Agora

## Version 1.3.1 (2026 年 3 月 22 日) - 修正字幕抓取錯誤。 - bugs fixed ---- ## Version 1.3.0 (2026 年 3 月 15 日) **中文說明：** 本次更新是架構性升級，引入「規則驅動解析引擎」，以後端雲端化的解析規則取代原先硬編碼的頁面解析邏輯。同時修正多項規則編輯器的顯示問題，並完成解析規則的雲端同步與社群共享。 - **規則驅動解析引擎**：頁面解析邏輯從硬編碼 CSS 選擇器改為動態載入後端 `ExtractionRule`。核心模組 `ruleExtractor.js` 置於 `atlas` 公用層，供 agora 與 fetcher 共享。 - **YouTube 播放頁套用預設解析規則**：影片播放頁自動匹配「YouTube Watch Page」規則（`https://www.youtube.com/watch`），擷取結構化數據。 - **解析規則雲端同步**：規則儲存於後端（`ag_catalog.extraction_rules`），預設公開共享，所有用戶皆可瀏覽與套用他人規則。 - **任意頁面自訂規則**：用戶可在 Rule Editor 對任何頁面設定解析規則，建立後即公開共享。 - **字幕面板開啟優化**：YouTube 播放頁字幕面板改為每個影片 URL 只觸發一次，若 5 秒後未開啟則優雅重試，解決捲動觸發重複執行導致畫面跳動的嚴重 bug。 - **joinChildren 顯示修正**：合併子欄位輸出時不再插入多餘空白，用戶可透過前/後置詞自行控制分隔樣式。 - **交錯模式支援 joinChildren**：交錯（interleaved）列表模式現可切換合併子欄位，各筆資料呈現於同一行。 --- **English Description:** This is an architectural upgrade introducing a **Rule-Driven Extraction Engine**, replacing hardcoded page parsing with dynamically loaded backend ExtractionRules. Several Rule Editor display issues are fixed, and extraction rules now sync to the cloud with community sharing enabled. - **Rule-Driven Extraction Engine**: Page parsing now loads `ExtractionRule` from the backend dynamically. Core module `ruleExtractor.js` lives in the shared `atlas` layer, usable by both agora and fetcher. - **YouTube Watch Page Default Rule**: The video watch page automatically matches the "YouTube Watch Page" rule (`https://www.youtube.com/watch`) and extracts structured data. - **Cloud Sync for Extraction Rules**: Rules are stored in the backend (`ag_catalog.extraction_rules`), publicly shared by default — all users can browse and apply each other's rules. - **Custom Rules on Any Page**: Users can define extraction rules for any webpage in the Rule Editor; created rules are publicly shared. - **Caption Panel Open Optimization**: The YouTube caption panel now triggers only once per video URL. If it hasn't opened after 5 seconds, it retries once gracefully — fixing a serious bug where scroll events caused repeated triggering and screen jitter. - **joinChildren Spacing Fix**: Merging child fields no longer inserts extra whitespace. Users control separators via prefix/suffix settings. - **Interleaved Mode joinChildren Support**: Interleaved list mode now supports toggling child field merging, rendering each record on a single line. ## Version 1.2.5 (2026 年 3 月 11 日) **中文說明：** 本次更新包含了 Youtube 字幕擷取選擇器的調整，以適應 Youtube 網站結構的變化，並在規則編輯器頁面中新增了下載按鈕。 - **Youtube 字幕擷取器更新**：更新了 Youtube 字幕擷取器的 CSS 選擇器，以確保在 Youtube 網站介面調整後，仍能正確擷取字幕內容。 - **規則下載功能新增**：在規則編輯器頁面 (`RuleEditorPage.jsx`) 中新增了下載按鈕，允許用戶將當前編輯的規則下載為 Markdown 檔案。 --- **English Description:** This update includes adjustments to the YouTube caption extraction selectors to adapt to changes in the YouTube website structure, and a new download button has been added to the Rule Editor page. - **YouTube Caption Extractor Update**: Updated the CSS selectors for the YouTube caption extractor to ensure correct caption content retrieval after changes to the YouTube website interface. - **Rule Download Functionality Added**: A download button has been added to the Rule Editor page (`RuleEditorPage.jsx`), allowing users to download the currently edited rule as a Markdown file. ## Version 1.2.4 (2026 年 2 月 16 日) **中文說明：** 本次更新強化了規則編輯器的元素選取體驗，並改善了擷取數據的預覽呈現。 - **選取器 ↑↓ 導航**：在 CSS 選擇器旁新增上下導航按鈕，可在選取元素後快速切換至父元素或子元素，無需重新選取。 - **元素索引篩選**：當 CSS 選擇器匹配多筆元素時，可手動輸入索引號碼 (nth) 指定只取第幾個。 - **數據預覽模式切換**：預覽區新增卡片/JSON 切換按鈕。卡片模式以垂直排列呈現每筆記錄的 key-value，更適合側邊欄的窄版面；JSON 模式顯示原始結構。 - **編號可選**：預覽輸出可切換是否附加行號。 - **輸出檔名改用網頁標題**：Markdown 下載與 PDF 列印的預設檔名改為當前網頁標題，而非規則名稱。 - **修復高亮殘留**：元素導航後的高亮效果現在會在 2 秒後自動清除，避免污染頁面樣式。 --- **English Description:** This update enhances the Rule Editor's element selection experience and improves extracted data preview. - **Selector ↑↓ Navigation**: Added up/down navigation buttons next to the CSS selector input, allowing users to quickly switch to parent or child elements after selection without re-picking. - **Element Index Filter**: When a CSS selector matches multiple elements, manually enter an index number (nth) to pick only the Nth match. - **Data Preview Mode Toggle**: Added card/JSON toggle in the preview section. Card mode displays each record's key-value pairs vertically, better suited for the narrow side panel; JSON mode shows the raw structure. - **Line Numbers Toggle**: Preview output can optionally include line numbers. - **Output Filename Uses Page Title**: Markdown download and PDF print now default to the current page title instead of the rule name. - **Fixed Highlight Residue**: Navigation highlight now auto-clears after 2 seconds, preventing page style pollution. --- ## Version 1.2.3 (2026 年 2 月 10 日) **中文說明：** 本次更新引入了強大的「規則編輯器」(Rule Editor)，允許用戶在特定網頁上自訂內容擷取規則。用戶可以透過視覺化方式選擇網頁元素，定義資料擷取欄位及類型，並將這些規則保存以供重複使用。此功能特別針對同一網址模式下但內容動態變化的文章頁面，實現了自動化內容擷取與輸出 (Markdown/PDF)。 - **自訂內容擷取規則**：用戶現在可以在網頁上定義並保存自訂的 CSS 選擇器規則，精準擷取所需的內容。 - **可重用規則**：保存的規則可以應用於符合 URL 模式的其他頁面，極大提升了資料擷取的效率和靈活性。 - **自動化文章內容擷取**：對於例如部落格、新聞網站等具有相似結構但內容不同的文章頁面，只需設定一次規則，即可自動擷取所有匹配的文章內容。 --- **English Description:** This update introduces a powerful "Rule Editor" that allows users to customize content extraction rules on specific web pages. Users can visually select web elements, define data extraction fields and types, and save these rules for reuse. This feature is particularly designed for article pages with dynamic content under the same URL pattern, enabling automated content extraction and output (Markdown/PDF). - **Customizable Content Extraction Rules**: Users can now define and save custom CSS selector rules on web pages to precisely extract desired content. - **Reusable Rules**: Saved rules can be applied to other pages matching the URL pattern, significantly improving data extraction efficiency and flexibility. - **Automated Article Content Extraction**: For example, on blogs, news websites, or similar structured pages with varying content, a single rule setup can automate the extraction of a…