feat(mobile): add LLM smart shuffle#271
Draft
xiongzikun0106 wants to merge 8 commits into
Draft
Conversation
Made-with: Cursor
Made-with: Cursor
Made-with: Cursor
SafeDep Report Summary
Package Details
This report is generated by SafeDep Github App |
|
@xiongzikun0106 is attempting to deploy a commit to the yanyao2333's projects Team on Vercel. A member of the Team first needs to authorize it. |
|
Important Review skippedDraft detected. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Note Reviews pausedUse the following commands to manage reviews:
Use the checkboxes below for quick actions:
Walkthrough该PR引入了一个LLM驱动的"智能随机播放"功能，用于音乐播放列表。包括数据库表用于存储LLM派生的轨道标签、配置模态框、智能队列生成服务、以及与现有同步管道的集成。 Changes
Sequence Diagram(s)sequenceDiagram
participant User as User
participant PlaylistUI as Playlist UI
participant SmartShuffleService as SmartShuffle Service
participant LLMApi as LLM API
participant Database as Database
participant QueueManager as Queue Manager
User->>PlaylistUI: Click Smart Shuffle Button
PlaylistUI->>SmartShuffleService: createSmartQueue(tracks, prompt)
SmartShuffleService->>Database: Fetch stored LLM tags for tracks
Database-->>SmartShuffleService: Track tags (confidence, categories)
alt Tags missing locally
SmartShuffleService->>SmartShuffleService: Generate tags from track metadata
end
SmartShuffleService->>SmartShuffleService: Score tracks based on preference
SmartShuffleService->>SmartShuffleService: Apply time bias & exploration level
SmartShuffleService->>SmartShuffleService: Stable shuffle & sort by score
SmartShuffleService-->>PlaylistUI: Reordered track list
PlaylistUI->>QueueManager: Enqueue smart queue
QueueManager-->>User: Start playback with smart order
sequenceDiagram
participant SyncProcess as Sync Process
participant Database as Database
participant SmartShuffleService as SmartShuffle Service
participant LLMApi as LLM API
participant TrackLlmTagsTable as track_llm_tags Table
SyncProcess->>Database: Sync playlist tracks
Database-->>SyncProcess: Transaction committed
SyncProcess->>SmartShuffleService: indexPlaylistTracks(playlistId, sourceInfo)
SmartShuffleService->>Database: Fetch all playlist tracks
Database-->>SmartShuffleService: Track list
SmartShuffleService->>SmartShuffleService: Batch tracks into LLM requests
SmartShuffleService->>LLMApi: Extract tags (language, genre, mood, etc.)
LLMApi-->>SmartShuffleService: Tag arrays with confidence
SmartShuffleService->>SmartShuffleService: Normalize tags & build indices
SmartShuffleService->>TrackLlmTagsTable: Upsert tag records with metadata
TrackLlmTagsTable-->>SmartShuffleService: Insert/Update complete
SmartShuffleService-->>SyncProcess: Indexing complete
SyncProcess->>SyncProcess: Emit progress update
Estimated code review effort🎯 4 (Complex) | ⏱️ ~50 minutes Possibly related PRs
Suggested labels
Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Collaborator
|
@CodeRabbit pause |
|
Warning Review the following alerts detected in dependencies. According to your organization's Security Policy, it is recommended to resolve "Warn" alerts. Learn more about Socket for GitHub.
|
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 9
🧹 Nitpick comments (4)
apps/mobile/drizzle/0019_smart_shuffle_tags.sql (1)
4-4: 建议在库层补充confidence范围约束。当前仅靠服务层 clamp，建议增加
CHECK (confidence >= 0 AND confidence <= 1)，避免旁路写入导致脏数据。🧱 建议修改
- `confidence` real DEFAULT 0 NOT NULL, + `confidence` real DEFAULT 0 NOT NULL CHECK (`confidence` >= 0 AND `confidence` <= 1),🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@apps/mobile/drizzle/0019_smart_shuffle_tags.sql` at line 4, Add a DB-level check constraint to enforce confidence is within [0,1] for the confidence column; update the column definition in the migration (the `confidence` real DEFAULT 0 NOT NULL` line) to include a CHECK (confidence >= 0 AND confidence <= 1) or add a named CONSTRAINT (e.g., CONSTRAINT check_confidence_range CHECK (confidence >= 0 AND confidence <= 1)) so out-of-band writes cannot insert values outside 0..1.apps/mobile/src/components/ModalRegistry.tsx (1)
68-73: 新增懒加载导入建议改为@/*别名。这两处新增
lazy(import(...))仍使用相对路径，建议与移动端导入规范统一，避免后续规则冲突。♻️ 建议修改
const LlmSmartShuffleSettingsModal = lazy( - () => import('./modals/settings/LlmSmartShuffleSettingsModal'), + () => import('@/components/modals/settings/LlmSmartShuffleSettingsModal'), ) const SmartShuffleModal = lazy( - () => import('./modals/playlist/SmartShuffleModal'), + () => import('@/components/modals/playlist/SmartShuffleModal'), )As per coding guidelines,
apps/mobile/src/**/*.{ts,tsx}: Use import aliases@/*for all imports in the mobile app instead of relative paths.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@apps/mobile/src/components/ModalRegistry.tsx` around lines 68 - 73, Replace the relative import paths in the lazy calls for LlmSmartShuffleSettingsModal and SmartShuffleModal so they use the app alias import style (e.g., '@/...') instead of './modals/...'; update the two lazy(() => import(...)) expressions for the symbols LlmSmartShuffleSettingsModal and SmartShuffleModal to import via the alias path that matches the mobile app convention.apps/mobile/src/features/playlist/local/components/LocalPlaylistHeader.tsx (1)
33-33: 建议将智能随机按钮限定在本地歌单，并将回调设为可选。现在是无条件渲染 + 必填回调，会让非本地场景也被迫接入该行为。建议按
playlist.type === 'local'控制展示，避免组件 API 过度耦合。♻️ 建议修改
