What problem does it solve?
Turn an HTML scraping intent into a valid, lint-clean KDL Schema DSL that extracts the right fields reliably, without hand-writing fragile parser code.
Core Features & Use Cases
- Declarative HTML scraping DSL generation: produces KDL v2.1 schemas for
(item)struct, (list)struct, (flat)struct, (table)struct, and (dict)struct using CSS-selector pipelines.
- Schema lint loop for correctness: iteratively fixes parse/type/struct issues based on linter output (text or JSON) until the schema is clean.
- Structured extraction patterns: supports table-style key/value extraction with
@table, @rows, @match, @value, and dynamic maps with @key/@value.
- Type conversions and fallbacks: enforces string/number/boolean flows using conversions like
to-int, to-float, to-bool, plus typed fallback for robustness.
- LLM-friendly workflow: guides schema iteration by mapping HTML structure to selectors and by using lint feedback as a tight validation mechanism.
Quick Start
Generate a lint-clean KDL HTML scraping schema by analyzing my provided HTML and extraction requirements, outputting a complete .kdl file that passes ssc-gen check -f json with no errors.