From 0917dc876613fd71c9726a34bf0138b4f5121be9 Mon Sep 17 00:00:00 2001
From: Benjamin <b.mpickford@outlook.com>
Date: Mon, 18 Nov 2019 06:18:01 +1000
Subject: Update markdown parsing library to github.com/gomarkdown/markdown
(#944)
---
.../github.com/gomarkdown/markdown/parser/aside.go | 73 +
.../gomarkdown/markdown/parser/attribute.go | 116 ++
.../github.com/gomarkdown/markdown/parser/block.go | 1978 ++++++++++++++++++++
.../gomarkdown/markdown/parser/callout.go | 29 +
.../gomarkdown/markdown/parser/caption.go | 70 +
.../gomarkdown/markdown/parser/citation.go | 86 +
.../github.com/gomarkdown/markdown/parser/esc.go | 20 +
.../gomarkdown/markdown/parser/figures.go | 119 ++
.../gomarkdown/markdown/parser/include.go | 129 ++
.../gomarkdown/markdown/parser/inline.go | 1284 +++++++++++++
.../gomarkdown/markdown/parser/matter.go | 36 +
.../gomarkdown/markdown/parser/options.go | 32 +
.../gomarkdown/markdown/parser/parser.go | 812 ++++++++
.../github.com/gomarkdown/markdown/parser/ref.go | 89 +
14 files changed, 4873 insertions(+)
create mode 100644 vendor/github.com/gomarkdown/markdown/parser/aside.go
create mode 100644 vendor/github.com/gomarkdown/markdown/parser/attribute.go
create mode 100644 vendor/github.com/gomarkdown/markdown/parser/block.go
create mode 100644 vendor/github.com/gomarkdown/markdown/parser/callout.go
create mode 100644 vendor/github.com/gomarkdown/markdown/parser/caption.go
create mode 100644 vendor/github.com/gomarkdown/markdown/parser/citation.go
create mode 100644 vendor/github.com/gomarkdown/markdown/parser/esc.go
create mode 100644 vendor/github.com/gomarkdown/markdown/parser/figures.go
create mode 100644 vendor/github.com/gomarkdown/markdown/parser/include.go
create mode 100644 vendor/github.com/gomarkdown/markdown/parser/inline.go
create mode 100644 vendor/github.com/gomarkdown/markdown/parser/matter.go
create mode 100644 vendor/github.com/gomarkdown/markdown/parser/options.go
create mode 100644 vendor/github.com/gomarkdown/markdown/parser/parser.go
create mode 100644 vendor/github.com/gomarkdown/markdown/parser/ref.go
(limited to 'vendor/github.com/gomarkdown/markdown/parser')
diff --git a/vendor/github.com/gomarkdown/markdown/parser/aside.go b/vendor/github.com/gomarkdown/markdown/parser/aside.go
new file mode 100644
index 00000000..96e25fe0
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/parser/aside.go
@@ -0,0 +1,73 @@
+package parser
+
+import (
+ "bytes"
+
+ "github.com/gomarkdown/markdown/ast"
+)
+
+// returns aisde prefix length
+func (p *Parser) asidePrefix(data []byte) int {
+ i := 0
+ n := len(data)
+ for i < 3 && i < n && data[i] == ' ' {
+ i++
+ }
+ if i+1 < n && data[i] == 'A' && data[i+1] == '>' {
+ if i+2 < n && data[i+2] == ' ' {
+ return i + 3
+ }
+ return i + 2
+ }
+ return 0
+}
+
+// aside ends with at least one blank line
+// followed by something without a aside prefix
+func (p *Parser) terminateAside(data []byte, beg, end int) bool {
+ if p.isEmpty(data[beg:]) <= 0 {
+ return false
+ }
+ if end >= len(data) {
+ return true
+ }
+ return p.asidePrefix(data[end:]) == 0 && p.isEmpty(data[end:]) == 0
+}
+
+// parse a aside fragment
+func (p *Parser) aside(data []byte) int {
+ var raw bytes.Buffer
+ beg, end := 0, 0
+ // identical to quote
+ for beg < len(data) {
+ end = beg
+ // Step over whole lines, collecting them. While doing that, check for
+ // fenced code and if one's found, incorporate it altogether,
+ // irregardless of any contents inside it
+ for end < len(data) && data[end] != '\n' {
+ if p.extensions&FencedCode != 0 {
+ if i := p.fencedCodeBlock(data[end:], false); i > 0 {
+ // -1 to compensate for the extra end++ after the loop:
+ end += i - 1
+ break
+ }
+ }
+ end++
+ }
+ end = skipCharN(data, end, '\n', 1)
+ if pre := p.asidePrefix(data[beg:]); pre > 0 {
+ // skip the prefix
+ beg += pre
+ } else if p.terminateAside(data, beg, end) {
+ break
+ }
+ // this line is part of the aside
+ raw.Write(data[beg:end])
+ beg = end
+ }
+
+ block := p.addBlock(&ast.Aside{})
+ p.block(raw.Bytes())
+ p.finalize(block)
+ return end
+}
diff --git a/vendor/github.com/gomarkdown/markdown/parser/attribute.go b/vendor/github.com/gomarkdown/markdown/parser/attribute.go
new file mode 100644
index 00000000..5fdb0709
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/parser/attribute.go
@@ -0,0 +1,116 @@
+package parser
+
+import (
+ "bytes"
+
+ "github.com/gomarkdown/markdown/ast"
+)
+
+// attribute parses a (potential) block attribute and adds it to p.
+func (p *Parser) attribute(data []byte) []byte {
+ if len(data) < 3 {
+ return data
+ }
+ i := 0
+ if data[i] != '{' {
+ return data
+ }
+ i++
+
+ // last character must be a } otherwise it's not an attribute
+ end := skipUntilChar(data, i, '\n')
+ if data[end-1] != '}' {
+ return data
+ }
+
+ i = skipSpace(data, i)
+ b := &ast.Attribute{Attrs: make(map[string][]byte)}
+
+ esc := false
+ quote := false
+ trail := 0
+Loop:
+ for ; i < len(data); i++ {
+ switch data[i] {
+ case ' ', '\t', '\f', '\v':
+ if quote {
+ continue
+ }
+ chunk := data[trail+1 : i]
+ if len(chunk) == 0 {
+ trail = i
+ continue
+ }
+ switch {
+ case chunk[0] == '.':
+ b.Classes = append(b.Classes, chunk[1:])
+ case chunk[0] == '#':
+ b.ID = chunk[1:]
+ default:
+ k, v := keyValue(chunk)
+ if k != nil && v != nil {
+ b.Attrs[string(k)] = v
+ } else {
+ // this is illegal in an attribute
+ return data
+ }
+ }
+ trail = i
+ case '"':
+ if esc {
+ esc = !esc
+ continue
+ }
+ quote = !quote
+ case '\\':
+ esc = !esc
+ case '}':
+ if esc {
+ esc = !esc
+ continue
+ }
+ chunk := data[trail+1 : i]
+ if len(chunk) == 0 {
+ return data
+ }
+ switch {
+ case chunk[0] == '.':
+ b.Classes = append(b.Classes, chunk[1:])
+ case chunk[0] == '#':
+ b.ID = chunk[1:]
+ default:
+ k, v := keyValue(chunk)
+ if k != nil && v != nil {
+ b.Attrs[string(k)] = v
+ } else {
+ return data
+ }
+ }
+ i++
+ break Loop
+ default:
+ esc = false
+ }
+ }
+
+ p.attr = b
+ return data[i:]
+}
+
+// key="value" quotes are mandatory.
+func keyValue(data []byte) ([]byte, []byte) {
+ chunk := bytes.SplitN(data, []byte{'='}, 2)
+ if len(chunk) != 2 {
+ return nil, nil
+ }
+ key := chunk[0]
+ value := chunk[1]
+
+ if len(value) < 3 || len(key) == 0 {
+ return nil, nil
+ }
+ if value[0] != '"' || value[len(value)-1] != '"' {
+ return key, nil
+ }
+ return key, value[1 : len(value)-1]
+}
diff --git a/vendor/github.com/gomarkdown/markdown/parser/block.go b/vendor/github.com/gomarkdown/markdown/parser/block.go
new file mode 100644
index 00000000..18a2dd89
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/parser/block.go
@@ -0,0 +1,1978 @@
+package parser
+
+import (
+ "bytes"
+ "html"
+ "regexp"
+ "strconv"
+ "unicode"
+
+ "github.com/gomarkdown/markdown/ast"
+)
+
+// Parsing block-level elements.
+
+const (
+ charEntity = "&(?:#x[a-f0-9]{1,8}|#[0-9]{1,8}|[a-z][a-z0-9]{1,31});"
+ escapable = "[!\"#$%&'()*+,./:;<=>?@[\\\\\\]^_`{|}~-]"
+)
+
+var (
+ reBackslashOrAmp = regexp.MustCompile("[\\&]")
+ reEntityOrEscapedChar = regexp.MustCompile("(?i)\\\\" + escapable + "|" + charEntity)
+
+ // blockTags is a set of tags that are recognized as HTML block tags.
+ // Any of these can be included in markdown text without special escaping.
+ blockTags = map[string]struct{}{
+ "blockquote": struct{}{},
+ "del": struct{}{},
+ "div": struct{}{},
+ "dl": struct{}{},
+ "fieldset": struct{}{},
+ "form": struct{}{},
+ "h1": struct{}{},
+ "h2": struct{}{},
+ "h3": struct{}{},
+ "h4": struct{}{},
+ "h5": struct{}{},
+ "h6": struct{}{},
+ "iframe": struct{}{},
+ "ins": struct{}{},
+ "math": struct{}{},
+ "noscript": struct{}{},
+ "ol": struct{}{},
+ "pre": struct{}{},
+ "p": struct{}{},
+ "script": struct{}{},
+ "style": struct{}{},
+ "table": struct{}{},
+ "ul": struct{}{},
+
+ // HTML5
+ "address": struct{}{},
+ "article": struct{}{},
+ "aside": struct{}{},
+ "canvas": struct{}{},
+ "figcaption": struct{}{},
+ "figure": struct{}{},
+ "footer": struct{}{},
+ "header": struct{}{},
+ "hgroup": struct{}{},
+ "main": struct{}{},
+ "nav": struct{}{},
+ "output": struct{}{},
+ "progress": struct{}{},
+ "section": struct{}{},
+ "video": struct{}{},
+ }
+)
+
+// sanitizeAnchorName returns a sanitized anchor name for the given text.
+// Taken from https://github.com/shurcooL/sanitized_anchor_name/blob/master/main.go#L14:1
+func sanitizeAnchorName(text string) string {
+ var anchorName []rune
+ var futureDash = false
+ for _, r := range text {
+ switch {
+ case unicode.IsLetter(r) || unicode.IsNumber(r):
+ if futureDash && len(anchorName) > 0 {
+ anchorName = append(anchorName, '-')
+ }
+ futureDash = false
+ anchorName = append(anchorName, unicode.ToLower(r))
+ default:
+ futureDash = true
+ }
+ }
+ return string(anchorName)
+}
+
+// Parse block-level data.
+// Note: this function and many that it calls assume that
+// the input buffer ends with a newline.
+func (p *Parser) block(data []byte) {
+ // this is called recursively: enforce a maximum depth
+ if p.nesting >= p.maxNesting {
+ return
+ }
+ p.nesting++
+
+ // parse out one block-level construct at a time
+ for len(data) > 0 {
+ // attributes that can be specific before a block element:
+ //
+ // {#id .class1 .class2 key="value"}
+ if p.extensions&Attributes != 0 {
+ data = p.attribute(data)
+ }
+
+ if p.extensions&Includes != 0 {
+ f := p.readInclude
+ path, address, consumed := p.isInclude(data)
+ if consumed == 0 {
+ path, address, consumed = p.isCodeInclude(data)
+ f = p.readCodeInclude
+ }
+ if consumed > 0 {
+ included := f(p.includeStack.Last(), path, address)
+ p.includeStack.Push(path)
+ p.block(included)
+ p.includeStack.Pop()
+ data = data[consumed:]
+ continue
+ }
+ }
+
+ // user supplied parser function
+ if p.Opts.ParserHook != nil {
+ node, blockdata, consumed := p.Opts.ParserHook(data)
+ if consumed > 0 {
+ data = data[consumed:]
+
+ if node != nil {
+ p.addBlock(node)
+ if blockdata != nil {
+ p.block(blockdata)
+ p.finalize(node)
+ }
+ }
+ continue
+ }
+ }
+
+ // prefixed heading:
+ //
+ // # Heading 1
+ // ## Heading 2
+ // ...
+ // ###### Heading 6
+ if p.isPrefixHeading(data) {
+ data = data[p.prefixHeading(data):]
+ continue
+ }
+
+ // prefixed special heading:
+ // (there are no levels.)
+ //
+ // .# Abstract
+ if p.isPrefixSpecialHeading(data) {
+ data = data[p.prefixSpecialHeading(data):]
+ continue
+ }
+
+ // block of preformatted HTML:
+ //
+ // <div>
+ // ...
+ // </div>
+ if data[0] == '<' {
+ if i := p.html(data, true); i > 0 {
+ data = data[i:]
+ continue
+ }
+ }
+
+ // title block
+ //
+ // % stuff
+ // % more stuff
+ // % even more stuff
+ if p.extensions&Titleblock != 0 {
+ if data[0] == '%' {
+ if i := p.titleBlock(data, true); i > 0 {
+ data = data[i:]
+ continue
+ }
+ }
+ }
+
+ // blank lines. note: returns the # of bytes to skip
+ if i := p.isEmpty(data); i > 0 {
+ data = data[i:]
+ continue
+ }
+
+ // indented code block:
+ //
+ // func max(a, b int) int {
+ // if a > b {
+ // return a
+ // }
+ // return b
+ // }
+ if p.codePrefix(data) > 0 {
+ data = data[p.code(data):]
+ continue
+ }
+
+ // fenced code block:
+ //
+ // ``` go
+ // func fact(n int) int {
+ // if n <= 1 {
+ // return n
+ // }
+ // return n * fact(n-1)
+ // }
+ // ```
+ if p.extensions&FencedCode != 0 {
+ if i := p.fencedCodeBlock(data, true); i > 0 {
+ data = data[i:]
+ continue
+ }
+ }
+
+ // horizontal rule:
+ //
+ // ------
+ // or
+ // ******
+ // or
+ // ______
+ if p.isHRule(data) {
+ p.addBlock(&ast.HorizontalRule{})
+ i := skipUntilChar(data, 0, '\n')
+ data = data[i:]
+ continue
+ }
+
+ // block quote:
+ //
+ // > A big quote I found somewhere
+ // > on the web
+ if p.quotePrefix(data) > 0 {
+ data = data[p.quote(data):]
+ continue
+ }
+
+ // aside:
+ //
+ // A> The proof is too large to fit
+ // A> in the margin.
