Rewrite reference processing code in preparation for opening/closing from comment references (#8261)

* Add a markdown stripper for mentions and xrefs * Improve comments * Small code simplification * Move reference code to modules/references * Fix typo * Make MarkdownStripper return [][]byte * Implement preliminary keywords parsing * Add FIXME comment * Fix comment * make fmt * Fix permissions check * Fix text assumptions * Fix imports * Fix lint, fmt * Fix unused import * Add missing export comment * Bypass revive on implemented interface * Move mdstripper into its own package * Support alphanumeric patterns * Refactor FindAllMentions * Move mentions test to references * Parse mentions from reference package * Refactor code to implement renderizable references * Fix typo * Move patterns and tests to the references package * Fix nil reference * Preliminary rendering attempt of closing keywords * Normalize names, comments, general tidy-up * Add CSS style for action keywords * Fix permission for admin and owner * Fix golangci-lint * Fix golangci-lint
2025-05-25 11:22:16 +00:00 · 2019-10-13 19:29:10 -03:00 · 2019-10-13 19:29:10 -03:00 · 15809d81f7
commit 15809d81f7
parent 6e3f51098b
16 changed files with 1123 additions and 438 deletions
--- a/modules/markup/html.go
+++ b/modules/markup/html.go
@ -15,6 +15,7 @@ import (
 	"code.gitea.io/gitea/modules/base"
 	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/references"
 	"code.gitea.io/gitea/modules/setting"
 	"code.gitea.io/gitea/modules/util"

@ -36,17 +37,6 @@ var (
 	// While fast, this is also incorrect and lead to false positives.
 	// TODO: fix invalid linking issue

-	// mentionPattern matches all mentions in the form of "@user"
-	mentionPattern = regexp.MustCompile(`(?:\s|^|\(|\[)(@[0-9a-zA-Z-_\.]+)(?:\s|$|\)|\])`)
-
-	// issueNumericPattern matches string that references to a numeric issue, e.g. #1287
-	issueNumericPattern = regexp.MustCompile(`(?:\s|^|\(|\[)(#[0-9]+)(?:\s|$|\)|\]|:|\.(\s|$))`)
-	// issueAlphanumericPattern matches string that references to an alphanumeric issue, e.g. ABC-1234
-	issueAlphanumericPattern = regexp.MustCompile(`(?:\s|^|\(|\[)([A-Z]{1,10}-[1-9][0-9]*)(?:\s|$|\)|\]|:|\.(\s|$))`)
-	// crossReferenceIssueNumericPattern matches string that references a numeric issue in a different repository
-	// e.g. gogits/gogs#12345
-	crossReferenceIssueNumericPattern = regexp.MustCompile(`(?:\s|^|\(|\[)([0-9a-zA-Z-_\.]+/[0-9a-zA-Z-_\.]+#[0-9]+)(?:\s|$|\)|\]|\.(\s|$))`)
-
 	// sha1CurrentPattern matches string that represents a commit SHA, e.g. d8a994ef243349f321568f9e36d5c3f444b99cae
 	// Although SHA1 hashes are 40 chars long, the regex matches the hash from 7 to 40 chars in length
 	// so that abbreviated hash links can be used as well. This matches git and github useability.
@ -70,6 +60,9 @@ var (
 	linkRegex, _ = xurls.StrictMatchingScheme("https?://")
 )

+// CSS class for action keywords (e.g. "closes: #1")
+const keywordClass = "issue-keyword"
+
 // regexp for full links to issues/pulls
 var issueFullPattern *regexp.Regexp

@ -99,17 +92,6 @@ func getIssueFullPattern() *regexp.Regexp {
 	return issueFullPattern
 }

-// FindAllMentions matches mention patterns in given content
-// and returns a list of found user names without @ prefix.
-func FindAllMentions(content string) []string {
-	mentions := mentionPattern.FindAllStringSubmatch(content, -1)
-	ret := make([]string, len(mentions))
-	for i, val := range mentions {
-		ret[i] = val[1][1:]
-	}
-	return ret
-}
-
 // IsSameDomain checks if given url string has the same hostname as current Gitea instance
 func IsSameDomain(s string) bool {
 	if strings.HasPrefix(s, "/") {
@ -142,7 +124,6 @@ var defaultProcessors = []processor{
 	linkProcessor,
 	mentionProcessor,
 	issueIndexPatternProcessor,
-	crossReferenceIssueIndexPatternProcessor,
 	sha1CurrentPatternProcessor,
 	emailAddressProcessor,
 }
@ -183,7 +164,6 @@ var commitMessageProcessors = []processor{
 	linkProcessor,
 	mentionProcessor,
 	issueIndexPatternProcessor,
-	crossReferenceIssueIndexPatternProcessor,
 	sha1CurrentPatternProcessor,
 	emailAddressProcessor,
 }
@ -217,7 +197,6 @@ var commitMessageSubjectProcessors = []processor{
 	linkProcessor,
 	mentionProcessor,
 	issueIndexPatternProcessor,
-	crossReferenceIssueIndexPatternProcessor,
 	sha1CurrentPatternProcessor,
 }

@ -330,6 +309,24 @@ func (ctx *postProcessCtx) textNode(node *html.Node) {
 	}
 }

+// createKeyword() renders a highlighted version of an action keyword
+func createKeyword(content string) *html.Node {
+	span := &html.Node{
+		Type: html.ElementNode,
+		Data: atom.Span.String(),
+		Attr: []html.Attribute{},
+	}
+	span.Attr = append(span.Attr, html.Attribute{Key: "class", Val: keywordClass})
+
+	text := &html.Node{
+		Type: html.TextNode,
+		Data: content,
+	}
+	span.AppendChild(text)
+
+	return span
+}
+
 func createLink(href, content, class string) *html.Node {
 	a := &html.Node{
 		Type: html.ElementNode,
@ -377,10 +374,16 @@ func createCodeLink(href, content, class string) *html.Node {
 	return a
 }

