new parser(-ish)

bun.lock

@@ -16,6 +16,8 @@
         "@lezer/highlight": "^1.2.1",
         "@lezer/lr": "^1.4.2",
         "@types/bun": "latest",
+        "diff": "^8.0.2",
+        "kleur": "^4.1.5",
       },
     },
   },
@@ -60,8 +62,12 @@
 
     "csstype": ["csstype@3.1.3", "", {}, "sha512-M1uQkMl8rQK/szD0LNhtqxIPLpimGm8sOBwU7lLnCpSbTyY3yeU1Vc7l4KT5zT4s/yOxHH5O7tIuuLOCnLADRw=="],
 
+    "diff": ["diff@8.0.2", "", {}, "sha512-sSuxWU5j5SR9QQji/o2qMvqRNYRDOcBTgsJ/DeCf4iSN4gW+gNMXM7wFIP+fdXZxoNiAnHUTGjCr+TSWXdRDKg=="],
+
     "hono": ["hono@4.10.4", "", {}, "sha512-YG/fo7zlU3KwrBL5vDpWKisLYiM+nVstBQqfr7gCPbSYURnNEP9BDxEMz8KfsDR9JX0lJWDRNc6nXX31v7ZEyg=="],
 
+    "kleur": ["kleur@4.1.5", "", {}, "sha512-o+NO+8WrRiQEE4/7nwRJhN1HWpVmJm511pBHUxPLtp0BUISzlBplORYSmTclCnJvQq2tKu/sgl3xVpkc7ZWuQQ=="],
+
     "reefvm": ["reefvm@git+https://git.nose.space/defunkt/reefvm#3e2e68b31f504347225a4d705c7568a0957d629e", { "peerDependencies": { "typescript": "^5" } }, "3e2e68b31f504347225a4d705c7568a0957d629e"],
 
     "style-mod": ["style-mod@4.1.3", "", {}, "sha512-i/n8VsZydrugj3Iuzll8+x/00GH2vnYsk1eomD8QiRrSAeW6ItbCQDtfXCeJHd0iwiNagqjQkvpvREEPtW3IoQ=="],

package.json

@@ -24,7 +24,9 @@
   "devDependencies": {
     "@lezer/highlight": "^1.2.1",
     "@lezer/lr": "^1.4.2",
-    "@types/bun": "latest"
+    "@types/bun": "latest",
+    "diff": "^8.0.2",
+    "kleur": "^4.1.5"
   },
   "prettier": {
     "semi": false,

src/parser/node.ts

@@ -0,0 +1,221 @@
+import { type Token, TokenType } from "./tokenizer2"
+
+export type NodeType =
+  | 'Program'
+  | 'Block'
+
+  | 'FunctionCall'
+  | 'FunctionCallOrIdentifier'
+  | 'FunctionCallWithBlock'
+  | 'PositionalArg'
+  | 'NamedArg'
+
+  | 'FunctionDef'
+  | 'Params'
+  | 'NamedParam'
+
+  | 'Null'
+  | 'Boolean'
+  | 'Number'
+  | 'String'
+  | 'StringFragment'
+  | 'CurlyString'
+  | 'DoubleQuote'
+  | 'EscapeSeq'
+  | 'Interpolation'
+  | 'Regex'
+  | 'Identifier'
+  | 'AssignableIdentifier'
+  | 'IdentifierBeforeDot'
+  | 'Word'
+  | 'Array'
+  | 'Dict'
+  | 'Comment'
+
+  | 'BinOp'
+  | 'ConditionalOp'
+  | 'ParenExpr'
+  | 'Assign'
+  | 'CompoundAssign'
+  | 'DotGet'
+  | 'PipeExpr'
+
+  | 'IfExpr'
+  | 'ElseIfExpr'
+  | 'ElseExpr'
+  | 'WhileExpr'
+  | 'TryExpr'
+  | 'CatchExpr'
+  | 'FinallyExpr'
+  | 'Throw'
+
+  | 'Eq'
+  | 'Modulo'
+  | 'Plus'
+  | 'Star'
+  | 'Slash'
+
+  | 'Import'
+  | 'Do'
+  | 'colon'
+  | 'keyword'
+  | 'operator'
+
+// TODO: remove this when we switch from lezer
+export const operators: Record<string, any> = {
+  // Logic
+  'and': 'And',
+  'or': 'Or',
+
+  // Bitwise
+  'band': 'Band',
+  'bor': 'Bor',
+  'bxor': 'Bxor',
+  '>>>': 'Ushr',
+  '>>': 'Shr',
+  '<<': 'Shl',
+
+  // Comparison
+  '>=': 'Gte',
+  '<=': 'Lte',
+  '>': 'Gt',
+  '<': 'Lt',
+  '!=': 'Neq',
+  '==': 'EqEq',
+
+  // Compound assignment operators
+  '??=': 'NullishEq',
+  '+=': 'PlusEq',
+  '-=': 'MinusEq',
+  '*=': 'StarEq',
+  '/=': 'SlashEq',
+  '%=': 'ModuloEq',
+
+  // Nullish coalescing
+  '??': 'NullishCoalesce',
+
+  // Math
+  '*': 'Star',
+  '**': 'StarStar',
+  '=': 'Eq',
+  '/': 'Slash',
+  '+': 'Plus',
+  '-': 'Minus',
+  '%': 'Modulo',
+
+  // Dotget
+  '.': 'Dot',
+
+  // Pipe
+  '|': 'operator',
+}
+
+export class SyntaxNode {
+  type: NodeType
+  from: number
+  to: number
+  parent: SyntaxNode | null
+  children: SyntaxNode[] = []
+
+  constructor(type: NodeType, from: number, to: number, parent: SyntaxNode | null = null) {
+    this.type = type
+    this.from = from
+    this.to = to
+    this.parent = parent
+  }
+
+  static from(token: Token, parent?: SyntaxNode): SyntaxNode {
+    return new SyntaxNode(TokenType[token.type] as NodeType, token.from, token.to, parent ?? null)
+  }
+
+  get name(): string {
+    return this.type
+  }
+
+  get isError(): boolean {
+    return false
+  }
+
+  get firstChild(): SyntaxNode | null {
+    return this.children[0] ?? null
+  }
+
+  get lastChild(): SyntaxNode | null {
+    return this.children.at(-1) ?? null
+  }
+
+  get nextSibling(): SyntaxNode | null {
+    if (!this.parent) return null
+    const siblings = this.parent.children
+    const index = siblings.indexOf(this)
+    return index >= 0 && index < siblings.length - 1 ? siblings[index + 1]! : null
+  }
+
+  get prevSibling(): SyntaxNode | null {
+    if (!this.parent) return null
+    const siblings = this.parent.children
+    const index = siblings.indexOf(this)
+    return index > 0 ? siblings[index - 1]! : null
+  }
+
+  add(node: SyntaxNode) {
+    node.parent = this
+    this.children.push(node)
+  }
+
+  push(...nodes: SyntaxNode[]): SyntaxNode {
+    nodes.forEach(child => child.parent = this)
+    this.children.push(...nodes)
+    return this
+  }
+
+  toString(): string {
+    return this.type
+  }
+}
+
+// Operator precedence (binding power) - higher = tighter binding
+export const precedence: Record<string, number> = {
+  // Logical
+  'or': 10,
+  'and': 20,
+
+  // Comparison
+  '==': 30,
+  '!=': 30,
+  '<': 30,
+  '>': 30,
+  '<=': 30,
+  '>=': 30,
+
+  // Nullish coalescing
+  '??': 35,
+
+  // Addition/Subtraction
+  '+': 40,
+  '-': 40,
+
+  // Multiplication/Division/Modulo
+  '*': 50,
+  '/': 50,
+  '%': 50,
+
+  // Bitwise
+  'band': 45,
+  'bor': 45,
+  'bxor': 45,
+  '<<': 45,
+  '>>': 45,
+  '>>>': 45,
+
+  // Exponentiation (right-associative)
+  '**': 60,
+}
+
+export const conditionals = new Set([
+  '==', '!=', '<', '>', '<=', '>=', '??', 'and', 'or'
+])
+
+export const compounds = [
+  '??=', '+=', '-=', '*=', '/=', '%='
+]