+ if p.extensions&Mmark != 0 {
+ if p.asidePrefix(data) > 0 {
+ data = data[p.aside(data):]
+ continue
+ }
+ }
+
+ // figure block:
+ //
+ // !---
+ // 
+ // 
+ // !---
+ if p.extensions&Mmark != 0 {
+ if i := p.figureBlock(data, true); i > 0 {
+ data = data[i:]
+ continue
+ }
+ }
+
+ // table:
+ //
+ // Name | Age | Phone
+ // ------|-----|---------
+ // Bob | 31 | 555-1234
+ // Alice | 27 | 555-4321
+ if p.extensions&Tables != 0 {
+ if i := p.table(data); i > 0 {
+ data = data[i:]
+ continue
+ }
+ }
+
+ // an itemized/unordered list:
+ //
+ // * Item 1
+ // * Item 2
+ //
+ // also works with + or -
+ if p.uliPrefix(data) > 0 {
+ data = data[p.list(data, 0, 0):]
+ continue
+ }
+
+ // a numbered/ordered list:
+ //
+ // 1. Item 1
+ // 2. Item 2
+ if i := p.oliPrefix(data); i > 0 {
+ start := 0
+ if i > 2 && p.extensions&OrderedListStart != 0 {
+ s := string(data[:i-2])
+ start, _ = strconv.Atoi(s)
+ if start == 1 {
+ start = 0
+ }
+ }
+ data = data[p.list(data, ast.ListTypeOrdered, start):]
+ continue
+ }
+
+ // definition lists:
+ //
+ // Term 1
+ // : Definition a
+ // : Definition b
+ //
+ // Term 2
+ // : Definition c
+ if p.extensions&DefinitionLists != 0 {
+ if p.dliPrefix(data) > 0 {
+ data = data[p.list(data, ast.ListTypeDefinition, 0):]
+ continue
+ }
+ }
+
+ if p.extensions&MathJax != 0 {
+ if i := p.blockMath(data); i > 0 {
+ data = data[i:]
+ continue
+ }
+ }
+
+ // document matters:
+ //
+ // {frontmatter}/{mainmatter}/{backmatter}
+ if p.extensions&Mmark != 0 {
+ if i := p.documentMatter(data); i > 0 {
+ data = data[i:]
+ continue
+ }
+ }
+
+ // anything else must look like a normal paragraph
+ // note: this finds underlined headings, too
+ idx := p.paragraph(data)
+ data = data[idx:]
+ }
+
+ p.nesting--
+}
+
+func (p *Parser) addBlock(n ast.Node) ast.Node {
+ p.closeUnmatchedBlocks()
+
+ if p.attr != nil {
+ if c := n.AsContainer(); c != nil {
+ c.Attribute = p.attr
+ }
+ if l := n.AsLeaf(); l != nil {
+ l.Attribute = p.attr
+ }
+ p.attr = nil
+ }
+ return p.addChild(n)
+}
+
+func (p *Parser) isPrefixHeading(data []byte) bool {
+ if data[0] != '#' {
+ return false
+ }
+
+ if p.extensions&SpaceHeadings != 0 {
+ level := skipCharN(data, 0, '#', 6)
+ if level == len(data) || data[level] != ' ' {
+ return false
+ }
+ }
+ return true
+}
+
+func (p *Parser) prefixHeading(data []byte) int {
+ level := skipCharN(data, 0, '#', 6)
+ i := skipChar(data, level, ' ')
+ end := skipUntilChar(data, i, '\n')
+ skip := end
+ id := ""
+ if p.extensions&HeadingIDs != 0 {
+ j, k := 0, 0
+ // find start/end of heading id
+ for j = i; j < end-1 && (data[j] != '{' || data[j+1] != '#'); j++ {
+ }
+ for k = j + 1; k < end && data[k] != '}'; k++ {
+ }
+ // extract heading id iff found
+ if j < end && k < end {
+ id = string(data[j+2 : k])
+ end = j
+ skip = k + 1
+ for end > 0 && data[end-1] == ' ' {
+ end--
+ }
+ }
+ }
+ for end > 0 && data[end-1] == '#' {
+ if isBackslashEscaped(data, end-1) {
+ break
+ }
+ end--
+ }
+ for end > 0 && data[end-1] == ' ' {
+ end--
+ }
+ if end > i {
+ if id == "" && p.extensions&AutoHeadingIDs != 0 {
+ id = sanitizeAnchorName(string(data[i:end]))
+ }
+ block := &ast.Heading{
+ HeadingID: id,
+ Level: level,
+ }
+ block.Content = data[i:end]
+ p.addBlock(block)
+ }
+ return skip
+}
+
+func (p *Parser) isPrefixSpecialHeading(data []byte) bool {
+ if p.extensions|Mmark == 0 {
+ return false
+ }
+ if len(data) < 4 {
+ return false
+ }
+ if data[0] != '.' {
+ return false
+ }
+ if data[1] != '#' {
+ return false
+ }
+ if data[2] == '#' { // we don't support level, so nack this.
+ return false
+ }
+
+ if p.extensions&SpaceHeadings != 0 {
+ if data[2] != ' ' {
+ return false
+ }
+ }
+ return true
+}
+
+func (p *Parser) prefixSpecialHeading(data []byte) int {
+ i := skipChar(data, 2, ' ') // ".#" skipped
+ end := skipUntilChar(data, i, '\n')
+ skip := end
+ id := ""
+ if p.extensions&HeadingIDs != 0 {
+ j, k := 0, 0
+ // find start/end of heading id
+ for j = i; j < end-1 && (data[j] != '{' || data[j+1] != '#'); j++ {
+ }
+ for k = j + 1; k < end && data[k] != '}'; k++ {
+ }
+ // extract heading id iff found
+ if j < end && k < end {
+ id = string(data[j+2 : k])
+ end = j
+ skip = k + 1
+ for end > 0 && data[end-1] == ' ' {
+ end--
+ }
+ }
+ }
+ for end > 0 && data[end-1] == '#' {
+ if isBackslashEscaped(data, end-1) {
+ break
+ }
+ end--
+ }
+ for end > 0 && data[end-1] == ' ' {
+ end--
+ }
+ if end > i {
+ if id == "" && p.extensions&AutoHeadingIDs != 0 {
+ id = sanitizeAnchorName(string(data[i:end]))
+ }
+ block := &ast.Heading{
+ HeadingID: id,
+ IsSpecial: true,
+ Level: 1, // always level 1.
+ }
+ block.Literal = data[i:end]
+ block.Content = data[i:end]
+ p.addBlock(block)
+ }
+ return skip
+}
+
+func (p *Parser) isUnderlinedHeading(data []byte) int {
+ // test of level 1 heading
+ if data[0] == '=' {
+ i := skipChar(data, 1, '=')
+ i = skipChar(data, i, ' ')
+ if i < len(data) && data[i] == '\n' {
+ return 1
+ }
+ return 0
+ }
+
+ // test of level 2 heading
+ if data[0] == '-' {
+ i := skipChar(data, 1, '-')
+ i = skipChar(data, i, ' ')
+ if i < len(data) && data[i] == '\n' {
+ return 2
+ }
+ return 0
+ }
+
+ return 0
+}
+
+func (p *Parser) titleBlock(data []byte, doRender bool) int {
+ if data[0] != '%' {
+ return 0
+ }
+ splitData := bytes.Split(data, []byte("\n"))
+ var i int
+ for idx, b := range splitData {
+ if !bytes.HasPrefix(b, []byte("%")) {
+ i = idx // - 1
+ break
+ }
+ }
+
+ data = bytes.Join(splitData[0:i], []byte("\n"))
+ consumed := len(data)
+ data = bytes.TrimPrefix(data, []byte("% "))
+ data = bytes.Replace(data, []byte("\n% "), []byte("\n"), -1)
+ block := &ast.Heading{
+ Level: 1,
+ IsTitleblock: true,
+ }
+ block.Content = data
+ p.addBlock(block)
+
+ return consumed
+}
+
+func (p *Parser) html(data []byte, doRender bool) int {
+ var i, j int
+
+ // identify the opening tag
+ if data[0] != '<' {
+ return 0
+ }
+ curtag, tagfound := p.htmlFindTag(data[1:])
+
+ // handle special cases
+ if !tagfound {
+ // check for an HTML comment
+ if size := p.htmlComment(data, doRender); size > 0 {
+ return size
+ }
+
+ // check for an <hr> tag
+ if size := p.htmlHr(data, doRender); size > 0 {
+ return size
+ }
+
+ // no special case recognized
+ return 0
+ }
+
+ // look for an unindented matching closing tag
+ // followed by a blank line
+ found := false
+ /*
+ closetag := []byte("\n</" + curtag + ">")
+ j = len(curtag) + 1
+ for !found {
+ // scan for a closing tag at the beginning of a line
+ if skip := bytes.Index(data[j:], closetag); skip >= 0 {
+ j += skip + len(closetag)
+ } else {
+ break
+ }
+
+ // see if it is the only thing on the line
+ if skip := p.isEmpty(data[j:]); skip > 0 {
+ // see if it is followed by a blank line/eof
+ j += skip
+ if j >= len(data) {
+ found = true
+ i = j
+ } else {
+ if skip := p.isEmpty(data[j:]); skip > 0 {
+ j += skip
+ found = true
+ i = j
+ }
+ }
+ }
+ }
+ */
+
+ // if not found, try a second pass looking for indented match
+ // but not if tag is "ins" or "del" (following original Markdown.pl)
+ if !found && curtag != "ins" && curtag != "del" {
+ i = 1
+ for i < len(data) {
+ i++
+ for i < len(data) && !(data[i-1] == '<' && data[i] == '/') {
+ i++
+ }
+
+ if i+2+len(curtag) >= len(data) {
+ break
+ }
+
+ j = p.htmlFindEnd(curtag, data[i-1:])
+
+ if j > 0 {
+ i += j - 1
+ found = true
+ break
+ }
+ }
+ }
+
+ if !found {
+ return 0
+ }
+
+ // the end of the block has been found
+ if doRender {
+ // trim newlines
+ end := backChar(data, i, '\n')
+ htmlBLock := &ast.HTMLBlock{ast.Leaf{Content: data[:end]}}
+ p.addBlock(htmlBLock)
+ finalizeHTMLBlock(htmlBLock)
+ }
+
+ return i
+}
+
+func finalizeHTMLBlock(block *ast.HTMLBlock) {
+ block.Literal = block.Content
+ block.Content = nil
+}
+
+// HTML comment, lax form
+func (p *Parser) htmlComment(data []byte, doRender bool) int {
+ i := p.inlineHTMLComment(data)
+ // needs to end with a blank line
+ if j := p.isEmpty(data[i:]); j > 0 {
+ size := i + j
+ if doRender {
+ // trim trailing newlines
+ end := backChar(data, size, '\n')
+ htmlBLock := &ast.HTMLBlock{ast.Leaf{Content: data[:end]}}
+ p.addBlock(htmlBLock)
+ finalizeHTMLBlock(htmlBLock)
+ }
+ return size
+ }
+ return 0
+}
+
+// HR, which is the only self-closing block tag considered
+func (p *Parser) htmlHr(data []byte, doRender bool) int {
+ if len(data) < 4 {
+ return 0
+ }
+ if data[0] != '<' || (data[1] != 'h' && data[1] != 'H') || (data[2] != 'r' && data[2] != 'R') {
+ return 0
+ }
+ if data[3] != ' ' && data[3] != '/' && data[3] != '>' {
+ // not an <hr> tag after all; at least not a valid one
+ return 0
+ }
+ i := 3
+ for i < len(data) && data[i] != '>' && data[i] != '\n' {
+ i++
+ }
+ if i < len(data) && data[i] == '>' {
+ i++
+ if j := p.isEmpty(data[i:]); j > 0 {
+ size := i + j
+ if doRender {
+ // trim newlines
+ end := backChar(data, size, '\n')
+ htmlBlock := &ast.HTMLBlock{ast.Leaf{Content: data[:end]}}
+ p.addBlock(htmlBlock)
+ finalizeHTMLBlock(htmlBlock)
+ }
+ return size
+ }
+ }
+ return 0
+}
+
+func (p *Parser) htmlFindTag(data []byte) (string, bool) {
+ i := skipAlnum(data, 0)
+ key := string(data[:i])
+ if _, ok := blockTags[key]; ok {
+ return key, true
+ }
+ return "", false
+}
+
+func (p *Parser) htmlFindEnd(tag string, data []byte) int {
+ // assume data[0] == '<' && data[1] == '/' already tested
+ if tag == "hr" {
+ return 2
+ }
+ // check if tag is a match
+ closetag := []byte("</" + tag + ">")
+ if !bytes.HasPrefix(data, closetag) {
+ return 0
+ }
+ i := len(closetag)
+
+ // check that the rest of the line is blank
+ skip := 0
+ if skip = p.isEmpty(data[i:]); skip == 0 {
+ return 0
+ }
+ i += skip
+ skip = 0
+
+ if i >= len(data) {
+ return i
+ }
+
+ if p.extensions&LaxHTMLBlocks != 0 {
+ return i
+ }
+ if skip = p.isEmpty(data[i:]); skip == 0 {
+ // following line must be blank
+ return 0
+ }
+
+ return i + skip
+}
+
+func (*Parser) isEmpty(data []byte) int {
+ // it is okay to call isEmpty on an empty buffer
+ if len(data) == 0 {
+ return 0
+ }
+
+ var i int
+ for i = 0; i < len(data) && data[i] != '\n'; i++ {
+ if data[i] != ' ' && data[i] != '\t' {
+ return 0
+ }
+ }
+ i = skipCharN(data, i, '\n', 1)
+ return i
+}
+
+func (*Parser) isHRule(data []byte) bool {
+ i := 0
+
+ // skip up to three spaces
+ for i < 3 && data[i] == ' ' {
+ i++
+ }
+
+ // look at the hrule char
+ if data[i] != '*' && data[i] != '-' && data[i] != '_' {
+ return false
+ }
+ c := data[i]
+
+ // the whole line must be the char or whitespace
+ n := 0
+ for i < len(data) && data[i] != '\n' {
+ switch {
+ case data[i] == c:
+ n++
+ case data[i] != ' ':
+ return false
+ }
+ i++
+ }
+
+ return n >= 3
+}
+
+// isFenceLine checks if there's a fence line (e.g., ``` or ``` go) at the beginning of data,
+// and returns the end index if so, or 0 otherwise. It also returns the marker found.
+// If syntax is not nil, it gets set to the syntax specified in the fence line.
+func isFenceLine(data []byte, syntax *string, oldmarker string) (end int, marker string) {
+ i, size := 0, 0
+
+ n := len(data)
+ // skip up to three spaces
+ for i < n && i < 3 && data[i] == ' ' {
+ i++
+ }
+
+ // check for the marker characters: ~ or `
+ if i >= n {
+ return 0, ""
+ }
+ if data[i] != '~' && data[i] != '`' {
+ return 0, ""
+ }
+
+ c := data[i]
+
+ // the whole line must be the same char or whitespace
+ for i < n && data[i] == c {
+ size++
+ i++
+ }
+
+ // the marker char must occur at least 3 times
+ if size < 3 {
+ return 0, ""
+ }
+ marker = string(data[i-size : i])
+
+ // if this is the end marker, it must match the beginning marker
+ if oldmarker != "" && marker != oldmarker {
+ return 0, ""
+ }
+
+ // TODO(shurcooL): It's probably a good idea to simplify the 2 code paths here
+ // into one, always get the syntax, and discard it if the caller doesn't care.
+ if syntax != nil {
+ syn := 0
+ i = skipChar(data, i, ' ')
+
+ if i >= n {
+ if i == n {
+ return i, marker
+ }
+ return 0, ""
+ }
+
+ syntaxStart := i
+
+ if data[i] == '{' {
+ i++
+ syntaxStart++
+
+ for i < n && data[i] != '}' && data[i] != '\n' {
+ syn++
+ i++
+ }
+
+ if i >= n || data[i] != '}' {
+ return 0, ""
+ }
+
+ // strip all whitespace at the beginning and the end
+ // of the {} block
+ for syn > 0 && isSpace(data[syntaxStart]) {
+ syntaxStart++
+ syn--
+ }
+
+ for syn > 0 && isSpace(data[syntaxStart+syn-1]) {
+ syn--
+ }
+
+ i++
+ } else {
+ for i < n && !isSpace(data[i]) {
+ syn++
+ i++
+ }
+ }
+
+ *syntax = string(data[syntaxStart : syntaxStart+syn])
+ }
+
+ i = skipChar(data, i, ' ')
+ if i >= n || data[i] != '\n' {
+ if i == n {
+ return i, marker
+ }
+ return 0, ""
+ }
+ return i + 1, marker // Take newline into account.