interface PlaylistHeaderProps { onClickPlayAll: () => void - onClickSmartShuffle: () => void + onClickSmartShuffle?: () => void onClickSync: () => void onClickCopyToLocalPlaylist: () => void } ... - <IconButton - mode='contained' - icon='shuffle-variant' - size={20} - onPress={onClickSmartShuffle} - testID='playlist-smart-shuffle' - /> + {playlist.type === 'local' && onClickSmartShuffle && ( + <IconButton + mode='contained' + icon='shuffle-variant' + size={20} + onPress={onClickSmartShuffle} + testID='playlist-smart-shuffle' + /> + )}Also applies to: 92-93, 358-364
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@apps/mobile/src/features/playlist/local/components/LocalPlaylistHeader.tsx` at line 33, The LocalPlaylistHeader prop onClickSmartShuffle is currently required and the smart-shuffle UI is always rendered; make the prop optional (onClickSmartShuffle?: () => void) and only render the smart-shuffle button when playlist.type === 'local' and onClickSmartShuffle is provided; update the LocalPlaylistHeader props/interface and all render sites that reference onClickSmartShuffle (including the other instances noted around the smart-shuffle button rendering) to handle it being undefined to avoid forcing non-local playlists to implement the callback.apps/mobile/src/lib/db/schema.ts (1)
201-201: 建议在数据库层约束confidence取值范围（0~1）。仅依赖服务层 clamp 容易被绕过；在 schema 加
check可以更好保证数据一致性。🛡️ 建议修改
(table) => [ index('track_llm_tags_source_idx').on(table.sourceType, table.sourceId), index('track_llm_tags_indexed_at_idx').on(table.indexedAt), + check( + 'track_llm_tags_confidence_range_check', + sql`confidence >= 0 AND confidence <= 1`, + ), ],Also applies to: 214-217
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@apps/mobile/src/lib/db/schema.ts` at line 201, Add a DB-level CHECK constraint to the confidence column(s) so values are always between 0 and 1: locate the confidence column definition(s) (real('confidence').notNull().default(0)) in schema.ts and modify the column/schema to include a CHECK constraint enforcing confidence >= 0 AND confidence <= 1; update the corresponding table definitions referenced around the other occurrences (lines ~214-217) and add a migration so the new constraint is applied to existing databases.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@apps/mobile/expo-plugins/withAbiFilters.js`:
- Around line 12-15: The current global regex used when building
contentsWithoutManagedBlock removes any ndk { abiFilters ... } block across the
file (affecting productFlavors/buildTypes); change the logic to only remove the
plugin-managed block by matching explicit markers the plugin inserts (e.g.,
BEGIN/END managed comments) or a unique sentinel the plugin writes, and update
the replacement in the code that computes contentsWithoutManagedBlock (and any
helper that writes the block) to search for and remove only that marked block so
other ndk blocks remain untouched.
In `@apps/mobile/src/components/modals/playlist/SmartShuffleModal.tsx`:
- Around line 47-49: The filter on tracksResult.value can throw when item.source
=== 'bilibili' but item.bilibiliMetadata is null/undefined; update the predicate
to use optional chaining and a safe default (e.g.
item.bilibiliMetadata?.videoIsValid ?? false) so the expression becomes: when
source is 'bilibili' check item.bilibiliMetadata?.videoIsValid (or false if
missing), otherwise keep true; adjust in the code that computes tracks
(references: tracks, tracksResult, item.bilibiliMetadata.videoIsValid).
In `@apps/mobile/src/components/modals/settings/LlmSmartShuffleSettingsModal.tsx`:
- Around line 78-100: The three TextInput components in
LlmSmartShuffleSettingsModal (the fields bound to llmBaseUrl, llmApiKey, and
llmModel) need keyboard behavior fixed: add autoCapitalize='none' and
autoCorrect={false} props to each TextInput (the one using llmBaseUrl, the one
using llmApiKey, and the one using llmModel) so URLs, API keys, and model names
are not auto-capitalized or autocorrected by the mobile keyboard.
- Around line 28-36: handleSave is persisting llmApiKey into MMKV without
encryption and the TextInput fields for llmBaseUrl and llmModel lack keyboard
protections; either enable MMKV encryption during app store initialization (call
storage.setEncryptionKey(...) from useAppStore/init code) or remove llmApiKey
from persisted state in setSettings, and add autoCorrect={false} and
autoCapitalize="none" to the TextInput components used for llmBaseUrl and
llmModel to prevent keyboard autocorrections; ensure secureTextEntry remains for
UI masking but do not rely on it for storage protection.
In `@apps/mobile/src/lib/facades/syncBilibiliPlaylist.ts`:
- Around line 303-308: The transaction callback passed to this.db.transaction is
missing the transaction handle and therefore
playlistService/trackService/artistService are using the root DB instead of the
transaction; change the async callback signature to accept the tx parameter
(e.g., async (tx) => { ... }) and replace withDB(this.db) calls with withDB(tx)
for playlistService, trackService and artistService inside that callback so all
writes run within the same transaction context (ensure the playlistId assignment
remains inside the transaction-returning block).
In `@apps/mobile/src/lib/services/llmSmartShuffleService.ts`:
- Around line 260-298: The LLM-returned trackId should not be trusted as the DB
primary key: before inserting into schema.trackLlmTags (the block building
values from indexes mapped from response.tracks) filter or remap entries using
the local context map (contextById) so you only insert tags for known contexts;
e.g., only keep indexes where contextById.has(index.trackId) or better yet, map
each response item to the context entry and use context.track.id (or
request-order mapping) as the authoritative trackId when constructing the insert
payload to avoid accidental overwrites.
