# Changelog
All notable changes to this project will be documented in this file.
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
## [1.4.1] - 2026-01-15
### Fixed
- Updated PyPI package metadata to accurately reflect 40 parsers
- Fixed version number discrepancy between PyPI and GitHub
- Enhanced project description with all 9 parser categories
### Documentation
- Updated README with accurate parser count and category breakdown
- Improved PyPI description to include all parser categories
- Synchronized version info across all documentation
## [1.4.0] - 2026-01-14
### Added
- **Link Extractor parser** for domain scraping workflows:
- `extract_link_extractor` - Extract all links from HTML pages
- Multi-level crawling support (depth 1-5)
- Internal/external link filtering
- Automatic deduplication
- 3 presets: `default` (single page, internal only), `deep_crawl` (multi-level), `all_links` (internal + external)
- Optimized for Agent Orchestration pattern
- **40 parsers total** (was 39)
- **Content category**: 3 parsers (was 2) - Article, Text, Link Extractors
### Changed
- Updated parser count in README and documentation
- Added link_extractor to medium timeout category (120s default)
- Enhanced input schema for link_extractor with preset enum validation
### Documentation
- Added domain scraping workflow examples
- Updated README with v1.4.0 features
- Added Agent Orchestration usage patterns
## [1.3.1] - 2026-01-12
### Fixed
- **Critical timeout issue** - Heavy parsers (GoogleAI, Google Search, Translate, Trends) were timing out before results were ready
- Increased default timeouts based on parser category:
- **Fast parsers (60s)**: Perplexity, ChatGPT, DeepAI, YouTube Suggest, Reddit, etc.
- **Medium parsers (120s)**: YouTube Search/Video, Pinterest, Article Extractor
- **Slow parsers (180s)**: GoogleAI, Google Search/Translate/Trends, DeepL, Bing, all Instagram/TikTok
### Changed
- **Improved exponential backoff** in `wait_for_result()`:
- Start delay: 1.0s → 1.5s (give A-Parser initial processing time)
- Max delay: 3.0s → 5.0s (reduce Redis polling frequency)
- Better backoff curve: 1.5s → 1.8s → 2.2s → 2.6s → 3.1s → 3.7s → 4.5s → 5s
- Default timeout in `wait_for_result()`: 90s → 180s
- Tool descriptions now show parser-specific default timeouts
### Added
- `PARSER_TIMEOUT_CATEGORIES` - Parser categorization by expected processing time
- `get_default_timeout(parser_id)` - Get appropriate timeout for each parser
## [1.3.0] - 2026-01-12
### Added
- **10 new parsers** across 4 categories (29 → 39 parsers):
- **Content (2)**: article_extractor (Mozilla Readability), text_extractor (2000 queries/min)
- **Social expansion (6)**: Instagram parsers (profile, post, tag, geo, search) + TikTok profile
- **Analytics (1)**: google_trends (keyword trends, interest data)
- **Visual (1)**: pinterest_search (4000+ queries/min image search)
- **39 parsers total** organized in 9 categories:
- **FreeAI (6)**: perplexity, googleai, chatgpt, kimi, deepai, copilot
- **YouTube (6)**: youtube_video, youtube_search, youtube_suggest, youtube_channel_videos, youtube_channel_about, youtube_comments
- **Social (10)**: telegram_group, reddit_posts, reddit_post_info, reddit_comments, instagram_profile, instagram_post, instagram_tag, instagram_geo, instagram_search, tiktok_profile
- **Translation (4)**: google_translate, deepl_translate, bing_translate, yandex_translate
- **Search Engines (8)**: google_search, yandex_search, bing_search, duckduckgo_search, baidu_search, yahoo_search, rambler_search, you_search
- **Content (2)**: article_extractor, text_extractor
- **Analytics (1)**: google_trends
- **Visual (1)**: pinterest_search
- **Net (1)**: http
- **Instagram features**:
- Profile parsing with followers, posts, bio extraction
- Post parsing with up to 1200 comments
- Hashtag and location-based post collection
- Search for profiles, hashtags, locations
- Note: Some Instagram parsers require authentication cookies
- **Content extraction features**:
- Article extraction using Mozilla Readability algorithm
- Automatic HTML tag cleaning
- Multi-page content parsing
- CloudFlare bypass support
- High-speed text extraction (2000 queries/min)
- **Analytics features**:
- Google Trends keyword analysis
- Interest over time and by region
- Related queries (top and rising)
- Multiple keyword comparison support
- **Visual content features**:
- Pinterest image search (4000+ queries/min)
- Image links, titles, descriptions
- Domain and author information
- Up to 100 pages per query
### Changed
- Total parser count: 29 → 39 (+34% increase)
- Category count: 6 → 9
- Updated documentation with new parser examples
## [1.2.0] - 2026-01-12
### Added
- **8 Search Engine parsers**: Google, Yandex, Bing, DuckDuckGo, Baidu, Yahoo, Rambler, You.com
- **29 parsers total** (was 21) - new SE category:
- **FreeAI (6)**: perplexity, googleai, chatgpt, kimi, deepai, copilot
- **YouTube (6)**: youtube_video, youtube_search, youtube_suggest, youtube_channel_videos, youtube_channel_about, youtube_comments
- **Social (4)**: telegram_group, reddit_posts, reddit_post_info, reddit_comments
- **Translation (4)**: google_translate, deepl_translate, bing_translate, yandex_translate
- **Search Engines (8)**: google_search, yandex_search, bing_search, duckduckgo_search, baidu_search, yahoo_search, rambler_search, you_search
- **Net (1)**: http (URL fetcher)
- **Search Engine features**:
- Support for all search operators (site:, inurl:, intitle:, etc.)