-// replaceContent takes a text node, and in its content it replaces a section of
-// it with the specified newNode. An example to visualize how this can work can
-// be found here: https://play.golang.org/p/5zP8NnHZ03s
+// replaceContent takes text node, and in its content it replaces a section of
+// it with the specified newNode.
 func replaceContent(node *html.Node, i, j int, newNode *html.Node) {
+	replaceContentList(node, i, j, []*html.Node{newNode})
+}
+
+// replaceContentList takes text node, and in its content it replaces a section of
+// it with the specified newNodes. An example to visualize how this can work can
+// be found here: https://play.golang.org/p/5zP8NnHZ03s
+func replaceContentList(node *html.Node, i, j int, newNodes []*html.Node) {
 	// get the data before and after the match
 	before := node.Data[:i]
 	after := node.Data[j:]
@ -392,7 +395,9 @@ func replaceContent(node *html.Node, i, j int, newNode *html.Node) {
 	// Get the current next sibling, before which we place the replaced data,
 	// and after that we place the new text node.
 	nextSibling := node.NextSibling
-	node.Parent.InsertBefore(newNode, nextSibling)
+	for _, n := range newNodes {
+		node.Parent.InsertBefore(n, nextSibling)
+	}
 	if after != "" {
 		node.Parent.InsertBefore(&html.Node{
 			Type: html.TextNode,
@ -402,13 +407,13 @@ func replaceContent(node *html.Node, i, j int, newNode *html.Node) {
 }

 func mentionProcessor(_ *postProcessCtx, node *html.Node) {
-	m := mentionPattern.FindStringSubmatchIndex(node.Data)
-	if m == nil {
+	// We replace only the first mention; other mentions will be addressed later
+	found, loc := references.FindFirstMentionBytes([]byte(node.Data))
+	if !found {
 		return
 	}
-	// Replace the mention with a link to the specified user.
-	mention := node.Data[m[2]:m[3]]
-	replaceContent(node, m[2], m[3], createLink(util.URLJoin(setting.AppURL, mention[1:]), mention, "mention"))
+	mention := node.Data[loc.Start:loc.End]
+	replaceContent(node, loc.Start, loc.End, createLink(util.URLJoin(setting.AppURL, mention[1:]), mention, "mention"))
 }

 func shortLinkProcessor(ctx *postProcessCtx, node *html.Node) {
@ -597,45 +602,44 @@ func issueIndexPatternProcessor(ctx *postProcessCtx, node *html.Node) {
 	if ctx.metas == nil {
 		return
 	}
-	// default to numeric pattern, unless alphanumeric is requested.
-	pattern := issueNumericPattern
+
+	var (
+		found bool
+		ref   *references.RenderizableReference
+	)
+
 	if ctx.metas["style"] == IssueNameStyleAlphanumeric {
-		pattern = issueAlphanumericPattern
-	}
-
-	match := pattern.FindStringSubmatchIndex(node.Data)
-	if match == nil {
-		return
-	}
-
-	id := node.Data[match[2]:match[3]]
-	var link *html.Node
-	if _, ok := ctx.metas["format"]; ok {
-		// Support for external issue tracker
-		if ctx.metas["style"] == IssueNameStyleAlphanumeric {
-			ctx.metas["index"] = id
-		} else {
-			ctx.metas["index"] = id[1:]
-		}
-		link = createLink(com.Expand(ctx.metas["format"], ctx.metas), id, "issue")
+		found, ref = references.FindRenderizableReferenceAlphanumeric(node.Data)
 	} else {
-		link = createLink(util.URLJoin(setting.AppURL, ctx.metas["user"], ctx.metas["repo"], "issues", id[1:]), id, "issue")
+		found, ref = references.FindRenderizableReferenceNumeric(node.Data)
 	}
-	replaceContent(node, match[2], match[3], link)
-}
-
-func crossReferenceIssueIndexPatternProcessor(ctx *postProcessCtx, node *html.Node) {
-	m := crossReferenceIssueNumericPattern.FindStringSubmatchIndex(node.Data)
-	if m == nil {
+	if !found {
 		return
 	}
-	ref := node.Data[m[2]:m[3]]

-	parts := strings.SplitN(ref, "#", 2)
-	repo, issue := parts[0], parts[1]
+	var link *html.Node
+	reftext := node.Data[ref.RefLocation.Start:ref.RefLocation.End]
+	if _, ok := ctx.metas["format"]; ok {
+		ctx.metas["index"] = ref.Issue
+		link = createLink(com.Expand(ctx.metas["format"], ctx.metas), reftext, "issue")
+	} else if ref.Owner == "" {
+		link = createLink(util.URLJoin(setting.AppURL, ctx.metas["user"], ctx.metas["repo"], "issues", ref.Issue), reftext, "issue")
+	} else {
+		link = createLink(util.URLJoin(setting.AppURL, ref.Owner, ref.Name, "issues", ref.Issue), reftext, "issue")
+	}

-	replaceContent(node, m[2], m[3],
-		createLink(util.URLJoin(setting.AppURL, repo, "issues", issue), ref, issue))
+	if ref.Action == references.XRefActionNone {
+		replaceContent(node, ref.RefLocation.Start, ref.RefLocation.End, link)
+		return
+	}
+
+	// Decorate action keywords
+	keyword := createKeyword(node.Data[ref.ActionLocation.Start:ref.ActionLocation.End])
+	spaces := &html.Node{
+		Type: html.TextNode,
+		Data: node.Data[ref.ActionLocation.End:ref.RefLocation.Start],
+	}
+	replaceContentList(node, ref.ActionLocation.Start, ref.RefLocation.End, []*html.Node{keyword, spaces, link})
 }

 // fullSha1PatternProcessor renders SHA containing URLs
--- a/modules/markup/html_internal_test.go
+++ b/modules/markup/html_internal_test.go
@ -239,34 +239,6 @@ func TestRender_FullIssueURLs(t *testing.T) {
 		`<a href="http://localhost:3000/gogits/gogs/issues/4" class="issue">#4</a>`)
 }