src/parser/parser2.ts

@@ -0,0 +1,899 @@
+import { Scanner, type Token, TokenType } from './tokenizer2'
+import { SyntaxNode, operators, precedence, conditionals, compounds } from './node'
+import { globals } from './tokenizer'
+import { parseString } from './stringParser'
+
+const $T = TokenType
+
+export const parse = (input: string): SyntaxNode => {
+  const parser = new Parser()
+  return parser.parse(input)
+}
+
+class Scope {
+  parent?: Scope
+  set = new Set<string>()
+
+  constructor(parent?: Scope) {
+    this.parent = parent
+
+    // no parent means this is global scope
+    if (!parent) for (const name of globals) this.add(name)
+  }
+
+  add(key: string) {
+    this.set.add(key)
+  }
+
+  has(key: string): boolean {
+    return this.set.has(key) || this.parent?.has(key) || false
+  }
+}
+
+export class Parser {
+  tokens: Token[] = []
+  pos = 0
+  inParens = 0
+  input = ''
+  scope = new Scope
+  inTestExpr = false
+
+  parse(input: string): SyntaxNode {
+    const scanner = new Scanner()
+    this.tokens = scanner.tokenize(input)
+    this.pos = 0
+    this.input = input
+    this.scope = new Scope()
+    this.inTestExpr = false
+
+    const node = new SyntaxNode('Program', 0, input.length)
+
+    while (!this.isEOF()) {
+      if (this.is($T.Newline) || this.is($T.Semicolon)) {
+        this.next()
+        continue
+      }
+
+      const prevPos = this.pos
+      const stmt = this.statement()
+      if (stmt) node.add(stmt)
+
+      if (this.pos === prevPos && !this.isEOF())
+        throw "parser didn't advance - you need to call next()\n\n       ${this.input}\n"
+    }
+
+    return node
+  }
+
+  // 
+  // parse foundation nodes - statements, expressions
+  //
+
+  // statement is a line of code
+  statement(): SyntaxNode | null {
+    if (this.is($T.Comment))
+      return this.comment()
+
+    while (this.is($T.Newline) || this.is($T.Semicolon))
+      this.next()
+
+    if (this.isEOF() || this.isExprEndKeyword())
+      return null
+
+    return this.expression()
+  }
+
+  // expressions can be found in four places:
+  // 1. line of code
+  // 2. right side of assignment
+  // 3. if/while conditions
+  // 4. inside (parens)
+  expression(allowPipe = true): SyntaxNode {
+    let expr
+
+    // x = value
+    if (this.is($T.Identifier) && (
+      this.nextIs($T.Operator, '=') || compounds.some(x => this.nextIs($T.Operator, x))
+    ))
+      expr = this.assign()
+
+    // if, while, do, etc
+    else if (this.is($T.Keyword))
+      expr = this.keywords()
+
+    // dotget
+    else if (this.nextIs($T.Operator, '.'))
+      expr = this.dotGetFunctionCall()
+
+    // echo hello world
+    else if (this.is($T.Identifier) && !this.nextIs($T.Operator) && !this.nextIsExprEnd())
+      expr = this.functionCall()
+
+    // bare-function-call
+    else if (this.is($T.Identifier) && this.nextIsExprEnd())
+      expr = this.functionCallOrIdentifier()
+
+    // everything else
+    else
+      expr = this.exprWithPrecedence()
+
+    // check for destructuring
+    if (expr.type === 'Array' && this.is($T.Operator, '='))
+      return this.destructure(expr)
+
+    // check for parens function call
+    // ex: (ref my-func) my-arg
+    if (expr.type === 'ParenExpr' && !this.isExprEnd())
+      expr = this.functionCall(expr)
+
+    // one | echo
+    if (allowPipe && this.isPipe())
+      return this.pipe(expr)
+
+    // regular
+    else
+      return expr
+  }
+
+  // piping | stuff | is | cool
+  pipe(left: SyntaxNode): SyntaxNode {
+    const canLookPastNewlines = this.inParens === 0
+    const parts: SyntaxNode[] = [left]
+
+    while (this.isPipe()) {
+      // consume newlines before pipe (only if not in parens)
+      if (canLookPastNewlines) {
+        while (this.is($T.Newline)) this.next()
+      }
+
+      const pipeOp = this.op('|')
+      pipeOp.type = 'operator'
+      parts.push(pipeOp)
+
+      // consume newlines after pipe (only if not in parens)
+      if (canLookPastNewlines) {
+        while (this.is($T.Newline)) this.next()
+      }
+
+      // parse right side - don't allow nested pipes
+      parts.push(this.expression(false))
+    }
+
+    const node = new SyntaxNode('PipeExpr', parts[0]!.from, parts.at(-1)!.to)
+    return node.push(...parts)
+  }
+
+  // Pratt parser - parses expressions with precedence climbing
+  // bp = binding precedence
+  exprWithPrecedence(minBp = 0): SyntaxNode {
+    let left = this.value()
+
+    // infix operators with precedence
+    while (this.is($T.Operator)) {
+      const op = this.current().value!
+      const bp = precedence[op]
+
+      // operator has lower precedence than required, stop
+      if (bp === undefined || bp < minBp) break
+
+      const opNode = this.op()
+
+      // right-associative operators (like **) use same bp, others use bp + 1
+      const nextMinBp = op === '**' ? bp : bp + 1
+
+      // parse right-hand side with higher precedence
+      const right = this.exprWithPrecedence(nextMinBp)
+
+      const nodeType = conditionals.has(op) ? 'ConditionalOp' : 'BinOp'
+      const node = new SyntaxNode(nodeType, left.from, right.to)
+
+      node.push(left, opNode, right)
+      left = node
+    }
+
+    return left
+  }
+
+  // if, while, do, etc
+  keywords(): SyntaxNode {
+    if (this.is($T.Keyword, 'if'))
+      return this.if()
+
+    if (this.is($T.Keyword, 'while'))
+      return this.while()
+
+    if (this.is($T.Keyword, 'do'))
+      return this.do()
+
+    if (this.is($T.Keyword, 'try'))
+      return this.try()
+
+    if (this.is($T.Keyword, 'throw'))
+      return this.throw()
+
+    if (this.is($T.Keyword, 'import'))
+      return this.import()
+
+    return this.expect($T.Keyword, 'if/while/do/import') as never
+  }
+
+  // value can be an atom or a (parens that gets turned into an atom)
+  // values are used in a few places:
+  // 1. function arguments
+  // 2. array/dict members
+  // 3. binary operations
+  // 4. anywhere an expression can be used
+  value(): SyntaxNode {
+    if (this.is($T.OpenParen))
+      return this.parens()
+
+    if (this.is($T.OpenBracket))
+      return this.arrayOrDict()
+
+    // dotget
+    if (this.nextIs($T.Operator, '.'))
+      return this.dotGet()
+
+    return this.atom()
+  }
+
+  //
+  // parse specific nodes
+  //
+
+  // [ 1 2 3 ]
+  array(): SyntaxNode {
+    const open = this.expect($T.OpenBracket)
+
+    const values = []
+    while (!this.is($T.CloseBracket) && !this.isEOF()) {
+      if (this.is($T.Semicolon) || this.is($T.Newline)) {
+        this.next()
+        continue
+      }
+
+      if (this.is($T.Comment)) {
+        values.push(this.comment())
+        continue
+      }
+
+      values.push(this.value())
+    }
+
+    const close = this.expect($T.CloseBracket)
+
+    const node = new SyntaxNode('Array', open.from, close.to)
+    return node.push(...values)
+  }
+
+  // which are we dealing with? ignores leading newlines and comments
+  arrayOrDict(): SyntaxNode {
+    let peek = 1
+    let curr = this.peek(peek++)
+    let isDict = false
+
+    while (curr && curr.type !== $T.CloseBracket) {
+      // definitely a dict
+      if (curr.type === $T.NamedArgPrefix) {
+        isDict = true
+        break
+      }
+
+      // empty dict
+      if (curr.type === $T.Operator && curr.value === '=') {
+        isDict = true
+        break
+      }
+
+      // probably an array
+      if (curr.type !== $T.Comment && curr.type !== $T.Semicolon && curr.type !== $T.Newline)
+        break
+
+      curr = this.peek(peek++)
+    }
+
+    return isDict ? this.dict() : this.array()
+  }
+
+  // x = true
+  assign(): SyntaxNode {
+    const ident = this.assignableIdentifier()
+    const opToken = this.current()!
+    const op = this.op()
+    const expr = this.expression()
+
+    const node = new SyntaxNode(
+      opToken.value === '=' ? 'Assign' : 'CompoundAssign',
+      ident.from,
+      expr.to
+    )
+
+    return node.push(ident, op, expr)
+  }
+
+  // identifier used in assignment (TODO: legacy lezer quirk)
+  assignableIdentifier(): SyntaxNode {
+    const token = this.expect($T.Identifier)
+    this.scope.add(token.value!)
+    const node = SyntaxNode.from(token)
+    node.type = 'AssignableIdentifier'
+    return node
+  }
+
+  // atoms are the basic building blocks: literals, identifiers, words
+  atom() {
+    if (this.is($T.String))
+      return this.string()
+
+    if (this.isAny($T.Null, $T.Boolean, $T.Number, $T.Identifier, $T.Word, $T.Regex))
+      return SyntaxNode.from(this.next())
+
+    const next = this.next()
+    throw `[atom] unexpected token ${TokenType[next.type]}: ${JSON.stringify(next)}\n\n       ${this.input}\n`
+  }
+
+  // blocks in if, do, special calls, etc
+  // `: something end` 
+  //
+  // `blockNode` determines whether we return [colon, BlockNode, end] or
+  // just a list of statements like [colon, stmt1, stmt2, end]
+  block(blockNode = true): SyntaxNode[] {
+    const stmts: SyntaxNode[] = []
+    const colon = this.colon()
+
+    while (!this.isExprEndKeyword() && !this.isEOF()) {
+      const stmt = this.statement()
+      if (stmt) stmts.push(stmt)
+    }
+
+    const out = [colon]
+
+    if (blockNode) {
+      const block = new SyntaxNode('Block', stmts[0]!.from, stmts.at(-1)!.to)
+      block.push(...stmts)
+      out.push(block)
+    } else {
+      out.push(...stmts)
+    }
+
+    return out
+  }
+
+  // catch err: block
+  catch(): SyntaxNode {
+    const keyword = this.keyword('catch')
+
+    let catchVar
+    if (this.is($T.Identifier))
+      catchVar = this.identifier()
+
+    const block = this.block()
+
+    const node = new SyntaxNode('CatchExpr', keyword.from, block.at(-1)!.to)
+
+    node.push(keyword)
+    if (catchVar) node.push(catchVar)
+    return node.push(...block)
+  }
+
+  // colon
+  colon(): SyntaxNode {
+    const colon = SyntaxNode.from(this.expect($T.Colon))
+    colon.type = 'colon' // TODO lezer legacy
+    return colon
+  }
+
+  // # comment
+  comment(): SyntaxNode {
+    return SyntaxNode.from(this.expect($T.Comment))
+  }
+
+  // [ a b c ] = [ 1 2 3 ]
+  destructure(array: SyntaxNode): SyntaxNode {
+    const eq = this.op('=')
+    const val = this.expression()
+
+    for (const ident of array.children) {
+      const varName = this.input.slice(ident.from, ident.to)
+      this.scope.add(varName)
+    }
+
+    const node = new SyntaxNode('Assign', array.from, val.to)
+    return node.push(array, eq, val)
+  }
+
+  // [ a=1 b=true c='three' ]
+  dict(): SyntaxNode {
+    const open = this.expect($T.OpenBracket)
+
+    // empty dict [=] or [ = ]
+    if (this.is($T.Operator, '=') && this.nextIs($T.CloseBracket)) {
+      const _op = this.next()
+      const close = this.next()
+      return new SyntaxNode('Dict', open.from, close.to)
+    }
+
+    const values = []
+    while (!this.is($T.CloseBracket) && !this.isEOF()) {
+      if (this.is($T.Semicolon) || this.is($T.Newline)) {
+        this.next()
+        continue
+      }
+
+      if (this.is($T.Comment)) {
+        values.push(this.comment())
+        continue
+      }
+
+      if (this.is($T.NamedArgPrefix))
+        values.push(this.namedArg())
+      else
+        values.push(this.value())
+    }
+
+    const close = this.expect($T.CloseBracket)
+
+    const node = new SyntaxNode('Dict', open.from, close.to)
+    return node.push(...values)
+  }
+
+  // FunctionDef `do x y: something end`
+  do(): SyntaxNode {
+    const doNode = this.keyword('do')
+    doNode.type = 'Do'
+    this.scope = new Scope(this.scope)
+
+    const params = []
+    while (!this.is($T.Colon) && !this.isExprEnd()) {
+      let varName = this.current().value!
+      if (varName.endsWith('=')) varName = varName.slice(0, varName.length - 1)
+      this.scope.add(varName)
+
+      let arg
+      if (this.is($T.Identifier))
+        arg = this.identifier()
+      else if (this.is($T.NamedArgPrefix))
+        arg = this.namedParam()
+      else
+        throw `[do] expected Identifier or NamedArgPrefix, got ${JSON.stringify(this.current())}\n\n       ${this.input}\n`
+
+      params.push(arg)
+    }
+
+    const block = this.block(false)
+    let catchNode, finalNode
+
+    if (this.is($T.Keyword, 'catch'))
+      catchNode = this.catch()
+
+    if (this.is($T.Keyword, 'finally'))
+      finalNode = this.finally()
+
+    let end = this.keyword('end')
+
+    let last = block.at(-1)
+    if (finalNode) last = finalNode.children.at(-1)!
+    else if (catchNode) last = catchNode.children.at(-1)!
+