+}
+
+// fencedCodeBlock returns the end index if data contains a fenced code block at the beginning,
+// or 0 otherwise. It writes to out if doRender is true, otherwise it has no side effects.
+// If doRender is true, a final newline is mandatory to recognize the fenced code block.
+func (p *Parser) fencedCodeBlock(data []byte, doRender bool) int {
+ var syntax string
+ beg, marker := isFenceLine(data, &syntax, "")
+ if beg == 0 || beg >= len(data) {
+ return 0
+ }
+
+ var work bytes.Buffer
+ work.WriteString(syntax)
+ work.WriteByte('\n')
+
+ for {
+ // safe to assume beg < len(data)
+
+ // check for the end of the code block
+ fenceEnd, _ := isFenceLine(data[beg:], nil, marker)
+ if fenceEnd != 0 {
+ beg += fenceEnd
+ break
+ }
+
+ // copy the current line
+ end := skipUntilChar(data, beg, '\n') + 1
+
+ // did we reach the end of the buffer without a closing marker?
+ if end >= len(data) {
+ return 0
+ }
+
+ // verbatim copy to the working buffer
+ if doRender {
+ work.Write(data[beg:end])
+ }
+ beg = end
+ }
+
+ if doRender {
+ codeBlock := &ast.CodeBlock{
+ IsFenced: true,
+ }
+ codeBlock.Content = work.Bytes() // TODO: get rid of temp buffer
+
+ if p.extensions&Mmark == 0 {
+ p.addBlock(codeBlock)
+ finalizeCodeBlock(codeBlock)
+ return beg
+ }
+
+ // Check for caption and if found make it a figure.
+ if captionContent, id, consumed := p.caption(data[beg:], []byte("Figure: ")); consumed > 0 {
+ figure := &ast.CaptionFigure{}
+ caption := &ast.Caption{}
+ figure.HeadingID = id
+ p.Inline(caption, captionContent)
+
+ p.addBlock(figure)
+ codeBlock.AsLeaf().Attribute = figure.AsContainer().Attribute
+ p.addChild(codeBlock)
+ finalizeCodeBlock(codeBlock)
+ p.addChild(caption)
+ p.finalize(figure)
+
+ beg += consumed
+
+ return beg
+ }
+
+ // Still here, normal block
+ p.addBlock(codeBlock)
+ finalizeCodeBlock(codeBlock)
+ }
+
+ return beg
+}
+
+func unescapeChar(str []byte) []byte {
+ if str[0] == '\\' {
+ return []byte{str[1]}
+ }
+ return []byte(html.UnescapeString(string(str)))
+}
+
+func unescapeString(str []byte) []byte {
+ if reBackslashOrAmp.Match(str) {
+ return reEntityOrEscapedChar.ReplaceAllFunc(str, unescapeChar)
+ }
+ return str
+}
+
+func finalizeCodeBlock(code *ast.CodeBlock) {
+ c := code.Content
+ if code.IsFenced {
+ newlinePos := bytes.IndexByte(c, '\n')
+ firstLine := c[:newlinePos]
+ rest := c[newlinePos+1:]
+ code.Info = unescapeString(bytes.Trim(firstLine, "\n"))
+ code.Literal = rest
+ } else {
+ code.Literal = c
+ }
+ code.Content = nil
+}
+
+func (p *Parser) table(data []byte) int {
+ i, columns, table := p.tableHeader(data)
+ if i == 0 {
+ return 0
+ }
+
+ p.addBlock(&ast.TableBody{})
+
+ for i < len(data) {
+ pipes, rowStart := 0, i
+ for ; i < len(data) && data[i] != '\n'; i++ {
+ if data[i] == '|' {
+ pipes++
+ }
+ }
+
+ if pipes == 0 {
+ i = rowStart
+ break
+ }
+
+ // include the newline in data sent to tableRow
+ i = skipCharN(data, i, '\n', 1)
+
+ if p.tableFooter(data[rowStart:i]) {
+ continue
+ }
+
+ p.tableRow(data[rowStart:i], columns, false)
+ }
+ if captionContent, id, consumed := p.caption(data[i:], []byte("Table: ")); consumed > 0 {
+ caption := &ast.Caption{}
+ p.Inline(caption, captionContent)
+
+ // Some switcheroo to re-insert the parsed table as a child of the captionfigure.
+ figure := &ast.CaptionFigure{}
+ figure.HeadingID = id
+ table2 := &ast.Table{}
+ // Retain any block level attributes.
+ table2.AsContainer().Attribute = table.AsContainer().Attribute
+ children := table.GetChildren()
+ ast.RemoveFromTree(table)
+
+ table2.SetChildren(children)
+ ast.AppendChild(figure, table2)
+ ast.AppendChild(figure, caption)
+
+ p.addChild(figure)
+ p.finalize(figure)
+
+ i += consumed
+ }
+
+ return i
+}
+
+// check if the specified position is preceded by an odd number of backslashes
+func isBackslashEscaped(data []byte, i int) bool {
+ backslashes := 0
+ for i-backslashes-1 >= 0 && data[i-backslashes-1] == '\\' {
+ backslashes++
+ }
+ return backslashes&1 == 1
+}
+
+// tableHeaders parses the header. If recognized it will also add a table.
+func (p *Parser) tableHeader(data []byte) (size int, columns []ast.CellAlignFlags, table ast.Node) {
+ i := 0
+ colCount := 1
+ for i = 0; i < len(data) && data[i] != '\n'; i++ {
+ if data[i] == '|' && !isBackslashEscaped(data, i) {
+ colCount++
+ }
+ }
+
+ // doesn't look like a table header
+ if colCount == 1 {
+ return
+ }
+
+ // include the newline in the data sent to tableRow
+ j := skipCharN(data, i, '\n', 1)
+ header := data[:j]
+
+ // column count ignores pipes at beginning or end of line
+ if data[0] == '|' {
+ colCount--
+ }
+ if i > 2 && data[i-1] == '|' && !isBackslashEscaped(data, i-1) {
+ colCount--
+ }
+
+ columns = make([]ast.CellAlignFlags, colCount)
+
+ // move on to the header underline
+ i++
+ if i >= len(data) {
+ return
+ }
+
+ if data[i] == '|' && !isBackslashEscaped(data, i) {
+ i++
+ }
+ i = skipChar(data, i, ' ')
+
+ // each column header is of form: / *:?-+:? *|/ with # dashes + # colons >= 3
+ // and trailing | optional on last column
+ col := 0
+ n := len(data)
+ for i < n && data[i] != '\n' {
+ dashes := 0
+
+ if data[i] == ':' {
+ i++
+ columns[col] |= ast.TableAlignmentLeft
+ dashes++
+ }
+ for i < n && data[i] == '-' {
+ i++
+ dashes++
+ }
+ if i < n && data[i] == ':' {
+ i++
+ columns[col] |= ast.TableAlignmentRight
+ dashes++
+ }
+ for i < n && data[i] == ' ' {
+ i++
+ }
+ if i == n {
+ return
+ }
+ // end of column test is messy
+ switch {
+ case dashes < 3:
+ // not a valid column
+ return
+
+ case data[i] == '|' && !isBackslashEscaped(data, i):
+ // marker found, now skip past trailing whitespace
+ col++
+ i++
+ for i < n && data[i] == ' ' {
+ i++
+ }
+
+ // trailing junk found after last column
+ if col >= colCount && i < len(data) && data[i] != '\n' {
+ return
+ }
+
+ case (data[i] != '|' || isBackslashEscaped(data, i)) && col+1 < colCount:
+ // something else found where marker was required
+ return
+
+ case data[i] == '\n':
+ // marker is optional for the last column
+ col++
+
+ default:
+ // trailing junk found after last column
+ return
+ }
+ }
+ if col != colCount {
+ return
+ }
+
+ table = &ast.Table{}
+ p.addBlock(table)
+ p.addBlock(&ast.TableHeader{})
+ p.tableRow(header, columns, true)
+ size = skipCharN(data, i, '\n', 1)
+ return
+}
+
+func (p *Parser) tableRow(data []byte, columns []ast.CellAlignFlags, header bool) {
+ p.addBlock(&ast.TableRow{})
+ i, col := 0, 0
+
+ if data[i] == '|' && !isBackslashEscaped(data, i) {
+ i++
+ }
+
+ n := len(data)
+ for col = 0; col < len(columns) && i < n; col++ {
+ for i < n && data[i] == ' ' {
+ i++
+ }
+
+ cellStart := i
+
+ for i < n && (data[i] != '|' || isBackslashEscaped(data, i)) && data[i] != '\n' {
+ i++
+ }
+
+ cellEnd := i
+
+ // skip the end-of-cell marker, possibly taking us past end of buffer
+ i++
+
+ for cellEnd > cellStart && cellEnd-1 < n && data[cellEnd-1] == ' ' {
+ cellEnd--
+ }
+
+ block := &ast.TableCell{
+ IsHeader: header,
+ Align: columns[col],
+ }
+ block.Content = data[cellStart:cellEnd]
+ p.addBlock(block)
+ }
+
+ // pad it out with empty columns to get the right number
+ for ; col < len(columns); col++ {
+ block := &ast.TableCell{
+ IsHeader: header,
+ Align: columns[col],
+ }
+ p.addBlock(block)
+ }
+
+ // silently ignore rows with too many cells
+}
+
+// tableFooter parses the (optional) table footer.
+func (p *Parser) tableFooter(data []byte) bool {
+ colCount := 1
+ for i := 0; i < len(data) && data[i] != '\n'; i++ {
+ if data[i] == '|' && !isBackslashEscaped(data, i) {
+ colCount++
+ continue
+ }
+ // remaining data must be the = character
+ if data[i] != '=' {
+ return false
+ }
+ }
+
+ // doesn't look like a table footer
+ if colCount == 1 {
+ return false
+ }
+
+ p.addBlock(&ast.TableFooter{})
+
+ return true
+}
+
+// returns blockquote prefix length
+func (p *Parser) quotePrefix(data []byte) int {
+ i := 0
+ n := len(data)
+ for i < 3 && i < n && data[i] == ' ' {
+ i++
+ }
+ if i < n && data[i] == '>' {
+ if i+1 < n && data[i+1] == ' ' {
+ return i + 2
+ }
+ return i + 1
+ }
+ return 0
+}
+
+// blockquote ends with at least one blank line
+// followed by something without a blockquote prefix
+func (p *Parser) terminateBlockquote(data []byte, beg, end int) bool {
+ if p.isEmpty(data[beg:]) <= 0 {
+ return false
+ }
+ if end >= len(data) {
+ return true
+ }
+ return p.quotePrefix(data[end:]) == 0 && p.isEmpty(data[end:]) == 0
+}
+
+// parse a blockquote fragment
+func (p *Parser) quote(data []byte) int {
+ var raw bytes.Buffer
+ beg, end := 0, 0
+ for beg < len(data) {
+ end = beg
+ // Step over whole lines, collecting them. While doing that, check for
+ // fenced code and if one's found, incorporate it altogether,
+ // irregardless of any contents inside it
+ for end < len(data) && data[end] != '\n' {
+ if p.extensions&FencedCode != 0 {
+ if i := p.fencedCodeBlock(data[end:], false); i > 0 {
+ // -1 to compensate for the extra end++ after the loop:
+ end += i - 1
+ break
+ }
+ }
+ end++
+ }
+ end = skipCharN(data, end, '\n', 1)
+ if pre := p.quotePrefix(data[beg:]); pre > 0 {
+ // skip the prefix
+ beg += pre
+ } else if p.terminateBlockquote(data, beg, end) {
+ break
+ }
+ // this line is part of the blockquote
+ raw.Write(data[beg:end])
+ beg = end
+ }
+
+ if p.extensions&Mmark == 0 {
+ block := p.addBlock(&ast.BlockQuote{})
+ p.block(raw.Bytes())
+ p.finalize(block)
+ return end
+ }
+
+ if captionContent, id, consumed := p.caption(data[end:], []byte("Quote: ")); consumed > 0 {
+ figure := &ast.CaptionFigure{}
+ caption := &ast.Caption{}
+ figure.HeadingID = id
+ p.Inline(caption, captionContent)
+
+ p.addBlock(figure) // this discard any attributes
+ block := &ast.BlockQuote{}
+ block.AsContainer().Attribute = figure.AsContainer().Attribute
+ p.addChild(block)
+ p.block(raw.Bytes())
+ p.finalize(block)
+
+ p.addChild(caption)
+ p.finalize(figure)
+
+ end += consumed
+
+ return end
+ }
+
+ block := p.addBlock(&ast.BlockQuote{})
+ p.block(raw.Bytes())
+ p.finalize(block)
+
+ return end
+}
+
+// returns prefix length for block code
+func (p *Parser) codePrefix(data []byte) int {
+ n := len(data)
+ if n >= 1 && data[0] == '\t' {
+ return 1
+ }
+ if n >= 4 && data[3] == ' ' && data[2] == ' ' && data[1] == ' ' && data[0] == ' ' {
+ return 4
+ }
+ return 0
+}
+
+func (p *Parser) code(data []byte) int {
+ var work bytes.Buffer
+
+ i := 0
+ for i < len(data) {
+ beg := i
+
+ i = skipUntilChar(data, i, '\n')
+ i = skipCharN(data, i, '\n', 1)
+
+ blankline := p.isEmpty(data[beg:i]) > 0
+ if pre := p.codePrefix(data[beg:i]); pre > 0 {
+ beg += pre
+ } else if !blankline {
+ // non-empty, non-prefixed line breaks the pre
+ i = beg
+ break
+ }
+
+ // verbatim copy to the working buffer
+ if blankline {
+ work.WriteByte('\n')
+ } else {
+ work.Write(data[beg:i])
+ }
+ }
+
+ // trim all the \n off the end of work
+ workbytes := work.Bytes()
+
+ eol := backChar(workbytes, len(workbytes), '\n')
+
+ if eol != len(workbytes) {
+ work.Truncate(eol)
+ }
+
+ work.WriteByte('\n')
+
+ codeBlock := &ast.CodeBlock{
+ IsFenced: false,
+ }
+ // TODO: get rid of temp buffer
+ codeBlock.Content = work.Bytes()
+ p.addBlock(codeBlock)
+ finalizeCodeBlock(codeBlock)
+
+ return i
+}
+
+// returns unordered list item prefix
+func (p *Parser) uliPrefix(data []byte) int {
+ // start with up to 3 spaces
+ i := skipCharN(data, 0, ' ', 3)
+
+ if i >= len(data)-1 {
+ return 0
+ }
+ // need one of {'*', '+', '-'} followed by a space or a tab
+ if (data[i] != '*' && data[i] != '+' && data[i] != '-') ||
+ (data[i+1] != ' ' && data[i+1] != '\t') {
+ return 0
+ }
+ return i + 2
+}
+
+// returns ordered list item prefix
+func (p *Parser) oliPrefix(data []byte) int {
+ // start with up to 3 spaces
+ i := skipCharN(data, 0, ' ', 3)
+
+ // count the digits
+ start := i
+ for i < len(data) && data[i] >= '0' && data[i] <= '9' {
+ i++
+ }
+ if start == i || i >= len(data)-1 {
+ return 0
+ }
+
+ // we need >= 1 digits followed by a dot and a space or a tab
+ if data[i] != '.' || !(data[i+1] == ' ' || data[i+1] == '\t') {
+ return 0
+ }
+ return i + 2
+}
+
+// returns definition list item prefix
+func (p *Parser) dliPrefix(data []byte) int {
+ if len(data) < 2 {
+ return 0
+ }
+ // need a ':' followed by a space or a tab
+ if data[0] != ':' || !(data[1] == ' ' || data[1] == '\t') {
+ return 0
+ }
+ i := skipChar(data, 0, ' ')
+ return i + 2
+}
+
+// parse ordered or unordered list block
+func (p *Parser) list(data []byte, flags ast.ListType, start int) int {
+ i := 0
+ flags |= ast.ListItemBeginningOfList
+ list := &ast.List{
+ ListFlags: flags,
+ Tight: true,
+ Start: start,
+ }
+ block := p.addBlock(list)
+
+ for i < len(data) {
+ skip := p.listItem(data[i:], &flags)
+ if flags&ast.ListItemContainsBlock != 0 {
+ list.Tight = false
+ }
+ i += skip
+ if skip == 0 || flags&ast.ListItemEndOfList != 0 {
+ break
+ }
+ flags &= ^ast.ListItemBeginningOfList
+ }
+
+ above := block.GetParent()
+ finalizeList(list)
+ p.tip = above
+ return i
+}
+
+// Returns true if the list item is not the same type as its parent list
+func (p *Parser) listTypeChanged(data []byte, flags *ast.ListType) bool {
+ if p.dliPrefix(data) > 0 && *flags&ast.ListTypeDefinition == 0 {
+ return true
+ } else if p.oliPrefix(data) > 0 && *flags&ast.ListTypeOrdered == 0 {
+ return true
+ } else if p.uliPrefix(data) > 0 && (*flags&ast.ListTypeOrdered != 0 || *flags&ast.ListTypeDefinition != 0) {
+ return true
+ }
+ return false
+}
+
+// Returns true if block ends with a blank line, descending if needed
+// into lists and sublists.