- Around line 81-89: parseJsonObject and callJson currently throw Errors which
violates the service-layer convention to use neverthrow Results; change both
functions to return neverthrow Result<T, Error> or ResultAsync<T, Error> instead
of throwing: on JSON parse success return ok(value), on failure return err(new
Error(...)) (for parseJsonObject include the fallback match logic but return err
if still invalid); ensure callers like indexTrackBatch and parsePreference
consume/chain these Results (using .map/.andThen/.mapErr or ResultAsync
equivalents) so error branches are composed rather than throwing. Use the
existing function names parseJsonObject and callJson so callers can swap to the
Result-returning APIs without changing call sites beyond handling the Result.
- Around line 190-202: The fetch call in llmSmartShuffleService.ts must be
protected with an AbortController timeout and the service must stop throwing
errors directly; change the fetch block to use an AbortController with a
configurable timeout and call controller.abort() on timeout, and refactor error
paths to use neverthrow's ResultAsync/okAsync/errAsync instead of throwing;
update the helper callJson function to return ResultAsync (wrap fetch + JSON
parsing errors into errAsync) and replace all direct throw Error usages in this
module (and call sites referenced by callJson) with returning errAsync(...) or
mapping to okAsync(...) so the public API of functions like the
LlmSmartShuffleService method returns ResultAsync<ResultType, ErrorType>.
In `@lefthook.yml`:
- Line 2: The lefthook configuration currently sets parallel: true which causes
race conditions between the format-codes hook (which rewrites and restages
files) and the oxlint/eslint hooks; change the configuration to run hooks
sequentially by removing or setting parallel: false and ensure ordering so that
the format-codes hook runs before oxlint/eslint (apply the same change to the
other hook groups referenced around lines 6-15) to eliminate the timing/race
issue.
---
Nitpick comments:
In `@apps/mobile/drizzle/0019_smart_shuffle_tags.sql`:
- Line 4: Add a DB-level check constraint to enforce confidence is within [0,1]
for the confidence column; update the column definition in the migration (the
`confidence` real DEFAULT 0 NOT NULL` line) to include a CHECK (confidence >= 0
AND confidence <= 1) or add a named CONSTRAINT (e.g., CONSTRAINT
check_confidence_range CHECK (confidence >= 0 AND confidence <= 1)) so
out-of-band writes cannot insert values outside 0..1.
In `@apps/mobile/src/components/ModalRegistry.tsx`:
- Around line 68-73: Replace the relative import paths in the lazy calls for
LlmSmartShuffleSettingsModal and SmartShuffleModal so they use the app alias
import style (e.g., '@/...') instead of './modals/...'; update the two lazy(()
=> import(...)) expressions for the symbols LlmSmartShuffleSettingsModal and
SmartShuffleModal to import via the alias path that matches the mobile app
convention.
In `@apps/mobile/src/features/playlist/local/components/LocalPlaylistHeader.tsx`:
- Line 33: The LocalPlaylistHeader prop onClickSmartShuffle is currently
required and the smart-shuffle UI is always rendered; make the prop optional
(onClickSmartShuffle?: () => void) and only render the smart-shuffle button when
playlist.type === 'local' and onClickSmartShuffle is provided; update the
LocalPlaylistHeader props/interface and all render sites that reference
onClickSmartShuffle (including the other instances noted around the
smart-shuffle button rendering) to handle it being undefined to avoid forcing
non-local playlists to implement the callback.
In `@apps/mobile/src/lib/db/schema.ts`:
- Line 201: Add a DB-level CHECK constraint to the confidence column(s) so
values are always between 0 and 1: locate the confidence column definition(s)
(real('confidence').notNull().default(0)) in schema.ts and modify the
column/schema to include a CHECK constraint enforcing confidence >= 0 AND
confidence <= 1; update the corresponding table definitions referenced around
the other occurrences (lines ~214-217) and add a migration so the new constraint
is applied to existing databases.
🪄 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: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: 36e0bff6-cc4f-4e54-859f-5bad1c83b61b
⛔ Files ignored due to path filters (1)
pnpm-lock.yamlis excluded by!**/pnpm-lock.yaml
📒 Files selected for processing (20)
apps/mobile/drizzle/0019_smart_shuffle_tags.sqlapps/mobile/drizzle/meta/_journal.jsonapps/mobile/drizzle/migrations.jsapps/mobile/expo-plugins/withAbiFilters.jsapps/mobile/package.jsonapps/mobile/src/app/playlist/local/[id].tsxapps/mobile/src/app/settings/playback.tsxapps/mobile/src/components/ModalRegistry.tsxapps/mobile/src/components/modals/playlist/SmartShuffleModal.tsxapps/mobile/src/components/modals/settings/LlmSmartShuffleSettingsModal.tsxapps/mobile/src/features/playlist/local/components/LocalPlaylistHeader.tsxapps/mobile/src/hooks/stores/useAppStore.tsapps/mobile/src/lib/db/schema.tsapps/mobile/src/lib/facades/syncBilibiliPlaylist.tsapps/mobile/src/lib/services/llmSmartShuffleService.tsapps/mobile/src/types/core/appStore.tsapps/mobile/src/types/navigation.tsapps/mobile/src/types/services/llmSmartShuffle.tslefthook.ymlscripts/lefthook-gitleaks.cjs
Comment on lines
+12
to
+15
| const contentsWithoutManagedBlock = contents.replace( | ||
| /\n\s*ndk\s*\{\s*\n\s*abiFilters\s+[^\n]+\n\s*\}/g, | ||
| '', | ||
| ) |
There was a problem hiding this comment.
