| 📦 Turbo Rust 713ea5c k33g 12h ago | 1 | package rustlang |
| 2 | |
| 3 | import "rickub.com/turbo-editors/turbo-core/syntax" |
| 4 | |
| 5 | // carry is what a line of Rust leaves open for the next one. |
| 6 | // |
| 7 | // Three constructs in Rust can cross a line break, and each of them has to be |
| 8 | // remembered exactly rather than guessed at: a block comment (which nests, so a |
| 9 | // depth and not a flag), a raw string (whose terminator is a quote followed by |
| 10 | // as many hashes as its opener had), and an ordinary string (which may contain |
| 11 | // a real newline). Anything else is decided from the line in front of you. |
| 12 | type carry struct { |
| 13 | // commentDepth is how many /* are still open. Rust nests block comments, |
| 14 | // so /* /* */ */ is one comment and a flag would end it one level early. |
| 15 | commentDepth int |
| 16 | // rawHashes is the number of # a raw string was opened with, and rawOpen |
| 17 | // says whether one is open at all — a raw string may be opened with none, |
| 18 | // so the count alone cannot say. |
| 19 | rawOpen bool |
| 20 | rawHashes int |
| 21 | // stringOpen says an ordinary "…" string ran past the end of a line. |
| 22 | stringOpen bool |
| 23 | } |
| 24 | |
| 25 | // Highlight colours Rust source. |
| 26 | // |
| 27 | // It is written against syntax.LineScanner, a line at a time, with the three |
| 28 | // multi-line constructs above threaded through carry. Rust has no tokeniser in |
| 29 | // the Go standard library the way Go does, so this is a scanner in the same |
| 30 | // style as the ones turbo-core ships for TOML, Markdown and shell. |
| 31 | // |
| 32 | // It is deliberately tolerant of broken input: source under the cursor is |
| 33 | // invalid most of the time it is being typed, and a highlighter that gives up |
| 34 | // is a highlighter that flickers off. |
| 35 | func Highlight(src string) [][]syntax.Span { |
| 36 | return syntax.ScanLines(src, scanLine) |
| 37 | } |
| 38 | |
| 39 | // scanLine colours one line and returns what it leaves open. |
| 40 | func scanLine(line []rune, open carry) ([]syntax.Span, carry) { |
| 41 | s := syntax.NewLineScanner(line) |
| 42 | |
| 43 | // Whatever ran past the end of the previous line is finished first: until |
| 44 | // it closes, nothing on this line is code. |
| 45 | if !finishCarried(s, &open) { |
| 46 | return s.Spans(), open |
| 47 | } |
| 48 | |
| 49 | for !s.AtEnd() { |
| 50 | scanToken(s, &open) |
| 51 | } |
| 52 | return s.Spans(), open |
| 53 | } |
| 54 | |
| 55 | // finishCarried closes whatever the previous line left open, and reports |
| 56 | // whether the rest of this line is code. |
| 57 | func finishCarried(s *syntax.LineScanner, open *carry) bool { |
| 58 | switch { |
| 59 | case open.commentDepth > 0: |
| 60 | return continueBlockComment(s, open) |
| 61 | case open.rawOpen: |
| 62 | return continueRawString(s, open) |
| 63 | case open.stringOpen: |
| 64 | return continueString(s, open) |
| 65 | } |
| 66 | return true |
| 67 | } |
| 68 | |
| 69 | // scanToken colours whatever starts at the scanner's position. |
| 70 | func scanToken(s *syntax.LineScanner, open *carry) { |
| 71 | r := s.Peek(0) |
| 72 | |
| 73 | switch { |
| 74 | case r == ' ' || r == '\t': |
| 75 | s.SkipSpaces() |
| 76 | case s.HasPrefix(0, "//"): |
| 77 | s.TakeRest(syntax.ClassComment) |
| 78 | case s.HasPrefix(0, "/*"): |
| 79 | startBlockComment(s, open) |
| 80 | case r == '#' && (s.Peek(1) == '[' || (s.Peek(1) == '!' && s.Peek(2) == '[')): |
| 81 | takeAttribute(s) |
| 82 | case isRawStringStart(s): |
| 83 | startRawString(s, open) |
| 84 | case isByteOrCharStart(s): |
| 85 | takeByteLiteral(s, open) |
| 86 | case r == '"': |
| 87 | startString(s, open) |
| 88 | case r == '\'': |
| 89 | takeQuoteOrLifetime(s) |
