| 📦 Turbo Golo d710c1b k33g 12h ago | 1 | package gololang |
| 2 | |
| 3 | import "rickub.com/turbo-editors/turbo-core/syntax" |
| 4 | |
| 5 | // carry is what a line of Golo leaves open for the next one: which construct, |
| 6 | // if any, the line ended inside. |
| 7 | // |
| 8 | // Four constructs may reach the next line, and the interpreter's own lexer is |
| 9 | // the authority on each. A block comment runs from one ---- to the next |
| 10 | // wherever that is. A "string", a """multi-line string""" and a 'character' |
| 11 | // are all read to their closing quote, and that quote may be on a later line: |
| 12 | // lexer.go reads a plain string with `for l.ch != '"' && l.ch != 0`, which |
| 13 | // stops at the quote or the end of the file and at nothing in between. So an |
| 14 | // unterminated string paints the rest of the file — and that is what the |
| 15 | // interpreter does with it too, which is the reason to carry rather than to |
| 16 | // stop at the line the way Turbo MoonBit does for a language whose grammar |
| 17 | // forbids the newline. |
| 18 | // |
| 19 | // None of the four nests, so a value saying which one is open is the whole of |
| 20 | // the state. A depth would be a claim the language does not make. |
| 21 | type carry int |
| 22 | |
| 23 | const ( |
| 24 | // nothingOpen is the state between constructs, and the zero value |
| 25 | // ScanLines starts the first line in. |
| 26 | nothingOpen carry = iota |
| 27 | // blockCommentOpen means a ---- was seen and its closing ---- was not. |
| 28 | blockCommentOpen |
| 29 | // stringOpen means a " was seen and its closing " was not. |
| 30 | stringOpen |
| 31 | // multilineStringOpen means a """ was seen and its closing """ was not. |
| 32 | multilineStringOpen |
| 33 | // charOpen means a ' was seen and its closing ' was not. |
| 34 | charOpen |
| 35 | ) |
| 36 | |
| 37 | // Highlight colours Golo source. |
| 38 | // |
| 39 | // It is written against syntax.LineScanner, a line at a time, with the |
| 40 | // interpreter's lexer/lexer.go as its specification. GoloScript's own lexer is |
| 41 | // a Go package, but its module is named `golo` and is not importable from |
| 42 | // another module, so the rules are carried here rather than called — and what |
| 43 | // the lexer reads as one token, this scanner colours as one span. |
| 44 | // |
| 45 | // It is deliberately tolerant of broken input: source under the cursor is |
| 46 | // invalid most of the time it is being typed, and a highlighter that gives up |
| 47 | // is a highlighter that flickers off. |
| 48 | // |
| 49 | // spans := gololang.Highlight("function main = |args| {\n println(\"hi\")\n}\n") |
| 50 | // // spans[0][0] covers "function" with syntax.ClassKeyword |
| 51 | func Highlight(src string) [][]syntax.Span { |
| 52 | return syntax.ScanLines(src, scanLine) |
| 53 | } |
| 54 | |
| 55 | // scanLine colours one line, finishing whatever the previous line left open |
| 56 | // before looking at anything new, and reports what this line leaves open. |
| 57 | func scanLine(line []rune, open carry) ([]syntax.Span, carry) { |
| 58 | s := syntax.NewLineScanner(line) |
| 59 | |
| 60 | open = finishOpen(s, open) |
| 61 | for !s.AtEnd() { |
| 62 | open = scanToken(s) |
| 63 | } |
| 64 | return s.Spans(), open |
| 65 | } |
| 66 | |
| 67 | // finishOpen colours the continuation of a construct opened on an earlier |
| 68 | // line, and says whether it is still open at the end of this one. With nothing |
| 69 | // open it does nothing. |
| 70 | func finishOpen(s *syntax.LineScanner, open carry) carry { |
| 71 | switch open { |
| 72 | case blockCommentOpen: |
| 73 | return stillOpen(syntax.FinishBlockComment(s, blockCommentMarker, syntax.ClassComment), open) |
| 74 | case multilineStringOpen: |
| 75 | return stillOpen(syntax.FinishBlockComment(s, multilineStringQuote, syntax.ClassString), open) |
| 76 | case stringOpen: |
| 77 | return finishQuoted(s, '"', syntax.ClassString, open) |