仅删除“受管块”，当前正则会误删其他 ndk 配置
Line [13] 的全局正则会删除文件中任意 ndk { abiFilters ... } 块，不限于 defaultConfig。这会误伤 productFlavors/buildTypes 的 ABI 配置，导致构建行为被悄然改变。建议改为只清理插件自己插入的受管块。
🔧 建议修复（用受管标记实现精准 upsert）
+const MANAGED_ABI_FILTERS_BLOCK_REGEX =
+ /\n?\s*\/\/ `@managed-by-withAbiFilters`:start[\s\S]*?\/\/ `@managed-by-withAbiFilters`:end\s*\n?/g
+
const upsertAbiFilters = (contents, abiFiltersString) => {
- const ndkBlock = ` ndk {
+ const ndkBlock = ` // `@managed-by-withAbiFilters`:start
+ ndk {
abiFilters ${abiFiltersString}
}
+ // `@managed-by-withAbiFilters`:end
`
const contentsWithoutManagedBlock = contents.replace(
- /\n\s*ndk\s*\{\s*\n\s*abiFilters\s+[^\n]+\n\s*\}/g,
+ MANAGED_ABI_FILTERS_BLOCK_REGEX,
'',
)🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@apps/mobile/expo-plugins/withAbiFilters.js` around lines 12 - 15, The current
global regex used when building contentsWithoutManagedBlock removes any ndk {
abiFilters ... } block across the file (affecting productFlavors/buildTypes);
change the logic to only remove the plugin-managed block by matching explicit
markers the plugin inserts (e.g., BEGIN/END managed comments) or a unique
sentinel the plugin writes, and update the replacement in the code that computes
contentsWithoutManagedBlock (and any helper that writes the block) to search for
and remove only that marked block so other ndk blocks remain untouched.
Comment on lines
+47
to
+49
| const tracks = tracksResult.value.filter((item) => | ||
| item.source === 'bilibili' ? item.bilibiliMetadata.videoIsValid : true, | ||
| ) |
There was a problem hiding this comment.
这里有潜在空引用风险。
当 item.source === 'bilibili' 且元数据缺失时会直接崩溃。建议加可选链保护。
🐛 建议修复
const tracks = tracksResult.value.filter((item) =>
- item.source === 'bilibili' ? item.bilibiliMetadata.videoIsValid : true,
+ item.source === 'bilibili'
+ ? (item.bilibiliMetadata?.videoIsValid ?? true)
+ : true,
)📝 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
| const tracks = tracksResult.value.filter((item) => | |
| item.source === 'bilibili' ? item.bilibiliMetadata.videoIsValid : true, | |
| ) | |
| const tracks = tracksResult.value.filter((item) => | |
| item.source === 'bilibili' | |
| ? (item.bilibiliMetadata?.videoIsValid ?? true) | |
| : true, | |
| ) |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@apps/mobile/src/components/modals/playlist/SmartShuffleModal.tsx` around
lines 47 - 49, The filter on tracksResult.value can throw when item.source ===
'bilibili' but item.bilibiliMetadata is null/undefined; update the predicate to
use optional chaining and a safe default (e.g.
item.bilibiliMetadata?.videoIsValid ?? false) so the expression becomes: when
source is 'bilibili' check item.bilibiliMetadata?.videoIsValid (or false if
missing), otherwise keep true; adjust in the code that computes tracks
(references: tracks, tracksResult, item.bilibiliMetadata.videoIsValid).
Comment on lines
+28
to
+36
| const handleSave = () => { | ||
| setSettings({ | ||
| enableLlmTagging, | ||
| allowLlmMetadataUpload, | ||
| llmBaseUrl: llmBaseUrl.trim() || 'https://api.openai.com/v1', | ||
| llmApiKey: llmApiKey.trim(), | ||
| llmModel: llmModel.trim() || 'gpt-4o-mini', | ||
| llmDefaultPreference: llmDefaultPreference.trim(), | ||
| }) |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo "== useAppStore implementation =="
fd -i 'useAppStore\.ts$' apps/mobile/src/hooks/stores -x sed -n '1,260p' {}
echo
echo "== persistence backend references =="
rg -n -C3 'persist\(|createJSONStorage|AsyncStorage|MMKV|SecureStore|EncryptedStorage|storage:' apps/mobile/src
echo
echo "== llmApiKey usage/persistence sites =="
rg -n -C2 '\bllmApiKey\b' apps/mobile/srcRepository: bbplayer-app/BBPlayer
Length of output: 22258
🏁 Script executed:
#!/bin/bash
echo "== Search for setEncryptionKey calls =="
rg -n 'setEncryptionKey\|setPassword\|encrypt' apps/mobile/src --type ts --type tsx
echo
echo "== Full LlmSmartShuffleSettingsModal.tsx file (to check TextInput props) =="
fd 'LlmSmartShuffleSettingsModal\.tsx$' apps/mobile/src -x cat -n {}
echo
echo "== MMKV initialization with any security setup =="
rg -n -B5 -A10 'createMMKV\(\)' apps/mobile/src/utils/mmkv.tsRepository: bbplayer-app/BBPlayer
Length of output: 5581
确认 llmApiKey 明文存储，且 URL/模型字段缺少键盘输入保护。
第 29-35 行将 API Key 持久化到 MMKV，但 MMKV 初始化时未启用加密（createMMKV() 无密钥参数，代码中无 setEncryptionKey() 调用）。此外：
- 第 78-85 行的
llmBaseUrl和第 94-101 行的llmModelTextInput 缺少autoCorrect={false}和autoCapitalize="none"属性，易被键盘自动修正破坏 URL 格式和模型名称。 - 第 92 行虽有
secureTextEntry掩盖了 UI 中的输入，但无法解决存储端的明文问题。
建议：为 MMKV 启用加密（在 useAppStore 初始化时调用 storage.setEncryptionKey()），或从持久化状态中排除 llmApiKey；同时在 URL 和模型字段添加 autoCorrect={false} 与 autoCapitalize="none"。
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@apps/mobile/src/components/modals/settings/LlmSmartShuffleSettingsModal.tsx`
around lines 28 - 36, handleSave is persisting llmApiKey into MMKV without
encryption and the TextInput fields for llmBaseUrl and llmModel lack keyboard
protections; either enable MMKV encryption during app store initialization (call
storage.setEncryptionKey(...) from useAppStore/init code) or remove llmApiKey
from persisted state in setSettings, and add autoCorrect={false} and
autoCapitalize="none" to the TextInput components used for llmBaseUrl and
llmModel to prevent keyboard autocorrections; ensure secureTextEntry remains for
UI masking but do not rely on it for storage protection.
