descheduler/vendor/github.com/gomarkdown/markdown/parser/inline.go

package parser

import (
	"bytes"
	"regexp"
	"strconv"

	"github.com/gomarkdown/markdown/ast"
)

// Parsing of inline elements

var (
	urlRe    = `((https?|ftp):\/\/|\/)[-A-Za-z0-9+&@#\/%?=~_|!:,.;\(\)]+`
	anchorRe = regexp.MustCompile(`^(<a\shref="` + urlRe + `"(\stitle="[^"<>]+")?\s?>` + urlRe + `<\/a>)`)

	// TODO: improve this regexp to catch all possible entities:
	htmlEntityRe = regexp.MustCompile(`&[a-z]{2,5};`)
)

// Inline parses text within a block.
// Each function returns the number of consumed chars.
func (p *Parser) Inline(currBlock ast.Node, data []byte) {
	// handlers might call us recursively: enforce a maximum depth
	if p.nesting >= p.maxNesting || len(data) == 0 {
		return
	}
	p.nesting++
	beg, end := 0, 0

	n := len(data)
	for end < n {
		handler := p.inlineCallback[data[end]]
		if handler == nil {
			end++
			continue
		}
		consumed, node := handler(p, data, end)
		if consumed == 0 {
			// no action from the callback
			end++
			continue
		}
		// copy inactive chars into the output
		ast.AppendChild(currBlock, newTextNode(data[beg:end]))
		if node != nil {
			ast.AppendChild(currBlock, node)
		}
		beg = end + consumed
		end = beg
	}

	if beg < n {
		if data[end-1] == '\n' {
			end--
		}
		ast.AppendChild(currBlock, newTextNode(data[beg:end]))
	}
	p.nesting--
}

// single and double emphasis parsing
func emphasis(p *Parser, data []byte, offset int) (int, ast.Node) {
	data = data[offset:]
	c := data[0]

	n := len(data)
	if n > 2 && data[1] != c {
		// whitespace cannot follow an opening emphasis;
		// strikethrough only takes two characters '~~'
		if isSpace(data[1]) {
			return 0, nil
		}
		if p.extensions&SuperSubscript != 0 && c == '~' {
			// potential subscript, no spaces, except when escaped, helperEmphasis does
			// not check that for us, so walk the bytes and check.
			ret := skipUntilChar(data[1:], 0, c)
			if ret == 0 {
				return 0, nil
			}
			ret++ // we started with data[1:] above.
			for i := 1; i < ret; i++ {
				if isSpace(data[i]) && !isEscape(data, i) {
					return 0, nil
				}
			}
			sub := &ast.Subscript{}
			sub.Literal = data[1:ret]
			return ret + 1, sub
		}
		ret, node := helperEmphasis(p, data[1:], c)
		if ret == 0 {
			return 0, nil
		}

		return ret + 1, node
	}

	if n > 3 && data[1] == c && data[2] != c {
		if isSpace(data[2]) {
			return 0, nil
		}
		ret, node := helperDoubleEmphasis(p, data[2:], c)
		if ret == 0 {
			return 0, nil
		}

		return ret + 2, node
	}

	if n > 4 && data[1] == c && data[2] == c && data[3] != c {
		if c == '~' || isSpace(data[3]) {
			return 0, nil
		}
		ret, node := helperTripleEmphasis(p, data, 3, c)
		if ret == 0 {
			return 0, nil
		}

		return ret + 3, node
	}

	return 0, nil
}

func codeSpan(p *Parser, data []byte, offset int) (int, ast.Node) {
	data = data[offset:]

	// count the number of backticks in the delimiter
	nb := skipChar(data, 0, '`')

	// find the next delimiter
	i, end := 0, 0
	for end = nb; end < len(data) && i < nb; end++ {
		if data[end] == '`' {
			i++
		} else {
			i = 0
		}
	}

	// no matching delimiter?
	if i < nb && end >= len(data) {
		return 0, nil
	}

	// trim outside whitespace
	fBegin := nb
	for fBegin < end && data[fBegin] == ' ' {
		fBegin++
	}

	fEnd := end - nb
	for fEnd > fBegin && data[fEnd-1] == ' ' {
		fEnd--
	}

	// render the code span
	if fBegin != fEnd {
		code := &ast.Code{}
		code.Literal = data[fBegin:fEnd]
		return end, code
	}

	return end, nil
}

// newline preceded by two spaces becomes <br>
func maybeLineBreak(p *Parser, data []byte, offset int) (int, ast.Node) {
	origOffset := offset
	offset = skipChar(data, offset, ' ')

	if offset < len(data) && data[offset] == '\n' {
		if offset-origOffset >= 2 {
			return offset - origOffset + 1, &ast.Hardbreak{}
		}
		return offset - origOffset, nil
	}
	return 0, nil
}