+    const node = new SyntaxNode('FunctionDef', doNode.from, last!.to)
+
+    node.add(doNode)
+
+    const paramsNode = new SyntaxNode(
+      'Params',
+      params[0]?.from ?? 0,
+      params.at(-1)?.to ?? 0
+    )
+
+    if (params.length) paramsNode.push(...params)
+    node.add(paramsNode)
+
+    this.scope = this.scope.parent!
+
+    node.push(...block)
+
+    if (catchNode) node.push(catchNode)
+    if (finalNode) node.push(finalNode)
+
+    return node.push(end)
+  }
+
+  // config.path
+  dotGet(): SyntaxNode {
+    const left = this.identifier()
+    const ident = this.input.slice(left.from, left.to)
+
+    // not in scope, just return Word
+    if (!this.scope.has(ident))
+      return this.word(left)
+
+    if (left.type === 'Identifier') left.type = 'IdentifierBeforeDot'
+
+    let parts = []
+    while (this.is($T.Operator, '.')) {
+      this.next()
+      parts.push(this.is($T.OpenParen) ? this.parens() : this.atom())
+    }
+
+    // TODO lezer legacy - we can do a flat DotGet if we remove this
+    const nodes = parts.length > 1 ? collapseDotGets(parts) : undefined
+
+    const node = new SyntaxNode('DotGet', left.from, parts.at(-1)!.to)
+    return nodes ? node.push(left, nodes!) : node.push(left, ...parts)
+  }
+
+  // dotget in a statement/expression (something.blah) or (something.blah arg1)
+  dotGetFunctionCall(): SyntaxNode {
+    const dotGet = this.dotGet()
+
+    // dotget not in scope, regular Word
+    if (dotGet.type === 'Word') return dotGet
+
+    if (this.isExprEnd())
+      return this.functionCallOrIdentifier(dotGet)
+    else
+      return this.functionCall(dotGet)
+  }
+
+  // can be used in functions or try block
+  finally(): SyntaxNode {
+    const keyword = this.keyword('finally')
+    const block = this.block()
+    const node = new SyntaxNode('FinallyExpr', keyword.from, block.at(-1)!.to)
+
+    return node.push(keyword, ...block)
+  }
+
+  // you're lookin at it
+  functionCall(fn?: SyntaxNode): SyntaxNode {
+    const ident = fn ?? this.identifier()
+
+    const args: SyntaxNode[] = []
+    while (!this.isExprEnd() && !this.is($T.Operator, '|')) {
+      if (this.is($T.NamedArgPrefix)) {
+        args.push(this.namedArg())
+      } else {
+        // 'do' is the only keyword allowed as a function argument
+        const val = this.is($T.Keyword, 'do') ? this.do() : this.value()
+        const arg = new SyntaxNode('PositionalArg', val.from, val.to)
+        arg.add(val)
+        args.push(arg)
+      }
+    }
+
+    const node = new SyntaxNode('FunctionCall', ident.from, (args.at(-1) || ident).to)
+    node.push(ident, ...args)
+
+    if (!this.inTestExpr && this.is($T.Colon)) {
+      const block = this.block()
+      const end = this.keyword('end')
+      const blockNode = new SyntaxNode('FunctionCallWithBlock', node.from, end.to)
+      return blockNode.push(node, ...block, end)
+    }
+
+    return node
+  }
+
+  // bare identifier in an expression
+  functionCallOrIdentifier(inner?: SyntaxNode) {
+    if (!inner && this.nextIs($T.Operator, '.')) {
+      inner = this.dotGet()
+
+      // if the dotGet was just a Word, bail
+      if (inner.type === 'Word') return inner
+    }
+
+    inner ??= this.identifier()
+
+    const wrapper = new SyntaxNode('FunctionCallOrIdentifier', inner.from, inner.to)
+    wrapper.push(inner)
+
+    if (!this.inTestExpr && this.is($T.Colon)) {
+      const block = this.block()
+      const end = this.keyword('end')
+      const node = new SyntaxNode('FunctionCallWithBlock', wrapper.from, end.to)
+      return node.push(wrapper, ...block, end)
+    }
+
+    return wrapper
+  }
+
+  // function and variable names
+  identifier(): SyntaxNode {
+    return SyntaxNode.from(this.expect($T.Identifier))
+  }
+
+  // if something: blah end
+  // if something: blah else: blah end
+  // if something: blah else if something: blah else: blah end
+  if(): SyntaxNode {
+    const ifNode = this.keyword('if')
+    const test = this.testExpr()
+    const ifBlock = this.block()
+
+    const node = new SyntaxNode('IfExpr', ifNode.from, ifBlock.at(-1)!.to)
+    node.push(ifNode, test)
+    node.push(...ifBlock)
+
+    while (this.is($T.Keyword, 'else') && this.nextIs($T.Keyword, 'if')) {
+      const elseWord = this.keyword('else')
+      const ifWord = this.keyword('if')
+      const elseIfTest = this.testExpr()
+      const elseIfBlock = this.block()
+      const elseIfNode = new SyntaxNode('ElseIfExpr', ifBlock.at(-1)!.from, elseIfBlock.at(-1)!.to)
+      elseIfNode.push(elseWord, ifWord, elseIfTest)
+      elseIfNode.push(...elseIfBlock)
+      node.push(elseIfNode)
+    }
+
+    if (this.is($T.Keyword, 'else') && this.nextIs($T.Colon)) {
+      const elseWord = this.keyword('else')
+      const elseBlock = this.block()
+      const elseNode = new SyntaxNode('ElseExpr', ifBlock.at(-1)!.from, elseBlock.at(-1)!.to)
+      elseNode.push(elseWord)
+      elseNode.push(...elseBlock)
+      node.push(elseNode)
+    }
+
+    return node.push(this.keyword('end'))
+  }
+
+  import(): SyntaxNode {
+    const keyword = this.keyword('import')
+
+    const args: SyntaxNode[] = []
+    while (!this.isExprEnd()) {
+      if (this.is($T.NamedArgPrefix)) {
+        const prefix = SyntaxNode.from(this.next())
+        const val = this.value()
+        const arg = new SyntaxNode('NamedArg', prefix.from, val.to)
+        arg.push(prefix, val)
+        args.push(arg)
+      } else {
+        args.push(this.identifier())
+      }
+    }
+
+    const node = new SyntaxNode('Import', keyword.from, args.at(-1)!.to)
+    node.add(keyword)
+    return node.push(...args)
+  }
+
+  // if, while, do, etc
+  keyword(name: string): SyntaxNode {
+    const node = SyntaxNode.from(this.expect($T.Keyword, name))
+    node.type = 'keyword' // TODO lezer legacy
+    return node
+  }
+
+  // abc= true
+  namedArg(): SyntaxNode {
+    const prefix = SyntaxNode.from(this.expect($T.NamedArgPrefix))
+    const val = this.value()
+    const node = new SyntaxNode('NamedArg', prefix.from, val.to)
+    return node.push(prefix, val)
+  }
+
+  // abc= null|true|123|'hi'
+  namedParam(): SyntaxNode {
+    const prefix = SyntaxNode.from(this.expect($T.NamedArgPrefix))
+    const val = this.value()
+
+    if (!['Null', 'Boolean', 'Number', 'String'].includes(val.type))
+      throw `[namedParam] default value must be Null|Bool|Num|Str, got ${val.type}\n\n       ${this.input}\n`
+
+    const node = new SyntaxNode('NamedParam', prefix.from, val.to)
+    return node.push(prefix, val)
+  }
+
+  // operators like + - =
+  op(op?: string): SyntaxNode {
+    const token = op ? this.expect($T.Operator, op) : this.expect($T.Operator)
+    const name = operators[token.value!]
+    if (!name) throw `[op] operator not registered: ${token.value!}\n\n       ${this.input}\n`
+    return new SyntaxNode(name, token.from, token.to)
+  }
+
+  // ( expressions in parens )
+  parens(): SyntaxNode {
+    this.inParens++
+    const open = this.expect($T.OpenParen)
+    const child = this.expression()
+    const close = this.expect($T.CloseParen)
+    this.inParens--
+
+    const node = new SyntaxNode('ParenExpr', open.from, close.to)
+    node.add(child)
+
+    return node
+  }
+
+  // 'hell yes' "hell no" { hell if i know }
+  string(): SyntaxNode {
+    const token = this.expect($T.String)
+    return parseString(this.input, token.from, token.to, this)
+  }
+
+  // if TEST: blah end
+  testExpr(): SyntaxNode {
+    this.inTestExpr = true
+    const expr = this.expression()
+    this.inTestExpr = false
+    return expr
+  }
+
+  // throw blah
+  throw(): SyntaxNode {
+    const keyword = this.keyword('throw')
+    const val = this.value()
+    const node = new SyntaxNode('Throw', keyword.from, val.to)
+    return node.push(keyword, val)
+  }
+
+  // try: blah catch e: blah end
+  try(): SyntaxNode {
+    const tryNode = this.keyword('try')
+    const tryBlock = this.block()
+    let last = tryBlock.at(-1)
+    let catchNode, finalNode
+
+    if (this.is($T.Keyword, 'catch'))
+      catchNode = this.catch()
+
+    if (this.is($T.Keyword, 'finally'))
+      finalNode = this.finally()
+
+    const end = this.keyword('end')
+
+    if (finalNode) last = finalNode.children.at(-1)
+    else if (catchNode) last = catchNode.children.at(-1)
+
+    const node = new SyntaxNode('TryExpr', tryNode.from, last!.to)
+    node.push(tryNode, ...tryBlock)
+
+    if (catchNode)
+      node.push(catchNode)
+
+    if (finalNode)
+      node.push(finalNode)
+
+    return node.push(end)
+  }
+
+  // while test: blah end
+  while(): SyntaxNode {
+    const keyword = this.keyword('while')
+    const test = this.testExpr()
+    const block = this.block()
+    const end = this.keyword('end')
+
+    const node = new SyntaxNode('WhileExpr', keyword.from, end.to)
+    return node.push(keyword, test, ...block, end)
+  }
+
+  // readme.txt (when `readme` isn't in scope)
+  word(start?: SyntaxNode): SyntaxNode {
+    const parts = [start ?? this.expect($T.Word)]
+
+    while (this.is($T.Operator, '.')) {
+      this.next()
+      if (this.isAny($T.Word, $T.Identifier, $T.Number))
+        parts.push(this.next())
+    }
+
+    return new SyntaxNode('Word', parts[0]!.from, parts.at(-1)!.to)
+  }
+
+  // 
+  // helpers
+  //
+
+  current(): Token {
+    return this.tokens[this.pos] || { type: TokenType.Newline, from: 0, to: 0 }
+  }
+
+  peek(offset = 1): Token | undefined {
+    return this.tokens[this.pos + offset]
+  }
+
+  // look past newlines to check for a specific token
+  peekPastNewlines(type: TokenType, value?: string): boolean {
+    let offset = 1
+    let peek = this.peek(offset)
+
+    while (peek && peek.type === $T.Newline)
+      peek = this.peek(++offset)
+
+    if (!peek || peek.type !== type) return false
+    if (value !== undefined && peek.value !== value) return false
+    return true
+  }
+
+  next(): Token {
+    const token = this.current()
+    this.pos++
+    return token
+  }
+
+  is(type: TokenType, value?: string): boolean {
+    const token = this.current()
+    if (!token || token.type !== type) return false
+    if (value !== undefined && token.value !== value) return false
+    return true
+  }
+
+  isAny(...type: TokenType[]): boolean {
+    return type.some(x => this.is(x))
+  }
+
+  nextIs(type: TokenType, value?: string): boolean {
+    const token = this.peek()
+    if (!token || token.type !== type) return false
+    if (value !== undefined && token.value !== value) return false
+    return true
+  }
+
+  nextIsAny(...type: TokenType[]): boolean {
+    return type.some(x => this.nextIs(x))
+  }
+
+  isExprEnd(): boolean {
+    return this.isAny($T.Colon, $T.Semicolon, $T.Newline, $T.CloseParen, $T.CloseBracket) ||
+      this.isExprEndKeyword() || !this.current()
+  }
+
+  nextIsExprEnd(): boolean {
+    // pipes act like expression end for function arg parsing
+    if (this.nextIs($T.Operator, '|'))
+      return true
+
+    return this.nextIsAny($T.Colon, $T.Semicolon, $T.Newline, $T.CloseBracket, $T.CloseParen) ||
+      this.nextIs($T.Keyword, 'end') || this.nextIs($T.Keyword, 'else') ||
+      this.nextIs($T.Keyword, 'catch') || this.nextIs($T.Keyword, 'finally') ||
+      !this.peek()
+  }
+
+  isExprEndKeyword(): boolean {
+    return this.is($T.Keyword, 'end') || this.is($T.Keyword, 'else') ||
+      this.is($T.Keyword, 'catch') || this.is($T.Keyword, 'finally')
+  }
+
+  isPipe(): boolean {
+    // inside parens, only look for pipes on same line (don't look past newlines)
+    const canLookPastNewlines = this.inParens === 0
+
+    return this.is($T.Operator, '|') ||
+      (canLookPastNewlines && this.peekPastNewlines($T.Operator, '|'))
+  }
+
+  expect(type: TokenType, value?: string): Token | never {
+    if (!this.is(type, value)) {
+      const token = this.current()
+      throw `expected ${TokenType[type]}${value ? ` "${value}"` : ''}, got ${TokenType[token?.type || 0]}${token?.value ? ` "${token.value}"` : ''} at position ${this.pos}\n\n       ${this.input}\n`
+    }
+    return this.next()
+  }
+
+  isEOF(): boolean {
+    return this.pos >= this.tokens.length
+  }
+}
+
+// TODO lezer legacy
+function collapseDotGets(origNodes: SyntaxNode[]): SyntaxNode {
+  const nodes = [...origNodes]
+  let right = nodes.pop()!
+
+  while (nodes.length > 0) {
+    const left = nodes.pop()!
+
+    if (left.type === 'Identifier') left.type = 'IdentifierBeforeDot'
+
+    const dot = new SyntaxNode("DotGet", left.from, right.to);
+    dot.push(left, right)
+
+    right = dot
+  }
+
+  return right
+}