-func TestRegExp_issueNumericPattern(t *testing.T) {
-	trueTestCases := []string{
-		"#1234",
-		"#0",
-		"#1234567890987654321",
-		"  #12",
-		"#12:",
-		"ref: #12: msg",
-	}
-	falseTestCases := []string{
-		"# 1234",
-		"# 0",
-		"# ",
-		"#",
-		"#ABC",
-		"#1A2B",
-		"",
-		"ABC",
-	}
-
-	for _, testCase := range trueTestCases {
-		assert.True(t, issueNumericPattern.MatchString(testCase))
-	}
-	for _, testCase := range falseTestCases {
-		assert.False(t, issueNumericPattern.MatchString(testCase))
-	}
-}
-
 func TestRegExp_sha1CurrentPattern(t *testing.T) {
 	trueTestCases := []string{
 		"d8a994ef243349f321568f9e36d5c3f444b99cae",
@ -325,70 +297,6 @@ func TestRegExp_anySHA1Pattern(t *testing.T) {
 	}
 }

-func TestRegExp_mentionPattern(t *testing.T) {
-	trueTestCases := []string{
-		"@Unknwon",
-		"@ANT_123",
-		"@xxx-DiN0-z-A..uru..s-xxx",
-		"   @lol   ",
-		" @Te-st",
-		"(@gitea)",
-		"[@gitea]",
-	}
-	falseTestCases := []string{
-		"@ 0",
-		"@ ",
-		"@",
-		"",
-		"ABC",
-		"/home/gitea/@gitea",
-		"\"@gitea\"",
-	}
-
-	for _, testCase := range trueTestCases {
-		res := mentionPattern.MatchString(testCase)
-		assert.True(t, res)
-	}
-	for _, testCase := range falseTestCases {
-		res := mentionPattern.MatchString(testCase)
-		assert.False(t, res)
-	}
-}
-
-func TestRegExp_issueAlphanumericPattern(t *testing.T) {
-	trueTestCases := []string{
-		"ABC-1234",
-		"A-1",
-		"RC-80",
-		"ABCDEFGHIJ-1234567890987654321234567890",
-		"ABC-123.",
-		"(ABC-123)",
-		"[ABC-123]",
-		"ABC-123:",
-	}
-	falseTestCases := []string{
-		"RC-08",
-		"PR-0",
-		"ABCDEFGHIJK-1",
-		"PR_1",
-		"",
-		"#ABC",
-		"",
-		"ABC",
-		"GG-",
-		"rm-1",
-		"/home/gitea/ABC-1234",
-		"MY-STRING-ABC-123",
-	}
-
-	for _, testCase := range trueTestCases {
-		assert.True(t, issueAlphanumericPattern.MatchString(testCase))
-	}
-	for _, testCase := range falseTestCases {
-		assert.False(t, issueAlphanumericPattern.MatchString(testCase))
-	}
-}
-
 func TestRegExp_shortLinkPattern(t *testing.T) {
 	trueTestCases := []string{
 		"[[stuff]]",
--- a/modules/markup/mdstripper/mdstripper.go
+++ b/modules/markup/mdstripper/mdstripper.go
@ -0,0 +1,260 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package mdstripper
+
+import (
+	"bytes"
+
+	"github.com/russross/blackfriday"
+)
+
+// MarkdownStripper extends blackfriday.Renderer
+type MarkdownStripper struct {
+	blackfriday.Renderer
+	links     []string
+	coallesce bool
+}
+
+const (
+	blackfridayExtensions = 0 |
+		blackfriday.EXTENSION_NO_INTRA_EMPHASIS |
+		blackfriday.EXTENSION_TABLES |
+		blackfriday.EXTENSION_FENCED_CODE |
+		blackfriday.EXTENSION_STRIKETHROUGH |
+		blackfriday.EXTENSION_NO_EMPTY_LINE_BEFORE_BLOCK |
+		blackfriday.EXTENSION_DEFINITION_LISTS |
+		blackfriday.EXTENSION_FOOTNOTES |
+		blackfriday.EXTENSION_HEADER_IDS |
+		blackfriday.EXTENSION_AUTO_HEADER_IDS |
+		// Not included in modules/markup/markdown/markdown.go;
+		// required here to process inline links
+		blackfriday.EXTENSION_AUTOLINK
+)
+
+//revive:disable:var-naming Implementing the Rendering interface requires breaking some linting rules
+
+// StripMarkdown parses markdown content by removing all markup and code blocks
+//	in order to extract links and other references
+func StripMarkdown(rawBytes []byte) (string, []string) {
+	stripper := &MarkdownStripper{
+		links: make([]string, 0, 10),
+	}
+	body := blackfriday.Markdown(rawBytes, stripper, blackfridayExtensions)
+	return string(body), stripper.GetLinks()
+}
+
+// StripMarkdownBytes parses markdown content by removing all markup and code blocks
+//	in order to extract links and other references
+func StripMarkdownBytes(rawBytes []byte) ([]byte, []string) {
+	stripper := &MarkdownStripper{
+		links: make([]string, 0, 10),
+	}
+	body := blackfriday.Markdown(rawBytes, stripper, blackfridayExtensions)
+	return body, stripper.GetLinks()
+}
+
+// block-level callbacks
+
+// BlockCode dummy function to proceed with rendering
+func (r *MarkdownStripper) BlockCode(out *bytes.Buffer, text []byte, infoString string) {
+	// Not rendered
+	r.coallesce = false
+}
+
+// BlockQuote dummy function to proceed with rendering
+func (r *MarkdownStripper) BlockQuote(out *bytes.Buffer, text []byte) {
+	// FIXME: perhaps it's better to leave out block quote for this?
+	r.processString(out, text, false)
+}
+
+// BlockHtml dummy function to proceed with rendering
+func (r *MarkdownStripper) BlockHtml(out *bytes.Buffer, text []byte) { //nolint
+	// Not rendered
+	r.coallesce = false
+}
+
+// Header dummy function to proceed with rendering
+func (r *MarkdownStripper) Header(out *bytes.Buffer, text func() bool, level int, id string) {
+	text()
+	r.coallesce = false
+}
+
+// HRule dummy function to proceed with rendering
+func (r *MarkdownStripper) HRule(out *bytes.Buffer) {
+	// Not rendered
+	r.coallesce = false
+}
+
+// List dummy function to proceed with rendering
+func (r *MarkdownStripper) List(out *bytes.Buffer, text func() bool, flags int) {
+	text()
+	r.coallesce = false
+}
+
+// ListItem dummy function to proceed with rendering
+func (r *MarkdownStripper) ListItem(out *bytes.Buffer, text []byte, flags int) {
+	r.processString(out, text, false)
+}
+
+// Paragraph dummy function to proceed with rendering
+func (r *MarkdownStripper) Paragraph(out *bytes.Buffer, text func() bool) {
+	text()
+	r.coallesce = false
+}
+
+// Table dummy function to proceed with rendering
+func (r *MarkdownStripper) Table(out *bytes.Buffer, header []byte, body []byte, columnData []int) {
+	r.processString(out, header, false)
+	r.processString(out, body, false)
+}
+
+// TableRow dummy function to proceed with rendering
+func (r *MarkdownStripper) TableRow(out *bytes.Buffer, text []byte) {
+	r.processString(out, text, false)
+}
+
+// TableHeaderCell dummy function to proceed with rendering