- High-performance parsing (up to 7000 queries/min for Google/Yandex)
- Related keywords, ads, snippets collection
- Multi-language support
### Changed
- Total parser count: 21 → 29
- Updated documentation with Search Engine examples
## [1.1.0] - 2026-01-12
### Changed
- **Short server name**: Use `@ayga` instead of `@ayga-mcp-client` for easier usage in Claude/VS Code
### Added
- **21 parsers total** (was 11) - full sync with redis_wrapper:
- **FreeAI (6)**: perplexity, googleai, chatgpt, kimi, deepai, copilot
- **YouTube (6)**: youtube_video, youtube_search, youtube_suggest, youtube_channel_videos, youtube_channel_about, youtube_comments
- **Social (4)**: telegram_group, reddit_posts, reddit_post_info, reddit_comments
- **Translation (4)**: google_translate, deepl_translate, bing_translate, yandex_translate
- **Net (1)**: http (URL fetcher)
- **Correct tool naming conventions**:
- FreeAI/Search: `search_{id}` (e.g., `search_perplexity`)
- YouTube video/comments: `parse_{id}` (e.g., `parse_youtube_video`)
- YouTube data: `get_{id}` (e.g., `get_youtube_suggest`)
- Translation: `translate_{id}` (e.g., `translate_google_translate`)
- Social scraping: `scrape_{id}` (e.g., `scrape_telegram_group`)
- HTTP: `fetch_{id}` (e.g., `fetch_http`)
- **Extended parameters support**:
- **YouTube parsers**: `preset`, `interface_language`, `subtitles_language`, `comments_pages`, `pages_count`, `sort`
- **Translation parsers**: `from_language` (default: "auto"), `to_language` (default: "en")
- **Social parsers**: `pages_count`, `sort`, `max_comments_count`, `max_empty_posts`
- All parsers: `preset` parameter support
- **Complete A-Parser name mappings**:
- YouTube: `SE::YouTube::Video`, `SE::YouTube::Suggest`, `JS::Example::Youtube::Channel::Videos`, etc.
- Social: `Telegram::GroupScraper`, `Reddit::Posts`, `Reddit::PostInfo`, `Reddit::Comments`
- Translation: `SE::Google::Translate`, `DeepL::Translator`, `SE::Bing::Translator`, `SE::Yandex::Translate`
### Changed
- Replaced hardcoded 11-parser list with complete 21-parser registry
- Tool generation now uses parser-specific prefixes from PARSERS list
- Input schemas dynamically generated based on parser category
- `submit_parser_task()` now passes extended options to A-Parser task format
### Removed
- Removed non-existent parsers: `claude`, `gemini`, `grok`, `deepseek`, `google_search`, `bing_search`, `duckduckgo`
### Breaking Changes
- Tool names changed for consistency:
- Old: `search_youtube_search` → New: `search_youtube_search` (unchanged)
- New tools added: `parse_youtube_video`, `translate_google_translate`, etc.
- Some parsers from v1.0.x no longer available (claude, gemini, grok, deepseek, google_search, bing_search, duckduckgo)
## [1.0.2] - 2026-01-10
### Fixed
- Synchronized `__version__` in `__init__.py` with package version (was 1.0.0, now 1.0.2)
- Removed unused `tools/` directory from package structure
### Maintenance
- Package cleanup and documentation consistency improvements
## [1.0.1] - 2026-01-10
### Fixed
- **Critical**: Fixed parser execution 404 errors
- Changed `submit_parser_task` to use correct Redis structures API endpoint: `/structures/list/aparser_redis_api/lpush`
- Changed `get_task_result` to use correct Redis KV API endpoint: `/kv/aparser_redis_api:{task_id}`
- Updated task submission to match A-Parser task format: `[taskId, parser, preset, query, {}, {}]`
- Added parser ID mapping (e.g., `perplexity` → `FreeAI::Perplexity`)
- Improved result parsing to handle A-Parser response format
### Technical Details
- Previously used non-existent `/parsers/{id}/execute` endpoint
- Now uses actual redis_wrapper implementation (Redis queue + KV storage)
- All parser tools (`search_perplexity`, `search_chatgpt`, etc.) now work correctly
## [1.0.0] - 2026-01-10
### Added
- Initial release of **ayga-mcp-client** (renamed from redis-mcp-client)
- MCP server with stdio transport
- 11+ AI parser tools (Perplexity, ChatGPT, Claude, Gemini, etc.)
- HTTP client for Redis API
- Authentication via username/password or API key
- Progressive backoff for task polling
- Metadata tools (list_parsers, get_parser_info, health_check)
- CLI entrypoint: `ayga-mcp-client`
- Support for Claude Desktop, VS Code Copilot, Cursor
- **Published to PyPI**: https://pypi.org/project/ayga-mcp-client/
### Changed
- Package renamed: `redis-mcp-client` → `ayga-mcp-client`
- Module renamed: `redis_mcp_client` → `ayga_mcp_client`
- CLI command: `redis-mcp-client` → `ayga-mcp-client`
- MCP Registry ID: `io.github.ozand/ayga-mcp-client`
### Documentation
- README with quick start guide
- Installation: `pip install ayga-mcp-client`
- Examples for Claude Desktop and VS Code