Comment on lines
+78
to
+100
| <TextInput | ||
| label='OpenAI 兼容 API 地址' | ||
| value={llmBaseUrl} | ||
| onChangeText={setLlmBaseUrl} | ||
| mode='outlined' | ||
| style={styles.input} | ||
| placeholder='https://api.openai.com/v1' | ||
| /> | ||
| <TextInput | ||
| label='API Key' | ||
| value={llmApiKey} | ||
| onChangeText={setLlmApiKey} | ||
| mode='outlined' | ||
| style={styles.input} | ||
| secureTextEntry | ||
| /> | ||
| <TextInput | ||
| label='模型名称' | ||
| value={llmModel} | ||
| onChangeText={setLlmModel} | ||
| mode='outlined' | ||
| style={styles.input} | ||
| placeholder='gpt-4o-mini' |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
# Check if the file exists and read the relevant lines
if [ -f "apps/mobile/src/components/modals/settings/LlmSmartShuffleSettingsModal.tsx" ]; then
echo "=== File found. Reading lines 70-110 for context ==="
sed -n '70,110p' "apps/mobile/src/components/modals/settings/LlmSmartShuffleSettingsModal.tsx"
else
echo "File not found"
fiRepository: bbplayer-app/BBPlayer
Length of output: 1082
🏁 Script executed:
# Also check the imports to verify TextInput source
head -30 "apps/mobile/src/components/modals/settings/LlmSmartShuffleSettingsModal.tsx"Repository: bbplayer-app/BBPlayer
Length of output: 1217
TextInput 字段需要禁用自动更正和自动大写。
这三个字段（llmBaseUrl、llmApiKey、llmModel）包含精确字符串，移动端键盘的自动更正和首字母大写会无声地破坏 URL、API 密钥或模型名称。需要添加 autoCapitalize='none' 和 autoCorrect={false} 到这些 TextInput 组件。
建议修改
<TextInput
label='OpenAI 兼容 API 地址'
value={llmBaseUrl}
onChangeText={setLlmBaseUrl}
mode='outlined'
style={styles.input}
placeholder='https://api.openai.com/v1'
+ autoCapitalize='none'
+ autoCorrect={false}
/>
<TextInput
label='API Key'
value={llmApiKey}
onChangeText={setLlmApiKey}
mode='outlined'
style={styles.input}
secureTextEntry
+ autoCapitalize='none'
+ autoCorrect={false}
/>
<TextInput
label='模型名称'
value={llmModel}
onChangeText={setLlmModel}
mode='outlined'
style={styles.input}
placeholder='gpt-4o-mini'
+ autoCapitalize='none'
+ autoCorrect={false}
/>📝 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
| <TextInput | |
| label='OpenAI 兼容 API 地址' | |
| value={llmBaseUrl} | |
| onChangeText={setLlmBaseUrl} | |
| mode='outlined' | |
| style={styles.input} | |
| placeholder='https://api.openai.com/v1' | |
| /> | |
| <TextInput | |
| label='API Key' | |
| value={llmApiKey} | |
| onChangeText={setLlmApiKey} | |
| mode='outlined' | |
| style={styles.input} | |
| secureTextEntry | |
| /> | |
| <TextInput | |
| label='模型名称' | |
| value={llmModel} | |
| onChangeText={setLlmModel} | |
| mode='outlined' | |
| style={styles.input} | |
| placeholder='gpt-4o-mini' | |
| <TextInput | |
| label='OpenAI 兼容 API 地址' | |
| value={llmBaseUrl} | |
| onChangeText={setLlmBaseUrl} | |
| mode='outlined' | |
| style={styles.input} | |
| placeholder='https://api.openai.com/v1' | |
| autoCapitalize='none' | |
| autoCorrect={false} | |
| /> | |
| <TextInput | |
| label='API Key' | |
| value={llmApiKey} | |
| onChangeText={setLlmApiKey} | |
| mode='outlined' | |
| style={styles.input} | |
| secureTextEntry | |
| autoCapitalize='none' | |
| autoCorrect={false} | |
| /> | |
| <TextInput | |
| label='模型名称' | |
| value={llmModel} | |
| onChangeText={setLlmModel} | |
| mode='outlined' | |
| style={styles.input} | |
| placeholder='gpt-4o-mini' | |
| autoCapitalize='none' | |
| autoCorrect={false} | |
| /> |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@apps/mobile/src/components/modals/settings/LlmSmartShuffleSettingsModal.tsx`
around lines 78 - 100, The three TextInput components in
LlmSmartShuffleSettingsModal (the fields bound to llmBaseUrl, llmApiKey, and
llmModel) need keyboard behavior fixed: add autoCapitalize='none' and
autoCorrect={false} props to each TextInput (the one using llmBaseUrl, the one
using llmApiKey, and the one using llmModel) so URLs, API keys, and model names
are not auto-capitalized or autocorrected by the mobile keyboard.