+func (r *MarkdownStripper) TableHeaderCell(out *bytes.Buffer, text []byte, flags int) {
+	r.processString(out, text, false)
+}
+
+// TableCell dummy function to proceed with rendering
+func (r *MarkdownStripper) TableCell(out *bytes.Buffer, text []byte, flags int) {
+	r.processString(out, text, false)
+}
+
+// Footnotes dummy function to proceed with rendering
+func (r *MarkdownStripper) Footnotes(out *bytes.Buffer, text func() bool) {
+	text()
+}
+
+// FootnoteItem dummy function to proceed with rendering
+func (r *MarkdownStripper) FootnoteItem(out *bytes.Buffer, name, text []byte, flags int) {
+	r.processString(out, text, false)
+}
+
+// TitleBlock dummy function to proceed with rendering
+func (r *MarkdownStripper) TitleBlock(out *bytes.Buffer, text []byte) {
+	r.processString(out, text, false)
+}
+
+// Span-level callbacks
+
+// AutoLink dummy function to proceed with rendering
+func (r *MarkdownStripper) AutoLink(out *bytes.Buffer, link []byte, kind int) {
+	r.processLink(out, link, []byte{})
+}
+
+// CodeSpan dummy function to proceed with rendering
+func (r *MarkdownStripper) CodeSpan(out *bytes.Buffer, text []byte) {
+	// Not rendered
+	r.coallesce = false
+}
+
+// DoubleEmphasis dummy function to proceed with rendering
+func (r *MarkdownStripper) DoubleEmphasis(out *bytes.Buffer, text []byte) {
+	r.processString(out, text, false)
+}
+
+// Emphasis dummy function to proceed with rendering
+func (r *MarkdownStripper) Emphasis(out *bytes.Buffer, text []byte) {
+	r.processString(out, text, false)
+}
+
+// Image dummy function to proceed with rendering
+func (r *MarkdownStripper) Image(out *bytes.Buffer, link []byte, title []byte, alt []byte) {
+	// Not rendered
+	r.coallesce = false
+}
+
+// LineBreak dummy function to proceed with rendering
+func (r *MarkdownStripper) LineBreak(out *bytes.Buffer) {
+	// Not rendered
+	r.coallesce = false
+}
+
+// Link dummy function to proceed with rendering
+func (r *MarkdownStripper) Link(out *bytes.Buffer, link []byte, title []byte, content []byte) {
+	r.processLink(out, link, content)
+}
+
+// RawHtmlTag dummy function to proceed with rendering
+func (r *MarkdownStripper) RawHtmlTag(out *bytes.Buffer, tag []byte) { //nolint
+	// Not rendered
+	r.coallesce = false
+}
+
+// TripleEmphasis dummy function to proceed with rendering
+func (r *MarkdownStripper) TripleEmphasis(out *bytes.Buffer, text []byte) {
+	r.processString(out, text, false)
+}
+
+// StrikeThrough dummy function to proceed with rendering
+func (r *MarkdownStripper) StrikeThrough(out *bytes.Buffer, text []byte) {
+	r.processString(out, text, false)
+}
+
+// FootnoteRef dummy function to proceed with rendering
+func (r *MarkdownStripper) FootnoteRef(out *bytes.Buffer, ref []byte, id int) {
+	// Not rendered
+	r.coallesce = false
+}
+
+// Low-level callbacks
+
+// Entity dummy function to proceed with rendering
+func (r *MarkdownStripper) Entity(out *bytes.Buffer, entity []byte) {
+	// FIXME: literal entities are not parsed; perhaps they should
+	r.coallesce = false
+}
+
+// NormalText dummy function to proceed with rendering
+func (r *MarkdownStripper) NormalText(out *bytes.Buffer, text []byte) {
+	r.processString(out, text, true)
+}
+
+// Header and footer
+
+// DocumentHeader dummy function to proceed with rendering
+func (r *MarkdownStripper) DocumentHeader(out *bytes.Buffer) {
+	r.coallesce = false
+}
+
+// DocumentFooter dummy function to proceed with rendering
+func (r *MarkdownStripper) DocumentFooter(out *bytes.Buffer) {
+	r.coallesce = false
+}
+
+// GetFlags returns rendering flags
+func (r *MarkdownStripper) GetFlags() int {
+	return 0
+}
+
+//revive:enable:var-naming
+
+func doubleSpace(out *bytes.Buffer) {
+	if out.Len() > 0 {
+		out.WriteByte('\n')
+	}
+}
+
+func (r *MarkdownStripper) processString(out *bytes.Buffer, text []byte, coallesce bool) {
+	// Always break-up words
+	if !coallesce || !r.coallesce {
+		doubleSpace(out)
+	}
+	out.Write(text)
+	r.coallesce = coallesce
+}
+func (r *MarkdownStripper) processLink(out *bytes.Buffer, link []byte, content []byte) {
+	// Links are processed out of band
+	r.links = append(r.links, string(link))
+	r.coallesce = false
+}
+
+// GetLinks returns the list of link data collected while parsing
+func (r *MarkdownStripper) GetLinks() []string {
+	return r.links
+}
--- a/modules/markup/mdstripper/mdstripper_test.go
+++ b/modules/markup/mdstripper/mdstripper_test.go
@ -0,0 +1,71 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package mdstripper
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestMarkdownStripper(t *testing.T) {
+	type testItem struct {
+		markdown      string
+		expectedText  []string
+		expectedLinks []string
+	}
+
+	list := []testItem{
+		{
+			`
+## This is a title
+
+This is [one](link) to paradise.
+This **is emphasized**.
+This: should coallesce.
+
+` + "```" + `
+This is a code block.
+This should not appear in the output at all.
+` + "```" + `
+
+* Bullet 1
+* Bullet 2
+
+A HIDDEN ` + "`" + `GHOST` + "`" + ` IN THIS LINE.
+		`,
+			[]string{
+				"This is a title",
+				"This is",
+				"to paradise.",
+				"This",
+				"is emphasized",
+				".",
+				"This: should coallesce.",
+				"Bullet 1",
+				"Bullet 2",
+				"A HIDDEN",
+				"IN THIS LINE.",
+			},
+			[]string{
+				"link",
+			}},
+	}
+
+	for _, test := range list {
+		text, links := StripMarkdown([]byte(test.markdown))
+		rawlines := strings.Split(text, "\n")
+		lines := make([]string, 0, len(rawlines))
+		for _, line := range rawlines {
+			line := strings.TrimSpace(line)
+			if line != "" {
+				lines = append(lines, line)
+			}
+		}
+		assert.EqualValues(t, test.expectedText, lines)
+		assert.EqualValues(t, test.expectedLinks, links)
+	}
+}
--- a/modules/markup/sanitizer.go
+++ b/modules/markup/sanitizer.go
@ -38,6 +38,9 @@ func NewSanitizer() {