src/parser/stringParser.ts

@@ -0,0 +1,258 @@
+import { SyntaxNode } from './node'
+
+/**
+ * Parse string contents into fragments, interpolations, and escape sequences.
+ *
+ * Input: full string including quotes, e.g. "'hello $name'"
+ * Output: SyntaxNode tree with StringFragment, Interpolation, EscapeSeq children
+ */
+export const parseString = (input: string, from: number, to: number, parser: any): SyntaxNode => {
+  const stringNode = new SyntaxNode('String', from, to)
+  const content = input.slice(from, to)
+
+  // Determine string type
+  const firstChar = content[0]
+
+  // Double-quoted strings: no interpolation or escapes
+  if (firstChar === '"') {
+    const fragment = new SyntaxNode('DoubleQuote', from, to)
+    stringNode.add(fragment)
+    return stringNode
+  }
+
+  // Curly strings: interpolation but no escapes
+  if (firstChar === '{') {
+    parseCurlyString(stringNode, input, from, to, parser)
+    return stringNode
+  }
+
+  // Single-quoted strings: interpolation and escapes
+  if (firstChar === "'") {
+    parseSingleQuoteString(stringNode, input, from, to, parser)
+    return stringNode
+  }
+
+  throw `Unknown string type starting with: ${firstChar}`
+}
+
+/**
+ * Parse single-quoted string: 'hello $name\n'
+ * Supports: interpolation ($var, $(expr)), escape sequences (\n, \$, etc)
+ */
+const parseSingleQuoteString = (stringNode: SyntaxNode, input: string, from: number, to: number, parser: any) => {
+  let pos = from + 1 // Skip opening '
+  let fragmentStart = pos
+
+  while (pos < to - 1) { // -1 to skip closing '
+    const char = input[pos]
+
+    // Escape sequence
+    if (char === '\\' && pos + 1 < to - 1) {
+      // Push accumulated fragment
+      if (pos > fragmentStart) {
+        const frag = new SyntaxNode('StringFragment', fragmentStart, pos)
+        stringNode.add(frag)
+      }
+
+      // Add escape sequence node
+      const escNode = new SyntaxNode('EscapeSeq', pos, pos + 2)
+      stringNode.add(escNode)
+
+      pos += 2
+      fragmentStart = pos
+      continue
+    }
+
+    // Interpolation
+    if (char === '$') {
+      // Push accumulated fragment
+      if (pos > fragmentStart) {
+        const frag = new SyntaxNode('StringFragment', fragmentStart, pos)
+        stringNode.add(frag)
+      }
+
+      pos++ // Skip $
+
+      // Parse interpolation content
+      if (input[pos] === '(') {
+        // Expression interpolation: $(expr)
+        const interpStart = pos - 1 // Include the $
+        const exprResult = parseInterpolationExpr(input, pos, parser)
+        const interpNode = new SyntaxNode('Interpolation', interpStart, exprResult.endPos)
+        interpNode.add(exprResult.node)
+        stringNode.add(interpNode)
+        pos = exprResult.endPos
+      } else {
+        // Variable interpolation: $name
+        const interpStart = pos - 1
+        const identEnd = findIdentifierEnd(input, pos, to - 1)
+        const identNode = new SyntaxNode('FunctionCallOrIdentifier', pos, identEnd)
+        const innerIdent = new SyntaxNode('Identifier', pos, identEnd)
+        identNode.add(innerIdent)
+
+        const interpNode = new SyntaxNode('Interpolation', interpStart, identEnd)
+        interpNode.add(identNode)
+        stringNode.add(interpNode)
+        pos = identEnd
+      }
+
+      fragmentStart = pos
+      continue
+    }
+
+    pos++
+  }
+
+  // Push final fragment
+  if (pos > fragmentStart && fragmentStart < to - 1) {
+    const frag = new SyntaxNode('StringFragment', fragmentStart, pos)
+    stringNode.add(frag)
+  }
+}
+
+/**
+ * Parse curly string: { hello $name }
+ * Supports: interpolation ($var, $(expr)), nested braces
+ * Does NOT support: escape sequences (raw content)
+ */
+const parseCurlyString = (stringNode: SyntaxNode, input: string, from: number, to: number, parser: any) => {
+  let pos = from + 1 // Skip opening {
+  let fragmentStart = from // Include the opening { in the fragment
+  let depth = 1
+
+  while (pos < to && depth > 0) {
+    const char = input[pos]
+
+    // Track brace nesting
+    if (char === '{') {
+      depth++
+      pos++
+      continue
+    }
+
+    if (char === '}') {
+      depth--
+      if (depth === 0) {
+        // Push final fragment including closing }
+        const frag = new SyntaxNode('CurlyString', fragmentStart, pos + 1)
+        stringNode.add(frag)
+        break
+      }
+      pos++
+      continue
+    }
+
+    // Interpolation
+    if (char === '$') {
+      // Push accumulated fragment
+      if (pos > fragmentStart) {
+        const frag = new SyntaxNode('CurlyString', fragmentStart, pos)
+        stringNode.add(frag)
+      }
+
+      pos++ // Skip $
+
+      // Parse interpolation content
+      if (input[pos] === '(') {
+        // Expression interpolation: $(expr)
+        const interpStart = pos - 1
+        const exprResult = parseInterpolationExpr(input, pos, parser)
+        const interpNode = new SyntaxNode('Interpolation', interpStart, exprResult.endPos)
+        interpNode.add(exprResult.node)
+        stringNode.add(interpNode)
+        pos = exprResult.endPos
+      } else {
+        // Variable interpolation: $name
+        const interpStart = pos - 1
+        const identEnd = findIdentifierEnd(input, pos, to)
+        const identNode = new SyntaxNode('FunctionCallOrIdentifier', pos, identEnd)
+        const innerIdent = new SyntaxNode('Identifier', pos, identEnd)
+        identNode.add(innerIdent)
+
+        const interpNode = new SyntaxNode('Interpolation', interpStart, identEnd)
+        interpNode.add(identNode)
+        stringNode.add(interpNode)
+        pos = identEnd
+      }
+
+      fragmentStart = pos
+      continue
+    }
+
+    pos++
+  }
+}
+
+/**
+ * Parse a parenthesized expression interpolation: $(a + b)
+ * Returns the parsed expression node and the position after the closing )
+ * pos is position of the opening ( in the full input string
+ */
+const parseInterpolationExpr = (input: string, pos: number, parser: any): { node: SyntaxNode, endPos: number } => {
+  // Find matching closing paren
+  let depth = 1
+  let start = pos
+  let end = pos + 1 // Start after opening (
+
+  while (end < input.length && depth > 0) {
+    if (input[end] === '(') depth++
+    if (input[end] === ')') {
+      depth--
+      if (depth === 0) break
+    }
+    end++
+  }
+
+  const exprContent = input.slice(start + 1, end) // Content between ( and )
+  const closeParen = end
+  end++ // Move past closing )
+
+  // Use the main parser to parse the expression
+  const exprNode = parser.parse(exprContent)
+
+  // Get the first real node (skip Program wrapper)
+  const innerNode = exprNode.firstChild || exprNode
+
+  // Adjust node positions: they're relative to exprContent, need to offset to full input
+  const offset = start + 1 // Position where exprContent starts in full input
+  adjustNodePositions(innerNode, offset)
+
+  // Wrap in ParenExpr - use positions in the full string
+  const parenNode = new SyntaxNode('ParenExpr', start, closeParen + 1)
+  parenNode.add(innerNode)
+
+  return { node: parenNode, endPos: end }
+}
+
+/**
+ * Recursively adjust all node positions by adding an offset
+ */
+const adjustNodePositions = (node: SyntaxNode, offset: number) => {
+  node.from += offset
+  node.to += offset
+
+  for (const child of node.children) {
+    adjustNodePositions(child, offset)
+  }
+}
+
+/**
+ * Find the end position of an identifier starting at pos
+ * Identifiers: lowercase letter or emoji, followed by letters/digits/dashes/emoji
+ */
+const findIdentifierEnd = (input: string, pos: number, maxPos: number): number => {
+  let end = pos
+
+  while (end < maxPos) {
+    const char = input[end]
+
+    // Stop at non-identifier characters
+    if (!/[a-z0-9\-?]/.test(char)) {
+      break
+    }
+
+    end++
+  }
+
+  return end
+}

src/parser/tests/basics.test.ts

@@ -810,44 +810,6 @@ describe('Nullish coalescing operator', () => {
   })
 })
 
-describe('DotGet whitespace sensitivity', () => {
-  test('no whitespace - DotGet works when identifier in scope', () => {
-    expect('basename = 5; basename.prop').toMatchTree(`
-      Assign
-        AssignableIdentifier basename
-        Eq =
-        Number 5
-      FunctionCallOrIdentifier
-        DotGet
-          IdentifierBeforeDot basename
-          Identifier prop`)
-  })
-
-  test('space before dot - NOT DotGet, parses as division', () => {
-    expect('basename = 5; basename / prop').toMatchTree(`
-      Assign
-        AssignableIdentifier basename
-        Eq =
-        Number 5
-      BinOp
-        Identifier basename
-        Slash /
-        Identifier prop`)
-  })
-
-  test('dot followed by slash is Word, not DotGet', () => {
-    expect('basename ./cool').toMatchTree(`
-      FunctionCall
-        Identifier basename
-        PositionalArg
-          Word ./cool`)
-  })
-
-  test('identifier not in scope with dot becomes Word', () => {
-    expect('readme.txt').toMatchTree(`Word readme.txt`)
-  })
-})
-
 describe('Comments', () => {
   test('are greedy', () => {
     expect(`
@@ -897,61 +859,6 @@ basename = 5 # very astute
   })
 })
 
-describe('Array destructuring', () => {
-  test('parses array pattern with two variables', () => {
-    expect('[ a b ] = [ 1 2 3 4]').toMatchTree(`
-      Assign
-        Array
-          Identifier a
-          Identifier b
-        Eq =
-        Array
-          Number 1
-          Number 2
-          Number 3
-          Number 4`)
-  })
-
-  test('parses array pattern with one variable', () => {
-    expect('[ x ] = [ 42 ]').toMatchTree(`
-      Assign
-        Array
-          Identifier x
-        Eq =
-        Array
-          Number 42`)
-  })
-
-  test('parses array pattern with emoji identifiers', () => {
-    expect('[ 🚀 💎 ] = [ 1 2 ]').toMatchTree(`
-      Assign
-        Array
-          Identifier 🚀
-          Identifier 💎
-        Eq =
-        Array
-          Number 1
-          Number 2`)
-  })
-
-  test('works with dotget', () => {
-    expect('[ a ] = [ [1 2 3] ]; a.1').toMatchTree(`
-      Assign
-        Array
-          Identifier a
-        Eq =
-        Array
-          Array
-            Number 1
-            Number 2
-            Number 3
-      FunctionCallOrIdentifier
-        DotGet
-          IdentifierBeforeDot a
-          Number 1`)
-  })
-})
-
 describe('Conditional ops', () => {
   test('or can be chained', () => {
     expect(`
@@ -1037,34 +944,3 @@ Assign
         `)
   })
 })
-
-describe('import', () => {
-  test('parses single import', () => {
-    expect(`import str`).toMatchTree(`
-      Import
-        keyword import
-        Identifier str
-      `)
-  })
-
-  test('parses multiple imports', () => {
-    expect(`import str math list`).toMatchTree(`
-      Import
-        keyword import
-        Identifier str
-        Identifier math
-        Identifier list
-      `)
-  })
-
-  test('parses named args', () => {
-    expect(`import str only=ends-with?`).toMatchTree(`
-      Import
-        keyword import
-        Identifier str
-        NamedArg
-          NamedArgPrefix only=
-          Identifier ends-with?
-      `)
-  })
-})

src/parser/tests/control-flow.test.ts

@@ -24,7 +24,8 @@ describe('if/else if/else', () => {
         Eq =
         IfExpr
           keyword if
-          Identifier x
+          FunctionCallOrIdentifier
+            Identifier x
           colon :
           Block
             Number 2
@@ -59,7 +60,8 @@ describe('if/else if/else', () => {
     end`).toMatchTree(`
       IfExpr
         keyword if
-        Identifier with-else
+        FunctionCallOrIdentifier
+          Identifier with-else
         colon :
         Block
           FunctionCallOrIdentifier
@@ -82,7 +84,8 @@ describe('if/else if/else', () => {
     end`).toMatchTree(`
       IfExpr
         keyword if
-        Identifier with-else-if
+        FunctionCallOrIdentifier
+          Identifier with-else-if
         colon :
         Block
           FunctionCallOrIdentifier
@@ -90,7 +93,8 @@ describe('if/else if/else', () => {
         ElseIfExpr
           keyword else
           keyword if
-          Identifier another-condition
+          FunctionCallOrIdentifier
+            Identifier another-condition
           colon :
           Block
             FunctionCallOrIdentifier
@@ -111,7 +115,8 @@ describe('if/else if/else', () => {
     end`).toMatchTree(`
       IfExpr
         keyword if
-        Identifier with-else-if-else
+        FunctionCallOrIdentifier
+          Identifier with-else-if-else
         colon :
         Block
           FunctionCallOrIdentifier
@@ -119,7 +124,8 @@ describe('if/else if/else', () => {
         ElseIfExpr
           keyword else
           keyword if
-          Identifier another-condition
+          FunctionCallOrIdentifier
+            Identifier another-condition
           colon :
           Block
             FunctionCallOrIdentifier
@@ -127,7 +133,8 @@ describe('if/else if/else', () => {
         ElseIfExpr
           keyword else
           keyword if
-          Identifier yet-another-condition
+          FunctionCallOrIdentifier
+            Identifier yet-another-condition
           colon :
           Block
             FunctionCallOrIdentifier
@@ -173,7 +180,7 @@ describe('if/else if/else', () => {
     `)
   })
 
-  test('parses function calls in if tests', () => {
+  test("parses paren'd function calls in if tests", () => {
     expect(`if (var? 'abc'): true end`).toMatchTree(`
       IfExpr
         keyword if
@@ -214,7 +221,7 @@ describe('if/else if/else', () => {
     `)
   })
 
-  test('parses function calls in else-if tests', () => {
+  test("parses paren'd function calls in else-if tests", () => {
     expect(`if false: true else if (var? 'abc'): true end`).toMatchTree(`
       IfExpr
         keyword if

src/parser/tests/destructuring.test.ts

@@ -0,0 +1,58 @@
+import { expect, describe, test } from 'bun:test'
+
+import '../shrimp.grammar' // Importing this so changes cause it to retest!
+
+describe('Array destructuring', () => {
+  test('parses array pattern with two variables', () => {
+    expect('[ a b ] = [ 1 2 3 4]').toMatchTree(`
+      Assign
+        Array
+          Identifier a
+          Identifier b
+        Eq =
+        Array
+          Number 1
+          Number 2
+          Number 3
+          Number 4`)
+  })
+
+  test('parses array pattern with one variable', () => {
+    expect('[ x ] = [ 42 ]').toMatchTree(`
+      Assign
+        Array
+          Identifier x
+        Eq =
+        Array
+          Number 42`)
+  })
+
+  test('parses array pattern with emoji identifiers', () => {
+    expect('[ 🚀 💎 ] = [ 1 2 ]').toMatchTree(`
+      Assign
+        Array
+          Identifier 🚀
+          Identifier 💎
+        Eq =
+        Array
+          Number 1
+          Number 2`)
+  })
+
+  test('works with dotget', () => {
+    expect('[ a ] = [ [1 2 3] ]; a.1').toMatchTree(`
+      Assign
+        Array
+          Identifier a
+        Eq =
+        Array
+          Array
+            Number 1
+            Number 2
+            Number 3
+      FunctionCallOrIdentifier
+        DotGet
+          IdentifierBeforeDot a
+          Number 1`)
+  })
+})

src/parser/tests/dot-get.test.ts

@@ -1,6 +1,44 @@
 import { describe, test, expect } from 'bun:test'
 import '../../testSetup'
 