Comment on lines
303
to
308
| return ResultAsync.fromPromise( | ||
| this.db.transaction(async () => { | ||
| (async () => { | ||
| const playlistId = await this.db.transaction(async () => { | ||
| const playlistSvc = this.playlistService.withDB(this.db) | ||
| const trackSvc = this.trackService.withDB(this.db) | ||
| const artistSvc = this.artistService.withDB(this.db) |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
cat -n apps/mobile/src/lib/facades/syncBilibiliPlaylist.ts | sed -n '295,320p'Repository: bbplayer-app/BBPlayer
Length of output: 1175
🏁 Script executed:
# Search for transaction usage patterns to understand the API
rg "\.transaction\(" apps/mobile/src --type ts --type tsx -A 3 | head -50Repository: bbplayer-app/BBPlayer
Length of output: 93
🏁 Script executed:
# Check the withDB method signature
rg "withDB\(" apps/mobile/src/lib/facades/syncBilibiliPlaylist.ts -B 2 -A 2Repository: bbplayer-app/BBPlayer
Length of output: 1052
🏁 Script executed:
# Search for more transaction patterns to understand full context
rg "\.transaction\(async" apps/mobile/src -B 2 -A 5 | head -80Repository: bbplayer-app/BBPlayer
Length of output: 6624
🏁 Script executed:
# Check the db.transaction method to understand its signature
fd -e ts -e tsx . apps/mobile/src | xargs grep -l "transaction.*async" | head -5Repository: bbplayer-app/BBPlayer
Length of output: 305
🏁 Script executed:
# Look for the database or transaction type definitions
rg "transaction\s*\(" apps/mobile/src -B 3 -A 8 | grep -A 10 -B 3 "async.*tx"Repository: bbplayer-app/BBPlayer
Length of output: 17439
事务句柄未传入，导致操作脱离事务上下文。
第 305 行的回调未接收 tx 参数，第 306-308 行继续使用 withDB(this.db) 而非事务句柄。这使得 playlist / track / artist 的写入不在事务范围内，任一步失败时将留下部分写入的不一致数据。
应改为：
事务参数修复
- const playlistId = await this.db.transaction(async () => {
- const playlistSvc = this.playlistService.withDB(this.db)
- const trackSvc = this.trackService.withDB(this.db)
- const artistSvc = this.artistService.withDB(this.db)
+ const playlistId = await this.db.transaction(async (tx) => {
+ const playlistSvc = this.playlistService.withDB(tx)
+ const trackSvc = this.trackService.withDB(tx)
+ const artistSvc = this.artistService.withDB(tx)📝 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
| return ResultAsync.fromPromise( | |
| this.db.transaction(async () => { | |
| (async () => { | |
| const playlistId = await this.db.transaction(async () => { | |
| const playlistSvc = this.playlistService.withDB(this.db) | |
| const trackSvc = this.trackService.withDB(this.db) | |
| const artistSvc = this.artistService.withDB(this.db) | |
| return ResultAsync.fromPromise( | |
| (async () => { | |
| const playlistId = await this.db.transaction(async (tx) => { | |
| const playlistSvc = this.playlistService.withDB(tx) | |
| const trackSvc = this.trackService.withDB(tx) | |
| const artistSvc = this.artistService.withDB(tx) |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@apps/mobile/src/lib/facades/syncBilibiliPlaylist.ts` around lines 303 - 308,
The transaction callback passed to this.db.transaction is missing the
transaction handle and therefore playlistService/trackService/artistService are
using the root DB instead of the transaction; change the async callback
signature to accept the tx parameter (e.g., async (tx) => { ... }) and replace
withDB(this.db) calls with withDB(tx) for playlistService, trackService and
artistService inside that callback so all writes run within the same transaction
context (ensure the playlistId assignment remains inside the
transaction-returning block).
Comment on lines
+81
to
+89
| function parseJsonObject<T>(content: string): T { | ||
| try { | ||
| return JSON.parse(content) as T | ||
| } catch { | ||
| const match = content.match(/\{[\s\S]*\}/) | ||
| if (!match) throw new Error('LLM 没有返回 JSON 对象') | ||
| return JSON.parse(match[0]) as T | ||
| } | ||
| } |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
# Check if the file exists and get its structure
fd -t f "llmSmartShuffleService" apps/mobile/src/lib/services/Repository: bbplayer-app/BBPlayer
Length of output: 120
🏁 Script executed:
#!/bin/bash
# Read the file to examine the error handling patterns
wc -l apps/mobile/src/lib/services/llmSmartShuffleService.tsRepository: bbplayer-app/BBPlayer
Length of output: 124
🏁 Script executed:
#!/bin/bash
# Read the specific sections mentioned in the review (lines 81-89 and 204-211)
sed -n '75,95p' apps/mobile/src/lib/services/llmSmartShuffleService.ts
echo "---"
sed -n '200,220p' apps/mobile/src/lib/services/llmSmartShuffleService.tsRepository: bbplayer-app/BBPlayer
Length of output: 1093
🏁 Script executed:
#!/bin/bash
# Search for all throw statements in the file
rg "throw\s+new\s+Error|throw\s+" apps/mobile/src/lib/services/llmSmartShuffleService.ts -nRepository: bbplayer-app/BBPlayer
Length of output: 217
🏁 Script executed:
#!/bin/bash
# Check if neverthrow is imported or used in the file
rg "neverthrow|Result|from.*neverthrow" apps/mobile/src/lib/services/llmSmartShuffleService.tsRepository: bbplayer-app/BBPlayer
Length of output: 273
服务层应使用 neverthrow Result 模式处理错误，不应直接抛异常。
parseJsonObject（第 86 行）和 callJson（第 205、211 行）直接抛出 Error，绕过了该目录约定的 neverthrow 错误处理模式。应改为这些函数返回 Result 或 ResultAsync，由调用方（如 indexTrackBatch 和 parsePreference）组合失败分支。按照编码规范："在 Facades 和 Services 中使用 neverthrow Result 类型处理所有错误——不应直接抛异常"。
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@apps/mobile/src/lib/services/llmSmartShuffleService.ts` around lines 81 - 89,
parseJsonObject and callJson currently throw Errors which violates the
service-layer convention to use neverthrow Results; change both functions to
return neverthrow Result<T, Error> or ResultAsync<T, Error> instead of throwing:
on JSON parse success return ok(value), on failure return err(new Error(...))
(for parseJsonObject include the fallback match logic but return err if still
invalid); ensure callers like indexTrackBatch and parsePreference consume/chain
these Results (using .map/.andThen/.mapErr or ResultAsync equivalents) so error
branches are composed rather than throwing. Use the existing function names
parseJsonObject and callJson so callers can swap to the Result-returning APIs
without changing call sites beyond handling the Result.
