Ronbun/cmd/web/kaku/parser.go

package kaku

import "strings"

// parser holds the lines being parsed and the current position, so each parse function can advance it without passing an index around.
type parser struct {
	lines []string
	pos   int
}

// Checks if at the end of lines
func (p *parser) done() bool {
	return p.pos >= len(p.lines)
}

// Returns the current line position
func (p *parser) current() string {
	return p.lines[p.pos]
}

// Moves the parser to the next line
func (p *parser) advance() {
	p.pos++
}

// Parse turns raw Kaku source into a Document, which has Node as children.
func Parse(input string) (doc *Document) {
	defer func() {
		if r := recover(); r != nil {
			doc = &Document{Children: []Node{&Paragraph{Children: []Node{&Text{Value: input}}}}}
		}
	}()

	p := &parser{lines: strings.Split(input, "\n")}
	var children []Node

	for !p.done() {
		switch classifyLine(p.current()) {
		case lineBlank:
			p.advance()
		case lineHR:
			children = append(children, &HR{})
			p.advance()
		case lineHeading:
			children = append(children, parseHeading(p.current()))
			p.advance()
		case lineCodeFence:
			children = append(children, parseCodeBlock(p))
		case lineBullet:
			children = append(children, parseList(p, "ul", "- "))
		case lineOrdered:
			children = append(children, parseList(p, "ol", "+ "))
		case lineDef:
			children = append(children, parseDefList(p))
		case lineKeyedTag:
			children = append(children, parseKeyedTagLine(p.current()))
			p.advance()
		case lineAside:
			children = append(children, parseAside(p.current()))
			p.advance()
		case lineTodo:
			children = append(children, parseTodoList(p))
		case lineHTML:
			children = append(children, &RawHTML{Content: p.current()})
			p.advance()
		default:
			children = append(children, parseParagraph(p))
		}
	}

	return &Document{Children: children}
}

// Returns Heading node
func parseHeading(line string) *Heading {
	trimmed := strings.TrimSpace(line)
	level := len(trimmed) - len(strings.TrimLeft(trimmed, "#"))
	text := strings.TrimSpace(trimmed[level:])
	return &Heading{Level: level, Children: parseInline(text)}
}

// Returns CodeBlock node and doesn't parse text between delimeters
func parseCodeBlock(p *parser) *CodeBlock {
	const fence = "```"
	lang := strings.TrimPrefix(strings.TrimSpace(p.current()), fence)
	p.advance() // skip opening fence

	var content []string
	for !p.done() {
		line := p.current()
		if strings.HasPrefix(strings.TrimSpace(line), fence) {
			break
		}
		content = append(content, line)
		p.advance()
	}
	if !p.done() {
		p.advance() // skip closing fence
	}
	return &CodeBlock{Language: lang, Content: strings.Join(content, "\n")}
}

// Joins consecutive plain lines into one paragraph node until a blank line or a different kind of block starts.
func parseParagraph(p *parser) *Paragraph {
	var text []string
	for !p.done() {
		if classifyLine(p.current()) != lineParagraph {
			break
		}
		text = append(text, strings.TrimSpace(p.current()))
		p.advance()
	}
	return &Paragraph{Children: parseInline(strings.Join(text, " "))}
}

// Groups consecutive lines starting with - or + as List node and run content through parseInline
func parseList(p *parser, kind, prefix string) *List {
	var items []*ListItem
	for !p.done() {
		line := strings.TrimSpace(p.current())
		if !strings.HasPrefix(line, prefix) {
			break
		}
		text := strings.TrimPrefix(line, prefix)
		items = append(items, &ListItem{Children: parseInline(text)})
		p.advance()
	}
	return &List{Kind: kind, Items: items}
}

// Groups consecutive "? term : definition" lines into a DefList node. Run content through parseInline.
func parseDefList(p *parser) *DefList {
	const prefix = "? "
	var items []*DefItem
	for !p.done() {
		line := strings.TrimSpace(p.current())
		if !strings.HasPrefix(line, prefix) {
			break
		}
		term, def, _ := strings.Cut(strings.TrimPrefix(line, prefix), ":")
		items = append(items, &DefItem{
			Term:       parseInline(strings.TrimSpace(term)),
			Definition: parseInline(strings.TrimSpace(def)),
		})
		p.advance()
	}
	return &DefList{Items: items}
}

// Builds a Quote, Link, Image, Video, or Audio node from a single "(keyword: ...)" line.
func parseKeyedTagLine(line string) Node {
	trimmed := strings.TrimSpace(line)
	inner := strings.TrimSuffix(strings.TrimPrefix(trimmed, "("), ")")

	switch {
	case strings.HasPrefix(inner, "quote:"):
		f := parseKeyedTag(inner, []string{"quote", "author", "source", "link"})
		return &Quote{
			Text:   parseInline(f["quote"]),
			Author: f["author"],
			Source: f["source"],
			Link:   f["link"],
		}
	case strings.HasPrefix(inner, "link:"):
		f := parseKeyedTag(inner, []string{"link", "text", "label", "title"})
		return &Link{
			URL:   f["link"],
			Text:  parseInline(f["text"]),
			Label: f["label"],
			Title: f["title"]}

	case strings.HasPrefix(inner, "image:"):
		f := parseKeyedTag(inner, []string{"image", "alt", "figcaption"})
		return &Image{
			URL:     f["image"],
			Alt:     f["alt"],
			Caption: parseInline(f["figcaption"])}

	case strings.HasPrefix(inner, "video:"):
		autoplay := strings.Contains(inner, "autoplay")
		cleaned := strings.Replace(inner, "autoplay", "", 1)
		f := parseKeyedTag(cleaned, []string{"video", "figcaption"})
		return &Video{
			URL:      strings.TrimSpace(f["video"]),
			Caption:  parseInline(f["figcaption"]),
			Autoplay: autoplay,
		}

	case strings.HasPrefix(inner, "audio:"):
		f := parseKeyedTag(inner, []string{"audio"})
		return &Audio{
			URL: f["audio"]}

	case strings.HasPrefix(inner, "details:"):
		f := parseKeyedTag(inner, []string{"details", "summary"})
		return &Details{
			Summary: parseInline(f["summary"]),
			Content: parseInline(f["details"]),
		}
	}
	return &Paragraph{Children: parseInline(line)}
}

// Builds an Aside node from a single "{ ... }" line.
func parseAside(line string) *Aside {
	trimmed := strings.TrimSpace(line)
	inner := strings.TrimSuffix(strings.TrimPrefix(trimmed, "{"), "}")
	return &Aside{Children: parseInline(inner)}
}

// Groups consecutive "[X] text" lines into one TodoList node.
func parseTodoList(p *parser) *TodoList {
	var items []*TodoItem
	for !p.done() {
		line := strings.TrimSpace(p.current())
		if classifyLine(line) != lineTodo {
			break
		}
		items = append(items, &TodoItem{
			State:    line[1],
			Children: parseInline(strings.TrimSpace(line[3:])),
		})
		p.advance()
	}
	return &TodoList{Items: items}
}