+describe('DotGet whitespace sensitivity', () => {
+  test('no whitespace - DotGet works when identifier in scope', () => {
+    expect('basename = 5; basename.prop').toMatchTree(`
+      Assign
+        AssignableIdentifier basename
+        Eq =
+        Number 5
+      FunctionCallOrIdentifier
+        DotGet
+          IdentifierBeforeDot basename
+          Identifier prop`)
+  })
+
+  test('space before dot - NOT DotGet, parses as division', () => {
+    expect('basename = 5; basename / prop').toMatchTree(`
+      Assign
+        AssignableIdentifier basename
+        Eq =
+        Number 5
+      BinOp
+        Identifier basename
+        Slash /
+        Identifier prop`)
+  })
+
+  test('dot followed by slash is Word, not DotGet', () => {
+    expect('basename ./cool').toMatchTree(`
+      FunctionCall
+        Identifier basename
+        PositionalArg
+          Word ./cool`)
+  })
+
+  test('identifier not in scope with dot becomes Word', () => {
+    expect('readme.txt').toMatchTree(`Word readme.txt`)
+  })
+})
+
 describe('DotGet', () => {
   test('readme.txt is Word when readme not in scope', () => {
     expect('readme.txt').toMatchTree(`Word readme.txt`)
@@ -199,7 +237,7 @@ end`).toMatchTree(`
     `)
   })
 
-  test("dot get doesn't work with spaces", () => {
+  test.skip("dot get doesn't work with spaces", () => {
     expect('obj . prop').toMatchTree(`
       FunctionCall
         Identifier obj

src/parser/tests/functions.test.ts

@@ -57,7 +57,7 @@ describe('calling functions', () => {
     `)
   })
 
-  test('Incomplete namedArg', () => {
+  test.skip('Incomplete namedArg', () => {
     expect('tail lines=').toMatchTree(`
       FunctionCall
         Identifier tail

src/parser/tests/import.test.ts

@@ -0,0 +1,34 @@
+import { expect, describe, test } from 'bun:test'
+
+import '../shrimp.grammar' // Importing this so changes cause it to retest!
+
+describe('import', () => {
+  test('parses single import', () => {
+    expect(`import str`).toMatchTree(`
+      Import
+        keyword import
+        Identifier str
+      `)
+  })
+
+  test('parses multiple imports', () => {
+    expect(`import str math list`).toMatchTree(`
+      Import
+        keyword import
+        Identifier str
+        Identifier math
+        Identifier list
+      `)
+  })
+
+  test('parses named args', () => {
+    expect(`import str only=ends-with?`).toMatchTree(`
+      Import
+        keyword import
+        Identifier str
+        NamedArg
+          NamedArgPrefix only=
+          Identifier ends-with?
+      `)
+  })
+})

src/parser/tests/literals.test.ts

@@ -375,10 +375,11 @@ describe('dict literals', () => {
     expect('[=]').toMatchTree(`
       Dict [=]
     `)
+  })
 
+  test('empty dict w whitespace', () => {
     expect('[ = ]').toMatchTree(`
-      Array
-        Word =
+      Dict [ = ]
     `)
   })
 

