Some upstream providers (e.g. certain OpenAI-compatible aggregators)
emit thinking/reasoning tokens with the same text written to both
delta.content and delta.reasoning_content simultaneously.
When metapi forwarded these chunks as-is, downstream clients would read
delta.content and display the internal reasoning text to end users —
appearing as word-by-word "thinking narration" in chat interfaces.
Fix: after resolving both contentDelta and reasoningDelta, suppress
contentDelta when it is non-empty and identical to reasoningDelta.
This covers the duplicate-write pattern without affecting any legitimate
case (a provider would never intentionally set content == reasoning).
feat: enhance proxy usage parser to include cache read and creation tokens
feat: add billing details to ProxyLog interface and implement related formatting functions
fix: correct mojibake issues in source files and add regression test for mojibake detection
chore: add SQL migration for billing details in proxy_logs table
test: improve site speed button tests to ensure proper state updates without DOM lookups
feat: add run-server script to manage server execution and compatibility checks
feat: create main application logic for desktop with window management, server handling, and auto-updater
test: add unit tests for desktop runtime helpers
feat: implement desktop server environment configuration and health check
test: add tests for server configuration and desktop routes
feat: add native module guard for better-sqlite3 compatibility
test: add tests for native module guard functionality
test: add comprehensive tests for account verification feedback handling
feat: implement account verification feedback helpers
chore: add TypeScript configuration for desktop builds
- Added createdAt timestamps to various check-in log entries in checkinService.ts for better tracking.
- Updated i18n translations in Models.tsx to improve user experience with accurate Chinese translations.
- Reduced default page size in ProgramLogs.tsx and ProxyLogs.tsx from 80 and 100 to 50 for better performance.
- Enhanced ProxyLogs.tsx with memoization for filtered logs and summary statistics.
- Implemented support for parsing Unix timestamps and PostgreSQL-style timezone offsets in checkinLogTime.ts.
- Added comprehensive tests for database proxy query wrapper in index.proxy-wrap.test.ts to ensure functionality.
- Implement tests for retrieving and saving runtime database configuration.
- Validate handling of invalid configuration payloads.
- Utilize Fastify for API testing with Vitest.
- Ensure proper cleanup of database state between tests.
- Implemented `normalizeMigrationInput` to standardize database migration input for sqlite, mysql, and postgres.
- Added `maskConnectionString` to obscure sensitive credentials in connection strings.
- Created comprehensive tests for migration input normalization and connection string masking.
- Established database schema creation logic for multiple dialects (sqlite, mysql, postgres).
- Developed functions for managing database transactions, including begin, commit, and rollback.
- Implemented data insertion logic from backup snapshots into the target database.
- Added new utility functions for building minimal JSON headers and normalizing responses for compatibility.
- Implemented detailed error handling for unsupported media types and endpoint downgrade scenarios.
- Introduced a new endpoint flow mechanism to manage retries and downgrades between different upstream endpoints.
- Enhanced tests for endpoint flow and error summarization to cover various edge cases and scenarios.
- Updated CSS styles for accounts table to improve layout and responsiveness.
- Refactored ModelTester component to handle incremental text updates more effectively.
- Implement tests for accounts login shield detection, verifying behavior when login fails with HTML JSON parse syntax error.
- Create tests for proxy log retention service, ensuring expired logs are deleted based on retention days and that cleanup is skipped when retention is disabled.
- Add tests for token router cache, validating that route snapshots are maintained within TTL until explicitly invalidated.
- Introduce necessary mock setups and database seeding for comprehensive testing.
- Implemented tests for the `/api/models/token-candidates` endpoint to verify model availability and token coverage scenarios.
- Added tests for `composeProxyLogMessage` to ensure proper formatting of log messages with downstream and upstream metadata.
- Created utility functions for parsing proxy log path metadata and handling missing token hints.
- Developed tests for the `Sites` and `TokenRoutes` components to validate UI behavior and interactions, including scrolling and group expansion.
- Introduced a custom hook `useAnimatedVisibility` for managing visibility animations in components.
- Enhanced the documentation link for site resources.
- Added support for tool calls in NormalizedFinalResponse type.
- Implemented functions to collect and serialize tool calls from OpenAI and Claude responses.
- Updated convertClaudeRequestToOpenAiBody to handle tool use and tool result messages.
- Enhanced normalizeUpstreamFinalResponse to include tool calls in the response structure.
- Introduced utility functions for safe JSON stringification and parsing.
- Improved conversion functions for OpenAI tools to Anthropic format and vice versa.
- Updated input handling in convertOpenAiBodyToMessagesBody and convertOpenAiBodyToResponsesBody to accommodate tool messages.
- Refactored serialization logic in serializeFinalResponse and buildSyntheticOpenAiChunks to include tool calls.
- Implemented `CliProxyApiAdapter` with detection logic for CLIProxyAPI.
- Added tests for `CliProxyApiAdapter` to verify detection and response handling.
- Created `OneHubAdapter` with methods for fetching models and user groups.
- Added tests for `OneHubAdapter` to ensure correct fallback and response parsing.
- Developed `Sub2ApiAdapter` with detection and user info fetching capabilities.
- Added comprehensive tests for `Sub2ApiAdapter` covering various scenarios.
- Introduced `TitleHintPlatform` type and detection logic based on HTML title.
- Implemented tests for title hint detection to identify platforms from URLs.
- Implement tests for TokenRouter downstream policy, covering allowedRouteIds and site weight multipliers.
- Add tests for TokenRouter patterns and model mapping, including regex pattern matching and model mapping precedence.
- Create tests for BrandIcon component to verify brand detection logic.
- Develop tests for Accounts page, focusing on rebind panel behavior and refresh action on error.
- Introduce fuzzy search utility tests to validate substring and subsequence matching.
- Implement navigation focus helpers tests for building and parsing focus paths.
- Add progressive render tests to ensure correct initial and next visible counts.
- Create route model candidates index tests to validate route candidate indexing and matching logic.
cita