markdown/Render.go

package markdown

import (
	"html"
	"strings"
	"unsafe"
)

var (
	headerStart = []string{"<h1>", "<h2>", "<h3>", "<h4>", "<h5>", "<h6>"}
	headerEnd   = []string{"</h1>", "</h2>", "</h3>", "</h4>", "</h5>", "</h6>"}
)

// renderer represents a Markdown to HTML renderer.
type renderer struct {
	out                []byte
	paragraphLevel     int
	quoteLevel         int
	listLevel          int
	olistLevel         int
	tableLevel         int
	codeLines          int
	tableHeaderWritten bool
	inCodeBlock        bool
}

// Render creates HTML from the supplied markdown text.
func Render(markdown string) string {
	var (
		r         renderer
		i         = 0
		lineStart = 0
	)

	r.out = make([]byte, 0, nextPowerOf2(uint32(len(markdown)+4)))

	for {
		if i > len(markdown) {
			r.closeAll()

			for range r.quoteLevel {
				r.WriteString("</blockquote>")
			}

			return unsafe.String(unsafe.SliceData(r.out), len(r.out))
		}

		if i != len(markdown) && markdown[i] != '\n' {
			i++
			continue
		}

		line := markdown[lineStart:i]
		lineStart = i + 1
		i++

		r.processLine(line)
	}
}

func (r *renderer) processLine(line string) {
	if r.inCodeBlock {
		if strings.HasPrefix(line, "```") {
			r.WriteString("</code></pre>")
			r.inCodeBlock = false
			r.codeLines = 0
		} else {
			if r.codeLines != 0 {
				r.WriteByte('\n')
			}

			r.WriteString(html.EscapeString(line))
			r.codeLines++
		}

		return
	}

	newQuoteLevel := 0

	for strings.HasPrefix(line, ">") {
		line = strings.TrimSpace(line[1:])
		newQuoteLevel++
	}

	if newQuoteLevel > r.quoteLevel {
		r.closeParagraphs()

		for range newQuoteLevel - r.quoteLevel {
			r.WriteString("<blockquote>")
		}
	} else if newQuoteLevel < r.quoteLevel {
		r.closeParagraphs()

		for range r.quoteLevel - newQuoteLevel {
			r.WriteString("</blockquote>")
		}
	}

	r.quoteLevel = newQuoteLevel

	if len(line) == 0 {
		r.closeAll()
		return
	}

	switch line[0] {
	case '#':
		r.closeAll()
		space := strings.IndexByte(line, ' ')

		if space > 0 && space <= 6 {
			r.WriteString(headerStart[space-1])
			r.writeText(line[space+1:])
			r.WriteString(headerEnd[space-1])
		}

		return

	case '-', '*':
		if strings.HasPrefix(line, "---") {
			r.WriteString("<hr>")
			return
		}

		if len(line) > 1 && line[1] == ' ' {
			line = line[2:]

			if r.listLevel == 0 {
				r.WriteString("<ul>")
				r.listLevel++
			}

			r.WriteString("<li>")
			r.writeText(line)
			r.WriteString("</li>")
			return
		}

	case '`':
		if strings.HasPrefix(line, "```") {
			language := line[3:]

			if !r.inCodeBlock {
				if language != "" {
					r.WriteString("<pre><code class=\"language-")
					r.WriteString(html.EscapeString(language))
					r.WriteString("\">")
				} else {
					r.WriteString("<pre><code>")
				}

				r.inCodeBlock = true
			}

			return
		}

	case '|':
		line = line[1:]

		if r.tableLevel == 0 {
			r.WriteString("<table><thead>")
			r.tableLevel++
		}

		column := 0

		for {
			pipe := strings.IndexByte(line, '|')

			if pipe == -1 {
				r.WriteString("</tr>")
				return
			}

			content := strings.TrimSpace(line[:pipe])

			if strings.HasPrefix(content, "---") {
				r.WriteString("</thead><tbody>")
				r.tableHeaderWritten = true
				return
			}

			if column == 0 {
				r.WriteString("<tr>")
			}

			if r.tableHeaderWritten {
				r.WriteString("<td>")
				r.writeText(content)
				r.WriteString("</td>")
			} else {
				r.WriteString("<th>")
				r.writeText(content)
				r.WriteString("</th>")
			}

			line = line[pipe+1:]
			column++
		}
	}

	pos := 0

	for pos < len(line) && line[pos] >= '0' && line[pos] <= '9' {
		pos++

		if pos < len(line) && (line[pos] == '.' || line[pos] == ')') {
			line = strings.TrimSpace(line[pos+1:])

			if r.olistLevel == 0 {
				r.WriteString("<ol>")
				r.olistLevel++
			}

			r.WriteString("<li>")
			r.writeText(line)
			r.WriteString("</li>")
			return
		}
	}

	if r.paragraphLevel == 0 {
		r.WriteString("<p>")
		r.paragraphLevel++
		r.writeText(line)
		return
	}

	r.WriteByte(' ')
	r.writeText(line)
}

// closeAll closes all open tags.
func (r *renderer) closeAll() {
	r.closeLists()
	r.closeParagraphs()
	r.closeTables()
}