+func endsWithBlankLine(block ast.Node) bool {
+ // TODO: figure this out. Always false now.
+ for block != nil {
+ //if block.lastLineBlank {
+ //return true
+ //}
+ switch block.(type) {
+ case *ast.List, *ast.ListItem:
+ block = ast.GetLastChild(block)
+ default:
+ return false
+ }
+ }
+ return false
+}
+
+func finalizeList(list *ast.List) {
+ items := list.Parent.GetChildren()
+ lastItemIdx := len(items) - 1
+ for i, item := range items {
+ isLastItem := i == lastItemIdx
+ // check for non-final list item ending with blank line:
+ if !isLastItem && endsWithBlankLine(item) {
+ list.Tight = false
+ break
+ }
+ // recurse into children of list item, to see if there are spaces
+ // between any of them:
+ subItems := item.GetParent().GetChildren()
+ lastSubItemIdx := len(subItems) - 1
+ for j, subItem := range subItems {
+ isLastSubItem := j == lastSubItemIdx
+ if (!isLastItem || !isLastSubItem) && endsWithBlankLine(subItem) {
+ list.Tight = false
+ break
+ }
+ }
+ }
+}
+
+// Parse a single list item.
+// Assumes initial prefix is already removed if this is a sublist.
+func (p *Parser) listItem(data []byte, flags *ast.ListType) int {
+ // keep track of the indentation of the first line
+ itemIndent := 0
+ if data[0] == '\t' {
+ itemIndent += 4
+ } else {
+ for itemIndent < 3 && data[itemIndent] == ' ' {
+ itemIndent++
+ }
+ }
+
+ var bulletChar byte = '*'
+ i := p.uliPrefix(data)
+ if i == 0 {
+ i = p.oliPrefix(data)
+ } else {
+ bulletChar = data[i-2]
+ }
+ if i == 0 {
+ i = p.dliPrefix(data)
+ // reset definition term flag
+ if i > 0 {
+ *flags &= ^ast.ListTypeTerm
+ }
+ }
+ if i == 0 {
+ // if in definition list, set term flag and continue
+ if *flags&ast.ListTypeDefinition != 0 {
+ *flags |= ast.ListTypeTerm
+ } else {
+ return 0
+ }
+ }
+
+ // skip leading whitespace on first line
+ i = skipChar(data, i, ' ')
+
+ // find the end of the line
+ line := i
+ for i > 0 && i < len(data) && data[i-1] != '\n' {
+ i++
+ }
+
+ // get working buffer
+ var raw bytes.Buffer
+
+ // put the first line into the working buffer
+ raw.Write(data[line:i])
+ line = i
+
+ // process the following lines
+ containsBlankLine := false
+ sublist := 0
+
+gatherlines:
+ for line < len(data) {
+ i++
+
+ // find the end of this line
+ for i < len(data) && data[i-1] != '\n' {
+ i++
+ }
+
+ // if it is an empty line, guess that it is part of this item
+ // and move on to the next line
+ if p.isEmpty(data[line:i]) > 0 {
+ containsBlankLine = true
+ line = i
+ continue
+ }
+
+ // calculate the indentation
+ indent := 0
+ indentIndex := 0
+ if data[line] == '\t' {
+ indentIndex++
+ indent += 4
+ } else {
+ for indent < 4 && line+indent < i && data[line+indent] == ' ' {
+ indent++
+ indentIndex++
+ }
+ }
+
+ chunk := data[line+indentIndex : i]
+
+ // evaluate how this line fits in
+ switch {
+ // is this a nested list item?
+ case (p.uliPrefix(chunk) > 0 && !p.isHRule(chunk)) || p.oliPrefix(chunk) > 0 || p.dliPrefix(chunk) > 0:
+
+ // to be a nested list, it must be indented more
+ // if not, it is either a different kind of list
+ // or the next item in the same list
+ if indent <= itemIndent {
+ if p.listTypeChanged(chunk, flags) {
+ *flags |= ast.ListItemEndOfList
+ } else if containsBlankLine {
+ *flags |= ast.ListItemContainsBlock
+ }
+
+ break gatherlines
+ }
+
+ if containsBlankLine {
+ *flags |= ast.ListItemContainsBlock
+ }
+
+ // is this the first item in the nested list?
+ if sublist == 0 {
+ sublist = raw.Len()
+ // in the case of dliPrefix we are too late and need to search back for the definition item, which
+ // should be on the previous line, we then adjust sublist to start there.
+ if p.dliPrefix(chunk) > 0 {
+ sublist = backUntilChar(raw.Bytes(), raw.Len()-1, '\n')
+ }
+ }
+
+ // is this a nested prefix heading?
+ case p.isPrefixHeading(chunk), p.isPrefixSpecialHeading(chunk):
+ // if the heading is not indented, it is not nested in the list
+ // and thus ends the list
+ if containsBlankLine && indent < 4 {
+ *flags |= ast.ListItemEndOfList
+ break gatherlines
+ }
+ *flags |= ast.ListItemContainsBlock
+
+ // anything following an empty line is only part
+ // of this item if it is indented 4 spaces
+ // (regardless of the indentation of the beginning of the item)
+ case containsBlankLine && indent < 4:
+ if *flags&ast.ListTypeDefinition != 0 && i < len(data)-1 {
+ // is the next item still a part of this list?
+ next := i
+ for next < len(data) && data[next] != '\n' {
+ next++
+ }
+ for next < len(data)-1 && data[next] == '\n' {
+ next++
+ }
+ if i < len(data)-1 && data[i] != ':' && next < len(data)-1 && data[next] != ':' {
+ *flags |= ast.ListItemEndOfList
+ }
+ } else {
+ *flags |= ast.ListItemEndOfList
+ }
+ break gatherlines
+
+ // a blank line means this should be parsed as a block
+ case containsBlankLine:
+ raw.WriteByte('\n')
+ *flags |= ast.ListItemContainsBlock
+ }
+
+ // if this line was preceded by one or more blanks,
+ // re-introduce the blank into the buffer
+ if containsBlankLine {
+ containsBlankLine = false
+ raw.WriteByte('\n')
+ }
+
+ // add the line into the working buffer without prefix
+ raw.Write(data[line+indentIndex : i])
+
+ line = i
+ }
+
+ rawBytes := raw.Bytes()
+
+ listItem := &ast.ListItem{
+ ListFlags: *flags,
+ Tight: false,
+ BulletChar: bulletChar,
+ Delimiter: '.', // Only '.' is possible in Markdown, but ')' will also be possible in CommonMark
+ }
+ p.addBlock(listItem)
+
+ // render the contents of the list item
+ if *flags&ast.ListItemContainsBlock != 0 && *flags&ast.ListTypeTerm == 0 {
+ // intermediate render of block item, except for definition term
+ if sublist > 0 {
+ p.block(rawBytes[:sublist])
+ p.block(rawBytes[sublist:])
+ } else {
+ p.block(rawBytes)
+ }
+ } else {
+ // intermediate render of inline item
+ para := &ast.Paragraph{}
+ if sublist > 0 {
+ para.Content = rawBytes[:sublist]
+ } else {
+ para.Content = rawBytes
+ }
+ p.addChild(para)
+ if sublist > 0 {
+ p.block(rawBytes[sublist:])
+ }
+ }
+ return line
+}
+
+// render a single paragraph that has already been parsed out
+func (p *Parser) renderParagraph(data []byte) {
+ if len(data) == 0 {
+ return
+ }
+
+ // trim leading spaces
+ beg := skipChar(data, 0, ' ')
+
+ end := len(data)
+ // trim trailing newline
+ if data[len(data)-1] == '\n' {
+ end--
+ }
+
+ // trim trailing spaces
+ for end > beg && data[end-1] == ' ' {
+ end--
+ }
+ para := &ast.Paragraph{}
+ para.Content = data[beg:end]
+ p.addBlock(para)
+}
+
+// blockMath handle block surround with $$
+func (p *Parser) blockMath(data []byte) int {
+ if len(data) <= 4 || data[0] != '$' || data[1] != '$' || data[2] == '$' {
+ return 0
+ }
+
+ // find next $$
+ var end int
+ for end = 2; end+1 < len(data) && (data[end] != '$' || data[end+1] != '$'); end++ {
+ }
+
+ // $$ not match
+ if end+1 == len(data) {
+ return 0
+ }
+
+ // render the display math
+ mathBlock := &ast.MathBlock{}
+ mathBlock.Literal = data[2:end]
+ p.addBlock(mathBlock)
+
+ return end + 2
+}
+
+func (p *Parser) paragraph(data []byte) int {
+ // prev: index of 1st char of previous line
+ // line: index of 1st char of current line
+ // i: index of cursor/end of current line
+ var prev, line, i int
+ tabSize := tabSizeDefault
+ if p.extensions&TabSizeEight != 0 {
+ tabSize = tabSizeDouble
+ }
+ // keep going until we find something to mark the end of the paragraph
+ for i < len(data) {
+ // mark the beginning of the current line
+ prev = line
+ current := data[i:]
+ line = i
+
+ // did we find a reference or a footnote? If so, end a paragraph
+ // preceding it and report that we have consumed up to the end of that
+ // reference:
+ if refEnd := isReference(p, current, tabSize); refEnd > 0 {
+ p.renderParagraph(data[:i])
+ return i + refEnd
+ }
+
+ // did we find a blank line marking the end of the paragraph?
+ if n := p.isEmpty(current); n > 0 {
+ // did this blank line followed by a definition list item?
+ if p.extensions&DefinitionLists != 0 {
+ if i < len(data)-1 && data[i+1] == ':' {
+ listLen := p.list(data[prev:], ast.ListTypeDefinition, 0)
+ return prev + listLen
+ }
+ }
+
+ p.renderParagraph(data[:i])
+ return i + n
+ }
+
+ // an underline under some text marks a heading, so our paragraph ended on prev line
+ if i > 0 {
+ if level := p.isUnderlinedHeading(current); level > 0 {
+ // render the paragraph
+ p.renderParagraph(data[:prev])
+
+ // ignore leading and trailing whitespace
+ eol := i - 1
+ for prev < eol && data[prev] == ' ' {
+ prev++
+ }
+ for eol > prev && data[eol-1] == ' ' {
+ eol--
+ }
+
+ id := ""
+ if p.extensions&AutoHeadingIDs != 0 {
+ id = sanitizeAnchorName(string(data[prev:eol]))
+ }
+
+ block := &ast.Heading{
+ Level: level,
+ HeadingID: id,
+ }
+ block.Content = data[prev:eol]
+ p.addBlock(block)
+
+ // find the end of the underline
+ return skipUntilChar(data, i, '\n')
+ }
+ }
+
+ // if the next line starts a block of HTML, then the paragraph ends here
+ if p.extensions&LaxHTMLBlocks != 0 {
+ if data[i] == '<' && p.html(current, false) > 0 {
+ // rewind to before the HTML block
+ p.renderParagraph(data[:i])
+ return i
+ }
+ }
+
+ // if there's a prefixed heading or a horizontal rule after this, paragraph is over
+ if p.isPrefixHeading(current) || p.isPrefixSpecialHeading(current) || p.isHRule(current) {
+ p.renderParagraph(data[:i])
+ return i
+ }
+
+ // if there's a fenced code block, paragraph is over
+ if p.extensions&FencedCode != 0 {
+ if p.fencedCodeBlock(current, false) > 0 {
+ p.renderParagraph(data[:i])
+ return i
+ }
+ }
+
+ // if there's a figure block, paragraph is over
+ if p.extensions&Mmark != 0 {
+ if p.figureBlock(current, false) > 0 {
+ p.renderParagraph(data[:i])
+ return i
+ }
+ }
+
+ // if there's a definition list item, prev line is a definition term
+ if p.extensions&DefinitionLists != 0 {
+ if p.dliPrefix(current) != 0 {
+ ret := p.list(data[prev:], ast.ListTypeDefinition, 0)
+ return ret + prev
+ }
+ }
+
+ // if there's a list after this, paragraph is over
+ if p.extensions&NoEmptyLineBeforeBlock != 0 {
+ if p.uliPrefix(current) != 0 ||
+ p.oliPrefix(current) != 0 ||
+ p.quotePrefix(current) != 0 ||
+ p.codePrefix(current) != 0 {
+ p.renderParagraph(data[:i])
+ return i
+ }
+ }
+
+ // otherwise, scan to the beginning of the next line
+ nl := bytes.IndexByte(data[i:], '\n')
+ if nl >= 0 {
+ i += nl + 1
+ } else {
+ i += len(data[i:])
+ }
+ }
+
+ p.renderParagraph(data[:i])
+ return i
+}
+
+// skipChar advances i as long as data[i] == c
+func skipChar(data []byte, i int, c byte) int {
+ n := len(data)
+ for i < n && data[i] == c {
+ i++
+ }
+ return i
+}
+
+// like skipChar but only skips up to max characters
+func skipCharN(data []byte, i int, c byte, max int) int {
+ n := len(data)
+ for i < n && max > 0 && data[i] == c {
+ i++
+ max--
+ }
+ return i
+}
+
+// skipUntilChar advances i as long as data[i] != c
+func skipUntilChar(data []byte, i int, c byte) int {
+ n := len(data)
+ for i < n && data[i] != c {
+ i++
+ }
+ return i
+}
+
+func skipAlnum(data []byte, i int) int {
+ n := len(data)
+ for i < n && isAlnum(data[i]) {
+ i++
+ }
+ return i
+}
+
+func skipSpace(data []byte, i int) int {
+ n := len(data)
+ for i < n && isSpace(data[i]) {
+ i++
+ }
+ return i
+}
+
+func backChar(data []byte, i int, c byte) int {
+ for i > 0 && data[i-1] == c {
+ i--
+ }
+ return i
+}
+
+func backUntilChar(data []byte, i int, c byte) int {
+ for i > 0 && data[i-1] != c {
+ i--
+ }
+ return i
+}
diff --git a/vendor/github.com/gomarkdown/markdown/parser/callout.go b/vendor/github.com/gomarkdown/markdown/parser/callout.go
new file mode 100644
index 00000000..15858aa9
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/parser/callout.go
@@ -0,0 +1,29 @@
+package parser
+
+import (
+ "bytes"
+ "strconv"
+)
+
+// IsCallout detects a callout in the following format: <<N>> Where N is a integer > 0.
