plaintext.go
| 1 | package markdown |
| 2 | |
| 3 | import ( |
| 4 | "fmt" |
| 5 | "regexp" |
| 6 | "strings" |
| 7 | |
| 8 | "github.com/yuin/goldmark" |
| 9 | "github.com/yuin/goldmark/ast" |
| 10 | "github.com/yuin/goldmark/extension" |
| 11 | extast "github.com/yuin/goldmark/extension/ast" |
| 12 | "github.com/yuin/goldmark/text" |
| 13 | ) |
| 14 | |
| 15 | // plaintextParser only needs the GFM AST, never HTML output. |
| 16 | var plaintextParser = goldmark.New(goldmark.WithExtensions(extension.GFM)).Parser() |
| 17 | |
| 18 | var ( |
| 19 | tagRE = regexp.MustCompile(`</?[a-zA-Z][^>]*>`) |
| 20 | blankRunRE = regexp.MustCompile(`\n{3,}`) |
| 21 | spaceRunRE = regexp.MustCompile(`\n+`) |
| 22 | ) |
| 23 | |
| 24 | // ToPlaintext converts markdown to plain text, keeping structure such as list |
| 25 | // prefixes and blockquote markers. Used for previews. |
| 26 | func ToPlaintext(md string) string { |
| 27 | src := []byte(md) |
| 28 | doc := plaintextParser.Parse(text.NewReader(src)) |
| 29 | out := renderPlain(doc, src) |
| 30 | // Inline raw HTML reaches the output verbatim, so drop tag-shaped runs. |
| 31 | out = tagRE.ReplaceAllString(out, "") |
| 32 | return strings.TrimSpace(blankRunRE.ReplaceAllString(out, "\n\n")) |
| 33 | } |
| 34 | |
| 35 | // renderPlain walks the children of n and concatenates their plain text. |
| 36 | func renderPlain(n ast.Node, src []byte) string { |
| 37 | var b strings.Builder |
| 38 | for c := n.FirstChild(); c != nil; c = c.NextSibling() { |
| 39 | b.WriteString(plainNode(c, src)) |
| 40 | } |
| 41 | return b.String() |
| 42 | } |
| 43 | |
| 44 | func plainNode(n ast.Node, src []byte) string { |
| 45 | switch node := n.(type) { |
| 46 | case *ast.Heading, *ast.Paragraph, *ast.TextBlock: |
| 47 | return renderPlain(n, src) + "\n\n" |
| 48 | |
| 49 | case *ast.FencedCodeBlock, *ast.CodeBlock: |
| 50 | return linesOf(n, src) + "\n" |
| 51 | |
| 52 | case *ast.Blockquote: |
| 53 | body := strings.TrimSpace(renderPlain(n, src)) |
| 54 | lines := strings.Split(body, "\n") |
| 55 | for i, l := range lines { |
| 56 | lines[i] = "> " + l |
| 57 | } |
| 58 | return strings.Join(lines, "\n") + "\n\n" |
| 59 | |
| 60 | case *ast.List: |
| 61 | return plainList(node, src) + "\n" |
| 62 | |
| 63 | case *ast.ThematicBreak: |
| 64 | return "---\n\n" |
| 65 | |
| 66 | case *ast.HTMLBlock: |
| 67 | return "" |
| 68 | |
| 69 | case *extast.Table: |
| 70 | return renderPlain(n, src) + "\n" |
| 71 | |
| 72 | case *extast.TableHeader, *extast.TableRow: |
| 73 | var cells []string |
| 74 | for c := n.FirstChild(); c != nil; c = c.NextSibling() { |
| 75 | cells = append(cells, strings.TrimSpace(renderPlain(c, src))) |
| 76 | } |
| 77 | return strings.Join(cells, " | ") + "\n" |
| 78 | |
| 79 | case *ast.Text: |
| 80 | out := string(node.Segment.Value(src)) |
| 81 | if node.SoftLineBreak() || node.HardLineBreak() { |
| 82 | out += "\n" |
| 83 | } |
| 84 | return out |
| 85 | |
| 86 | case *ast.String: |
| 87 | return string(node.Value) |
| 88 | |
| 89 | case *ast.CodeSpan: |
| 90 | return "`" + renderPlain(n, src) + "`" |
| 91 | |
| 92 | case *ast.AutoLink: |
| 93 | return string(node.URL(src)) |
| 94 | |
| 95 | case *ast.RawHTML: |
| 96 | return "" |
| 97 | |
| 98 | case *ast.Image: |
| 99 | label := strings.TrimSpace(string(node.Title)) |
| 100 | if label == "" { |
| 101 | label = strings.TrimSpace(renderPlain(n, src)) |
| 102 | } |
| 103 | if label == "" { |
| 104 | return "[Image]" |
| 105 | } |
| 106 | return "[Image: " + label + "]" |
| 107 | |
| 108 | default: |
| 109 | return renderPlain(n, src) |
| 110 | } |
| 111 | } |
| 112 | |
| 113 | // plainList prefixes each item with its bullet or number. |
| 114 | func plainList(list *ast.List, src []byte) string { |
| 115 | var b strings.Builder |
| 116 | number := list.Start |
| 117 | if number == 0 { |
| 118 | number = 1 |
| 119 | } |
| 120 | for item := list.FirstChild(); item != nil; item = item.NextSibling() { |
| 121 | prefix := "- " |
| 122 | if list.IsOrdered() { |
| 123 | prefix = fmt.Sprintf("%d. ", number) |
| 124 | number++ |
| 125 | } |
| 126 | body := renderPlain(item, src) |
| 127 | if cb, ok := item.FirstChild().(*ast.TextBlock); ok { |
| 128 | if box, ok := cb.FirstChild().(*extast.TaskCheckBox); ok { |
| 129 | if box.IsChecked { |
| 130 | prefix += "[x] " |
| 131 | } else { |
| 132 | prefix += "[ ] " |
| 133 | } |
| 134 | } |
| 135 | } |
| 136 | body = spaceRunRE.ReplaceAllString(strings.TrimSpace(body), " ") |
| 137 | b.WriteString(prefix + body + "\n") |
| 138 | } |
| 139 | return b.String() |
| 140 | } |
| 141 | |
| 142 | // linesOf returns the raw source text of a block node. |
| 143 | func linesOf(n ast.Node, src []byte) string { |
| 144 | var b strings.Builder |
| 145 | for i := 0; i < n.Lines().Len(); i++ { |
| 146 | seg := n.Lines().At(i) |
| 147 | b.Write(seg.Value(src)) |
| 148 | } |
| 149 | return b.String() |
| 150 | } |
| 151 |