src/parser/tests/tokens.test.ts

@@ -0,0 +1,729 @@
+import { expect, describe, test } from 'bun:test'
+
+describe('constant types', () => {
+  test('null', () => {
+    expect(`null`).toBeToken('Null')
+  })
+
+  test('boolean', () => {
+    expect(`true`).toMatchToken('Boolean', 'true')
+    expect(`false`).toMatchToken('Boolean', 'false')
+  })
+})
+
+describe('numbers', () => {
+  test('non-numbers', () => {
+    expect(`1st`).toMatchToken('Word', '1st')
+    expect(`1_`).toMatchToken('Word', '1_')
+    expect(`100.`).toMatchTokens(
+      { type: 'Number', value: '100' },
+      { type: 'Operator', value: '.' },
+    )
+  })
+
+  test('simple numbers', () => {
+    expect(`1`).toMatchToken('Number', '1')
+    expect(`200`).toMatchToken('Number', '200')
+    expect(`5.20`).toMatchToken('Number', '5.20')
+    expect(`0.20`).toMatchToken('Number', '0.20')
+    expect(`-20`).toMatchToken('Number', '-20')
+    expect(`+20`).toMatchToken('Number', '+20')
+    expect(`-2134.34`).toMatchToken('Number', '-2134.34')
+    expect(`+20.5325`).toMatchToken('Number', '+20.5325')
+    expect(`1_000`).toMatchToken('Number', '1_000')
+    expect(`53_232_220`).toMatchToken('Number', '53_232_220')
+  })
+
+  test('binary numbers', () => {
+    expect('0b110').toMatchToken('Number', '0b110')
+  })
+
+  test('hex numbers', () => {
+    expect('0xdeadbeef').toMatchToken('Number', '0xdeadbeef')
+    expect('0x02d3f4').toMatchToken('Number', '0x02d3f4')
+  })
+
+  test('hex numbers uppercase', () => {
+    expect('0xFF').toMatchToken('Number', '0xFF')
+  })
+
+  test('octal numbers', () => {
+    expect('0o644').toMatchToken('Number', '0o644')
+    expect('0o055').toMatchToken('Number', '0o055')
+  })
+
+  test('negative binary', () => {
+    expect('-0b110').toMatchToken('Number', '-0b110')
+  })
+
+  test('negative hex', () => {
+    expect('-0xFF').toMatchToken('Number', '-0xFF')
+  })
+
+  test('negative octal', () => {
+    expect('-0o755').toMatchToken('Number', '-0o755')
+  })
+
+  test('positive prefix binary', () => {
+    expect('+0b110').toMatchToken('Number', '+0b110')
+  })
+
+  test('positive prefix hex', () => {
+    expect('+0xFF').toMatchToken('Number', '+0xFF')
+  })
+
+  test('positive prefix octal', () => {
+    expect('+0o644').toMatchToken('Number', '+0o644')
+  })
+
+  test('underscores in number', () => {
+    expect(`1_000`).toMatchToken('Number', '1_000')
+    expect(`1_0`).toMatchToken('Number', '1_0')
+    expect('0b11_0').toMatchToken('Number', '0b11_0')
+    expect('0xdead_beef').toMatchToken('Number', '0xdead_beef')
+    expect('0o64_4').toMatchToken('Number', '0o64_4')
+  })
+})
+
+describe('identifiers', () => {
+  test('regular', () => {
+    expect('name').toBeToken('Identifier')
+    expect('bobby-mcgee').toBeToken('Identifier')
+    expect('starts-with?').toBeToken('Identifier')
+    expect('📢').toMatchToken('Identifier', '📢')
+    expect(' 📢 ').toMatchToken('Identifier', '📢')
+    expect(' oink-🐷-oink').toMatchToken('Identifier', 'oink-🐷-oink')
+    expect('$').toMatchToken('Identifier', '$')
+    expect('$cool').toMatchToken('Identifier', '$cool')
+  })
+
+  test('one character identifiers', () => {
+    expect('a').toMatchToken('Identifier', 'a')
+    expect('z').toMatchToken('Identifier', 'z')
+    expect('$').toMatchToken('Identifier', '$')
+    expect('📢').toMatchToken('Identifier', '📢')
+    expect('?').toBeToken('Word') // ? alone is not valid identifier start
+  })
+
+  test('two character identifiers', () => {
+    expect('ab').toMatchToken('Identifier', 'ab')
+    expect('a1').toMatchToken('Identifier', 'a1')
+    expect('a-').toMatchToken('Identifier', 'a-')
+    expect('a?').toMatchToken('Identifier', 'a?') // ? valid at end
+    expect('ab?').toMatchToken('Identifier', 'ab?')
+  })
+
+  test('three+ character identifiers', () => {
+    expect('abc').toMatchToken('Identifier', 'abc')
+    expect('a-b').toMatchToken('Identifier', 'a-b')
+    expect('a1b').toMatchToken('Identifier', 'a1b')
+    expect('abc?').toMatchToken('Identifier', 'abc?') // ? valid at end
+    expect('a-b-c?').toMatchToken('Identifier', 'a-b-c?')
+  })
+
+  test('edge cases', () => {
+    expect('-bobby-mcgee').toBeToken('Word')
+    expect('starts-with??').toMatchToken('Identifier', 'starts-with??')
+    expect('starts?with?').toMatchToken('Identifier', 'starts?with?')
+    expect('a??b').toMatchToken('Identifier', 'a??b')
+    expect('oink-oink!').toBeToken('Word')
+    expect('dog#pound').toMatchToken('Word', 'dog#pound')
+    expect('http://website.com').toMatchToken('Word', 'http://website.com')
+    expect('school$cool').toMatchToken('Identifier', 'school$cool')
+    expect('EXIT:').toMatchTokens(
+      { type: 'Word', value: 'EXIT' },
+      { type: 'Colon' },
+    )
+    expect(`if y == 1: 'cool' end`).toMatchTokens(
+      { type: 'Keyword', value: 'if' },
+      { type: 'Identifier', value: 'y' },
+      { type: 'Operator', value: '==' },
+      { type: 'Number', value: '1' },
+      { type: 'Colon' },
+      { type: 'String', value: `'cool'` },
+      { type: 'Keyword', value: 'end' },
+    )
+  })
+})
+
+describe('paths', () => {
+  test('starting with ./', () => {
+    expect('./tmp').toMatchToken('Word', './tmp')
+  })
+
+  test('starting with /', () => {
+    expect('/home/chris/dev').toMatchToken('Word', '/home/chris/dev')
+  })
+
+  test('identifiers with dots tokenize separately', () => {
+    expect('readme.txt').toMatchTokens(
+      { type: 'Identifier', value: 'readme' },
+      { type: 'Operator', value: '.' },
+      { type: 'Identifier', value: 'txt' },
+    )
+  })
+
+  test('words (non-identifiers) consume dots', () => {
+    expect('README.md').toMatchToken('Word', 'README.md')
+  })
+
+  test('all sorts of weird stuff', () => {
+    expect('dog#pound').toMatchToken('Word', 'dog#pound')
+    expect('my/kinda/place').toMatchToken('my/kinda/place')
+    expect('file://%/$##/@40!/index.php').toMatchToken('Word', 'file://%/$##/@40!/index.php')
+  })
+})
+
+describe('strings', () => {
+  test('single quoted', () => {
+    expect(`'hello world'`).toMatchToken('String', `'hello world'`)
+    expect(`'it\\'s a beautiful world'`).toMatchToken("'it\\'s a beautiful world'")
+  })
+
+  test('double quoted', () => {
+    expect(`"hello world"`).toMatchToken('String', `"hello world"`)
+    expect(`"it's a beautiful world"`).toMatchToken('String', `"it's a beautiful world"`)
+  })
+
+  test('empty strings', () => {
+    expect(`''`).toMatchToken('String', `''`)
+    expect(`""`).toMatchToken('String', `""`)
+  })
+
+  test('escape sequences', () => {
+    expect(`'hello\\nworld'`).toMatchToken('String', `'hello\\nworld'`)
+    expect(`'tab\\there'`).toMatchToken('String', `'tab\\there'`)
+    expect(`'quote\\''`).toMatchToken('String', `'quote\\''`)
+    expect(`'backslash\\\\'`).toMatchToken('String', `'backslash\\\\'`)
+    expect(`'dollar\\$sign'`).toMatchToken('String', `'dollar\\$sign'`)
+  })
+
+  test('unclosed strings - error case', () => {
+    // These should either fail or produce unexpected results
+    expect(`'hello`).toMatchToken('String', `'hello`)
+    expect(`"world`).toMatchToken('String', `"world`)
+  })
+})
+
+describe('curly strings', () => {
+  test('curly quoted', () => {
+    expect('{ one two three }').toMatchToken('String', `{ one two three }`)
+  })
+
+  test('work on multiple lines', () => {
+    expect(`{
+      one
+      two
+      three }`).toMatchToken('String', `{
+      one
+      two
+      three }`)
+  })
+
+  test('can contain other curlies', () => {
+    expect(`{ { one }
+      two
+      { three } }`).toMatchToken('String', `{ { one }
+      two
+      { three } }`)
+  })
+
+  test('empty curly string', () => {
+    expect('{}').toMatchToken('String', '{}')
+  })
+
+  test('unclosed curly string - error case', () => {
+    // Should either fail or produce unexpected results
+    expect('{ hello').toMatchToken('String', '{ hello')
+    expect('{ nested { unclosed }').toMatchToken('String', '{ nested { unclosed }')
+  })
+})
+
+describe('operators', () => {
+  test('math operators', () => {
+    // assignment
+    expect('=').toMatchToken('Operator', '=')
+
+    // logic
+    expect('or').toMatchToken('Operator', 'or')
+    expect('and').toMatchToken('Operator', 'and')
+
+    // bitwise
+    expect('band').toMatchToken('Operator', 'band')
+    expect('bor').toMatchToken('Operator', 'bor')
+    expect('bxor').toMatchToken('Operator', 'bxor')
+    expect('>>>').toMatchToken('Operator', '>>>')
+    expect('>>').toMatchToken('Operator', '>>')
+    expect('<<').toMatchToken('Operator', '<<')
+
+    // compound assignment
+    expect('??=').toMatchToken('Operator', '??=')
+    expect('+=').toMatchToken('Operator', '+=')
+    expect('-=').toMatchToken('Operator', '-=')
+    expect('*=').toMatchToken('Operator', '*=')
+    expect('/=').toMatchToken('Operator', '/=')
+    expect('%=').toMatchToken('Operator', '%=')
+
+    // nullish
+    expect('??').toMatchToken('Operator', '??')
+
+    // math
+    expect('**').toMatchToken('Operator', '**')
+    expect('*').toMatchToken('Operator', '*')
+    expect('/').toMatchToken('Operator', '/')
+    expect('+').toMatchToken('Operator', '+')
+    expect('-').toMatchToken('Operator', '-')
+    expect('%').toMatchToken('Operator', '%')
+
+    // comparison
+    expect('>=').toMatchToken('Operator', '>=')
+    expect('<=').toMatchToken('Operator', '<=')
+    expect('!=').toMatchToken('Operator', '!=')
+    expect('==').toMatchToken('Operator', '==')
+    expect('>').toMatchToken('Operator', '>')
+    expect('<').toMatchToken('Operator', '<')
+
+    // property access
+    expect('.').toMatchToken('Operator', '.')
+  })
+})
+
+describe('keywords', () => {
+  test('keywords', () => {
+    expect(`import`).toMatchToken('Keyword', 'import')
+
+    expect(`end`).toMatchToken('Keyword', 'end')
+    expect(`do`).toMatchToken('Keyword', 'do')
+
+    expect(`while`).toMatchToken('Keyword', 'while')
+
+    expect(`if`).toMatchToken('Keyword', 'if')
+    expect(`else`).toMatchToken('Keyword', 'else')
+
+    expect(`try`).toMatchToken('Keyword', 'try')
+    expect(`catch`).toMatchToken('Keyword', 'catch')
+    expect(`finally`).toMatchToken('Keyword', 'finally')
+    expect(`throw`).toMatchToken('Keyword', 'throw')
+  })
+})
+
+describe('regex', () => {
+  test('use double slash', () => {
+    expect(`//[0-9]+//`).toMatchToken('Regex', '//[0-9]+//')
+  })
+})
+
+describe('punctuation', () => {
+  test('underscore', () => {
+    expect(`_`).toBeToken('Underscore')
+    expect(`__`).toMatchToken('Word', '__')
+  })
+
+  test('semicolon', () => {
+    expect(`;`).toBeToken('Semicolon')
+  })
+
+  test('newline', () => {
+    expect('\n').toBeToken('Newline')
+  })
+
+  test('colon', () => {
+    expect(':').toBeToken('Colon')
+  })
+})
+
+describe('comments', () => {
+  test('comments', () => {
+    expect(`# hey friends`).toMatchToken('Comment', '# hey friends')
+    expect(`#hey-friends`).toMatchToken('Comment', '#hey-friends')
+  })
+})
+
+describe('brackets', () => {
+  test('parens', () => {
+    expect(`(`).toBeToken('OpenParen')
+    expect(`)`).toBeToken('CloseParen')
+  })
+
+  test('staples', () => {
+    expect(`[`).toBeToken('OpenBracket')
+    expect(`]`).toBeToken('CloseBracket')
+  })
+})
+
+describe('multiple tokens', () => {
+  test('constants work fine', () => {
+    expect(`null true false`).toMatchTokens(
+      { type: 'Null' },
+      { type: 'Boolean', value: 'true' },
+      { type: 'Boolean', value: 'false' },
+    )
+  })
+
+  test('numbers', () => {
+    expect(`100 -400.42 null`).toMatchTokens(
+      { type: 'Number', value: '100' },
+      { type: 'Number', value: '-400.42' },
+      { type: 'Null' },
+    )
+  })
+
+  test('whitespace', () => {
+    expect(`
+      'hello world'
+
+      'goodbye world'
+    `).toMatchTokens(
+      { type: 'Newline' },
+      { type: 'String', value: "'hello world'" },
+      { type: 'Newline' },
+      { type: 'Newline' },
+      { type: 'String', value: "'goodbye world'" },
+      { type: 'Newline' },
+    )
+  })
+
+  test('newline in parens is ignored', () => {
+    expect(`(
+      'hello world'
+
+      'goodbye world'
+     )`).toMatchTokens(
+      { type: 'OpenParen' },
+      { type: 'String', value: "'hello world'" },
+      { type: 'String', value: "'goodbye world'" },
+      { type: 'CloseParen' },
+    )
+  })
+
+  test('newline in brackets is ignored', () => {
+    expect(`[
+      a b
+c d
+
+e
+
+f
+
+    ]`).toMatchTokens(
+      { type: 'OpenBracket' },
+      { type: 'Identifier', value: "a" },
+      { type: 'Identifier', value: "b" },
+      { type: 'Identifier', value: "c" },
+      { type: 'Identifier', value: "d" },
+      { type: 'Identifier', value: "e" },
+      { type: 'Identifier', value: "f" },
+      { type: 'CloseBracket' },
+    )
+  })
+
+  test('function call', () => {
+    expect('echo hello world').toMatchTokens(
+      { type: 'Identifier', value: 'echo' },
+      { type: 'Identifier', value: 'hello' },
+      { type: 'Identifier', value: 'world' },
+    )
+  })
+
+  test('assignment', () => {
+    expect('x = 5').toMatchTokens(
+      { type: 'Identifier', value: 'x' },
+      { type: 'Operator', value: '=' },
+      { type: 'Number', value: '5' },
+    )
+  })
+
+  test('math expression', () => {
+    expect('1 + 2 * 3').toMatchTokens(
+      { type: 'Number', value: '1' },
+      { type: 'Operator', value: '+' },
+      { type: 'Number', value: '2' },
+      { type: 'Operator', value: '*' },
+      { type: 'Number', value: '3' },
+    )
+  })
+
+  test('inline comment', () => {
+    expect('x = 5 # set x').toMatchTokens(
+      { type: 'Identifier', value: 'x' },
+      { type: 'Operator', value: '=' },
+      { type: 'Number', value: '5' },
+      { type: 'Comment', value: '# set x' },
+    )
+  })
+
+  test('line comment', () => {
+    expect('x = 5 \n# hello\n set x').toMatchTokens(
+      { type: 'Identifier', value: 'x' },
+      { type: 'Operator', value: '=' },
+      { type: 'Number', value: '5' },
+      { type: 'Newline' },
+      { type: 'Comment', value: '# hello' },
+      { type: 'Newline' },
+      { type: 'Identifier', value: 'set' },
+      { type: 'Identifier', value: 'x' },
+    )
+  })
+
+  test('colons separate tokens', () => {
+    expect('x do: y').toMatchTokens(
+      { type: 'Identifier', value: 'x' },
+      { type: 'Keyword', value: 'do' },
+      { type: 'Colon' },
+      { type: 'Identifier', value: 'y' },
+    )
+
+    expect('x: y').toMatchTokens(
+      { type: 'Identifier', value: 'x' },
+      { type: 'Colon' },
+      { type: 'Identifier', value: 'y' },
+    )
+
+    expect('5: y').toMatchTokens(
+      { type: 'Number', value: '5' },
+      { type: 'Colon' },
+      { type: 'Identifier', value: 'y' },
+    )
+
+
+    expect(`if (var? 'abc'): y`).toMatchTokens(
+      { type: 'Keyword', value: 'if' },
+      { type: 'OpenParen' },
+      { type: 'Identifier', value: 'var?' },
+      { type: 'String', value: `'abc'` },
+      { type: 'CloseParen' },
+      { type: 'Colon' },
+      { type: 'Identifier', value: 'y' },
+    )
+
+    expect(`
+do x:
+  y
+end`).toMatchTokens(
+      { type: 'Newline' },
+      { type: 'Keyword', value: 'do' },
+      { type: 'Identifier', value: 'x' },
+      { type: 'Colon' },
+      { type: 'Newline' },
+      { type: 'Identifier', value: 'y' },
+      { type: 'Newline' },
+      { type: 'Keyword', value: 'end' },
+    )
+  })
+
+  test('semicolons separate statements', () => {
+    expect('x; y').toMatchTokens(
+      { type: 'Identifier', value: 'x' },
+      { type: 'Semicolon' },
+      { type: 'Identifier', value: 'y' },
+    )
+  })
+
+  test('semicolons in parens', () => {
+    expect('(x; y)').toMatchTokens(
+      { type: 'OpenParen' },
+      { type: 'Identifier', value: 'x' },
+      { type: 'Semicolon' },
+      { type: 'Identifier', value: 'y' },
+      { type: 'CloseParen' },
+    )
+  })
+
+  test('dot operator beginning word with slash', () => {
+    expect(`(basename ./cool)`).toMatchTokens(
+      { 'type': 'OpenParen' },
+      { 'type': 'Identifier', 'value': 'basename' },
+      { 'type': 'Word', 'value': './cool' },
+      { 'type': 'CloseParen' }
+    )
+  })
+
+  test('dot word after identifier with space', () => {
+    expect(`expand-path .git`).toMatchTokens(
+      { 'type': 'Identifier', 'value': 'expand-path' },
+      { 'type': 'Word', 'value': '.git' },
+    )
+  })
+
+  test('dot operator after identifier without space', () => {
+    expect(`config.path`).toMatchTokens(
+      { 'type': 'Identifier', 'value': 'config' },
+      { 'type': 'Operator', 'value': '.' },
+      { 'type': 'Identifier', 'value': 'path' },
+    )
+  })
+})
+
+describe('nesting edge cases', () => {
+  test('deeply nested parens', () => {
+    expect('((nested))').toMatchTokens(
+      { type: 'OpenParen' },
+      { type: 'OpenParen' },
+      { type: 'Identifier', value: 'nested' },
+      { type: 'CloseParen' },
+      { type: 'CloseParen' },
+    )
+  })
+
+  test('mixed nesting', () => {
+    expect('([combo])').toMatchTokens(
+      { type: 'OpenParen' },
+      { type: 'OpenBracket' },
+      { type: 'Identifier', value: 'combo' },
+      { type: 'CloseBracket' },
+      { type: 'CloseParen' },
+    )
+  })
+})
+
+describe('invalid numbers that should be words', () => {
+  test('invalid binary', () => {
+    expect('0b2').toMatchToken('Word', '0b2')
+    expect('0b123').toMatchToken('Word', '0b123')
+  })
+
+  test('invalid octal', () => {
+    expect('0o8').toMatchToken('Word', '0o8')
+    expect('0o999').toMatchToken('Word', '0o999')
+  })
+
+  test('invalid hex', () => {
+    expect('0xGGG').toMatchToken('Word', '0xGGG')
+    expect('0xZZZ').toMatchToken('Word', '0xZZZ')
+  })
+
+  test('multiple decimal points', () => {
+    expect('1.2.3').toMatchToken('Word', '1.2.3')
+  })
+})
+
+describe('unicode and emoji', () => {
+  test('greek letters', () => {
+    expect('αβγ').toMatchToken('Identifier', 'αβγ')
+    expect('delta-δ').toMatchToken('Identifier', 'delta-δ')
+  })
+
+  test('math symbols', () => {
+    expect('∑').toMatchToken('Identifier', '∑')
+    expect('∏').toMatchToken('Identifier', '∏')
+  })
+
+  test('CJK characters', () => {
+    expect('你好').toMatchToken('Identifier', '你好')
+    expect('こんにちは').toMatchToken('Identifier', 'こんにちは')
+  })
+})
+
+describe('empty and whitespace input', () => {
+  test('empty string', () => {
+    expect('').toMatchTokens()
+  })
+
+  test('only whitespace', () => {
+    expect('   ').toMatchTokens()
+  })
+
+  test('only tabs', () => {
+    expect('\t\t\t').toMatchTokens()
+  })
+
+  test('only newlines', () => {
+    expect('\n\n\n').toMatchTokens(
+      { type: 'Newline' },
+      { type: 'Newline' },
+      { type: 'Newline' },
+    )
+  })
+})
+
+describe('named args', () => {
+  test("don't need spaces", () => {
+    expect(`named=arg`).toMatchTokens(
+      { type: 'NamedArgPrefix', value: 'named=' },
+      { type: 'Identifier', value: 'arg' },
+    )
+  })
+
+  test("can have spaces", () => {
+    expect(`named= arg`).toMatchTokens(
+      { type: 'NamedArgPrefix', value: 'named=' },
+      { type: 'Identifier', value: 'arg' },
+    )
+  })
+
+  test("can include numbers", () => {
+    expect(`named123= arg`).toMatchTokens(
+      { type: 'NamedArgPrefix', value: 'named123=' },
+      { type: 'Identifier', value: 'arg' },
+    )
+  })
+})
+
+describe('dot operator', () => {
+  test('standalone dot', () => {
+    expect('.').toMatchToken('Operator', '.')
+  })
+
+  test('dot between identifiers tokenizes as separate tokens', () => {
+    expect('config.path').toMatchTokens(
+      { type: 'Identifier', value: 'config' },
+      { type: 'Operator', value: '.' },
+      { type: 'Identifier', value: 'path' },
+    )
+  })
+
+  test('dot with number', () => {
+    expect('array.0').toMatchTokens(
+      { type: 'Identifier', value: 'array' },
+      { type: 'Operator', value: '.' },
+      { type: 'Number', value: '0' },
+    )
+  })
+
+  test('chained dots', () => {
+    expect('a.b.c').toMatchTokens(
+      { type: 'Identifier', value: 'a' },
+      { type: 'Operator', value: '.' },
+      { type: 'Identifier', value: 'b' },
+      { type: 'Operator', value: '.' },
+      { type: 'Identifier', value: 'c' },
+    )
+  })
+
+  test('identifier-like paths tokenize separately', () => {
+    expect('readme.txt').toMatchTokens(
+      { type: 'Identifier', value: 'readme' },
+      { type: 'Operator', value: '.' },
+      { type: 'Identifier', value: 'txt' },
+    )
+  })
+
+  test('word-like paths remain as single token', () => {
+    expect('./file.txt').toMatchToken('Word', './file.txt')
+    expect('README.TXT').toMatchToken('Word', 'README.TXT')
+  })
+
+  test('dot with paren expression', () => {
+    expect('obj.(1 + 2)').toMatchTokens(
+      { type: 'Identifier', value: 'obj' },
+      { type: 'Operator', value: '.' },
+      { type: 'OpenParen' },
+      { type: 'Number', value: '1' },
+      { type: 'Operator', value: '+' },
+      { type: 'Number', value: '2' },
+      { type: 'CloseParen' },
+    )
+  })
+
+  test('chained dot with paren expression', () => {
+    expect('obj.items.(i)').toMatchTokens(
+      { type: 'Identifier', value: 'obj' },
+      { type: 'Operator', value: '.' },
+      { type: 'Identifier', value: 'items' },
+      { type: 'Operator', value: '.' },
+      { type: 'OpenParen' },
+      { type: 'Identifier', value: 'i' },
+      { type: 'CloseParen' },
+    )
+  })
+})

