indent-parser

A typescript parser for Indent Markup Langauge.

API

import { parse, parseFile } from "indent-parser";

const { roots } = parse(source);
// roots: IndentNode[]
// IndentNode = { kind: string; value?: string | number | boolean | RefValue; attrs: Record<string, AttrValue>; children: IndentNode[] }
// AttrValue = string | number | boolean | RefValue
// RefValue = { type: "ref", raw: string }

// Reads `entry.inml` from disk and resolves any include lines within it
// (and transitively within included files) before building the tree.
const { roots: rootsFromDisk } = parseFile("entry.inml");

parse and parseFile throw IndentParseError (with line and, for parseFile, a file property) on malformed input.

Wire format

encodeWireFormat/decodeWireFormat convert between normal 4-space-indented Indent source and a compact single-line "wire format" that replaces newline + indentation-change with escape sequences. This is a pure text transform (no tokenizing/parsing) meant for transmitting .inml structure without embedded raw newlines.

import { encodeWireFormat, decodeWireFormat } from "indent-parser";

const wire = encodeWireFormat(source);
const roundTripped = decodeWireFormat(wire);
// roundTripped === source (modulo normalizing line endings to `\n`)

Escape sequences, one character per depth level crossed:

• \+ — indent one level deeper than the previous line.
• \- — dedent one level shallower than the previous line.
• \= — same depth as the previous line (a sibling); no indentation change needs to be communicated, but the marker is still required so the wire format never contains a raw embedded newline.

Runs of \+/\- are used for multi-level jumps, e.g. going from depth 0 to depth 2 encodes as \+\+, and back down to depth 0 as \-\-. Blank lines and comment-only (;...) lines carry no indentation of their own, so they are encoded as a same-depth (\=) transition.

decodeWireFormat throws IndentParseError if a \- run would dedent past depth 0.

Development

npm install
npm run build
npm test