| 90 | case syntax.IsDigit(r): |
| 91 | takeNumber(s) |
| 92 | case syntax.IsLetter(r) || r == '_': |
| 93 | takeWord(s) |
| 94 | case r == ':': |
| 95 | // A path separator and a type annotation are both structure rather |
| 96 | // than computation, so they go with the brackets and the commas. ":" |
| 97 | // is an operator rune, so this has to come first. |
| 98 | s.TakeWhile(syntax.ClassPunctuation, func(c rune) bool { return c == ':' }) |
| 99 | case s.HasPrefix(0, ".."): |
| 100 | // A range really is an operation, and "." is a punctuation rune, so |
| 101 | // this has to come first too. |
| 102 | s.Take(rangeWidth(s), syntax.ClassOperator) |
| 103 | case syntax.IsOperatorRune(r): |
| 104 | s.TakeWhile(syntax.ClassOperator, syntax.IsOperatorRune) |
| 105 | case syntax.IsPunctuationRune(r) || r == '#' || r == '@' || r == '$': |
| 106 | s.Take(1, syntax.ClassPunctuation) |
| 107 | default: |
| 108 | // A rune nothing here claims — an emoji in an identifier, say — is |
| 109 | // stepped over uncoloured rather than guessed at. |
| 110 | s.Advance(1) |
| 111 | } |
| 112 | } |
| 113 | |
| 114 | // --- comments --------------------------------------------------------------- |
| 115 | |
| 116 | // startBlockComment colours a /* that opens on this line, counting the nesting |
| 117 | // Rust allows. |
| 118 | func startBlockComment(s *syntax.LineScanner, open *carry) { |
| 119 | start := s.Pos() |
| 120 | s.Advance(2) |
| 121 | open.commentDepth = 1 |
| 122 | |
| 123 | consumeComment(s, open) |
| 124 | s.Emit(start, s.Pos(), syntax.ClassComment) |
| 125 | } |
| 126 | |
| 127 | // continueBlockComment colours the rest of a comment opened on an earlier line, |
| 128 | // and reports whether the line has code after it. |
| 129 | func continueBlockComment(s *syntax.LineScanner, open *carry) bool { |
| 130 | consumeComment(s, open) |
| 131 | s.Emit(0, s.Pos(), syntax.ClassComment) |
| 132 | return open.commentDepth == 0 && !s.AtEnd() |
| 133 | } |
| 134 | |
| 135 | // consumeComment runs to the end of the comment or to the end of the line, |
| 136 | // keeping the nesting depth in step. |
| 137 | func consumeComment(s *syntax.LineScanner, open *carry) { |
| 138 | for !s.AtEnd() { |
| 139 | switch { |
| 140 | case s.HasPrefix(0, "/*"): |
| 141 | open.commentDepth++ |
| 142 | s.Advance(2) |
| 143 | case s.HasPrefix(0, "*/"): |
| 144 | open.commentDepth-- |
| 145 | s.Advance(2) |
| 146 | if open.commentDepth == 0 { |
| 147 | return |
| 148 | } |
| 149 | default: |
| 150 | s.Advance(1) |
| 151 | } |
| 152 | } |
| 153 | } |
| 154 | |
| 155 | // --- attributes ------------------------------------------------------------- |
| 156 | |
| 157 | // takeAttribute colours #[derive(Debug)] and its inner form #![no_std]. |
| 158 | // |
| 159 | // An attribute that runs past the end of its line is coloured to the end and |
| 160 | // not carried: unlike a comment or a string, an unclosed attribute is nearly |
| 161 | // always a half-typed one, and carrying it would paint the rest of the file. |
| 162 | func takeAttribute(s *syntax.LineScanner) { |
| 163 | start := s.Pos() |
| 164 | |
| 165 | // Step over the # and the ! of the inner form, so that the bracket |
| 166 | // matching below starts where the brackets actually are. Counting from the |
| 167 | // # instead ends #![no_std] at its second rune, which is a bug this had. |
| 168 | s.Advance(1) |
| 169 | if s.Peek(0) == '!' { |
| 170 | s.Advance(1) |
| 171 | } |
| 172 | |
| 173 | depth := 0 |
| 174 | for !s.AtEnd() { |
| 175 | switch s.Peek(0) { |
| 176 | case '[': |
| 177 | depth++ |
| 178 | case ']': |
| 179 | depth-- |
| 180 | } |
| 181 | s.Advance(1) |
| 182 | if depth == 0 { |
| 183 | break |
| 184 | } |
| 185 | } |
| 186 | s.Emit(start, s.Pos(), syntax.ClassAttribute) |
| 187 | } |