src/parser/tokenizer2.ts

@@ -0,0 +1,569 @@
+const DEBUG = process.env.DEBUG || false
+
+export type Token = {
+  type: TokenType
+  value?: string,
+  from: number,
+  to: number,
+}
+
+export enum TokenType {
+  Comment,
+
+  Keyword,
+  Operator,
+
+  Newline,
+  Semicolon,
+  Colon,
+  Underscore,
+
+  OpenParen,
+  CloseParen,
+  OpenBracket,
+  CloseBracket,
+
+  Identifier,
+  Word,
+  NamedArgPrefix,
+
+  Null,
+  Boolean,
+  Number,
+  String,
+  Regex,
+}
+
+const valueTokens = new Set([
+  TokenType.Comment,
+  TokenType.Keyword, TokenType.Operator,
+  TokenType.Identifier, TokenType.Word, TokenType.NamedArgPrefix,
+  TokenType.Boolean, TokenType.Number, TokenType.String, TokenType.Regex
+])
+
+const operators = new Set([
+  // assignment
+  '=',
+
+  // logic
+  'or',
+  'and',
+
+  // bitwise
+  'band',
+  'bor',
+  'bxor',
+  '>>>',
+  '>>',
+  '<<',
+
+  // compound assignment
+  '??=',
+  '+=',
+  '-=',
+  '*=',
+  '/=',
+  '%=',
+
+  // nullish
+  '??',
+
+  // math
+  '**',
+  '*',
+  '/',
+  '+',
+  '-',
+  '%',
+
+  // comparison
+  '>=',
+  '<=',
+  '!=',
+  '==',
+  '>',
+  '<',
+
+  // property access
+  '.',
+
+  // pipe
+  '|',
+])
+
+const keywords = new Set([
+  'import',
+  'end',
+  'do',
+  'if',
+  'while',
+  'if',
+  'else',
+  'try',
+  'catch',
+  'finally',
+  'throw',
+])
+
+// helper
+function c(strings: TemplateStringsArray, ...values: any[]) {
+  return strings.reduce((result, str, i) => result + str + (values[i] ?? ""), "").charCodeAt(0)
+}
+
+function s(c: number): string {
+  return String.fromCharCode(c)
+}
+
+export class Scanner {
+  input = ''
+  pos = 0
+  start = 0
+  char = 0
+  prev = 0
+  inParen = 0
+  inBracket = 0
+  tokens: Token[] = []
+  prevIsWhitespace = true
+
+  reset() {
+    this.input = ''
+    this.pos = 0
+    this.start = 0
+    this.char = 0
+    this.prev = 0
+    this.tokens.length = 0
+    this.prevIsWhitespace = true
+  }
+
+  peek(count = 0): number {
+    return getFullCodePoint(this.input, this.pos + count)
+  }
+
+  next(): number {
+    this.prevIsWhitespace = isWhitespace(this.char)
+    this.prev = this.char
+    this.char = this.peek()
+    this.pos += getCharSize(this.char)
+
+    return this.char
+  }
+
+  push(type: TokenType, from?: number, to?: number) {
+    from ??= this.start
+    to ??= this.pos - getCharSize(this.char)
+    if (to < from) to = from
+
+    this.tokens.push(Object.assign({}, {
+      type,
+      from,
+      to,
+    }, valueTokens.has(type) ? { value: this.input.slice(from, to) } : {}))
+
+    if (DEBUG) {
+      const tok = this.tokens.at(-1)
+      console.log(`≫ PUSH(${from},${to})`, TokenType[tok?.type || 0], '—', tok?.value)
+    }
+
+    this.start = this.pos
+  }
+
+  pushChar(type: TokenType) {
+    this.push(type, this.pos - 1, this.pos)
+  }
+
+  // turn shrimp code into shrimp tokens that get fed into the parser
+  tokenize(input: string): Token[] {
+    this.reset()
+    this.input = input
+    this.next()
+
+    while (this.char > 0) {
+      const char = this.char
+
+      if (char === c`#`) {
+        this.readComment()
+        continue
+      }
+
+      if (isBracket(char)) {
+        this.readBracket()
+        continue
+      }
+
+      if (isStringDelim(char)) {
+        this.readString(char)
+        continue
+      }
+
+      if (char === c`{`) {
+        this.readCurlyString()
+        continue
+      }
+
+      if (isIdentStart(char)) {
+        this.readWordOrIdent(true) // true = started with identifier char
+        continue
+      }
+
+      if (isDigit(char) || ((char === c`-` || char === c`+`) && isDigit(this.peek()))) {
+        this.readNumber()
+        continue
+      }
+
+      if (char === c`:`) {
+        this.pushChar(TokenType.Colon)
+        this.next()
+        continue
+      }
+
+      // whitespace-sensitive dot as operator (property access) only after identifier/number
+      if (char === c`.`) {
+        if (this.canBeDotGet(this.tokens.at(-1))) {
+          this.pushChar(TokenType.Operator)
+          this.next()
+          continue
+        }
+      }
+
+      if (char === c`/` && this.peek() === c`/`) {
+        this.readRegex()
+        continue
+      }
+
+      if (isWordChar(char)) {
+        this.readWordOrIdent(false) // false = didn't start with identifier char
+        continue
+      }
+
+      if (char === c`\n`) {
+        if (this.inParen === 0 && this.inBracket === 0)
+          this.pushChar(TokenType.Newline)
+        this.next()
+        continue
+      }
+
+      if (char === c`;`) {
+        this.pushChar(TokenType.Semicolon)
+        this.next()
+        continue
+      }
+
+      this.next()
+    }
+
+    return this.tokens
+  }
+
+  readComment() {
+    this.start = this.pos - 1
+    while (this.char !== c`\n` && this.char > 0) this.next()
+    this.push(TokenType.Comment)
+  }
+
+  readBracket() {
+    switch (this.char) {
+      case c`(`:
+        this.inParen++
+        this.pushChar(TokenType.OpenParen); break
+      case c`)`:
+        this.inParen--
+        this.pushChar(TokenType.CloseParen); break
+      case c`[`:
+        this.inBracket++
+        this.pushChar(TokenType.OpenBracket); break
+      case c`]`:
+        this.inBracket--
+        this.pushChar(TokenType.CloseBracket); break
+    }
+    this.next()
+  }
+
+  readString(delim: number) {
+    this.start = this.pos - 1
+    this.next() // skip opening delim
+    while (this.char > 0 && (this.char !== delim || (this.char === delim && this.prev === c`\\`)))
+      this.next()
+    this.next() // skip closing delim
+
+    this.push(TokenType.String)
+  }
+
+  readCurlyString() {
+    let depth = 1
+    this.next()
+
+    while (depth > 0 && this.char > 0) {
+      if (this.char === c`{`) depth++
+      if (this.char === c`}`) depth--
+      this.next()
+    }
+
+    this.push(TokenType.String)
+  }
+
+  readWordOrIdent(startedWithIdentChar: boolean) {
+    this.start = this.pos - getCharSize(this.char)
+
+    while (isWordChar(this.char)) {
+      // stop at colon if followed by whitespace (e.g., 'do x: echo x end')
+      if (this.char === c`:`) {
+        const nextCh = this.peek()
+        if (isWhitespace(nextCh) || nextCh === 0) break
+      }
+
+      // stop at equal sign (named arg) - but only if what we've read so far is an identifier
+      if (this.char === c`=`) {
+        const soFar = this.input.slice(this.start, this.pos - getCharSize(this.char))
+        if (isIdentifer(soFar)) {
+          this.next()
+          break
+        }
+      }
+
+      // stop at dot only if it would create a valid property access
+      // AND only if we started with an identifier character (not for Words like README.txt)
+      if (startedWithIdentChar && this.char === c`.`) {
+        const nextCh = this.peek()
+        if (isIdentStart(nextCh) || isDigit(nextCh) || nextCh === c`(`) {
+          const soFar = this.input.slice(this.start, this.pos - getCharSize(this.char))
+          if (isIdentifer(soFar)) break
+        }
+      }
+
+      this.next()
+    }
+
+    const word = this.input.slice(this.start, this.pos - getCharSize(this.char))
+
+    // classify the token based on what we read
+    if (word === '_')
+      this.pushChar(TokenType.Underscore)
+
+    else if (word === 'null')
+      this.push(TokenType.Null)
+
+    else if (word === 'true' || word === 'false')
+      this.push(TokenType.Boolean)
+
+    else if (isKeyword(word))
+      this.push(TokenType.Keyword)
+
+    else if (isOperator(word))
+      this.push(TokenType.Operator)
+
+    else if (isIdentifer(word))
+      this.push(TokenType.Identifier)
+
+    else if (word.endsWith('='))
+      this.push(TokenType.NamedArgPrefix)
+
+    else
+      this.push(TokenType.Word)
+  }
+
+  readNumber() {
+    this.start = this.pos - 1
+    while (isWordChar(this.char)) {
+      // stop at dot unless it's part of the number
+      if (this.char === c`.`) {
+        const nextCh = this.peek()
+        if (!isDigit(nextCh)) break
+      }
+
+      // stop at colon
+      if (this.char === c`:`) {
+        const nextCh = this.peek()
+        if (isWhitespace(nextCh) || nextCh === 0) break
+      }
+      this.next()
+    }
+    const ident = this.input.slice(this.start, this.pos - 1)
+    this.push(isNumber(ident) ? TokenType.Number : TokenType.Word)
+  }
+
+  readRegex() {
+    this.start = this.pos - 1
+    this.next() // skip 2nd /
+
+    while (this.char > 0) {
+      if (this.char === c`/` && this.peek() === c`/`) {
+        this.next() // skip /
+        this.next() // skip /
+        this.push(TokenType.Regex)
+        break
+      }
+
+      this.next()
+    }
+  }
+
+  canBeDotGet(lastToken?: Token): boolean {
+    return !this.prevIsWhitespace && !!lastToken &&
+      (lastToken.type === TokenType.Identifier ||
+        lastToken.type === TokenType.Number ||
+        lastToken.type === TokenType.CloseParen ||
+        lastToken.type === TokenType.CloseBracket)
+  }
+}
+
+const isNumber = (word: string): boolean => {
+  // regular number
+  if (/^[+-]?\d+(_?\d+)*(\.(\d+(_?\d+)*))?$/.test(word))
+    return true
+
+  // binary
+  if (/^[+-]?0b[01]+(_?[01]+)*(\.[01](_?[01]*))?$/.test(word))
+    return true
+
+  // octal
+  if (/^[+-]?0o[0-7]+(_?[0-7]+)*(\.[0-7](_?[0-7]*))?$/.test(word))
+    return true
+
+  // hex
+  if (/^[+-]?0x[0-9a-f]+([0-9a-f]_?[0-9a-f]+)*(\.([0-9a-f]_?[0-9a-f]*))?$/i.test(word))
+    return true
+
+  return false
+}
+
+const isIdentifer = (s: string): boolean => {
+  if (s.length === 0) return false
+
+  let pos = 0
+  const chars = []
+  while (pos < s.length) {
+    const out = getFullCodePoint(s, pos)
+    pos += getCharSize(out)
+    chars.push(out)
+  }
+
+  if (chars.length === 1)
+    return isIdentStart(chars[0]!)
+  else if (chars.length === 2)
+    return isIdentStart(chars[0]!) && isIdentEnd(chars[1]!)
+  else
+    return isIdentStart(chars[0]!) &&
+      chars.slice(1, chars.length - 1).every(isIdentChar) &&
+      isIdentEnd(chars.at(-1)!)
+}
+
+const isStringDelim = (ch: number): boolean => {
+  return ch === c`'` || ch === c`"`
+}
+
+const isIdentStart = (char: number | string): boolean => {
+  let ch = typeof char === 'string' ? char.charCodeAt(0) : char
+  return isLowercaseLetter(ch) || isEmojiOrUnicode(ch) || ch === 36 /* $ */
+}
+
+const isIdentChar = (char: number | string): boolean => {
+  let ch = typeof char === 'string' ? char.charCodeAt(0) : char
+  return isIdentStart(ch) || isDigit(ch) || ch === 45 /* - */ || ch === 63 /* ? */
+}
+
+const isIdentEnd = (char: number | string): boolean => {
+  return isIdentChar(char)
+}
+
+const isLowercaseLetter = (ch: number): boolean => {
+  return ch >= 97 && ch <= 122 // a-z
+}
+
+const isDigit = (ch: number): boolean => {
+  return ch >= 48 && ch <= 57 // 0-9
+}
+
+const isWhitespace = (ch: number): boolean => {
+  return ch === 32 /* space */ || ch === 9 /* tab */ ||
+    ch === 13 /* \r */ || ch === 10 /* \n */ ||
+    ch === -1 || ch === 0 /* EOF */
+}
+
+const isWordChar = (ch: number): boolean => {
+  return (
+    !isWhitespace(ch) &&
+    ch !== 10 /* \n */ &&
+    ch !== 59 /* ; */ &&
+    ch !== 41 /* ) */ &&
+    ch !== 93 /* ] */ &&
+    ch !== -1 /* EOF */
+  )
+}
+
+const isOperator = (word: string): boolean => {
+  return operators.has(word)
+}
+
+const isKeyword = (word: string): boolean => {
+  return keywords.has(word)
+}
+
+const isBracket = (char: number): boolean => {
+  return char === c`(` || char === c`)` || char === c`[` || char === c`]`
+}
+
+const getCharSize = (ch: number) =>
+  (ch > 0xffff ? 2 : 1) // emoji takes 2 UTF-16 code units
+
+const getFullCodePoint = (input: string, pos: number): number => {
+  const ch = input[pos]?.charCodeAt(0) || 0
+
+  // Check if this is a high surrogate (0xD800-0xDBFF)
+  if (ch >= 0xd800 && ch <= 0xdbff) {
+    const low = input[pos + 1]?.charCodeAt(0) || 0
+    // Check if next is low surrogate (0xDC00-0xDFFF)
+    if (low >= 0xdc00 && low <= 0xdfff) {
+      // Combine surrogate pair into full code point
+      return 0x10000 + ((ch & 0x3ff) << 10) + (low & 0x3ff)
+    }
+  }
+
+  return ch
+}
+
+const isEmojiOrUnicode = (ch: number): boolean => {
+  return (
+    // Basic Emoticons
+    (ch >= 0x1f600 && ch <= 0x1f64f) ||
+    // Miscellaneous Symbols and Pictographs
+    (ch >= 0x1f300 && ch <= 0x1f5ff) ||
+    // Transport and Map Symbols
+    (ch >= 0x1f680 && ch <= 0x1f6ff) ||
+    // Regional Indicator Symbols (flags)
+    (ch >= 0x1f1e6 && ch <= 0x1f1ff) ||
+    // Miscellaneous Symbols (hearts, stars, weather)
+    (ch >= 0x2600 && ch <= 0x26ff) ||
+    // Dingbats (scissors, pencils, etc)
+    (ch >= 0x2700 && ch <= 0x27bf) ||
+    // Supplemental Symbols and Pictographs (newer emojis)
+    (ch >= 0x1f900 && ch <= 0x1f9ff) ||
+    // Symbols and Pictographs Extended-A (newest emojis)
+    (ch >= 0x1fa70 && ch <= 0x1faff) ||
+    // Various Asian Characters with emoji presentation
+    (ch >= 0x1f018 && ch <= 0x1f270) ||
+    // Variation Selectors (for emoji presentation)
+    (ch >= 0xfe00 && ch <= 0xfe0f) ||
+    // Additional miscellaneous items
+    (ch >= 0x238c && ch <= 0x2454) ||
+    // Combining Diacritical Marks for Symbols
+    (ch >= 0x20d0 && ch <= 0x20ff) ||
+    // Latin-1 Supplement (includes ², ³, ¹ and other special chars)
+    (ch >= 0x00a0 && ch <= 0x00ff) ||
+    // Greek and Coptic (U+0370-U+03FF)
+    (ch >= 0x0370 && ch <= 0x03ff) ||
+    // Mathematical Alphanumeric Symbols (U+1D400-U+1D7FF)
+    (ch >= 0x1d400 && ch <= 0x1d7ff) ||
+    // Mathematical Operators (U+2200-U+22FF)
+    (ch >= 0x2200 && ch <= 0x22ff) ||
+    // Superscripts and Subscripts (U+2070-U+209F)
+    (ch >= 0x2070 && ch <= 0x209f) ||
+    // Arrows (U+2190-U+21FF)
+    (ch >= 0x2190 && ch <= 0x21ff) ||
+    // Hiragana (U+3040-U+309F)
+    (ch >= 0x3040 && ch <= 0x309f) ||
+    // Katakana (U+30A0-U+30FF)
+    (ch >= 0x30a0 && ch <= 0x30ff) ||
+    // CJK Unified Ideographs (U+4E00-U+9FFF)
+    (ch >= 0x4e00 && ch <= 0x9fff)
+  )
+}