Comment on lines
+190
to
+202
| const response = await fetch(url, { | ||
| method: 'POST', | ||
| headers, | ||
| body: JSON.stringify({ | ||
| model: settings.llmModel.trim(), | ||
| temperature: 0.1, | ||
| response_format: { type: 'json_object' }, | ||
| messages: [ | ||
| { role: 'system', content: system }, | ||
| { role: 'user', content: user }, | ||
| ], | ||
| }), | ||
| }) |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
cat -n apps/mobile/src/lib/services/llmSmartShuffleService.ts | sed -n '180,220p'Repository: bbplayer-app/BBPlayer
Length of output: 1585
🏁 Script executed:
rg "callJson" apps/mobile/src/lib/services/llmSmartShuffleService.ts -A 3 -B 1Repository: bbplayer-app/BBPlayer
Length of output: 675
🏁 Script executed:
head -n 50 apps/mobile/src/lib/services/llmSmartShuffleService.tsRepository: bbplayer-app/BBPlayer
Length of output: 1130
🏁 Script executed:
grep -n "neverthrow\|Result" apps/mobile/src/lib/services/llmSmartShuffleService.ts | head -20Repository: bbplayer-app/BBPlayer
Length of output: 289
🏁 Script executed:
rg "callJson" apps/mobile/src/lib/services/llmSmartShuffleService.ts -B 2 -A 8Repository: bbplayer-app/BBPlayer
Length of output: 1065
🏁 Script executed:
cat -n apps/mobile/src/lib/services/llmSmartShuffleService.ts | sed -n '220,260p'Repository: bbplayer-app/BBPlayer
Length of output: 1334
🏁 Script executed:
grep -n "try\|catch\|throw" apps/mobile/src/lib/services/llmSmartShuffleService.tsRepository: bbplayer-app/BBPlayer
Length of output: 313
🏁 Script executed:
cat -n apps/mobile/src/lib/services/llmSmartShuffleService.ts | sed -n '320,345p'Repository: bbplayer-app/BBPlayer
Length of output: 946
🏁 Script executed:
cat -n apps/mobile/src/lib/services/llmSmartShuffleService.ts | sed -n '345,380p'Repository: bbplayer-app/BBPlayer
Length of output: 1380
🏁 Script executed:
ls -la apps/mobile/src/lib/services/Repository: bbplayer-app/BBPlayer
Length of output: 890
🏁 Script executed:
head -n 100 apps/mobile/src/lib/services/playlistService.ts | grep -A 5 "import.*neverthrow\|Result"Repository: bbplayer-app/BBPlayer
Length of output: 746
添加请求超时，并改用 neverthrow 进行错误处理。
该方法在请求卡住时会导致同步流程永久挂起，需要添加 AbortController 超时保护。同时，根据服务层编码规范，需要将错误处理改为 neverthrow 的 Result 模式，而非直接抛异�常。
当前代码在行 205、211 和 82、86 中直接 throw Error，违反了服务层不应直接抛异常的要求。建议将 callJson 改为返回 ResultAsync，并将错误处理统一为 Result 模式。
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@apps/mobile/src/lib/services/llmSmartShuffleService.ts` around lines 190 -
202, The fetch call in llmSmartShuffleService.ts must be protected with an
AbortController timeout and the service must stop throwing errors directly;
change the fetch block to use an AbortController with a configurable timeout and
call controller.abort() on timeout, and refactor error paths to use neverthrow's
ResultAsync/okAsync/errAsync instead of throwing; update the helper callJson
function to return ResultAsync (wrap fetch + JSON parsing errors into errAsync)
and replace all direct throw Error usages in this module (and call sites
referenced by callJson) with returning errAsync(...) or mapping to okAsync(...)
so the public API of functions like the LlmSmartShuffleService method returns
ResultAsync<ResultType, ErrorType>.
Comment on lines
+260
to
+298
| const indexes = (response.tracks ?? []) | ||
| .map((item) => { | ||
| if (!item || typeof item !== 'object') return null | ||
| const row = item as Record<string, unknown> | ||
| const trackId = Number(row.trackId) | ||
| if (!Number.isFinite(trackId)) return null | ||
| const index: TrackTagIndex = { | ||
| trackId, | ||
| tags: normalizeTags(row.tags), | ||
| confidence: | ||
| typeof row.confidence === 'number' | ||
| ? Math.max(0, Math.min(1, row.confidence)) | ||
| : 0, | ||
| } | ||
| if (typeof row.reason === 'string') index.reason = row.reason | ||
| return index | ||
| }) | ||
| .filter((item): item is TrackTagIndex => item !== null) | ||
|
|
||
| if (indexes.length === 0) return | ||
|
|
||
| const contextById = new Map(contexts.map((item) => [item.track.id, item])) | ||
| await db | ||
| .insert(schema.trackLlmTags) | ||
| .values( | ||
| indexes.map((index) => { | ||
| const context = contextById.get(index.trackId) | ||
| return { | ||
| trackId: index.trackId, | ||
| tags: index.tags, | ||
| confidence: index.confidence, | ||
| reason: index.reason, | ||
| model: settings.llmModel.trim(), | ||
| sourceType: context?.sourceType, | ||
| sourceId: context?.sourceId, | ||
| sourceSyncedAt: context?.sourceSyncedAt, | ||
| indexedAt: new Date(), | ||
| } | ||
| }), |
There was a problem hiding this comment.