 		// Custom URL-Schemes
 		sanitizer.policy.AllowURLSchemes(setting.Markdown.CustomURLSchemes...)
+
+		// Allow keyword markup
+		sanitizer.policy.AllowAttrs("class").Matching(regexp.MustCompile(`^` + keywordClass + `$`)).OnElements("span")
 	})
 }

--- a/modules/references/references.go
+++ b/modules/references/references.go
@ -0,0 +1,322 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package references
+
+import (
+	"net/url"
+	"regexp"
+	"strconv"
+	"strings"
+	"sync"
+
+	"code.gitea.io/gitea/modules/markup/mdstripper"
+	"code.gitea.io/gitea/modules/setting"
+)
+
+var (
+	// validNamePattern performs only the most basic validation for user or repository names
+	// Repository name should contain only alphanumeric, dash ('-'), underscore ('_') and dot ('.') characters.
+	validNamePattern = regexp.MustCompile(`^[a-z0-9_.-]+$`)
+
+	// NOTE: All below regex matching do not perform any extra validation.
+	// Thus a link is produced even if the linked entity does not exist.
+	// While fast, this is also incorrect and lead to false positives.
+	// TODO: fix invalid linking issue
+
+	// mentionPattern matches all mentions in the form of "@user"
+	mentionPattern = regexp.MustCompile(`(?:\s|^|\(|\[)(@[0-9a-zA-Z-_\.]+)(?:\s|$|\)|\])`)
+	// issueNumericPattern matches string that references to a numeric issue, e.g. #1287
+	issueNumericPattern = regexp.MustCompile(`(?:\s|^|\(|\[)(#[0-9]+)(?:\s|$|\)|\]|:|\.(\s|$))`)
+	// issueAlphanumericPattern matches string that references to an alphanumeric issue, e.g. ABC-1234
+	issueAlphanumericPattern = regexp.MustCompile(`(?:\s|^|\(|\[)([A-Z]{1,10}-[1-9][0-9]*)(?:\s|$|\)|\]|:|\.(\s|$))`)
+	// crossReferenceIssueNumericPattern matches string that references a numeric issue in a different repository
+	// e.g. gogits/gogs#12345
+	crossReferenceIssueNumericPattern = regexp.MustCompile(`(?:\s|^|\(|\[)([0-9a-zA-Z-_\.]+/[0-9a-zA-Z-_\.]+#[0-9]+)(?:\s|$|\)|\]|\.(\s|$))`)
+
+	// Same as GitHub. See
+	// https://help.github.com/articles/closing-issues-via-commit-messages
+	issueCloseKeywords  = []string{"close", "closes", "closed", "fix", "fixes", "fixed", "resolve", "resolves", "resolved"}
+	issueReopenKeywords = []string{"reopen", "reopens", "reopened"}
+
+	issueCloseKeywordsPat, issueReopenKeywordsPat *regexp.Regexp
+
+	giteaHostInit sync.Once
+	giteaHost     string
+)
+
+// XRefAction represents the kind of effect a cross reference has once is resolved
+type XRefAction int64
+
+const (
+	// XRefActionNone means the cross-reference is simply a comment
+	XRefActionNone XRefAction = iota // 0
+	// XRefActionCloses means the cross-reference should close an issue if it is resolved
+	XRefActionCloses // 1
+	// XRefActionReopens means the cross-reference should reopen an issue if it is resolved
+	XRefActionReopens // 2
+	// XRefActionNeutered means the cross-reference will no longer affect the source
+	XRefActionNeutered // 3
+)
+
+// IssueReference contains an unverified cross-reference to a local issue or pull request
+type IssueReference struct {
+	Index  int64
+	Owner  string
+	Name   string
+	Action XRefAction
+}
+
+// RenderizableReference contains an unverified cross-reference to with rendering information
+type RenderizableReference struct {
+	Issue          string
+	Owner          string
+	Name           string
+	RefLocation    *RefSpan
+	Action         XRefAction
+	ActionLocation *RefSpan
+}
+
+type rawReference struct {
+	index          int64
+	owner          string
+	name           string
+	action         XRefAction
+	issue          string
+	refLocation    *RefSpan
+	actionLocation *RefSpan
+}
+
+func rawToIssueReferenceList(reflist []*rawReference) []IssueReference {
+	refarr := make([]IssueReference, len(reflist))
+	for i, r := range reflist {
+		refarr[i] = IssueReference{
+			Index:  r.index,
+			Owner:  r.owner,
+			Name:   r.name,
+			Action: r.action,
+		}
+	}
+	return refarr
+}
+
+// RefSpan is the position where the reference was found within the parsed text
+type RefSpan struct {
+	Start int
+	End   int
+}
+
+func makeKeywordsPat(keywords []string) *regexp.Regexp {
+	return regexp.MustCompile(`(?i)(?:\s|^|\(|\[)(` + strings.Join(keywords, `|`) + `):? $`)
+}
+
+func init() {
+	issueCloseKeywordsPat = makeKeywordsPat(issueCloseKeywords)
+	issueReopenKeywordsPat = makeKeywordsPat(issueReopenKeywords)
+}
+
+// getGiteaHostName returns a normalized string with the local host name, with no scheme or port information
+func getGiteaHostName() string {
+	giteaHostInit.Do(func() {
+		if uapp, err := url.Parse(setting.AppURL); err == nil {
+			giteaHost = strings.ToLower(uapp.Host)
+		} else {