// newline without two spaces works when HardLineBreak is enabled
func lineBreak(p *Parser, data []byte, offset int) (int, ast.Node) {
	if p.extensions&HardLineBreak != 0 {
		return 1, &ast.Hardbreak{}
	}
	return 0, nil
}

type linkType int

const (
	linkNormal linkType = iota
	linkImg
	linkDeferredFootnote
	linkInlineFootnote
	linkCitation
)

func isReferenceStyleLink(data []byte, pos int, t linkType) bool {
	if t == linkDeferredFootnote {
		return false
	}
	return pos < len(data)-1 && data[pos] == '[' && data[pos+1] != '^'
}

func maybeImage(p *Parser, data []byte, offset int) (int, ast.Node) {
	if offset < len(data)-1 && data[offset+1] == '[' {
		return link(p, data, offset)
	}
	return 0, nil
}

func maybeInlineFootnoteOrSuper(p *Parser, data []byte, offset int) (int, ast.Node) {
	if offset < len(data)-1 && data[offset+1] == '[' {
		return link(p, data, offset)
	}

	if p.extensions&SuperSubscript != 0 {
		ret := skipUntilChar(data[offset:], 1, '^')
		if ret == 0 {
			return 0, nil
		}
		for i := offset; i < offset+ret; i++ {
			if isSpace(data[i]) && !isEscape(data, i) {
				return 0, nil
			}
		}
		sup := &ast.Superscript{}
		sup.Literal = data[offset+1 : offset+ret]
		return ret + 1, sup
	}

	return 0, nil
}

// '[': parse a link or an image or a footnote or a citation
func link(p *Parser, data []byte, offset int) (int, ast.Node) {
	// no links allowed inside regular links, footnote, and deferred footnotes
	if p.insideLink && (offset > 0 && data[offset-1] == '[' || len(data)-1 > offset && data[offset+1] == '^') {
		return 0, nil
	}

	var t linkType
	switch {
	// special case: ![^text] == deferred footnote (that follows something with
	// an exclamation point)
	case p.extensions&Footnotes != 0 && len(data)-1 > offset && data[offset+1] == '^':
		t = linkDeferredFootnote
	// ![alt] == image
	case offset >= 0 && data[offset] == '!':
		t = linkImg
		offset++
	// [@citation], [@-citation], [@?citation], [@!citation]
	case p.extensions&Mmark != 0 && len(data)-1 > offset && data[offset+1] == '@':
		t = linkCitation
	// [text] == regular link
	// ^[text] == inline footnote
	// [^refId] == deferred footnote
	case p.extensions&Footnotes != 0:
		if offset >= 0 && data[offset] == '^' {
			t = linkInlineFootnote
			offset++
		} else if len(data)-1 > offset && data[offset+1] == '^' {
			t = linkDeferredFootnote
		}
	default:
		t = linkNormal
	}

	data = data[offset:]

	if t == linkCitation {
		return citation(p, data, 0)
	}

	var (
		i                               = 1
		noteID                          int
		title, link, linkID, altContent []byte
		textHasNl                       = false
	)

	if t == linkDeferredFootnote {
		i++
	}

	// look for the matching closing bracket
	for level := 1; level > 0 && i < len(data); i++ {
		switch {
		case data[i] == '\n':
			textHasNl = true

		case data[i-1] == '\\':
			continue

		case data[i] == '[':
			level++

		case data[i] == ']':
			level--
			if level <= 0 {
				i-- // compensate for extra i++ in for loop
			}
		}
	}

	if i >= len(data) {
		return 0, nil
	}

	txtE := i
	i++
	var footnoteNode ast.Node

	// skip any amount of whitespace or newline
	// (this is much more lax than original markdown syntax)
	i = skipSpace(data, i)

	// inline style link
	switch {
	case i < len(data) && data[i] == '(':
		// skip initial whitespace
		i++

		i = skipSpace(data, i)

		linkB := i
		brace := 0

		// look for link end: ' " )
	findlinkend:
		for i < len(data) {
			switch {
			case data[i] == '\\':
				i += 2

			case data[i] == '(':
				brace++
				i++

			case data[i] == ')':
				if brace <= 0 {
					break findlinkend
				}
				brace--
				i++

			case data[i] == '\'' || data[i] == '"':
				break findlinkend

			default:
				i++
			}
		}

		if i >= len(data) {
			return 0, nil
		}
		linkE := i

		// look for title end if present
		titleB, titleE := 0, 0
		if data[i] == '\'' || data[i] == '"' {
			i++
			titleB = i
			titleEndCharFound := false

		findtitleend:
			for i < len(data) {
				switch {
				case data[i] == '\\':
					i++

				case data[i] == data[titleB-1]: // matching title delimiter
					titleEndCharFound = true

				case titleEndCharFound && data[i] == ')':
					break findtitleend
				}
				i++
			}

			if i >= len(data) {
				return 0, nil
			}

			// skip whitespace after title
			titleE = i - 1
			for titleE > titleB && isSpace(data[titleE]) {
				titleE--
			}

