| 🛟 Updated. 28d5985 k33g 19h ago | 1 | // Fenced code blocks: what an agent writes in Markdown, and the one place a |
| 2 | // scanner can be pointed at with confidence. |
| 3 | |
| 4 | package acp |
| 5 | |
| 6 | import ( |
| 7 | "strings" |
| 8 | |
| 📦 Turbo Core f3ade8d k33g 12h ago | 9 | "rickub.com/turbo-editors/turbo-core/syntax" |
| 🛟 Updated. 28d5985 k33g 19h ago | 10 | ) |
| 11 | |
| 12 | // Block is a run of an entry's text that is either prose or code. |
| 13 | // |
| 14 | // An agent answers in Markdown, so a fenced block is a real thing in what it |
| 15 | // sends and the one place a scanner can be pointed at with confidence. |
| 16 | type Block struct { |
| 17 | // Language is what the fence named, or LanguageNone for prose and for a |
| 18 | // fence that named nothing. |
| 19 | Language syntax.Language |
| 20 | // Code says this block came out of a fence. |
| 21 | Code bool |
| 22 | // Text is the block's own text, without the fence markers. |
| 23 | Text string |
| 24 | } |
| 25 | |
| 26 | // SplitBlocks separates fenced code from the prose around it. |
| 27 | // |
| 28 | // Only the *fence* decides. A scanner is never guessed at from the shape of |
| 29 | // the text: a highlighter that is wrong is worse than one that is quiet, which |
| 30 | // is the rule every scanner in this library already follows. |
| 31 | // |
| 32 | // blocks := acp.SplitBlocks("here:\n```go\nx := 1\n```\n") |
| 33 | // blocks[1].Language // "go" |
| 34 | func SplitBlocks(text string) []Block { |
| 35 | var blocks []Block |
| 36 | var current []string |
| 37 | inFence, language := false, syntax.LanguageNone |
| 38 | |
| 39 | flush := func(code bool, lang syntax.Language) { |
| 40 | if len(current) == 0 { |
| 41 | return |
| 42 | } |
| 43 | blocks = append(blocks, Block{Language: lang, Code: code, Text: strings.Join(current, "\n")}) |
| 44 | current = nil |
| 45 | } |
| 46 | |
| 47 | for _, line := range strings.Split(text, "\n") { |
| 48 | if fence, named := fenceOf(line); fence { |
| 49 | if inFence { |
| 50 | flush(true, language) |
| 51 | inFence, language = false, syntax.LanguageNone |
| 52 | continue |
| 53 | } |
| 54 | flush(false, syntax.LanguageNone) |
| 55 | inFence, language = true, named |
| 56 | continue |
| 57 | } |
| 58 | current = append(current, line) |
| 59 | } |
| 60 | |
| 61 | // A fence nobody closed is still code: an agent cut off mid-answer leaves |
| 62 | // one, and drawing the rest as prose would change colour halfway down a |
| 63 | // function for no reason a reader can see. |
| 64 | flush(inFence, language) |
| 65 | return blocks |
| 66 | } |
| 67 | |
| 68 | // fenceOf reports whether a line opens or closes a fence, and which language |
| 69 | // it named. |
| 70 | func fenceOf(line string) (bool, syntax.Language) { |
| 71 | trimmed := strings.TrimSpace(line) |
| 72 | if !strings.HasPrefix(trimmed, "```") { |
| 73 | return false, syntax.LanguageNone |
| 74 | } |
| 75 | |
| 76 | name := strings.ToLower(strings.TrimSpace(strings.TrimPrefix(trimmed, "```"))) |
| 77 | if name == "" { |
| 78 | return true, syntax.LanguageNone |
| 79 | } |
| 80 | return true, languageNamed(name) |
| 81 | } |
| 82 | |
| 83 | // languageNamed returns the registered language a fence's tag means, or |
| 84 | // LanguageNone when no scanner claims it. |
| 85 | // |
| 86 | // The tag is matched against what the editor actually knows rather than |
| 87 | // against a list kept here, so an editor that registers Rust colours a ```rust |
| 88 | // block with no change to this file. The few aliases below are the ones a |
| 89 | // model reaches for by habit. |
| 90 | func languageNamed(name string) syntax.Language { |
| 91 | if alias, ok := fenceAliases[name]; ok { |
| 92 | name = alias |
| 93 | } |
| 94 | |
| 95 | wanted := syntax.Language(name) |
| 96 | for _, known := range syntax.Registered() { |
| 97 | if known == wanted { |
| 98 | return known |
| 99 | } |
| 100 | } |
| 101 | return syntax.LanguageNone |
| 102 | } |
| 103 | |
| 104 | // fenceAliases maps the tags models write onto the names languages register |
| 105 | // themselves under. |
| 106 | var fenceAliases = map[string]string{ |
| 107 | "golang": "go", |
| 108 | "sh": "bash", |
| 109 | "shell": "bash", |
| 110 | "zsh": "bash", |
| 111 | "console": "bash", |
| 112 | "js": "javascript", |
| 113 | "mjs": "javascript", |
| 114 | "yml": "yaml", |
| 115 | "md": "markdown", |
| 116 | "docker": "dockerfile", |
| 117 | "containerfile": "dockerfile", |
| 118 | } |