+			giteaHost = ""
+		}
+	})
+	return giteaHost
+}
+
+// FindAllMentionsMarkdown matches mention patterns in given content and
+// returns a list of found unvalidated user names **not including** the @ prefix.
+func FindAllMentionsMarkdown(content string) []string {
+	bcontent, _ := mdstripper.StripMarkdownBytes([]byte(content))
+	locations := FindAllMentionsBytes(bcontent)
+	mentions := make([]string, len(locations))
+	for i, val := range locations {
+		mentions[i] = string(bcontent[val.Start+1 : val.End])
+	}
+	return mentions
+}
+
+// FindAllMentionsBytes matches mention patterns in given content
+// and returns a list of locations for the unvalidated user names, including the @ prefix.
+func FindAllMentionsBytes(content []byte) []RefSpan {
+	mentions := mentionPattern.FindAllSubmatchIndex(content, -1)
+	ret := make([]RefSpan, len(mentions))
+	for i, val := range mentions {
+		ret[i] = RefSpan{Start: val[2], End: val[3]}
+	}
+	return ret
+}
+
+// FindFirstMentionBytes matches the first mention in then given content
+// and returns the location of the unvalidated user name, including the @ prefix.
+func FindFirstMentionBytes(content []byte) (bool, RefSpan) {
+	mention := mentionPattern.FindSubmatchIndex(content)
+	if mention == nil {
+		return false, RefSpan{}
+	}
+	return true, RefSpan{Start: mention[2], End: mention[3]}
+}
+
+// FindAllIssueReferencesMarkdown strips content from markdown markup
+// and returns a list of unvalidated references found in it.
+func FindAllIssueReferencesMarkdown(content string) []IssueReference {
+	return rawToIssueReferenceList(findAllIssueReferencesMarkdown(content))
+}
+
+func findAllIssueReferencesMarkdown(content string) []*rawReference {
+	bcontent, links := mdstripper.StripMarkdownBytes([]byte(content))
+	return findAllIssueReferencesBytes(bcontent, links)
+}
+
+// FindAllIssueReferences returns a list of unvalidated references found in a string.
+func FindAllIssueReferences(content string) []IssueReference {
+	return rawToIssueReferenceList(findAllIssueReferencesBytes([]byte(content), []string{}))
+}
+
+// FindRenderizableReferenceNumeric returns the first unvalidated reference found in a string.
+func FindRenderizableReferenceNumeric(content string) (bool, *RenderizableReference) {
+	match := issueNumericPattern.FindStringSubmatchIndex(content)
+	if match == nil {
+		if match = crossReferenceIssueNumericPattern.FindStringSubmatchIndex(content); match == nil {
+			return false, nil
+		}
+	}
+	r := getCrossReference([]byte(content), match[2], match[3], false)
+	if r == nil {
+		return false, nil
+	}
+
+	return true, &RenderizableReference{
+		Issue:          r.issue,
+		Owner:          r.owner,
+		Name:           r.name,
+		RefLocation:    r.refLocation,
+		Action:         r.action,
+		ActionLocation: r.actionLocation,
+	}
+}
+
+// FindRenderizableReferenceAlphanumeric returns the first alphanumeric unvalidated references found in a string.
+func FindRenderizableReferenceAlphanumeric(content string) (bool, *RenderizableReference) {
+	match := issueAlphanumericPattern.FindStringSubmatchIndex(content)
+	if match == nil {
+		return false, nil
+	}
+
+	action, location := findActionKeywords([]byte(content), match[2])
+
+	return true, &RenderizableReference{
+		Issue:          string(content[match[2]:match[3]]),
+		RefLocation:    &RefSpan{Start: match[2], End: match[3]},
+		Action:         action,
+		ActionLocation: location,
+	}
+}
+
+// FindAllIssueReferencesBytes returns a list of unvalidated references found in a byte slice.
+func findAllIssueReferencesBytes(content []byte, links []string) []*rawReference {
+
+	ret := make([]*rawReference, 0, 10)
+
+	matches := issueNumericPattern.FindAllSubmatchIndex(content, -1)
+	for _, match := range matches {
+		if ref := getCrossReference(content, match[2], match[3], false); ref != nil {
+			ret = append(ret, ref)
+		}
+	}
+
+	matches = crossReferenceIssueNumericPattern.FindAllSubmatchIndex(content, -1)
+	for _, match := range matches {
+		if ref := getCrossReference(content, match[2], match[3], false); ref != nil {
+			ret = append(ret, ref)
+		}
+	}
+
+	localhost := getGiteaHostName()
+	for _, link := range links {
+		if u, err := url.Parse(link); err == nil {
+			// Note: we're not attempting to match the URL scheme (http/https)
+			host := strings.ToLower(u.Host)
+			if host != "" && host != localhost {
+				continue
+			}
+			parts := strings.Split(u.EscapedPath(), "/")
+			// /user/repo/issues/3
+			if len(parts) != 5 || parts[0] != "" {
+				continue
+			}
+			if parts[3] != "issues" && parts[3] != "pulls" {
+				continue
+			}
