(s.to_string())),
+                _ if s
+                    .bytes()
+                    .all(|b| b.is_ascii_lowercase() || b.is_ascii_digit() || b == b'-') =>
+                {
+                    Ok(Self::Basic(s.to_string()))
+                }
+                _ => Err(NoteDeckError::InvalidInput(format!(
+                    "invalid basic timeline key '{s}'"
+                ))),
             }

📝 Committable suggestion

‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.

Suggested change

	pub fn parse(s: &str) -> Result<Self, crate::error::NoteDeckError> {
	use crate::error::NoteDeckError;
	if s.is_empty() {
	return Err(NoteDeckError::InvalidInput(
	"timeline key must not be empty".to_string(),
	));
	}
	if s.len() > 256 {
	return Err(NoteDeckError::InvalidInput(
	"timeline key exceeds 256 bytes".to_string(),
	));
	}
	if s.bytes().any(|b| b < 0x20 || b == 0x7F) {
	return Err(NoteDeckError::InvalidInput(
	"timeline key contains control characters".to_string(),
	));
	}
	if let Some((prefix, id)) = s.split_once(':') {
	if id.is_empty() {
	return Err(NoteDeckError::InvalidInput(format!(
	"timeline key '{prefix}:' has empty id"
	)));
	}
	let id = id.to_string();
	match prefix {
	"user-list" => Ok(Self::UserList { list_id: id }),
	"antenna" => Ok(Self::Antenna { antenna_id: id }),
	"channel" => Ok(Self::Channel { channel_id: id }),
	"role" => Ok(Self::Role { role_id: id }),
	"clip" => Ok(Self::Clip { clip_id: id }),
	"user" => Ok(Self::UserNotes { user_id: id }),
	_ => Err(NoteDeckError::InvalidInput(format!(
	"unknown timeline key prefix '{prefix}'"
	))),
	}
	} else {
	match s {
	"mentions" => Ok(Self::Mentions),
	"specified" => Ok(Self::Specified),
	"favorites" => Ok(Self::Favorites),
	_ if RESERVED_PREFIXES.contains(&s) => Err(NoteDeckError::InvalidInput(format!(
	"bare reserved timeline key '{s}' (id required)"
	))),
	_ => Ok(Self::Basic(s.to_string())),
	}
	}
	}
	pub fn parse(s: &str) -> Result<Self, crate::error::NoteDeckError> {
	use crate::error::NoteDeckError;
	if s.is_empty() {
	return Err(NoteDeckError::InvalidInput(
	"timeline key must not be empty".to_string(),
	));
	}
	if s.len() > 256 {
	return Err(NoteDeckError::InvalidInput(
	"timeline key exceeds 256 bytes".to_string(),
	));
	}
	if s.bytes().any(|b| b < 0x20 || b == 0x7F) {
	return Err(NoteDeckError::InvalidInput(
	"timeline key contains control characters".to_string(),
	));
	}
	if let Some((prefix, id)) = s.split_once(':') {
	if id.is_empty() {
	return Err(NoteDeckError::InvalidInput(
	format!("timeline key '{prefix}:' has empty id")
	));
	}
	let id = id.to_string();
	match prefix {
	"user-list" => Ok(Self::UserList { list_id: id }),
	"antenna" => Ok(Self::Antenna { antenna_id: id }),
	"channel" => Ok(Self::Channel { channel_id: id }),
	"role" => Ok(Self::Role { role_id: id }),
	"clip" => Ok(Self::Clip { clip_id: id }),
	"user" => Ok(Self::UserNotes { user_id: id }),
	_ => Err(NoteDeckError::InvalidInput(format!(
	"unknown timeline key prefix '{prefix}'"
	))),
	}
	} else {
	match s {
	"mentions" => Ok(Self::Mentions),
	"specified" => Ok(Self::Specified),
	"favorites" => Ok(Self::Favorites),
	_ if RESERVED_PREFIXES.contains(&s) => Err(NoteDeckError::InvalidInput(format!(
	"bare reserved timeline key '{s}' (id required)"
	))),
	_ if s
	.bytes()
	.all(|b| b.is_ascii_lowercase() || b.is_ascii_digit() || b == b'-') =>
	{
	Ok(Self::Basic(s.to_string()))
	}
	_ => Err(NoteDeckError::InvalidInput(format!(
	"invalid basic timeline key '{s}'"
	))),
	}
	}
	}

🤖 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 `@src/models.rs` around lines 513 - 559, Update TimelineKey::parse so the
fallback Basic branch accepts only non-empty names composed of lowercase ASCII
letters, digits, and hyphens. Reject any other character with
NoteDeckError::InvalidInput before constructing Basic, while preserving the
existing prefixed-key parsing and reserved-name handling.