Implemented a DOM tree transformation system for extracting and translating text content. Added logic for grouping text into semantic blocks to preserve context during translation.
What's New
- HTML document parsing — extracting all text content from the page
- Grouping related text into semantic blocks for more accurate translation
- Basic static content translation — automatic translation of text and meta tags
- Context preservation — the algorithm considers element relationships when grouping
Technical Details
The transformer analyzes the page's DOM tree, extracts text nodes, and groups them by semantic proximity. This allows sending not individual words, but entire phrases and paragraphs for translation, which significantly improves translation quality.
More about the implementation: Automatic AI translation and website multilingualism
Result
The system is capable of translating static content on Tilda websites while preserving structure and context.