+			// Note: closing/reopening keywords not supported with URLs
+			bytes := []byte(parts[1] + "/" + parts[2] + "#" + parts[4])
+			if ref := getCrossReference(bytes, 0, len(bytes), true); ref != nil {
+				ref.refLocation = nil
+				ret = append(ret, ref)
+			}
+		}
+	}
+
+	return ret
+}
+
+func getCrossReference(content []byte, start, end int, fromLink bool) *rawReference {
+	refid := string(content[start:end])
+	parts := strings.Split(refid, "#")
+	if len(parts) != 2 {
+		return nil
+	}
+	repo, issue := parts[0], parts[1]
+	index, err := strconv.ParseInt(issue, 10, 64)
+	if err != nil {
+		return nil
+	}
+	if repo == "" {
+		if fromLink {
+			// Markdown links must specify owner/repo
+			return nil
+		}
+		action, location := findActionKeywords(content, start)
+		return &rawReference{
+			index:          index,
+			action:         action,
+			issue:          issue,
+			refLocation:    &RefSpan{Start: start, End: end},
+			actionLocation: location,
+		}
+	}
+	parts = strings.Split(strings.ToLower(repo), "/")
+	if len(parts) != 2 {
+		return nil
+	}
+	owner, name := parts[0], parts[1]
+	if !validNamePattern.MatchString(owner) || !validNamePattern.MatchString(name) {
+		return nil
+	}
+	action, location := findActionKeywords(content, start)
+	return &rawReference{
+		index:          index,
+		owner:          owner,
+		name:           name,
+		action:         action,
+		issue:          issue,
+		refLocation:    &RefSpan{Start: start, End: end},
+		actionLocation: location,
+	}
+}
+
+func findActionKeywords(content []byte, start int) (XRefAction, *RefSpan) {
+	m := issueCloseKeywordsPat.FindSubmatchIndex(content[:start])
+	if m != nil {
+		return XRefActionCloses, &RefSpan{Start: m[2], End: m[3]}
+	}
+	m = issueReopenKeywordsPat.FindSubmatchIndex(content[:start])
+	if m != nil {
+		return XRefActionReopens, &RefSpan{Start: m[2], End: m[3]}
+	}
+	return XRefActionNone, nil
+}
--- a/modules/references/references_test.go
+++ b/modules/references/references_test.go
@ -0,0 +1,296 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package references
+
+import (
+	"testing"
+
+	"code.gitea.io/gitea/modules/setting"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestFindAllIssueReferences(t *testing.T) {
+
+	type result struct {
+		Index          int64
+		Owner          string
+		Name           string
+		Issue          string
+		Action         XRefAction
+		RefLocation    *RefSpan
+		ActionLocation *RefSpan
+	}
+
+	type testFixture struct {
+		input    string
+		expected []result
+	}
+
+	fixtures := []testFixture{
+		{
+			"Simply closes: #29 yes",
+			[]result{
+				{29, "", "", "29", XRefActionCloses, &RefSpan{Start: 15, End: 18}, &RefSpan{Start: 7, End: 13}},
+			},
+		},
+		{
+			"#123 no, this is a title.",
+			[]result{},
+		},
+		{
+			" #124 yes, this is a reference.",
+			[]result{
+				{124, "", "", "124", XRefActionNone, &RefSpan{Start: 0, End: 4}, nil},
+			},
+		},
+		{
+			"```\nThis is a code block.\n#723 no, it's a code block.```",
+			[]result{},
+		},
+		{
+			"This `#724` no, it's inline code.",
+			[]result{},
+		},
+		{
+			"This user3/repo4#200 yes.",
+			[]result{
+				{200, "user3", "repo4", "200", XRefActionNone, &RefSpan{Start: 5, End: 20}, nil},
+			},
+		},
+		{
+			"This [one](#919) no, this is a URL fragment.",
+			[]result{},
+		},
+		{
+			"This [two](/user2/repo1/issues/921) yes.",
+			[]result{
+				{921, "user2", "repo1", "921", XRefActionNone, nil, nil},
+			},
+		},
+		{
+			"This [three](/user2/repo1/pulls/922) yes.",
+			[]result{
+				{922, "user2", "repo1", "922", XRefActionNone, nil, nil},
+			},
+		},
+		{
+			"This [four](http://gitea.com:3000/user3/repo4/issues/203) yes.",
+			[]result{
+				{203, "user3", "repo4", "203", XRefActionNone, nil, nil},
+			},
+		},
+		{
+			"This [five](http://github.com/user3/repo4/issues/204) no.",
+			[]result{},
+		},
+		{
+			"This http://gitea.com:3000/user4/repo5/201 no, bad URL.",
+			[]result{},
+		},
+		{
+			"This http://gitea.com:3000/user4/repo5/pulls/202 yes.",
+			[]result{
+				{202, "user4", "repo5", "202", XRefActionNone, nil, nil},
+			},
+		},
+		{
+			"This http://GiTeA.COM:3000/user4/repo6/pulls/205 yes.",
+			[]result{
+				{205, "user4", "repo6", "205", XRefActionNone, nil, nil},
+			},
+		},
+		{
+			"Reopens #15 yes",
+			[]result{
+				{15, "", "", "15", XRefActionReopens, &RefSpan{Start: 8, End: 11}, &RefSpan{Start: 0, End: 7}},
+			},
+		},
+		{
+			"This closes #20 for you yes",
+			[]result{
+				{20, "", "", "20", XRefActionCloses, &RefSpan{Start: 12, End: 15}, &RefSpan{Start: 5, End: 11}},
+			},
+		},
+		{
+			"Do you fix user6/repo6#300 ? yes",