// closeParagraphs closes open paragraphs.
func (r *renderer) closeParagraphs() {
	for range r.paragraphLevel {
		r.WriteString("</p>")
	}

	r.paragraphLevel = 0
}

// closeLists closes open lists.
func (r *renderer) closeLists() {
	for range r.listLevel {
		r.WriteString("</ul>")
	}

	for range r.olistLevel {
		r.WriteString("</ol>")
	}

	r.listLevel = 0
	r.olistLevel = 0
}

// closeTables closes open tables.
func (r *renderer) closeTables() {
	for range r.tableLevel {
		r.WriteString("</tbody></table>")
	}

	r.tableLevel = 0
	r.tableHeaderWritten = false
}

// writeText converts inline markdown to HTML.
func (r *renderer) writeText(markdown string) {
	var (
		tokenStart    = 0
		searchStart   = 0
		linkTextStart = -1
		linkTextEnd   = -1
		linkIsImage   = false
		emStart       = -1
		strongStart   = -1
		strikeStart   = -1
	)

begin:
	for {
		i := strings.IndexAny(markdown[searchStart:], "[]()`*_~!")

		if i == -1 {
			r.WriteString(html.EscapeString(markdown[tokenStart:]))
			return
		}

		i += searchStart
		searchStart = i + 1

		switch markdown[i] {
		case '[':
			r.WriteString(html.EscapeString(markdown[tokenStart:i]))
			tokenStart = i
			linkTextStart = i

		case ']':
			linkTextEnd = i

		case '(':
			if linkTextStart == -1 || linkTextEnd == -1 {
				continue
			}

			level := 1

			for {
				pos := strings.IndexAny(markdown[searchStart:], "()")

				if pos == -1 {
					goto begin
				}

				switch markdown[searchStart+pos] {
				case '(':
					level++
				case ')':
					level--

					if level == 0 {
						urlEnd := searchStart + pos
						searchStart = urlEnd + 1

						linkText := markdown[linkTextStart+1 : linkTextEnd]
						linkURL := markdown[i+1 : urlEnd]

						if linkIsImage {
							r.WriteString("<img src=\"")
							r.WriteString(sanitizeURL(linkURL))
							r.WriteString("\" alt=\"")
							r.WriteString(html.EscapeString(linkText))
							r.WriteString("\">")
						} else {
							r.WriteString("<a href=\"")
							r.WriteString(sanitizeURL(linkURL))
							r.WriteString("\">")
							r.WriteString(html.EscapeString(linkText))
							r.WriteString("</a>")
						}

						linkTextStart = -1
						linkTextEnd = -1
						tokenStart = urlEnd + 1
						goto begin
					}
				}

				searchStart += pos + 1
			}

		case '`':
			end := strings.IndexByte(markdown[searchStart:], '`')

			if end == -1 {
				continue
			}

			r.WriteString(html.EscapeString(markdown[tokenStart:i]))
			r.WriteString("<code>")
			r.WriteString(html.EscapeString(markdown[searchStart : searchStart+end]))
			r.WriteString("</code>")

			searchStart += end + 1
			tokenStart = searchStart

		case '*', '_':
			if i == emStart {
				strongStart = i + 1
				emStart = -1
			} else if strongStart != -1 {
				r.WriteString("<strong>")
				r.WriteString(html.EscapeString(markdown[strongStart:i]))
				r.WriteString("</strong>")
				strongStart = -1
				tokenStart = i + 2
				searchStart = tokenStart
			} else if emStart != -1 {
				r.WriteString("<em>")
				r.WriteString(html.EscapeString(markdown[emStart:i]))
				r.WriteString("</em>")
				emStart = -1
				tokenStart = i + 1
			} else {
				r.WriteString(html.EscapeString(markdown[tokenStart:i]))
				tokenStart = i
				emStart = i + 1
			}

		case '~':
			if i+1 >= len(markdown) || markdown[i+1] != '~' {
				continue
			}

			if strikeStart != -1 {
				r.WriteString("<del>")
				r.WriteString(html.EscapeString(markdown[strikeStart:i]))
				r.WriteString("</del>")
				strikeStart = -1
				tokenStart = i + 2
			} else {
				r.WriteString(html.EscapeString(markdown[tokenStart:i]))
				tokenStart = i
				strikeStart = i + 2
			}

		case '!':
			if i+1 >= len(markdown) || markdown[i+1] != '[' {
				continue
			}

			r.WriteString(html.EscapeString(markdown[tokenStart:i]))
			tokenStart = i
			linkTextStart = i + 1
			searchStart++
			linkIsImage = true
		}
	}
}

// sanitizeURL makes a URL safe to use as the value for a `href` attribute.
func sanitizeURL(linkURL string) string {
	linkURL = strings.TrimSpace(linkURL)

	if strings.HasPrefix(strings.ToLower(linkURL), "javascript:") {
		return ""
	}

	return html.EscapeString(linkURL)
}

// WriteByte adds a single byte to the output.
func (r *renderer) WriteByte(b byte) error {
	r.out = append(r.out, b)
	return nil
}

// WriteString adds a string to the output.
func (r *renderer) WriteString(text string) (int, error) {
	r.out = append(r.out, text...)
	return len(text), nil
}

// nextPowerOf2 calculates the next 32-bit power of 2.
func nextPowerOf2(x uint32) uint32 {
	x--
	x |= x >> 1
	x |= x >> 2
	x |= x >> 4
	x |= x >> 8
	x |= x >> 16
	x++
	return x
}