One API, every format
HTML, Markdown, YAML, JSON, CSS, CSV, GraphQL, TOML, XML, MDX, SQL — all behind the same parse() function with the same return shape and the same traversal API. Write your parse logic once and it works everywhere.
Lanexio Parser
One function. Eleven languages. Zero copies. Lanexio Parser is a modular, universal parsing framework built around a single, auto-detecting entry point. Hand it bytes — it identifies the format, routes to the right grammar, and returns a standardized, zero-copy syntax tree. Your application logic never needs to know what language it just parsed.
Zero-copy flat AST
Sixteen bytes per node. One ArrayBuffer for the entire tree. No per-node object allocation. No string copying. No GC pressure. O(1) subtree skip in preorder DFS. Fast because there’s nothing to allocate.
Install only what you need
Eleven independent packages under @lanexio/parser-grammar-*. Need HTML and Markdown? Install two. Don’t need SQL? Don’t install it. The universal entry point discovers what’s available at runtime.
Never throws. Never panics.
Malformed input produces LexError nodes — never an exception. Backed by 20 billion fuzz cases across 13 continuous-soak harnesses. Pass any byte sequence of any length and get a valid tree back.
Runs everywhere
No DOM dependency. No Node-specific APIs in the parse path. No framework imports. Browser, server, edge worker, test runner — same code, same guarantees.
Conformance with receipts
100% html5lib. 100% CommonMark + GFM. 100% yaml-test-suite. Every grammar carries its numbers publicly, and every number is an exact-count gate — if it regresses, the build fails.
Eleven grammars, one contract
Section titled “Eleven grammars, one contract”| Format | Package | Highlights |
|---|---|---|
| HTML | @lanexio/parser-grammar-html | 100% html5lib, all 23 insertion modes, adoption agency, sanitization |
| Markdown | @lanexio/parser-grammar-markdown | 100% CommonMark 0.31.2 + GFM, serializer, text-content extractor |
| MDX | @lanexio/parser-grammar-mdx | Markdown + JSX + ESM + expressions, built on the Markdown engine |
| YAML | @lanexio/parser-grammar-yaml | 100% yaml-test-suite (268 structural, 82 error rejection) |
| JSON | @lanexio/parser-grammar-json | RFC 8259, JSONC, JSON5 — three dialects in one package |
| CSS | @lanexio/parser-grammar-css | CSS Syntax Level 3, value validation, multi-stylesheet support |
| CSV | @lanexio/parser-grammar-csv | RFC 4180, configurable delimiters and quoting, CSV + TSV |
| GraphQL | @lanexio/parser-grammar-graphql | Auto-detects executable vs SDL, lenient parsing |
| TOML | @lanexio/parser-grammar-toml | TOML v1.0.0 and v1.1.0 with version selection |
| XML | @lanexio/parser-grammar-xml | XML 1.0 5th Edition, optional DTD validation, multi-byte encodings |
| SQL | @lanexio/parser-grammar-sql | Dialect-aware: PostgreSQL, MySQL, SQLite, T-SQL |
Start here
Section titled “Start here” Install and parse in under a minute pnpm, npm, or yarn. Install the packages you need and get your first parse tree.
Universal API `parse()`, `createParser()`, `register()`, `detectLanguage()` -- the whole story.
Flat AST architecture 16-byte nodes, zero-copy traversal, O(1) subtree skip, and the Shared Buffer Protocol.
Stability guarantees Never-throw, panic-free, 3-token recovery. 20B fuzz cases, zero failures.
All grammar reference docs Exact exports, options, return shapes, types, and layer contracts per package.
Conformance numbers Per-grammar conformance with exact-count gates and public test corpora.