+func IsCallout(data []byte) (id []byte, consumed int) {
+ if !bytes.HasPrefix(data, []byte("<<")) {
+ return nil, 0
+ }
+ start := 2
+ end := bytes.Index(data[start:], []byte(">>"))
+ if end < 0 {
+ return nil, 0
+ }
+
+ b := data[start : start+end]
+ b = bytes.TrimSpace(b)
+ i, err := strconv.Atoi(string(b))
+ if err != nil {
+ return nil, 0
+ }
+ if i <= 0 {
+ return nil, 0
+ }
+ return b, start + end + 2 // 2 for >>
+}
diff --git a/vendor/github.com/gomarkdown/markdown/parser/caption.go b/vendor/github.com/gomarkdown/markdown/parser/caption.go
new file mode 100644
index 00000000..54d3f741
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/parser/caption.go
@@ -0,0 +1,70 @@
+package parser
+
+import (
+ "bytes"
+)
+
+// caption checks for a caption, it returns the caption data and a potential "headingID".
+func (p *Parser) caption(data, caption []byte) ([]byte, string, int) {
+ if !bytes.HasPrefix(data, caption) {
+ return nil, "", 0
+ }
+ j := len(caption)
+ data = data[j:]
+ end := p.linesUntilEmpty(data)
+
+ data = data[:end]
+
+ id, start := captionID(data)
+ if id != "" {
+ return data[:start], id, end + j
+ }
+
+ return data, "", end + j
+}
+
+// linesUntilEmpty scans lines up to the first empty line.
+func (p *Parser) linesUntilEmpty(data []byte) int {
+ line, i := 0, 0
+
+ for line < len(data) {
+ i++
+
+ // find the end of this line
+ for i < len(data) && data[i-1] != '\n' {
+ i++
+ }
+
+ if p.isEmpty(data[line:i]) == 0 {
+ line = i
+ continue
+ }
+
+ break
+ }
+ return i
+}
+
+// captionID checks if the caption *ends* in {#....}. If so the text after {# is taken to be
+// the ID/anchor of the entire figure block.
+func captionID(data []byte) (string, int) {
+ end := len(data)
+
+ j, k := 0, 0
+ // find start/end of heading id
+ for j = 0; j < end-1 && (data[j] != '{' || data[j+1] != '#'); j++ {
+ }
+ for k = j + 1; k < end && data[k] != '}'; k++ {
+ }
+ // remains must be whitespace.
+ for l := k + 1; l < end; l++ {
+ if !isSpace(data[l]) {
+ return "", 0
+ }
+ }
+
+ if j > 0 && k > 0 && j+2 < k {
+ return string(data[j+2 : k]), j
+ }
+ return "", 0
+}
diff --git a/vendor/github.com/gomarkdown/markdown/parser/citation.go b/vendor/github.com/gomarkdown/markdown/parser/citation.go
new file mode 100644
index 00000000..8ea1fbee
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/parser/citation.go
@@ -0,0 +1,86 @@
+package parser
+
+import (
+ "bytes"
+
+ "github.com/gomarkdown/markdown/ast"
+)
+
+// citation parses a citation. In its most simple form [@ref], we allow multiple
+// being separated by semicolons and a sub reference inside ala pandoc: [@ref, p. 23].
+// Each citation can have a modifier: !, ? or - wich mean:
+//
+// ! - normative
+// ? - formative
+// - - suppressed
+//
+// The suffix starts after a comma, we strip any whitespace before and after. If the output
+// allows for it, this can be rendered.
+func citation(p *Parser, data []byte, offset int) (int, ast.Node) {
+ // look for the matching closing bracket
+ i := offset + 1
+ for level := 1; level > 0 && i < len(data); i++ {
+ switch {
+ case data[i] == '\n':
+ // no newlines allowed.
+ return 0, nil
+
+ 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
+ }
+
+ node := &ast.Citation{}
+
+ citations := bytes.Split(data[1:i], []byte(";"))
+ for _, citation := range citations {
+ var suffix []byte
+ citation = bytes.TrimSpace(citation)
+ j := 0
+ if citation[j] != '@' {
+ // not a citation, drop out entirely.
+ return 0, nil
+ }
+ if c := bytes.Index(citation, []byte(",")); c > 0 {
+ part := citation[:c]
+ suff := citation[c+1:]
+ part = bytes.TrimSpace(part)
+ suff = bytes.TrimSpace(suff)
+
+ citation = part
+ suffix = suff
+ }
+
+ citeType := ast.CitationTypeInformative
+ j = 1
+ switch citation[j] {
+ case '!':
+ citeType = ast.CitationTypeNormative
+ j++
+ case '?':
+ citeType = ast.CitationTypeInformative
+ j++
+ case '-':
+ citeType = ast.CitationTypeSuppressed
+ j++
+ }
+ node.Destination = append(node.Destination, citation[j:])
+ node.Type = append(node.Type, citeType)
+ node.Suffix = append(node.Suffix, suffix)
+ }
+
+ return i + 1, node
+}
diff --git a/vendor/github.com/gomarkdown/markdown/parser/esc.go b/vendor/github.com/gomarkdown/markdown/parser/esc.go
new file mode 100644
index 00000000..0a79aa35
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/parser/esc.go
@@ -0,0 +1,20 @@
+package parser
+
+// isEscape returns true if byte i is prefixed by an odd number of backslahses.
+func isEscape(data []byte, i int) bool {
+ if i == 0 {
+ return false
+ }
+ if i == 1 {
+ return data[0] == '\\'
+ }
+ j := i - 1
+ for ; j >= 0; j-- {
+ if data[j] != '\\' {
+ break
+ }
+ }
+ j++
+ // odd number of backslahes means escape
+ return (i-j)%2 != 0
+}
diff --git a/vendor/github.com/gomarkdown/markdown/parser/figures.go b/vendor/github.com/gomarkdown/markdown/parser/figures.go
new file mode 100644
index 00000000..6615449c
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/parser/figures.go
@@ -0,0 +1,119 @@
+package parser
+
+import (
+ "bytes"
+
+ "github.com/gomarkdown/markdown/ast"
+)
+
+// sFigureLine checks if there's a figure line (e.g., !--- ) at the beginning of data,
+// and returns the end index if so, or 0 otherwise.
+func sFigureLine(data []byte, oldmarker string) (end int, marker string) {
+ i, size := 0, 0
+
+ n := len(data)
+ // skip up to three spaces
+ for i < n && i < 3 && data[i] == ' ' {
+ i++
+ }
+
+ // check for the marker characters: !
+ if i+1 >= n {
+ return 0, ""
+ }
+ if data[i] != '!' || data[i+1] != '-' {
+ return 0, ""
+ }
+ i++
+
+ c := data[i] // i.e. the -
+
+ // the whole line must be the same char or whitespace
+ for i < n && data[i] == c {
+ size++
+ i++
+ }
+
+ // the marker char must occur at least 3 times
+ if size < 3 {
+ return 0, ""
+ }
+ marker = string(data[i-size : i])
+
+ // if this is the end marker, it must match the beginning marker
+ if oldmarker != "" && marker != oldmarker {
+ return 0, ""
+ }
+
+ // there is no syntax modifier although it might be an idea to re-use this space for something?
+
+ i = skipChar(data, i, ' ')
+ if i >= n || data[i] != '\n' {
+ if i == n {
+ return i, marker
+ }
+ return 0, ""
+ }
+ return i + 1, marker // Take newline into account.
+}
+
+// figureBlock returns the end index if data contains a figure block at the beginning,
+// or 0 otherwise. It writes to out if doRender is true, otherwise it has no side effects.
+// If doRender is true, a final newline is mandatory to recognize the figure block.
+func (p *Parser) figureBlock(data []byte, doRender bool) int {
+ beg, marker := sFigureLine(data, "")
+ if beg == 0 || beg >= len(data) {
+ return 0
+ }
+
+ var raw bytes.Buffer
+
+ for {
+ // safe to assume beg < len(data)
+
+ // check for the end of the code block
+ figEnd, _ := sFigureLine(data[beg:], marker)
+ if figEnd != 0 {
+ beg += figEnd
+ break
+ }
+
+ // copy the current line
+ end := skipUntilChar(data, beg, '\n') + 1
+
+ // did we reach the end of the buffer without a closing marker?
+ if end >= len(data) {
+ return 0
+ }
+
+ // verbatim copy to the working buffer
+ if doRender {
+ raw.Write(data[beg:end])
+ }
+ beg = end
+ }
+
+ if !doRender {
+ return beg
+ }
+
+ figure := &ast.CaptionFigure{}
+ p.addBlock(figure)
+ p.block(raw.Bytes())
+
+ defer p.finalize(figure)
+
+ if captionContent, id, consumed := p.caption(data[beg:], []byte("Figure: ")); consumed > 0 {
+ caption := &ast.Caption{}
+ p.Inline(caption, captionContent)
+
+ figure.HeadingID = id
+
+ p.addChild(caption)
+
+ beg += consumed
+ }
+
+ p.finalize(figure)
+ return beg
+}
diff --git a/vendor/github.com/gomarkdown/markdown/parser/include.go b/vendor/github.com/gomarkdown/markdown/parser/include.go
new file mode 100644
index 00000000..2448a685
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/parser/include.go
@@ -0,0 +1,129 @@
+package parser
+
+import (
+ "bytes"
+ "path"
+ "path/filepath"
+)
+
+// isInclude parses {{...}}[...], that contains a path between the {{, the [...] syntax contains
+// an address to select which lines to include. It is treated as an opaque string and just given
+// to readInclude.
+func (p *Parser) isInclude(data []byte) (filename string, address []byte, consumed int) {
+ i := skipCharN(data, 0, ' ', 3) // start with up to 3 spaces
+ if len(data[i:]) < 3 {
+ return "", nil, 0
+ }
+ if data[i] != '{' || data[i+1] != '{' {
+ return "", nil, 0
+ }
+ start := i + 2
+
+ // find the end delimiter
+ i = skipUntilChar(data, i, '}')
+ if i+1 >= len(data) {
+ return "", nil, 0
+ }
+ end := i
+ i++
+ if data[i] != '}' {
+ return "", nil, 0
+ }
+ filename = string(data[start:end])
+
+ if i+1 < len(data) && data[i+1] == '[' { // potential address specification
+ start := i + 2
+
+ end = skipUntilChar(data, start, ']')
+ if end >= len(data) {
+ return "", nil, 0
+ }
+ address = data[start:end]
+ return filename, address, end + 1
+ }
+
+ return filename, address, i + 1
+}
+
+func (p *Parser) readInclude(from, file string, address []byte) []byte {
+ if p.Opts.ReadIncludeFn != nil {
+ return p.Opts.ReadIncludeFn(from, file, address)
+ }
+
+ return nil
+}
+
+// isCodeInclude parses <{{...}} which is similar to isInclude the returned bytes are, however wrapped in a code block.
+func (p *Parser) isCodeInclude(data []byte) (filename string, address []byte, consumed int) {
+ i := skipCharN(data, 0, ' ', 3) // start with up to 3 spaces
+ if len(data[i:]) < 3 {
+ return "", nil, 0
+ }
+ if data[i] != '<' {
+ return "", nil, 0
+ }
+ start := i
+
+ filename, address, consumed = p.isInclude(data[i+1:])
+ if consumed == 0 {
+ return "", nil, 0
+ }
+ return filename, address, start + consumed + 1
+}
+
+// readCodeInclude acts like include except the returned bytes are wrapped in a fenced code block.
+func (p *Parser) readCodeInclude(from, file string, address []byte) []byte {
+ data := p.readInclude(from, file, address)
+ if data == nil {
+ return nil
+ }
+ ext := path.Ext(file)
+ buf := &bytes.Buffer{}
+ buf.Write([]byte("```"))
+ if ext != "" { // starts with a dot
+ buf.WriteString(" " + ext[1:] + "\n")
+ } else {
+ buf.WriteByte('\n')
+ }
+ buf.Write(data)
+ buf.WriteString("```\n")
+ return buf.Bytes()
+}
+
+// incStack hold the current stack of chained includes. Each value is the containing
+// path of the file being parsed.
+type incStack struct {
+ stack []string
+}
+
+func newIncStack() *incStack {
+ return &incStack{stack: []string{}}
+}
+
+// Push updates i with new.
+func (i *incStack) Push(new string) {
+ if path.IsAbs(new) {
+ i.stack = append(i.stack, path.Dir(new))
+ return
+ }
+ last := ""
+ if len(i.stack) > 0 {
+ last = i.stack[len(i.stack)-1]
+ }
+ i.stack = append(i.stack, path.Dir(filepath.Join(last, new)))
+}
+
+// Pop pops the last value.
+func (i *incStack) Pop() {
+ if len(i.stack) == 0 {
+ return
+ }
+ i.stack = i.stack[:len(i.stack)-1]
+}
+
+func (i *incStack) Last() string {
+ if len(i.stack) == 0 {
+ return ""
+ }
+ return i.stack[len(i.stack)-1]
+}
diff --git a/vendor/github.com/gomarkdown/markdown/parser/inline.go b/vendor/github.com/gomarkdown/markdown/parser/inline.go
new file mode 100644
index 00000000..81766b85
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/parser/inline.go
@@ -0,0 +1,1284 @@
+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 offset + ret, 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
+
+ // look for link end: ' " )
+ findlinkend:
+ for i < len(data) {
+ switch {
+ case data[i] == '\\':
+ i += 2
+
+ case data[i] == ')' || 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
+
+ findtitleend:
+ for i < len(data) {
+ switch {
+ case data[i] == '\\':
+ i += 2
+
+ case data[i] == ')':
+ break findtitleend
+
+ default:
+ 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
+ }
+
+ // 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,
+ Title: title,
+ NoteID: noteID,
+ Footnote: footnoteNode,
+ }
+ if t == linkDeferredFootnote {
+ link.DeferredID = data[2:txtE]
+ }
+ if t == linkInlineFootnote {
+ i++
+ }
+ return i, link
+
+ default:
+ return 0, nil
+ }
+}
+
+func (p *Parser) inlineHTMLComment(data []byte) int {
+ if len(data) < 5 {
+ return 0
+ }
+ if data[0] != '<' || data[1] != '!' || data[2] != '-' || data[3] != '-' {
+ return 0
+ }
+ i := 5
+ // scan for an end-of-comment marker, across lines if necessary
+ for i < len(data) && !(data[i-2] == '-' && data[i-1] == '-' && data[i] == '>') {
+ i++
+ }
+ // no end-of-comment marker
+ if i >= len(data) {
+ return 0
+ }
+ return i + 1
+}
+
+func stripMailto(link []byte) []byte {
+ if bytes.HasPrefix(link, []byte("mailto://")) {
+ return link[9:]
+ } else if bytes.HasPrefix(link, []byte("mailto:")) {
+ return link[7:]
+ } else {
+ return link
+ }
+}
+
+// autolinkType specifies a kind of autolink that gets detected.
+type autolinkType int
+
+// These are the possible flag values for the autolink renderer.
+const (
+ notAutolink autolinkType = iota
+ normalAutolink
+ emailAutolink
+)
+
+// '<' when tags or autolinks are allowed
+func leftAngle(p *Parser, data []byte, offset int) (int, ast.Node) {
+ data = data[offset:]
+
+ if p.extensions&Mmark != 0 {
+ id, consumed := IsCallout(data)
+ if consumed > 0 {
+ node := &ast.Callout{}
+ node.ID = id
+ return consumed, node
+ }
+ }
+
+ altype, end := tagLength(data)
+ if size := p.inlineHTMLComment(data); size > 0 {
+ end = size
+ }
+ if end <= 2 {
+ return end, nil
+ }
+ if altype == notAutolink {
+ htmlTag := &ast.HTMLSpan{}
+ htmlTag.Literal = data[:end]
+ return end, htmlTag
+ }
+
+ var uLink bytes.Buffer
+ unescapeText(&uLink, data[1:end+1-2])
+ if uLink.Len() <= 0 {
+ return end, nil
+ }
+ link := uLink.Bytes()
+ node := &ast.Link{
+ Destination: link,
+ }
+ if altype == emailAutolink {
+ node.Destination = append([]byte("mailto:"), link...)