src/testSetup.ts

@@ -1,12 +1,16 @@
 import { expect } from 'bun:test'
+import { diffLines } from 'diff'
+import color from 'kleur'
+import { Scanner, TokenType, type Token } from '#parser/tokenizer2'
 import { parser } from '#parser/shrimp'
 import { setGlobals } from '#parser/tokenizer'
+import { parse } from '#parser/parser2'
 import { globals as prelude } from '#prelude'
 import { $ } from 'bun'
 import { assert, errorMessage } from '#utils/utils'
 import { Compiler } from '#compiler/compiler'
 import { run, VM } from 'reefvm'
-import { treeToString, VMResultToValue } from '#utils/tree'
+import { treeToString2, treeToString, VMResultToValue } from '#utils/tree'
 
 const regenerateParser = async () => {
   let generate = true
@@ -37,6 +41,9 @@ declare module 'bun:test' {
     toFailParse(): T
     toEvaluateTo(expected: unknown, globals?: Record<string, any>): Promise<T>
     toFailEvaluation(): Promise<T>
+    toBeToken(expected: string): T
+    toMatchToken(typeOrValue: string, value?: string): T
+    toMatchTokens(...tokens: { type: string, value?: string }[]): T
   }
 }
 
@@ -46,8 +53,8 @@ expect.extend({
 
     const allGlobals = { ...prelude, ...(globals || {}) }
     setGlobals(Object.keys(allGlobals))
-    const tree = parser.parse(received)
-    const actual = treeToString(tree, received)
+    const tree = parse(received)
+    const actual = treeToString2(tree, received)
     const normalizedExpected = trimWhitespace(expected)
 
     try {
@@ -144,8 +151,107 @@ expect.extend({
       }
     }
   },
+  toBeToken(received: unknown, expected: string) {
+    assert(typeof received === 'string', 'toBeToken can only be used with string values')
+
+    try {
+      const tokens = tokenize(received)
+      const value = tokens[0] as Token
+      const target = TokenType[expected as keyof typeof TokenType]
+
+      if (!value) {
+        return {
+          message: () => `Expected token type to be ${expected}, but got ${value}`,
+          pass: false,
+        }
+      }
+
+      return {
+        message: () => `Expected token type to be ${expected}, but got ${TokenType[value.type]}`,
+        pass: value.type === target
+      }
+    } catch (error) {
+      return {
+        message: () => `Tokenization failed: ${errorMessage(error)}`,
+        pass: false,
+      }
+    }
+  },
+  toMatchToken(received: unknown, typeOrValue: string, value?: string) {
+    assert(typeof received === 'string', 'toMatchToken can only be used with string values')
+    const expectedValue = value ? value : typeOrValue
+    const expectedType = value ? typeOrValue : undefined
+
+    try {
+      const tokens = tokenize(received)
+      const token = tokens[0] as Token
+
+      if (!token) {
+        return {
+          message: () => `Expected token to be ${expectedValue.replaceAll('\n', '\\n')}, got ${token}`,
+          pass: false,
+        }
+      }
+
+      if (expectedType && TokenType[expectedType as keyof typeof TokenType] !== token.type) {
+        return {
+          message: () => `Expected token to be ${expectedType}, but got ${TokenType[token.type]}`,
+          pass: false
+        }
+      }
+
+      return {
+        message: () => `Expected token to be ${expectedValue.replaceAll('\n', '\\n')}, but got ${token.value}`,
+        pass: token.value === expectedValue
+      }
+    } catch (error) {
+      return {
+        message: () => `Tokenization failed: ${errorMessage(error)} `,
+        pass: false,
+      }
+    }
+  },
+  toMatchTokens(received: unknown, ...tokens: { type: string, value?: string }[]) {
+    assert(typeof received === 'string', 'toMatchTokens can only be used with string values')
+
+    try {
+      const result = tokenize(received).map(t => toHumanToken(t))
+
+      if (result.length === 0 && tokens.length > 0) {
+        return {
+          message: () => `Expected tokens ${JSON.stringify(tokens)}, got nothing`,
+          pass: false,
+        }
+      }
+
+      const expected = JSON.stringify(tokens, null, 2)
+      const actual = JSON.stringify(result, null, 2)
+
+      return {
+        message: () => `Tokens don't match: \n\n${diff(actual, expected)}`,
+        pass: expected == actual
+      }
+    } catch (error) {
+      return {
+        message: () => `Tokenization failed: ${errorMessage(error)} `,
+        pass: false,
+      }
+    }
+  }
 })
 
+const tokenize = (code: string): Token[] => {
+  const scanner = new Scanner
+  return scanner.tokenize(code)
+}
+
+const toHumanToken = (tok: Token): { type: string, value?: string } => {
+  return {
+    type: TokenType[tok.type],
+    value: tok.value
+  }
+}
+
 const trimWhitespace = (str: string): string => {
   const lines = str.split('\n').filter((line) => line.trim().length > 0)
   const firstLine = lines[0]
@@ -157,10 +263,33 @@ const trimWhitespace = (str: string): string => {
       if (!line.startsWith(leadingWhitespace)) {
         let foundWhitespace = line.match(/^(\s*)/)?.[1] || ''
         throw new Error(
-          `Line has inconsistent leading whitespace: "${line}" (found "${foundWhitespace}", expected "${leadingWhitespace}")`
+          `Line has inconsistent leading whitespace: "${line}"(found "${foundWhitespace}", expected "${leadingWhitespace}")`
         )
       }
       return line.slice(leadingWhitespace.length)
     })
     .join('\n')
 }
+
+const diff = (a: string, b: string): string => {
+  const expected = a.trim()
+  const actual = b.trim()
+  const lines = []
+
+  if (expected !== actual) {
+    const changes = diffLines(actual, expected)
+    for (const part of changes) {
+      const sign = part.added ? "+" : part.removed ? "-" : " "
+      let line = sign + part.value
+      if (part.added) {
+        line = color.green(line)
+      } else if (part.removed) {
+        line = color.red(line)
+      }
+
+      lines.push(line.endsWith("\n") || line.endsWith("\n\u001b[39m") ? line : line + "\n")
+    }
+  }
+
+  return lines.join('\n')
+}

src/utils/tree.ts

@@ -1,5 +1,38 @@
 import { Tree, TreeCursor } from '@lezer/common'
 import { type Value, fromValue } from 'reefvm'
+import { SyntaxNode } from '#parser/node'
+
+const nodeToString = (node: SyntaxNode, input: string, depth = 0): string => {
+  const indent = '  '.repeat(depth)
+  const text = input.slice(node.from, node.to)
+  const nodeName = node.name
+
+  if (node.firstChild) {
+    return `${indent}${nodeName}`
+  } else {
+    // Only strip quotes from whole String nodes (legacy DoubleQuote), not StringFragment/EscapeSeq/CurlyString
+    const cleanText = nodeName === 'String' ? text.slice(1, -1) : text
+    return `${indent}${nodeName} ${cleanText}`
+  }
+}
+
+export const treeToString2 = (tree: SyntaxNode, input: string, depth = 0): string => {
+  let lines = []
+  let node: SyntaxNode | null = tree
+
+  if (node.name === 'Program') node = node.firstChild
+
+  while (node) {
+    lines.push(nodeToString(node, input, depth))
+
+    if (node.firstChild)
+      lines.push(treeToString2(node.firstChild, input, depth + 1))
+
+    node = node.nextSibling
+  }
+
+  return lines.join('\n')
+}
 
 export const treeToString = (tree: Tree, input: string): string => {
   const lines: string[] = []