parsers

Fast, safe, from-scratch input parsers — a C++-authored stdlib module (like socket and hashlib): the parsers use templates, concepts, and std::variant, which are beyond the current .purr subset. (requests is the first stdlib module written in pure cheatah.)

• parsers.json — a fast, SIMD-accelerated JSON parser:

  • import parsers.json.Parser as Parser — the reusable DOM parser: pooled zero-copy views or a self-contained owning Document; iterative grammar with a 1000-level depth cap (heap frames, never C++ recursion); a compile-time Validate switch; SIMD whitespace/string scanning.

  • the typed reader read<T>() parses straight into schema'd structs — purrc synthesizes the schema for .purr structs, so json.read(text, q) works on any struct you define.

• parsers.url — import parsers.url.Parser as Parser — the http(s) URL parser (scheme://host[:port][/path][?query]).

• parsers.html — import parsers.html — HTML escaping (parsers.html.escape / unescape) plus a tolerant tokenizing parser (parsers.html.parse returns the parse events as data): the rough equivalent of Python's html module + html.parser.

• parsers.xml — import parsers.xml — a tolerant XML reader that parses into a slab DOM navigated by integer node id (no pointers): parsers.xml.parse(text) then find/findall/iter + attr/text. Iterative (no stack overflow at any depth) and lenient. Built to feed cheatah's own tooling (e.g. reading Doxygen XML).

The module runs clean under ASan + UBSan and Valgrind against malformed input: truncated literals and containers, wrong closers, bad escapes and numbers, nesting past the depth cap.