+ }
+ ast.AppendChild(node, newTextNode(stripMailto(link)))
+ return end, node
+}
+
+// '\\' backslash escape
+var escapeChars = []byte("\\`*_{}[]()#+-.!:|&<>~")
+
+func escape(p *Parser, data []byte, offset int) (int, ast.Node) {
+ data = data[offset:]
+
+ if len(data) <= 1 {
+ return 2, nil
+ }
+
+ if p.extensions&NonBlockingSpace != 0 && data[1] == ' ' {
+ return 2, &ast.NonBlockingSpace{}
+ }
+
+ if p.extensions&BackslashLineBreak != 0 && data[1] == '\n' {
+ return 2, &ast.Hardbreak{}
+ }
+
+ if bytes.IndexByte(escapeChars, data[1]) < 0 {
+ return 0, nil
+ }
+
+ return 2, newTextNode(data[1:2])
+}
+
+func unescapeText(ob *bytes.Buffer, src []byte) {
+ i := 0
+ for i < len(src) {
+ org := i
+ for i < len(src) && src[i] != '\\' {
+ i++
+ }
+
+ if i > org {
+ ob.Write(src[org:i])
+ }
+
+ if i+1 >= len(src) {
+ break
+ }
+
+ ob.WriteByte(src[i+1])
+ i += 2
+ }
+}
+
+// '&' escaped when it doesn't belong to an entity
+// valid entities are assumed to be anything matching &#?[A-Za-z0-9]+;
+func entity(p *Parser, data []byte, offset int) (int, ast.Node) {
+ data = data[offset:]
+
+ end := skipCharN(data, 1, '#', 1)
+ end = skipAlnum(data, end)
+
+ if end < len(data) && data[end] == ';' {
+ end++ // real entity
+ } else {
+ return 0, nil // lone '&'
+ }
+
+ ent := data[:end]
+ // undo & escaping or it will be converted to &amp; by another
+ // escaper in the renderer
+ if bytes.Equal(ent, []byte("&")) {
+ ent = []byte{'&'}
+ }
+
+ return end, newTextNode(ent)
+}
+
+func linkEndsWithEntity(data []byte, linkEnd int) bool {
+ entityRanges := htmlEntityRe.FindAllIndex(data[:linkEnd], -1)
+ return entityRanges != nil && entityRanges[len(entityRanges)-1][1] == linkEnd
+}
+
+// hasPrefixCaseInsensitive is a custom implementation of
+// strings.HasPrefix(strings.ToLower(s), prefix)
+// we rolled our own because ToLower pulls in a huge machinery of lowercasing
+// anything from Unicode and that's very slow. Since this func will only be
+// used on ASCII protocol prefixes, we can take shortcuts.
+func hasPrefixCaseInsensitive(s, prefix []byte) bool {
+ if len(s) < len(prefix) {
+ return false
+ }
+ delta := byte('a' - 'A')
+ for i, b := range prefix {
+ if b != s[i] && b != s[i]+delta {
+ return false
+ }
+ }
+ return true
+}
+
+var protocolPrefixes = [][]byte{
+ []byte("http://"),
+ []byte("https://"),
+ []byte("ftp://"),
+ []byte("file://"),
+ []byte("mailto:"),
+}
+
+const shortestPrefix = 6 // len("ftp://"), the shortest of the above
+
+func maybeAutoLink(p *Parser, data []byte, offset int) (int, ast.Node) {
+ // quick check to rule out most false hits
+ if p.insideLink || len(data) < offset+shortestPrefix {
+ return 0, nil
+ }
+ for _, prefix := range protocolPrefixes {
+ endOfHead := offset + 8 // 8 is the len() of the longest prefix
+ if endOfHead > len(data) {
+ endOfHead = len(data)
+ }
+ if hasPrefixCaseInsensitive(data[offset:endOfHead], prefix) {
+ return autoLink(p, data, offset)
+ }
+ }
+ return 0, nil
+}
+
+func autoLink(p *Parser, data []byte, offset int) (int, ast.Node) {
+ // Now a more expensive check to see if we're not inside an anchor element
+ anchorStart := offset
+ offsetFromAnchor := 0
+ for anchorStart > 0 && data[anchorStart] != '<' {
+ anchorStart--
+ offsetFromAnchor++
+ }
+
+ anchorStr := anchorRe.Find(data[anchorStart:])
+ if anchorStr != nil {
+ anchorClose := &ast.HTMLSpan{}
+ anchorClose.Literal = anchorStr[offsetFromAnchor:]
+ return len(anchorStr) - offsetFromAnchor, anchorClose
+ }
+
+ // scan backward for a word boundary
+ rewind := 0
+ for offset-rewind > 0 && rewind <= 7 && isLetter(data[offset-rewind-1]) {
+ rewind++
+ }
+ if rewind > 6 { // longest supported protocol is "mailto" which has 6 letters
+ return 0, nil
+ }
+
+ origData := data
+ data = data[offset-rewind:]
+
+ if !isSafeLink(data) {
+ return 0, nil
+ }
+
+ linkEnd := 0
+ for linkEnd < len(data) && !isEndOfLink(data[linkEnd]) {
+ linkEnd++
+ }
+
+ // Skip punctuation at the end of the link
+ if (data[linkEnd-1] == '.' || data[linkEnd-1] == ',') && data[linkEnd-2] != '\\' {
+ linkEnd--
+ }
+
+ // But don't skip semicolon if it's a part of escaped entity:
+ if data[linkEnd-1] == ';' && data[linkEnd-2] != '\\' && !linkEndsWithEntity(data, linkEnd) {
+ linkEnd--
+ }
+
+ // See if the link finishes with a punctuation sign that can be closed.
+ var copen byte
+ switch data[linkEnd-1] {
+ case '"':
+ copen = '"'
+ case '\'':
+ copen = '\''
+ case ')':
+ copen = '('
+ case ']':
+ copen = '['
+ case '}':
+ copen = '{'
+ default:
+ copen = 0
+ }
+
+ if copen != 0 {
+ bufEnd := offset - rewind + linkEnd - 2
+
+ openDelim := 1
+
+ /* Try to close the final punctuation sign in this same line;
+ * if we managed to close it outside of the URL, that means that it's
+ * not part of the URL. If it closes inside the URL, that means it
+ * is part of the URL.
+ *
+ * Examples:
+ *
+ * foo http://www.pokemon.com/Pikachu_(Electric) bar
+ * => http://www.pokemon.com/Pikachu_(Electric)
+ *
+ * foo (http://www.pokemon.com/Pikachu_(Electric)) bar
+ * => http://www.pokemon.com/Pikachu_(Electric)
+ *
+ * foo http://www.pokemon.com/Pikachu_(Electric)) bar
+ * => http://www.pokemon.com/Pikachu_(Electric))
+ *
+ * (foo http://www.pokemon.com/Pikachu_(Electric)) bar
+ * => foo http://www.pokemon.com/Pikachu_(Electric)
+ */
+
+ for bufEnd >= 0 && origData[bufEnd] != '\n' && openDelim != 0 {
+ if origData[bufEnd] == data[linkEnd-1] {
+ openDelim++
+ }
+
+ if origData[bufEnd] == copen {
+ openDelim--
+ }
+
+ bufEnd--
+ }
+
+ if openDelim == 0 {
+ linkEnd--
+ }
+ }
+
+ var uLink bytes.Buffer
+ unescapeText(&uLink, data[:linkEnd])
+
+ if uLink.Len() > 0 {
+ node := &ast.Link{
+ Destination: uLink.Bytes(),
+ }
+ ast.AppendChild(node, newTextNode(uLink.Bytes()))
+ return linkEnd, node
+ }
+
+ return linkEnd, nil
+}
+
+func isEndOfLink(char byte) bool {
+ return isSpace(char) || char == '<'
+}
+
+var validUris = [][]byte{[]byte("http://"), []byte("https://"), []byte("ftp://"), []byte("mailto://")}
+var validPaths = [][]byte{[]byte("/"), []byte("./"), []byte("../")}
+
+func isSafeLink(link []byte) bool {
+ nLink := len(link)
+ for _, path := range validPaths {
+ nPath := len(path)
+ linkPrefix := link[:nPath]
+ if nLink >= nPath && bytes.Equal(linkPrefix, path) {
+ if nLink == nPath {
+ return true
+ } else if isAlnum(link[nPath]) {
+ return true
+ }
+ }
+ }
+
+ for _, prefix := range validUris {
+ // TODO: handle unicode here
+ // case-insensitive prefix test
+ nPrefix := len(prefix)
+ if nLink > nPrefix {
+ linkPrefix := bytes.ToLower(link[:nPrefix])
+ if bytes.Equal(linkPrefix, prefix) && isAlnum(link[nPrefix]) {
+ return true
+ }
+ }
+ }
+
+ return false
+}
+
+// return the length of the given tag, or 0 is it's not valid
+func tagLength(data []byte) (autolink autolinkType, end int) {
+ var i, j int
+
+ // a valid tag can't be shorter than 3 chars
+ if len(data) < 3 {
+ return notAutolink, 0
+ }
+
+ // begins with a '<' optionally followed by '/', followed by letter or number
+ if data[0] != '<' {
+ return notAutolink, 0
+ }
+ if data[1] == '/' {
+ i = 2
+ } else {
+ i = 1
+ }
+
+ if !isAlnum(data[i]) {
+ return notAutolink, 0
+ }
+
+ // scheme test
+ autolink = notAutolink
+
+ // try to find the beginning of an URI
+ for i < len(data) && (isAlnum(data[i]) || data[i] == '.' || data[i] == '+' || data[i] == '-') {
+ i++
+ }
+
+ if i > 1 && i < len(data) && data[i] == '@' {
+ if j = isMailtoAutoLink(data[i:]); j != 0 {
+ return emailAutolink, i + j
+ }
+ }
+
+ if i > 2 && i < len(data) && data[i] == ':' {
+ autolink = normalAutolink
+ i++
+ }
+
+ // complete autolink test: no whitespace or ' or "
+ switch {
+ case i >= len(data):
+ autolink = notAutolink
+ case autolink != notAutolink:
+ j = i
+
+ for i < len(data) {
+ if data[i] == '\\' {
+ i += 2
+ } else if data[i] == '>' || data[i] == '\'' || data[i] == '"' || isSpace(data[i]) {
+ break
+ } else {
+ i++
+ }
+
+ }
+
+ if i >= len(data) {
+ return autolink, 0
+ }
+ if i > j && data[i] == '>' {
+ return autolink, i + 1
+ }
+
+ // one of the forbidden chars has been found
+ autolink = notAutolink
+ }
+ i += bytes.IndexByte(data[i:], '>')
+ if i < 0 {
+ return autolink, 0
+ }
+ return autolink, i + 1
+}
+
+// look for the address part of a mail autolink and '>'
+// this is less strict than the original markdown e-mail address matching
+func isMailtoAutoLink(data []byte) int {
+ nb := 0
+
+ // address is assumed to be: [-@._a-zA-Z0-9]+ with exactly one '@'
+ for i, c := range data {
+ if isAlnum(c) {
+ continue
+ }
+
+ switch c {
+ case '@':
+ nb++
+
+ case '-', '.', '_':
+ break
+
+ case '>':
+ if nb == 1 {
+ return i + 1
+ }
+ return 0
+ default:
+ return 0
+ }
+ }
+
+ return 0
+}
+
+// look for the next emph char, skipping other constructs
+func helperFindEmphChar(data []byte, c byte) int {
+ i := 0
+
+ for i < len(data) {
+ for i < len(data) && data[i] != c && data[i] != '`' && data[i] != '[' {
+ i++
+ }
+ if i >= len(data) {
+ return 0
+ }
+ // do not count escaped chars
+ if i != 0 && data[i-1] == '\\' {
+ i++
+ continue
+ }
+ if data[i] == c {
+ return i
+ }
+
+ if data[i] == '`' {
+ // skip a code span
+ tmpI := 0
+ i++
+ for i < len(data) && data[i] != '`' {
+ if tmpI == 0 && data[i] == c {
+ tmpI = i
+ }
+ i++
+ }
+ if i >= len(data) {
+ return tmpI
+ }
+ i++
+ } else if data[i] == '[' {
+ // skip a link
+ tmpI := 0
+ i++
+ for i < len(data) && data[i] != ']' {
+ if tmpI == 0 && data[i] == c {
+ tmpI = i
+ }
+ i++
+ }
+ i++
+ for i < len(data) && (data[i] == ' ' || data[i] == '\n') {
+ i++
+ }
+ if i >= len(data) {
+ return tmpI
+ }
+ if data[i] != '[' && data[i] != '(' { // not a link
+ if tmpI > 0 {
+ return tmpI
+ }
+ continue
+ }
+ cc := data[i]
+ i++
+ for i < len(data) && data[i] != cc {
+ if tmpI == 0 && data[i] == c {
+ return i
+ }
+ i++
+ }
+ if i >= len(data) {
+ return tmpI
+ }
+ i++
+ }
+ }
+ return 0
+}
+
+func helperEmphasis(p *Parser, data []byte, c byte) (int, ast.Node) {
+ i := 0
+
+ // skip one symbol if coming from emph3
+ if len(data) > 1 && data[0] == c && data[1] == c {
+ i = 1
+ }
+
+ for i < len(data) {
+ length := helperFindEmphChar(data[i:], c)
+ if length == 0 {
+ return 0, nil
+ }
+ i += length
+ if i >= len(data) {
+ return 0, nil
+ }
+
+ if i+1 < len(data) && data[i+1] == c {
+ i++
+ continue
+ }
+
+ if data[i] == c && !isSpace(data[i-1]) {
+
+ if p.extensions&NoIntraEmphasis != 0 {
+ if !(i+1 == len(data) || isSpace(data[i+1]) || isPunctuation(data[i+1])) {
+ continue
+ }
+ }
+
+ emph := &ast.Emph{}
+ p.Inline(emph, data[:i])
+ return i + 1, emph
+ }
+ }
+
+ return 0, nil
+}
+
+func helperDoubleEmphasis(p *Parser, data []byte, c byte) (int, ast.Node) {
+ i := 0
+
+ for i < len(data) {
+ length := helperFindEmphChar(data[i:], c)
+ if length == 0 {
+ return 0, nil
+ }
+ i += length
+
+ if i+1 < len(data) && data[i] == c && data[i+1] == c && i > 0 && !isSpace(data[i-1]) {
+ var node ast.Node = &ast.Strong{}
+ if c == '~' {
+ node = &ast.Del{}
+ }
+ p.Inline(node, data[:i])
+ return i + 2, node
+ }
+ i++
+ }
+ return 0, nil
+}
+
+func helperTripleEmphasis(p *Parser, data []byte, offset int, c byte) (int, ast.Node) {
+ i := 0
+ origData := data
+ data = data[offset:]
+
+ for i < len(data) {
+ length := helperFindEmphChar(data[i:], c)
+ if length == 0 {
+ return 0, nil
+ }
+ i += length
+
+ // skip whitespace preceded symbols
+ if data[i] != c || isSpace(data[i-1]) {
+ continue
+ }
+
+ switch {
+ case i+2 < len(data) && data[i+1] == c && data[i+2] == c:
+ // triple symbol found
+ strong := &ast.Strong{}
+ em := &ast.Emph{}
+ ast.AppendChild(strong, em)
+ p.Inline(em, data[:i])
+ return i + 3, strong
+ case i+1 < len(data) && data[i+1] == c:
+ // double symbol found, hand over to emph1
+ length, node := helperEmphasis(p, origData[offset-2:], c)
+ if length == 0 {
+ return 0, nil
+ }
+ return length - 2, node
+ default:
+ // single symbol found, hand over to emph2
+ length, node := helperDoubleEmphasis(p, origData[offset-1:], c)
+ if length == 0 {
+ return 0, nil
+ }
+ return length - 1, node
+ }
+ }
+ return 0, nil
+}
+
+// math handle inline math wrapped with '$'
+func math(p *Parser, data []byte, offset int) (int, ast.Node) {
+ data = data[offset:]
+
+ // too short, or block math
+ if len(data) <= 2 || data[1] == '$' {
+ return 0, nil
+ }
+
+ // find next '$'
+ var end int
+ for end = 1; end < len(data) && data[end] != '$'; end++ {
+ }
+
+ // $ not match
+ if end == len(data) {
+ return 0, nil
+ }
+
+ // create inline math node
+ math := &ast.Math{}
+ math.Literal = data[1:end]
+ return end + 1, math
+}
+
+func newTextNode(d []byte) *ast.Text {
+ return &ast.Text{ast.Leaf{Literal: d}}
+}
+
+func normalizeURI(s []byte) []byte {
+ return s // TODO: implement
+}
diff --git a/vendor/github.com/gomarkdown/markdown/parser/matter.go b/vendor/github.com/gomarkdown/markdown/parser/matter.go
new file mode 100644
index 00000000..92686357
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/parser/matter.go
@@ -0,0 +1,36 @@
+package parser
+
+import (
+ "bytes"
+
+ "github.com/gomarkdown/markdown/ast"
+)
+
+func (p *Parser) documentMatter(data []byte) int {
+ if data[0] != '{' {
+ return 0
+ }
+
+ consumed := 0
+ matter := ast.DocumentMatterNone
+ if bytes.HasPrefix(data, []byte("{frontmatter}")) {
+ consumed = len("{frontmatter}")
+ matter = ast.DocumentMatterFront
+ }
+ if bytes.HasPrefix(data, []byte("{mainmatter}")) {
+ consumed = len("{mainmatter}")
+ matter = ast.DocumentMatterMain
+ }
+ if bytes.HasPrefix(data, []byte("{backmatter}")) {
+ consumed = len("{backmatter}")
+ matter = ast.DocumentMatterBack
+ }
+ if consumed == 0 {
+ return 0
+ }
+ node := &ast.DocumentMatter{Matter: matter}
+ p.addBlock(node)
+ p.finalize(node)
+
+ return consumed
+}
diff --git a/vendor/github.com/gomarkdown/markdown/parser/options.go b/vendor/github.com/gomarkdown/markdown/parser/options.go
new file mode 100644
index 00000000..d3d0c088
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/parser/options.go
@@ -0,0 +1,32 @@
+package parser
+
+import (
+ "github.com/gomarkdown/markdown/ast"
+)
+
+// Flags control optional behavior of parser.