不要信任模型回传的 trackId 作为写库主键。
这里会把 LLM 返回的 trackId 直接拿去 upsert；只要模型串号到任意已有 ID，就可能覆盖别的歌曲标签。写入前至少要过滤 contextById.has(index.trackId)，更稳妥的是完全不要让模型决定主键，而是按请求顺序或本地映射回填。
最低限度先把越界的 `trackId` 过滤掉
- const indexes = (response.tracks ?? [])
+ const contextById = new Map(contexts.map((item) => [item.track.id, item]))
+ const indexes = (response.tracks ?? [])
.map((item) => {
if (!item || typeof item !== 'object') return null
const row = item as Record<string, unknown>
const trackId = Number(row.trackId)
if (!Number.isFinite(trackId)) return null
@@
})
.filter((item): item is TrackTagIndex => item !== null)
+ .filter((item) => contextById.has(item.trackId))
if (indexes.length === 0) return
-
- const contextById = new Map(contexts.map((item) => [item.track.id, item]))📝 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
| const indexes = (response.tracks ?? []) | |
| .map((item) => { | |
| if (!item || typeof item !== 'object') return null | |
| const row = item as Record<string, unknown> | |
| const trackId = Number(row.trackId) | |
| if (!Number.isFinite(trackId)) return null | |
| const index: TrackTagIndex = { | |
| trackId, | |
| tags: normalizeTags(row.tags), | |
| confidence: | |
| typeof row.confidence === 'number' | |
| ? Math.max(0, Math.min(1, row.confidence)) | |
| : 0, | |
| } | |
| if (typeof row.reason === 'string') index.reason = row.reason | |
| return index | |
| }) | |
| .filter((item): item is TrackTagIndex => item !== null) | |
| if (indexes.length === 0) return | |
| const contextById = new Map(contexts.map((item) => [item.track.id, item])) | |
| await db | |
| .insert(schema.trackLlmTags) | |
| .values( | |
| indexes.map((index) => { | |
| const context = contextById.get(index.trackId) | |
| return { | |
| trackId: index.trackId, | |
| tags: index.tags, | |
| confidence: index.confidence, | |
| reason: index.reason, | |
| model: settings.llmModel.trim(), | |
| sourceType: context?.sourceType, | |
| sourceId: context?.sourceId, | |
| sourceSyncedAt: context?.sourceSyncedAt, | |
| indexedAt: new Date(), | |
| } | |
| }), | |
| const contextById = new Map(contexts.map((item) => [item.track.id, item])) | |
| const indexes = (response.tracks ?? []) | |
| .map((item) => { | |
| if (!item || typeof item !== 'object') return null | |
| const row = item as Record<string, unknown> | |
| const trackId = Number(row.trackId) | |
| if (!Number.isFinite(trackId)) return null | |
| const index: TrackTagIndex = { | |
| trackId, | |
| tags: normalizeTags(row.tags), | |
| confidence: | |
| typeof row.confidence === 'number' | |
| ? Math.max(0, Math.min(1, row.confidence)) | |
| : 0, | |
| } | |
| if (typeof row.reason === 'string') index.reason = row.reason | |
| return index | |
| }) | |
| .filter((item): item is TrackTagIndex => item !== null) | |
| .filter((item) => contextById.has(item.trackId)) | |
| if (indexes.length === 0) return | |
| await db | |
| .insert(schema.trackLlmTags) | |
| .values( | |
| indexes.map((index) => { | |
| const context = contextById.get(index.trackId) | |
| return { | |
| trackId: index.trackId, | |
| tags: index.tags, | |
| confidence: index.confidence, | |
| reason: index.reason, | |
| model: settings.llmModel.trim(), | |
| sourceType: context?.sourceType, | |
| sourceId: context?.sourceId, | |
| sourceSyncedAt: context?.sourceSyncedAt, | |
| indexedAt: new Date(), | |
| } | |
| }), |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@apps/mobile/src/lib/services/llmSmartShuffleService.ts` around lines 260 -
298, The LLM-returned trackId should not be trusted as the DB primary key:
before inserting into schema.trackLlmTags (the block building values from
indexes mapped from response.tracks) filter or remap entries using the local
context map (contextById) so you only insert tags for known contexts; e.g., only
keep indexes where contextById.has(index.trackId) or better yet, map each
response item to the context entry and use context.track.id (or request-order
mapping) as the authoritative trackId when constructing the insert payload to
avoid accidental overwrites.
| @@ -2,22 +2,17 @@ pre-commit: | |||
| parallel: true | |||
There was a problem hiding this comment.
格式化与 lint 并行执行存在竞态风险。
当前 parallel: true 下，format-codes 会改写并重新暂存文件，而 oxlint/eslint 可能已并行读取旧内容，导致 pre-commit 结果偶发失败或不一致。建议改为顺序执行。
🔧 建议修复
pre-commit:
- parallel: true
+ parallel: false
commands:
gitleaks:
run: node scripts/lefthook-gitleaks.cjsAlso applies to: 6-15
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@lefthook.yml` at line 2, The lefthook configuration currently sets parallel:
true which causes race conditions between the format-codes hook (which rewrites
and restages files) and the oxlint/eslint hooks; change the configuration to run
hooks sequentially by removing or setting parallel: false and ensure ordering so
that the format-codes hook runs before oxlint/eslint (apply the same change to
the other hook groups referenced around lines 6-15) to eliminate the timing/race
issue.
✅ Actions performedReviews paused. |
Made-with: Cursor
Generate LLM datasets during favorite sync. Split large sort payloads to avoid context overflow. Made-with: Cursor
Made-with: Cursor
Made-with: Cursor
|
Review the following changes in direct dependencies. Learn more about Socket for GitHub.
|
Improve local title-based tags for smart shuffle and clean up Android warning sources that are tracked in the repo. Made-with: Cursor
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
2 participants
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
cross-env, fixing generated Android ABI filter insertion, and making Lefthook/gitleaks execution work without fragile inline shell parsing.Feature Details
SmartShuffleModalandLlmSmartShuffleSettingsModalso users can start smart shuffle from playlist UI and configure the feature from playback settings.LlmSmartShuffleServiceplusTrackTagIndexstorage to normalize LLM responses, clamp confidence values, persist tags, and support future smart ordering logic.0019_smart_shuffle_tagsand schema updates for smart shuffle tag indexing.Tooling / Android Notes
withAbiFiltersso Expo prebuild insertsndk.abiFiltersintodefaultConfigsafely without corruptingbuildConfigFieldinandroid/app/build.gradle.cross-envto mobile scripts soexpo startandexpo run:androidwork consistently on Windows shells.Verification
origin/devcontains the intended smart shuffle, Android prebuild, and tooling changes../gradlew.bat projectsfromapps/mobile/androidsuccessfully.Summary by CodeRabbit
发布说明
新功能
改进
Chores