| 78 | case charOpen: |
| 79 | return finishQuoted(s, '\'', syntax.ClassChar, open) |
| 80 | default: |
| 81 | return nothingOpen |
| 82 | } |
| 83 | } |
| 84 | |
| 85 | // stillOpen turns "did it close?" into what the next line should be told. |
| 86 | func stillOpen(closed bool, open carry) carry { |
| 87 | if closed { |
| 88 | return nothingOpen |
| 89 | } |
| 90 | return open |
| 91 | } |
| 92 | |
| 93 | // blockCommentMarker opens and closes a block comment. The lexer asks for four |
| 94 | // dashes exactly: three are an operator run, and a fifth is part of the text. |
| 95 | const blockCommentMarker = "----" |
| 96 | |
| 97 | // scanToken colours whatever starts at the scanner's position, and reports |
| 98 | // which construct, if any, ran off the end of the line. |
| 99 | // |
| 100 | // The order of the cases is the design, and three of them are load-bearing: |
| 101 | // |
| 102 | // - ---- is tested before the operators, because - is an operator rune and |
| 103 | // the run would otherwise be coloured as one. |
| 104 | // - """ is tested before ", because both begin with a quote and what tells |
| 105 | // them apart is the two runes after it. The lexer makes the same check in |
| 106 | // the same order. |
| 107 | // - A digit is tested before a word, and a word before a dot, so that 1.5 is |
| 108 | // one number and hello.World is a name, a dot and a name. |
| 109 | func scanToken(s *syntax.LineScanner) carry { |
| 110 | r := s.Peek(0) |
| 111 | |
| 112 | switch { |
| 113 | case r == ' ' || r == '\t': |
| 114 | s.SkipSpaces() |
| 115 | case r == '#': |
| 116 | // # opens a comment that runs to the end of the line. A shebang is |
| 117 | // one too: #!/usr/bin/env golo is how a script is run as a command, |
| 118 | // and the interpreter reads it as a comment because that is what it |
| 119 | // is. |
| 120 | s.TakeRest(syntax.ClassComment) |
| 121 | case s.HasPrefix(0, blockCommentMarker): |
| 122 | return stillOpen(syntax.OpenBlockComment(s, blockCommentMarker, blockCommentMarker, syntax.ClassComment), blockCommentOpen) |
| 123 | case s.HasPrefix(0, multilineStringQuote): |
| 124 | return takeMultilineString(s) |
| 125 | case r == '"': |
| 126 | return takeQuoted(s, '"', syntax.ClassString, stringOpen) |
| 127 | case r == '\'': |
| 128 | return takeQuoted(s, '\'', syntax.ClassChar, charOpen) |
| 129 | case syntax.IsDigit(r): |
| 130 | takeNumber(s) |
| 131 | case isIdentifierStart(r): |
| 132 | takeWord(s) |
| 133 | case r == '.': |
| 134 | takeDot(s) |
| 135 | case r == '$': |
| 136 | // $ joins a union to one of its variants in `augment Shape$Circle`. |
| 137 | // It is structure rather than computation, so it is punctuation like |
| 138 | // the dot in a module path. |
| 139 | s.Take(1, syntax.ClassPunctuation) |
| 140 | case syntax.IsOperatorRune(r): |
| 141 | s.TakeWhile(syntax.ClassOperator, syntax.IsOperatorRune) |
| 142 | case syntax.IsPunctuationRune(r): |
| 143 | s.Take(1, syntax.ClassPunctuation) |
| 144 | default: |
| 145 | // A rune nothing here claims — a stray control character, a symbol |
| 146 | // outside the ranges the lexer accepts in a name — is stepped over |
| 147 | // uncoloured rather than guessed at. |
| 148 | s.Advance(1) |
| 149 | } |
| 150 | return nothingOpen |
| 151 | } |
| 152 | |
| 153 | // takeDot colours a dot and whatever the language says belongs with it. |
| 154 | // |
| 155 | // A run of dots is an operator — .. is a range and ... marks a variadic |
| 156 | // parameter — and a dot on its own is punctuation: the separator in a module |
| 157 | // path such as hello.World, or in a decimal that takeNumber did not claim |
| 158 | // because no digit came before it. |
| 159 | func takeDot(s *syntax.LineScanner) { |
| 160 | if s.Peek(1) == '.' { |
| 161 | s.TakeWhile(syntax.ClassOperator, func(r rune) bool { return r == '.' }) |
| 162 | return |
| 163 | } |
| 164 | s.Take(1, syntax.ClassPunctuation) |
| 165 | } |