+type Flags int
+
+// Options is a collection of supplementary parameters tweaking the behavior of various parts of the parser.
+type Options struct {
+ ParserHook BlockFunc
+ ReadIncludeFn ReadIncludeFunc
+
+ Flags Flags // Flags allow customizing parser's behavior
+}
+
+// Parser renderer configuration options.
+const (
+ FlagsNone Flags = 0
+ SkipFootnoteList Flags = 1 << iota // Skip adding the footnote list (regardless if they are parsed)
+)
+
+// BlockFunc allows to registration of a parser function. If successful it
+// returns an ast.Node, a buffer that should be parsed as a block and the the number of bytes consumed.
+type BlockFunc func(data []byte) (ast.Node, []byte, int)
+
+// ReadIncludeFunc should read the file under path and returns the read bytes,
+// from will be set to the name of the current file being parsed. Initially
+// this will be empty. address is the optional address specifier of which lines
+// of the file to return. If this function is not set no data will be read.
+type ReadIncludeFunc func(from, path string, address []byte) []byte
diff --git a/vendor/github.com/gomarkdown/markdown/parser/parser.go b/vendor/github.com/gomarkdown/markdown/parser/parser.go
new file mode 100644
index 00000000..c7302dfd
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/parser/parser.go
@@ -0,0 +1,812 @@
+/*
+Package parser implements parser for markdown text that generates AST (abstract syntax tree).
+*/
+package parser
+
+import (
+ "bytes"
+ "fmt"
+ "strings"
+ "unicode/utf8"
+
+ "github.com/gomarkdown/markdown/ast"
+)
+
+// Extensions is a bitmask of enabled parser extensions.
+type Extensions int
+
+// Bit flags representing markdown parsing extensions.
+// Use | (or) to specify multiple extensions.
+const (
+ NoExtensions Extensions = 0
+ NoIntraEmphasis Extensions = 1 << iota // Ignore emphasis markers inside words
+ Tables // Parse tables
+ FencedCode // Parse fenced code blocks
+ Autolink // Detect embedded URLs that are not explicitly marked
+ Strikethrough // Strikethrough text using ~~test~~
+ LaxHTMLBlocks // Loosen up HTML block parsing rules
+ SpaceHeadings // Be strict about prefix heading rules
+ HardLineBreak // Translate newlines into line breaks
+ NonBlockingSpace // Translate backspace spaces into line non-blocking spaces
+ TabSizeEight // Expand tabs to eight spaces instead of four
+ Footnotes // Pandoc-style footnotes
+ NoEmptyLineBeforeBlock // No need to insert an empty line to start a (code, quote, ordered list, unordered list) block
+ HeadingIDs // specify heading IDs with {#id}
+ Titleblock // Titleblock ala pandoc
+ AutoHeadingIDs // Create the heading ID from the text
+ BackslashLineBreak // Translate trailing backslashes into line breaks
+ DefinitionLists // Parse definition lists
+ MathJax // Parse MathJax
+ OrderedListStart // Keep track of the first number used when starting an ordered list.
+ Attributes // Block Attributes
+ SuperSubscript // Super- and subscript support: 2^10^, H~2~O.
+ EmptyLinesBreakList // 2 empty lines break out of list
+ Includes // Support including other files.
+ Mmark // Support Mmark syntax, see https://mmark.nl/syntax
+
+ CommonExtensions Extensions = NoIntraEmphasis | Tables | FencedCode |
+ Autolink | Strikethrough | SpaceHeadings | HeadingIDs |
+ BackslashLineBreak | DefinitionLists | MathJax
+)
+
+// The size of a tab stop.
+const (
+ tabSizeDefault = 4
+ tabSizeDouble = 8
+)
+
+// for each character that triggers a response when parsing inline data.
+type inlineParser func(p *Parser, data []byte, offset int) (int, ast.Node)
+
+// ReferenceOverrideFunc is expected to be called with a reference string and
+// return either a valid Reference type that the reference string maps to or
+// nil. If overridden is false, the default reference logic will be executed.
+// See the documentation in Options for more details on use-case.
+type ReferenceOverrideFunc func(reference string) (ref *Reference, overridden bool)
+
+// Parser is a type that holds extensions and the runtime state used by
+// Parse, and the renderer. You can not use it directly, construct it with New.
+type Parser struct {
+
+ // ReferenceOverride is an optional function callback that is called every
+ // time a reference is resolved. It can be set before starting parsing.
+ //
+ // In Markdown, the link reference syntax can be made to resolve a link to
+ // a reference instead of an inline URL, in one of the following ways:
+ //
+ // * [link text][refid]
+ // * [refid][]
+ //
+ // Usually, the refid is defined at the bottom of the Markdown document. If
+ // this override function is provided, the refid is passed to the override
+ // function first, before consulting the defined refids at the bottom. If
+ // the override function indicates an override did not occur, the refids at
+ // the bottom will be used to fill in the link details.
+ ReferenceOverride ReferenceOverrideFunc
+
+ Opts Options
+
+ // after parsing, this is AST root of parsed markdown text
+ Doc ast.Node
+
+ extensions Extensions
+
+ refs map[string]*reference
+ refsRecord map[string]struct{}
+ inlineCallback [256]inlineParser
+ nesting int
+ maxNesting int
+ insideLink bool
+ indexCnt int // incremented after every index
+
+ // Footnotes need to be ordered as well as available to quickly check for
+ // presence. If a ref is also a footnote, it's stored both in refs and here
+ // in notes. Slice is nil if footnotes not enabled.
+ notes []*reference
+
+ tip ast.Node // = doc
+ oldTip ast.Node
+ lastMatchedContainer ast.Node // = doc
+ allClosed bool
+
+ // Attributes are attached to block level elements.
+ attr *ast.Attribute
+
+ includeStack *incStack
+}
+
+// New creates a markdown parser with CommonExtensions.
+//
+// You can then call `doc := p.Parse(markdown)` to parse markdown document
+// and `markdown.Render(doc, renderer)` to convert it to another format with
+// a renderer.
+func New() *Parser {
+ return NewWithExtensions(CommonExtensions)
+}
+
+// NewWithExtensions creates a markdown parser with given extensions.
+func NewWithExtensions(extension Extensions) *Parser {
+ p := Parser{
+ refs: make(map[string]*reference),
+ refsRecord: make(map[string]struct{}),
+ maxNesting: 16,
+ insideLink: false,
+ Doc: &ast.Document{},
+ extensions: extension,
+ allClosed: true,
+ includeStack: newIncStack(),
+ }
+ p.tip = p.Doc
+ p.oldTip = p.Doc
+ p.lastMatchedContainer = p.Doc
+
+ p.inlineCallback[' '] = maybeLineBreak
+ p.inlineCallback['*'] = emphasis
+ p.inlineCallback['_'] = emphasis
+ if p.extensions&Strikethrough != 0 {
+ p.inlineCallback['~'] = emphasis
+ }
+ p.inlineCallback['`'] = codeSpan
+ p.inlineCallback['\n'] = lineBreak
+ p.inlineCallback['['] = link
+ p.inlineCallback['<'] = leftAngle
+ p.inlineCallback['\\'] = escape
+ p.inlineCallback['&'] = entity
+ p.inlineCallback['!'] = maybeImage
+ if p.extensions&Mmark != 0 {
+ p.inlineCallback['('] = maybeShortRefOrIndex
+ }
+ p.inlineCallback['^'] = maybeInlineFootnoteOrSuper
+ if p.extensions&Autolink != 0 {
+ p.inlineCallback['h'] = maybeAutoLink
+ p.inlineCallback['m'] = maybeAutoLink
+ p.inlineCallback['f'] = maybeAutoLink
+ p.inlineCallback['H'] = maybeAutoLink
+ p.inlineCallback['M'] = maybeAutoLink
+ p.inlineCallback['F'] = maybeAutoLink
+ }
+ if p.extensions&MathJax != 0 {
+ p.inlineCallback['$'] = math
+ }
+
+ return &p
+}
+
+func (p *Parser) getRef(refid string) (ref *reference, found bool) {
+ if p.ReferenceOverride != nil {
+ r, overridden := p.ReferenceOverride(refid)
+ if overridden {
+ if r == nil {
+ return nil, false
+ }
+ return &reference{
+ link: []byte(r.Link),
+ title: []byte(r.Title),
+ noteID: 0,
+ hasBlock: false,
+ text: []byte(r.Text)}, true
+ }
+ }
+ // refs are case insensitive
+ ref, found = p.refs[strings.ToLower(refid)]
+ return ref, found
+}
+
+func (p *Parser) isFootnote(ref *reference) bool {
+ _, ok := p.refsRecord[string(ref.link)]
+ return ok
+}
+
+func (p *Parser) finalize(block ast.Node) {
+ p.tip = block.GetParent()
+}
+
+func (p *Parser) addChild(node ast.Node) ast.Node {
+ for !canNodeContain(p.tip, node) {
+ p.finalize(p.tip)
+ }
+ ast.AppendChild(p.tip, node)
+ p.tip = node
+ return node
+}
+
+func canNodeContain(n ast.Node, v ast.Node) bool {
+ switch n.(type) {
+ case *ast.List:
+ return isListItem(v)
+ case *ast.Document, *ast.BlockQuote, *ast.Aside, *ast.ListItem, *ast.CaptionFigure:
+ return !isListItem(v)
+ case *ast.Table:
+ switch v.(type) {
+ case *ast.TableHeader, *ast.TableBody, *ast.TableFooter:
+ return true
+ default:
+ return false
+ }
+ case *ast.TableHeader, *ast.TableBody, *ast.TableFooter:
+ _, ok := v.(*ast.TableRow)
+ return ok
+ case *ast.TableRow:
+ _, ok := v.(*ast.TableCell)
+ return ok
+ }
+ return false
+}
+
+func (p *Parser) closeUnmatchedBlocks() {
+ if p.allClosed {
+ return
+ }
+ for p.oldTip != p.lastMatchedContainer {
+ parent := p.oldTip.GetParent()
+ p.finalize(p.oldTip)
+ p.oldTip = parent
+ }
+ p.allClosed = true
+}
+
+// Reference represents the details of a link.
+// See the documentation in Options for more details on use-case.
+type Reference struct {
+ // Link is usually the URL the reference points to.
+ Link string
+ // Title is the alternate text describing the link in more detail.
+ Title string
+ // Text is the optional text to override the ref with if the syntax used was
+ // [refid][]
+ Text string
+}
+
+// Parse generates AST (abstract syntax tree) representing markdown document.
+//
+// The result is a root of the tree whose underlying type is *ast.Document
+//
+// You can then convert AST to html using html.Renderer, to some other format
+// using a custom renderer or transform the tree.
+func (p *Parser) Parse(input []byte) ast.Node {
+ p.block(input)
+ // Walk the tree and finish up some of unfinished blocks
+ for p.tip != nil {
+ p.finalize(p.tip)
+ }
+ // Walk the tree again and process inline markdown in each block
+ ast.WalkFunc(p.Doc, func(node ast.Node, entering bool) ast.WalkStatus {
+ switch node.(type) {
+ case *ast.Paragraph, *ast.Heading, *ast.TableCell:
+ p.Inline(node, node.AsContainer().Content)
+ node.AsContainer().Content = nil
+ }
+ return ast.GoToNext
+ })
+
+ if p.Opts.Flags&SkipFootnoteList == 0 {
+ p.parseRefsToAST()
+ }
+ return p.Doc
+}
+
+func (p *Parser) parseRefsToAST() {
+ if p.extensions&Footnotes == 0 || len(p.notes) == 0 {
+ return
+ }
+ p.tip = p.Doc
+ list := &ast.List{
+ IsFootnotesList: true,
+ ListFlags: ast.ListTypeOrdered,
+ }
+ p.addBlock(&ast.Footnotes{})
+ block := p.addBlock(list)
+ flags := ast.ListItemBeginningOfList
+ // Note: this loop is intentionally explicit, not range-form. This is
+ // because the body of the loop will append nested footnotes to p.notes and
+ // we need to process those late additions. Range form would only walk over
+ // the fixed initial set.
+ for i := 0; i < len(p.notes); i++ {
+ ref := p.notes[i]
+ p.addChild(ref.footnote)
+ block := ref.footnote
+ listItem := block.(*ast.ListItem)
+ listItem.ListFlags = flags | ast.ListTypeOrdered
+ listItem.RefLink = ref.link
+ if ref.hasBlock {
+ flags |= ast.ListItemContainsBlock
+ p.block(ref.title)
+ } else {
+ p.Inline(block, ref.title)
+ }
+ flags &^= ast.ListItemBeginningOfList | ast.ListItemContainsBlock
+ }
+ above := list.Parent
+ finalizeList(list)
+ p.tip = above
+
+ ast.WalkFunc(block, func(node ast.Node, entering bool) ast.WalkStatus {
+ switch node.(type) {
+ case *ast.Paragraph, *ast.Heading:
+ p.Inline(node, node.AsContainer().Content)
+ node.AsContainer().Content = nil
+ }
+ return ast.GoToNext
+ })
+}
+
+//
+// Link references
+//
+// This section implements support for references that (usually) appear
+// as footnotes in a document, and can be referenced anywhere in the document.
