Add support for markdown to HTML conversion (matrix). Closes #663 (#670)

This uses our own gomatrix lib with the SendHTML function which
adds HTML to formatted_body in matrix.
golang-commonmark is used to convert markdown into valid HTML.

vendor/gitlab.com/golang-commonmark/markdown/parser_block.go

@@ -0,0 +1,159 @@
+// Copyright 2015 The Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package markdown
+
+import (
+	"bytes"
+	"unicode/utf8"
+)
+
+type ParserBlock struct{}
+
+type BlockRule func(*StateBlock, int, int, bool) bool
+
+var blockRules []BlockRule
+
+var nl = []byte{'\n'}
+
+func normalizeNewlines(src []byte) ([]byte, int) {
+	if bytes.IndexByte(src, '\r') == -1 {
+		return src, bytes.Count(src, nl)
+	}
+	n := 0
+	buf := make([]byte, 0, len(src))
+	for i := 0; i < len(src); i++ {
+		switch ch := src[i]; ch {
+		case '\n':
+			n++
+			buf = append(buf, '\n')
+		case '\r':
+			buf = append(buf, '\n')
+			n++
+			if i < len(src)-1 && src[i+1] == '\n' {
+				i++
+			}
+		default:
+			buf = append(buf, ch)
+		}
+	}
+	return buf, n
+}
+
+func (b ParserBlock) Parse(src []byte, md *Markdown, env *Environment) []Token {
+	src, n := normalizeNewlines(src)
+	if len(src) == 0 || src[len(src)-1] != '\n' {
+		n++
+	}
+	n++
+
+	indentFound := false
+	start := 0
+	indent := 0
+	offset := 0
+
+	mem := make([]int, 0, n*5)
+	bMarks := mem[0:0:n]
+	eMarks := mem[n : n : n*2]
+	tShift := mem[n*2 : n*2 : n*3]
+	sCount := mem[n*3 : n*3 : n*4]
+	bsCount := mem[n*4 : n*4 : n*5]
+
+	_, lastRuneLen := utf8.DecodeLastRune(src)
+	lastRunePos := len(src) - lastRuneLen
+	for pos, r := range string(src) {
+		if !indentFound {
+			if runeIsSpace(r) {
+				indent++
+				if r == '\t' {
+					offset += 4 - offset%4
+				} else {
+					offset++
+				}
+				continue
+			}
+			indentFound = true
+		}
+
+		if r == '\n' || pos == lastRunePos {
+			if r != '\n' {
+				pos = len(src)
+			}
+			bMarks = append(bMarks, start)
+			eMarks = append(eMarks, pos)
+			tShift = append(tShift, indent)
+			sCount = append(sCount, offset)
+			bsCount = append(bsCount, 0)
+
+			indentFound = false
+			indent = 0
+			offset = 0
+			start = pos + 1
+		}
+	}
+
+	bMarks = append(bMarks, len(src))
+	eMarks = append(eMarks, len(src))
+	tShift = append(tShift, 0)
+	sCount = append(sCount, 0)
+	bsCount = append(bsCount, 0)
+
+	var s StateBlock
+	s.BMarks = bMarks
+	s.EMarks = eMarks
+	s.TShift = tShift
+	s.SCount = sCount
+	s.BSCount = bsCount
+	s.LineMax = n - 1
+	s.Src = string(src)
+	s.Md = md
+	s.Env = env
+
+	b.Tokenize(&s, s.Line, s.LineMax)
+
+	return s.Tokens
+}
+
+func (ParserBlock) Tokenize(s *StateBlock, startLine, endLine int) {
+	line := startLine
+	hasEmptyLines := false
+	maxNesting := s.Md.MaxNesting
+
+	for line < endLine {
+		line = s.SkipEmptyLines(line)
+		s.Line = line
+		if line >= endLine {
+			break
+		}
+
+		if s.SCount[line] < s.BlkIndent {
+			break
+		}
+
+		if s.Level >= maxNesting {
+			s.Line = endLine
+			break
+		}
+
+		for _, r := range blockRules {
+			if r(s, line, endLine, false) {
+				break
+			}
+		}
+
+		s.Tight = !hasEmptyLines
+
+		if s.IsLineEmpty(s.Line - 1) {
+			hasEmptyLines = true
+		}
+
+		line = s.Line
+
+		if line < endLine && s.IsLineEmpty(line) {
+			hasEmptyLines = true
+			line++
+			s.Line = line
+		}
+	}
+}