			// check for closing quote presence
			if data[titleE] != '\'' && data[titleE] != '"' {
				titleB, titleE = 0, 0
				linkE = i
			}
		}

		// remove whitespace at the end of the link
		for linkE > linkB && isSpace(data[linkE-1]) {
			linkE--
		}

		// remove optional angle brackets around the link
		if data[linkB] == '<' {
			linkB++
		}
		if data[linkE-1] == '>' {
			linkE--
		}

		// build escaped link and title
		if linkE > linkB {
			link = data[linkB:linkE]
		}

		if titleE > titleB {
			title = data[titleB:titleE]
		}

		i++

	// reference style link
	case isReferenceStyleLink(data, i, t):
		var id []byte
		altContentConsidered := false

		// look for the id
		i++
		linkB := i
		i = skipUntilChar(data, i, ']')

		if i >= len(data) {
			return 0, nil
		}
		linkE := i

		// find the reference
		if linkB == linkE {
			if textHasNl {
				var b bytes.Buffer

				for j := 1; j < txtE; j++ {
					switch {
					case data[j] != '\n':
						b.WriteByte(data[j])
					case data[j-1] != ' ':
						b.WriteByte(' ')
					}
				}

				id = b.Bytes()
			} else {
				id = data[1:txtE]
				altContentConsidered = true
			}
		} else {
			id = data[linkB:linkE]
		}

		// find the reference with matching id
		lr, ok := p.getRef(string(id))
		if !ok {
			return 0, nil
		}

		// keep link and title from reference
		linkID = id
		link = lr.link
		title = lr.title
		if altContentConsidered {
			altContent = lr.text
		}
		i++

	// shortcut reference style link or reference or inline footnote
	default:
		var id []byte

		// craft the id
		if textHasNl {
			var b bytes.Buffer

			for j := 1; j < txtE; j++ {
				switch {
				case data[j] != '\n':
					b.WriteByte(data[j])
				case data[j-1] != ' ':
					b.WriteByte(' ')
				}
			}

			id = b.Bytes()
		} else {
			if t == linkDeferredFootnote {
				id = data[2:txtE] // get rid of the ^
			} else {
				id = data[1:txtE]
			}
		}

		footnoteNode = &ast.ListItem{}
		if t == linkInlineFootnote {
			// create a new reference
			noteID = len(p.notes) + 1

			var fragment []byte
			if len(id) > 0 {
				if len(id) < 16 {
					fragment = make([]byte, len(id))
				} else {
					fragment = make([]byte, 16)
				}
				copy(fragment, slugify(id))
			} else {
				fragment = append([]byte("footnote-"), []byte(strconv.Itoa(noteID))...)
			}

			ref := &reference{
				noteID:   noteID,
				hasBlock: false,
				link:     fragment,
				title:    id,
				footnote: footnoteNode,
			}

			p.notes = append(p.notes, ref)
			p.refsRecord[string(ref.link)] = struct{}{}

			link = ref.link
			title = ref.title
		} else {
			// find the reference with matching id
			lr, ok := p.getRef(string(id))
			if !ok {
				return 0, nil
			}

			if t == linkDeferredFootnote && !p.isFootnote(lr) {
				lr.noteID = len(p.notes) + 1
				lr.footnote = footnoteNode
				p.notes = append(p.notes, lr)
				p.refsRecord[string(lr.link)] = struct{}{}
			}

			// keep link and title from reference
			link = lr.link
			// if inline footnote, title == footnote contents
			title = lr.title
			noteID = lr.noteID
			if len(lr.text) > 0 {
				altContent = lr.text
			}
		}

		// rewind the whitespace
		i = txtE + 1
	}

	var uLink []byte
	if t == linkNormal || t == linkImg {
		if len(link) > 0 {
			var uLinkBuf bytes.Buffer
			unescapeText(&uLinkBuf, link)
			uLink = uLinkBuf.Bytes()
		}

		// links need something to click on and somewhere to go
		if len(uLink) == 0 || (t == linkNormal && txtE <= 1) {
			return 0, nil
		}
	}

	// call the relevant rendering function
	switch t {
	case linkNormal:
		link := &ast.Link{
			Destination: normalizeURI(uLink),
			Title:       title,
			DeferredID:  linkID,
		}
		if len(altContent) > 0 {
			ast.AppendChild(link, newTextNode(altContent))
		} else {
			// links cannot contain other links, so turn off link parsing
			// temporarily and recurse
			insideLink := p.insideLink
			p.insideLink = true
			p.Inline(link, data[1:txtE])
			p.insideLink = insideLink
		}
		return i, link

	case linkImg:
		image := &ast.Image{
			Destination: uLink,
			Title:       title,
		}
		ast.AppendChild(image, newTextNode(data[1:txtE]))
		return i + 1, image

	case linkInlineFootnote, linkDeferredFootnote:
		link := &ast.Link{
			Destination: link,