+			[]result{
+				{300, "user6", "repo6", "300", XRefActionCloses, &RefSpan{Start: 11, End: 26}, &RefSpan{Start: 7, End: 10}},
+			},
+		},
+		{
+			"For 999 #1235 no keyword, but yes",
+			[]result{
+				{1235, "", "", "1235", XRefActionNone, &RefSpan{Start: 8, End: 13}, nil},
+			},
+		},
+		{
+			"Which abc. #9434 same as above",
+			[]result{
+				{9434, "", "", "9434", XRefActionNone, &RefSpan{Start: 11, End: 16}, nil},
+			},
+		},
+		{
+			"This closes #600 and reopens #599",
+			[]result{
+				{600, "", "", "600", XRefActionCloses, &RefSpan{Start: 12, End: 16}, &RefSpan{Start: 5, End: 11}},
+				{599, "", "", "599", XRefActionReopens, &RefSpan{Start: 29, End: 33}, &RefSpan{Start: 21, End: 28}},
+			},
+		},
+	}
+
+	// Save original value for other tests that may rely on it
+	prevURL := setting.AppURL
+	setting.AppURL = "https://gitea.com:3000/"
+
+	for _, fixture := range fixtures {
+		expraw := make([]*rawReference, len(fixture.expected))
+		for i, e := range fixture.expected {
+			expraw[i] = &rawReference{
+				index:          e.Index,
+				owner:          e.Owner,
+				name:           e.Name,
+				action:         e.Action,
+				issue:          e.Issue,
+				refLocation:    e.RefLocation,
+				actionLocation: e.ActionLocation,
+			}
+		}
+		expref := rawToIssueReferenceList(expraw)
+		refs := FindAllIssueReferencesMarkdown(fixture.input)
+		assert.EqualValues(t, expref, refs, "Failed to parse: {%s}", fixture.input)
+		rawrefs := findAllIssueReferencesMarkdown(fixture.input)
+		assert.EqualValues(t, expraw, rawrefs, "Failed to parse: {%s}", fixture.input)
+	}
+
+	// Restore for other tests that may rely on the original value
+	setting.AppURL = prevURL
+
+	type alnumFixture struct {
+		input          string
+		issue          string
+		refLocation    *RefSpan
+		action         XRefAction
+		actionLocation *RefSpan
+	}
+
+	alnumFixtures := []alnumFixture{
+		{
+			"This ref ABC-123 is alphanumeric",
+			"ABC-123", &RefSpan{Start: 9, End: 16},
+			XRefActionNone, nil,
+		},
+		{
+			"This closes ABCD-1234 alphanumeric",
+			"ABCD-1234", &RefSpan{Start: 12, End: 21},
+			XRefActionCloses, &RefSpan{Start: 5, End: 11},
+		},
+	}
+
+	for _, fixture := range alnumFixtures {
+		found, ref := FindRenderizableReferenceAlphanumeric(fixture.input)
+		if fixture.issue == "" {
+			assert.False(t, found, "Failed to parse: {%s}", fixture.input)
+		} else {
+			assert.True(t, found, "Failed to parse: {%s}", fixture.input)
+			assert.Equal(t, fixture.issue, ref.Issue, "Failed to parse: {%s}", fixture.input)
+			assert.Equal(t, fixture.refLocation, ref.RefLocation, "Failed to parse: {%s}", fixture.input)
+			assert.Equal(t, fixture.action, ref.Action, "Failed to parse: {%s}", fixture.input)
+			assert.Equal(t, fixture.actionLocation, ref.ActionLocation, "Failed to parse: {%s}", fixture.input)
+		}
+	}
+}
+
+func TestRegExp_mentionPattern(t *testing.T) {
+	trueTestCases := []string{
+		"@Unknwon",
+		"@ANT_123",
+		"@xxx-DiN0-z-A..uru..s-xxx",
+		"   @lol   ",
+		" @Te-st",
+		"(@gitea)",
+		"[@gitea]",
+	}
+	falseTestCases := []string{
+		"@ 0",
+		"@ ",
+		"@",
+		"",
+		"ABC",
+		"/home/gitea/@gitea",
+		"\"@gitea\"",
+	}
+
+	for _, testCase := range trueTestCases {
+		res := mentionPattern.MatchString(testCase)
+		assert.True(t, res)
+	}
+	for _, testCase := range falseTestCases {
+		res := mentionPattern.MatchString(testCase)
+		assert.False(t, res)
+	}
+}
+
+func TestRegExp_issueNumericPattern(t *testing.T) {
+	trueTestCases := []string{
+		"#1234",
+		"#0",
+		"#1234567890987654321",
+		"  #12",
+		"#12:",
+		"ref: #12: msg",
+	}
+	falseTestCases := []string{
+		"# 1234",
+		"# 0",
+		"# ",
+		"#",
+		"#ABC",
+		"#1A2B",
+		"",
+		"ABC",
+	}
+
+	for _, testCase := range trueTestCases {
+		assert.True(t, issueNumericPattern.MatchString(testCase))
+	}
+	for _, testCase := range falseTestCases {
+		assert.False(t, issueNumericPattern.MatchString(testCase))
+	}
+}
+
+func TestRegExp_issueAlphanumericPattern(t *testing.T) {
+	trueTestCases := []string{
+		"ABC-1234",
+		"A-1",
+		"RC-80",
+		"ABCDEFGHIJ-1234567890987654321234567890",
+		"ABC-123.",
+		"(ABC-123)",
+		"[ABC-123]",
+		"ABC-123:",
+	}
+	falseTestCases := []string{
+		"RC-08",
+		"PR-0",
+		"ABCDEFGHIJK-1",
+		"PR_1",
+		"",
+		"#ABC",
+		"",
+		"ABC",
+		"GG-",
+		"rm-1",
+		"/home/gitea/ABC-1234",
+		"MY-STRING-ABC-123",
+	}
+
+	for _, testCase := range trueTestCases {
+		assert.True(t, issueAlphanumericPattern.MatchString(testCase))
+	}
+	for _, testCase := range falseTestCases {
+		assert.False(t, issueAlphanumericPattern.MatchString(testCase))
+	}
+}