Syntax.Source

Source identity for the versioned syntax artifacts: source files and half-open byte spans.

Byte offsets are the canonical position vocabulary (line and column are projections for people, never a second identity), and these are the Prism-side types the token and surface-syntax exports decode into.

Types

`Span`

type Span = Span { lo: Int, hi: Int } deriving (Eq, Show)

A half-open byte range [lo, hi) into one source text.

span_len(Span { lo = 3, hi = 8 })

`SourceFile`

type SourceFile = SourceFile {
  digest: String,
  text: String
} deriving (Eq, Show)

One source file as an artifact embeds it: the exact text and its digest. Every span in an artifact indexes this text, so a persisted document needs no external file.

Functions and Values

`span_len`

span_len : (Syntax.Source.Span) -> Int

The number of bytes a span covers.

`span_valid`

span_valid : (Syntax.Source.Span) -> Bool

Whether a span is well formed: non-negative start, start at or before end.

(span_valid(Span { lo = 2, hi = 2 }), span_valid(Span { lo = 5, hi = 4 }))

(true, false)

`span_contains`

span_contains : (Syntax.Source.Span, Int) -> Bool

Whether a byte offset falls inside the span (half-open, so the end offset is outside).

`line_col`

line_col : (String, Int) -> (Int, Int)

The one-based line and column of a byte offset, as a projection over the text. Offsets past the end clamp to the final position.

Both the offset and the column it reports are counted in bytes, the same vocabulary spans use, so the projection agrees with the identity it projects from. Counting the walk in characters instead would read a byte offset as a character index and land somewhere else entirely on a line holding any multi-byte character, besides re-walking the text on every step.

line_col("ab\ncd", 4)

(2, 2)

Keyboard shortcuts

Prism