Files
lambda/pkg/token/scan.go
M.V. Hutz aca197ef51 refactor: simplify iterator.Try and remove unnecessary backtracking (#47)
## Description

`iterator.Try` previously copied the entire iterator and synced it back on success, causing an unnecessary heap allocation on every call.
This PR simplifies `Try` to save and restore the index directly, and removes the now-unused `Copy` and `Sync` methods.

- Rewrite `ScanRune` and `ParseRawToken` as peek-then-advance, eliminating the need for `Try` at leaf level.
- Remove redundant `Try` wrappers from `parseExpression`, `parseAbstraction`, `parseApplication`, `parseLet`, and `parseToken`, which are already disambiguated by their callers.
- Keep `Try` only where true backtracking is needed: `parseStatement`, which must choose between `parseLet` and `parseDeclare`.
- Fix pre-existing panic in saccharine `parseExpression` when the iterator is exhausted (added `Done()` guard).

### Decisions

- `Try` now operates on the original iterator instead of a copy, removing the confusing pattern where the callback's `i` was a different object than the caller's `i`.
- Removed `parseSoftBreak` and `parseHardBreak` helper functions since `ParseRawToken` no longer needs `Try` wrapping.

## Benefits

- Eliminates a heap allocation per `Try` call.
- Reduces nesting and indirection in all parse functions.
- Makes the code easier to follow by removing the shadow-`i` pattern.
- `Try` is now only used at genuine choice points in the grammar.

## Checklist

- [x] Code follows conventional commit format.
- [x] Branch follows naming convention (`<type>/<description>`). Always use underscores.
- [x] Tests pass (if applicable).
- [x] Documentation updated (if applicable).

Reviewed-on: #47
Co-authored-by: M.V. Hutz <git@maximhutz.me>
Co-committed-by: M.V. Hutz <git@maximhutz.me>
2026-02-12 01:04:26 +00:00

75 lines
2.1 KiB
Go

package token
import (
"errors"
"fmt"
"unicode"
"git.maximhutz.com/max/lambda/pkg/iterator"
)
// IsVariable determines whether a rune can be a valid variable character.
func IsVariable(r rune) bool {
return unicode.IsLetter(r) || unicode.IsNumber(r) || r == '_'
}
// ScanRune consumes the next rune from the iterator if it satisfies the
// predicate.
// Returns an error if the iterator is exhausted or the rune does not match.
func ScanRune(i *iterator.Iterator[rune], expected func(rune) bool) (rune, error) {
r, err := i.Get()
if err != nil {
return r, err
}
if !expected(r) {
return r, fmt.Errorf("got unexpected rune %v'", r)
}
i.Forward()
return r, nil
}
// ScanCharacter consumes the next rune from the iterator if it matches the
// expected rune exactly.
// Returns an error if the iterator is exhausted or the rune does not match.
func ScanCharacter(i *iterator.Iterator[rune], expected rune) (rune, error) {
return ScanRune(i, func(r rune) bool { return r == expected })
}
// ScanAtom scans a contiguous sequence of variable characters into a single
// atom token.
// The first rune has already been consumed and is passed in.
func ScanAtom[T Type](i *iterator.Iterator[rune], first rune, typ T, column int) *Token[T] {
atom := []rune{first}
for {
if r, err := ScanRune(i, IsVariable); err != nil {
break
} else {
atom = append(atom, r)
}
}
return NewAtom(typ, string(atom), column)
}
// Scan tokenizes an input string using a language-specific scanToken function.
// The scanToken function is called repeatedly until the input is exhausted.
// It returns nil (no token, no error) for skippable input like whitespace.
// Errors are accumulated and returned joined at the end.
func Scan[T Type](input string, scanToken func(*iterator.Iterator[rune]) (*Token[T], error)) ([]Token[T], error) {
i := iterator.Of([]rune(input))
tokens := []Token[T]{}
errorList := []error{}
for !i.Done() {
token, err := scanToken(i)
if err != nil {
errorList = append(errorList, err)
} else if token != nil {
tokens = append(tokens, *token)
}
}
return tokens, errors.Join(errorList...)
}