+// The basic format is:
+//
+// [1]: http://www.google.com/ "Google"
+// [2]: http://www.github.com/ "Github"
+//
+// Anywhere in the document, the reference can be linked by referring to its
+// label, i.e., 1 and 2 in this example, as in:
+//
+// This library is hosted on [Github][2], a git hosting site.
+//
+// Actual footnotes as specified in Pandoc and supported by some other Markdown
+// libraries such as php-markdown are also taken care of. They look like this:
+//
+// This sentence needs a bit of further explanation.[^note]
+//
+// [^note]: This is the explanation.
+//
+// Footnotes should be placed at the end of the document in an ordered list.
+// Inline footnotes such as:
+//
+// Inline footnotes^[Not supported.] also exist.
+//
+// are not yet supported.
+
+// reference holds all information necessary for a reference-style links or
+// footnotes.
+//
+// Consider this markdown with reference-style links:
+//
+// [link][ref]
+//
+// [ref]: /url/ "tooltip title"
+//
+// It will be ultimately converted to this HTML:
+//
+// <p><a href=\"/url/\" title=\"title\">link</a></p>
+//
+// And a reference structure will be populated as follows:
+//
+// p.refs["ref"] = &reference{
+// link: "/url/",
+// title: "tooltip title",
+// }
+//
+// Alternatively, reference can contain information about a footnote. Consider
+// this markdown:
+//
+// Text needing a footnote.[^a]
+//
+// [^a]: This is the note
+//
+// A reference structure will be populated as follows:
+//
+// p.refs["a"] = &reference{
+// link: "a",
+// title: "This is the note",
+// noteID: <some positive int>,
+// }
+//
+// TODO: As you can see, it begs for splitting into two dedicated structures
+// for refs and for footnotes.
+type reference struct {
+ link []byte
+ title []byte
+ noteID int // 0 if not a footnote ref
+ hasBlock bool
+ footnote ast.Node // a link to the Item node within a list of footnotes
+
+ text []byte // only gets populated by refOverride feature with Reference.Text
+}
+
+func (r *reference) String() string {
+ return fmt.Sprintf("{link: %q, title: %q, text: %q, noteID: %d, hasBlock: %v}",
+ r.link, r.title, r.text, r.noteID, r.hasBlock)
+}
+
+// Check whether or not data starts with a reference link.
+// If so, it is parsed and stored in the list of references
+// (in the render struct).
+// Returns the number of bytes to skip to move past it,
+// or zero if the first line is not a reference.
+func isReference(p *Parser, data []byte, tabSize int) int {
+ // up to 3 optional leading spaces
+ if len(data) < 4 {
+ return 0
+ }
+ i := 0
+ for i < 3 && data[i] == ' ' {
+ i++
+ }
+
+ noteID := 0
+
+ // id part: anything but a newline between brackets
+ if data[i] != '[' {
+ return 0
+ }
+ i++
+ if p.extensions&Footnotes != 0 {
+ if i < len(data) && data[i] == '^' {
+ // we can set it to anything here because the proper noteIds will
+ // be assigned later during the second pass. It just has to be != 0
+ noteID = 1
+ i++
+ }
+ }
+ idOffset := i
+ for i < len(data) && data[i] != '\n' && data[i] != '\r' && data[i] != ']' {
+ i++
+ }
+ if i >= len(data) || data[i] != ']' {
+ return 0
+ }
+ idEnd := i
+ // footnotes can have empty ID, like this: [^], but a reference can not be
+ // empty like this: []. Break early if it's not a footnote and there's no ID
+ if noteID == 0 && idOffset == idEnd {
+ return 0
+ }
+ // spacer: colon (space | tab)* newline? (space | tab)*
+ i++
+ if i >= len(data) || data[i] != ':' {
+ return 0
+ }
+ i++
+ for i < len(data) && (data[i] == ' ' || data[i] == '\t') {
+ i++
+ }
+ if i < len(data) && (data[i] == '\n' || data[i] == '\r') {
+ i++
+ if i < len(data) && data[i] == '\n' && data[i-1] == '\r' {
+ i++
+ }
+ }
+ for i < len(data) && (data[i] == ' ' || data[i] == '\t') {
+ i++
+ }
+ if i >= len(data) {
+ return 0
+ }
+
+ var (
+ linkOffset, linkEnd int
+ titleOffset, titleEnd int
+ lineEnd int
+ raw []byte
+ hasBlock bool
+ )
+
+ if p.extensions&Footnotes != 0 && noteID != 0 {
+ linkOffset, linkEnd, raw, hasBlock = scanFootnote(p, data, i, tabSize)
+ lineEnd = linkEnd
+ } else {
+ linkOffset, linkEnd, titleOffset, titleEnd, lineEnd = scanLinkRef(p, data, i)
+ }
+ if lineEnd == 0 {
+ return 0
+ }
+
+ // a valid ref has been found
+
+ ref := &reference{
+ noteID: noteID,
+ hasBlock: hasBlock,
+ }
+
+ if noteID > 0 {
+ // reusing the link field for the id since footnotes don't have links
+ ref.link = data[idOffset:idEnd]
+ // if footnote, it's not really a title, it's the contained text
+ ref.title = raw
+ } else {
+ ref.link = data[linkOffset:linkEnd]
+ ref.title = data[titleOffset:titleEnd]
+ }
+
+ // id matches are case-insensitive
+ id := string(bytes.ToLower(data[idOffset:idEnd]))
+
+ p.refs[id] = ref
+
+ return lineEnd
+}
+
+func scanLinkRef(p *Parser, data []byte, i int) (linkOffset, linkEnd, titleOffset, titleEnd, lineEnd int) {
+ // link: whitespace-free sequence, optionally between angle brackets
+ if data[i] == '<' {
+ i++
+ }
+ linkOffset = i
+ for i < len(data) && data[i] != ' ' && data[i] != '\t' && data[i] != '\n' && data[i] != '\r' {
+ i++
+ }
+ linkEnd = i
+ if linkEnd < len(data) && data[linkOffset] == '<' && data[linkEnd-1] == '>' {
+ linkOffset++
+ linkEnd--
+ }
+
+ // optional spacer: (space | tab)* (newline | '\'' | '"' | '(' )
+ for i < len(data) && (data[i] == ' ' || data[i] == '\t') {
+ i++
+ }
+ if i < len(data) && data[i] != '\n' && data[i] != '\r' && data[i] != '\'' && data[i] != '"' && data[i] != '(' {
+ return
+ }
+
+ // compute end-of-line
+ if i >= len(data) || data[i] == '\r' || data[i] == '\n' {
+ lineEnd = i
+ }
+ if i+1 < len(data) && data[i] == '\r' && data[i+1] == '\n' {
+ lineEnd++
+ }
+
+ // optional (space|tab)* spacer after a newline
+ if lineEnd > 0 {
+ i = lineEnd + 1
+ for i < len(data) && (data[i] == ' ' || data[i] == '\t') {
+ i++
+ }
+ }
+
+ // optional title: any non-newline sequence enclosed in '"() alone on its line
+ if i+1 < len(data) && (data[i] == '\'' || data[i] == '"' || data[i] == '(') {
+ i++
+ titleOffset = i
+
+ // look for EOL
+ for i < len(data) && data[i] != '\n' && data[i] != '\r' {
+ i++
+ }
+ if i+1 < len(data) && data[i] == '\n' && data[i+1] == '\r' {
+ titleEnd = i + 1
+ } else {
+ titleEnd = i
+ }
+
+ // step back
+ i--
+ for i > titleOffset && (data[i] == ' ' || data[i] == '\t') {
+ i--
+ }
+ if i > titleOffset && (data[i] == '\'' || data[i] == '"' || data[i] == ')') {
+ lineEnd = titleEnd
+ titleEnd = i
+ }
+ }
+
+ return
+}
+
+// The first bit of this logic is the same as Parser.listItem, but the rest
+// is much simpler. This function simply finds the entire block and shifts it
+// over by one tab if it is indeed a block (just returns the line if it's not).
+// blockEnd is the end of the section in the input buffer, and contents is the
+// extracted text that was shifted over one tab. It will need to be rendered at
+// the end of the document.
+func scanFootnote(p *Parser, data []byte, i, indentSize int) (blockStart, blockEnd int, contents []byte, hasBlock bool) {
+ if i == 0 || len(data) == 0 {
+ return
+ }
+
+ // skip leading whitespace on first line
+ for i < len(data) && data[i] == ' ' {
+ i++
+ }
+
+ blockStart = i
+
+ // find the end of the line
+ blockEnd = i
+ for i < len(data) && data[i-1] != '\n' {
+ i++
+ }
+
+ // get working buffer
+ var raw bytes.Buffer
+
+ // put the first line into the working buffer
+ raw.Write(data[blockEnd:i])
+ blockEnd = i
+
+ // process the following lines
+ containsBlankLine := false
+
+gatherLines:
+ for blockEnd < len(data) {
+ i++
+
+ // find the end of this line
+ for i < len(data) && data[i-1] != '\n' {
+ i++
+ }
+
+ // if it is an empty line, guess that it is part of this item
+ // and move on to the next line
+ if p.isEmpty(data[blockEnd:i]) > 0 {
+ containsBlankLine = true
+ blockEnd = i
+ continue
+ }
+
+ n := 0
+ if n = isIndented(data[blockEnd:i], indentSize); n == 0 {
+ // this is the end of the block.
+ // we don't want to include this last line in the index.
+ break gatherLines
+ }
+
+ // if there were blank lines before this one, insert a new one now
+ if containsBlankLine {
+ raw.WriteByte('\n')
+ containsBlankLine = false
+ }
+
+ // get rid of that first tab, write to buffer
+ raw.Write(data[blockEnd+n : i])
+ hasBlock = true
+
+ blockEnd = i
+ }
+
+ if data[blockEnd-1] != '\n' {
+ raw.WriteByte('\n')
+ }
+
+ contents = raw.Bytes()
+
+ return
+}
+
+// isPunctuation returns true if c is a punctuation symbol.
+func isPunctuation(c byte) bool {
+ for _, r := range []byte("!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~") {
+ if c == r {
+ return true
+ }
+ }
+ return false
+}
+
+// isSpace returns true if c is a white-space charactr
+func isSpace(c byte) bool {
+ return c == ' ' || c == '\t' || c == '\n' || c == '\r' || c == '\f' || c == '\v'
+}
+
+// isLetter returns true if c is ascii letter
+func isLetter(c byte) bool {
+ return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z')
+}
+
+// isAlnum returns true if c is a digit or letter
+// TODO: check when this is looking for ASCII alnum and when it should use unicode
+func isAlnum(c byte) bool {
+ return (c >= '0' && c <= '9') || isLetter(c)
+}
+
+// TODO: this is not used
+// Replace tab characters with spaces, aligning to the next TAB_SIZE column.
+// always ends output with a newline
+func expandTabs(out *bytes.Buffer, line []byte, tabSize int) {
+ // first, check for common cases: no tabs, or only tabs at beginning of line
+ i, prefix := 0, 0
+ slowcase := false
+ for i = 0; i < len(line); i++ {
+ if line[i] == '\t' {
+ if prefix == i {
+ prefix++
+ } else {
+ slowcase = true
+ break
+ }
+ }
+ }
+
+ // no need to decode runes if all tabs are at the beginning of the line
+ if !slowcase {
+ for i = 0; i < prefix*tabSize; i++ {
+ out.WriteByte(' ')
+ }
+ out.Write(line[prefix:])
+ return
+ }
+
+ // the slow case: we need to count runes to figure out how
+ // many spaces to insert for each tab
+ column := 0
+ i = 0
+ for i < len(line) {
+ start := i
+ for i < len(line) && line[i] != '\t' {
+ _, size := utf8.DecodeRune(line[i:])
+ i += size
+ column++
+ }
+
+ if i > start {
+ out.Write(line[start:i])
+ }
+
+ if i >= len(line) {
+ break
+ }
+
+ for {
+ out.WriteByte(' ')
+ column++
+ if column%tabSize == 0 {
+ break
+ }
+ }
+
+ i++
+ }
+}
+
+// Find if a line counts as indented or not.
+// Returns number of characters the indent is (0 = not indented).
+func isIndented(data []byte, indentSize int) int {
+ if len(data) == 0 {
+ return 0
+ }
+ if data[0] == '\t' {
+ return 1
+ }
+ if len(data) < indentSize {
+ return 0
+ }
+ for i := 0; i < indentSize; i++ {
+ if data[i] != ' ' {
+ return 0
+ }
+ }
+ return indentSize
+}
+
+// Create a url-safe slug for fragments
+func slugify(in []byte) []byte {
+ if len(in) == 0 {
+ return in
+ }
+ out := make([]byte, 0, len(in))
+ sym := false
+
+ for _, ch := range in {
+ if isAlnum(ch) {
+ sym = false
+ out = append(out, ch)
+ } else if sym {
+ continue
+ } else {
+ out = append(out, '-')
+ sym = true
+ }
+ }
+ var a, b int
+ var ch byte
+ for a, ch = range out {
+ if ch != '-' {
+ break
+ }
+ }
+ for b = len(out) - 1; b > 0; b-- {
+ if out[b] != '-' {
+ break
+ }
+ }
+ return out[a : b+1]
+}
+
+func isListItem(d ast.Node) bool {
+ _, ok := d.(*ast.ListItem)
+ return ok
+}
diff --git a/vendor/github.com/gomarkdown/markdown/parser/ref.go b/vendor/github.com/gomarkdown/markdown/parser/ref.go
new file mode 100644
index 00000000..0b59a196
--- /dev/null
+++ b/vendor/github.com/gomarkdown/markdown/parser/ref.go
@@ -0,0 +1,89 @@
+package parser
+
+import (
+ "bytes"
+ "fmt"
+
+ "github.com/gomarkdown/markdown/ast"
+)
+
+// parse '(#r)', where r does not contain spaces. Or.
+// (!item) (!item, subitem), for an index, (!!item) signals primary.
+func maybeShortRefOrIndex(p *Parser, data []byte, offset int) (int, ast.Node) {
+ if len(data[offset:]) < 4 {
+ return 0, nil
+ }
+ // short ref first
+ data = data[offset:]
+ i := 1
+ switch data[i] {
+ case '#': // cross ref
+ i++
+ Loop:
+ for i < len(data) {
+ c := data[i]
+ switch {
+ case c == ')':
+ break Loop
+ case !isAlnum(c):
+ if c == '_' || c == '-' || c == ':' {
+ i++
+ continue
+ }
+ i = 0
+ break Loop
+ }
+ i++
+ }
+ if i >= len(data) {
+ return 0, nil
+ }
+ if data[i] != ')' {
+ return 0, nil
+ }
+
+ id := data[2:i]
+ node := &ast.CrossReference{}
+ node.Destination = id
+
+ return i + 1, node
+
+ case '!': // index
+ i++
+ start := i
+ i = skipUntilChar(data, start, ')')
+
+ // did we reach the end of the buffer without a closing marker?
+ if i >= len(data) {
+ return 0, nil
+ }
+
+ if len(data[start:i]) < 1 {
+ return 0, nil
+ }
+
+ idx := &ast.Index{}
+
+ idx.ID = fmt.Sprintf("idxref:%d", p.indexCnt)
+ p.indexCnt++
+
+ idx.Primary = data[start] == '!'
+ buf := data[start:i]
+
+ if idx.Primary {
+ buf = buf[1:]
+ }
+ items := bytes.Split(buf, []byte(","))
+ switch len(items) {
+ case 1:
+ idx.Item = bytes.TrimSpace(items[0])
+ return i + 1, idx
+ case 2:
+ idx.Item = bytes.TrimSpace(items[0])
+ idx.Subitem = bytes.TrimSpace(items[1])
+ return i + 1, idx
+ }
+ }
+
+ return 0, nil
+}
--
cgit v1.2.3