[bugfix] Fix emphasis being added to emoji shortcodes with markdown parsing (#856)

* fix underscored emoji shortcodes being emphasized

* remove footnote parsing from md

internal/processing/status/create_test.go

@@ -24,6 +24,8 @@ import (
 
 	"github.com/stretchr/testify/suite"
 	"github.com/superseriousbusiness/gotosocial/internal/api/model"
+	"github.com/superseriousbusiness/gotosocial/internal/db"
+	"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
 )
 
 type StatusCreateTestSuite struct {
@@ -98,6 +100,45 @@ func (suite *StatusCreateTestSuite) TestProcessContentWarningWithHTMLEscapedQuot
 	suite.Equal("\"test\"", apiStatus.SpoilerText)
 }
 
+func (suite *StatusCreateTestSuite) TestProcessStatusMarkdownWithUnderscoreEmoji() {
+	ctx := context.Background()
+
+	// update the shortcode of the rainbow emoji to surround it in underscores
+	if err := suite.db.UpdateWhere(ctx, []db.Where{{Key: "shortcode", Value: "rainbow"}}, "shortcode", "_rainbow_", &gtsmodel.Emoji{}); err != nil {
+		suite.FailNow(err.Error())
+	}
+
+	creatingAccount := suite.testAccounts["local_account_1"]
+	creatingApplication := suite.testApplications["application_1"]
+
+	statusCreateForm := &model.AdvancedStatusCreateForm{
+		StatusCreateRequest: model.StatusCreateRequest{
+			Status:      "poopoo peepee :_rainbow_:",
+			MediaIDs:    []string{},
+			Poll:        nil,
+			InReplyToID: "",
+			Sensitive:   false,
+			Visibility:  model.VisibilityPublic,
+			ScheduledAt: "",
+			Language:    "en",
+			Format:      model.StatusFormatMarkdown,
+		},
+		AdvancedVisibilityFlagsForm: model.AdvancedVisibilityFlagsForm{
+			Federated: nil,
+			Boostable: nil,
+			Replyable: nil,
+			Likeable:  nil,
+		},
+	}
+
+	apiStatus, err := suite.status.Create(ctx, creatingAccount, creatingApplication, statusCreateForm)
+	suite.NoError(err)
+	suite.NotNil(apiStatus)
+
+	suite.Equal("<p>poopoo peepee :_rainbow_:</p>", apiStatus.Content)
+	suite.NotEmpty(apiStatus.Emojis)
+}
+
 func TestStatusCreateTestSuite(t *testing.T) {
 	suite.Run(t, new(StatusCreateTestSuite))
 }

internal/processing/status/util.go

@@ -302,7 +302,7 @@ func (p *processor) ProcessContent(ctx context.Context, form *apimodel.AdvancedS
 	case apimodel.StatusFormatPlain:
 		formatted = p.formatter.FromPlain(ctx, form.Status, status.Mentions, status.Tags)
 	case apimodel.StatusFormatMarkdown:
-		formatted = p.formatter.FromMarkdown(ctx, form.Status, status.Mentions, status.Tags)
+		formatted = p.formatter.FromMarkdown(ctx, form.Status, status.Mentions, status.Tags, status.Emojis)
 	default:
 		return fmt.Errorf("format %s not recognised as a valid status format", form.Format)
 	}

internal/text/formatter.go

@@ -30,7 +30,7 @@ type Formatter interface {
 	// FromPlain parses an HTML text from a plaintext.
 	FromPlain(ctx context.Context, plain string, mentions []*gtsmodel.Mention, tags []*gtsmodel.Tag) string
 	// FromMarkdown parses an HTML text from a markdown-formatted text.
-	FromMarkdown(ctx context.Context, md string, mentions []*gtsmodel.Mention, tags []*gtsmodel.Tag) string
+	FromMarkdown(ctx context.Context, md string, mentions []*gtsmodel.Mention, tags []*gtsmodel.Tag, emojis []*gtsmodel.Emoji) string
 
 	// ReplaceTags takes a piece of text and a slice of tags, and returns the same text with the tags nicely formatted as hrefs.
 	ReplaceTags(ctx context.Context, in string, tags []*gtsmodel.Tag) string

internal/text/markdown.go

@@ -22,6 +22,7 @@ import (
 	"bytes"
 	"context"
 	"io"
+	"strings"
 
 	"github.com/russross/blackfriday/v2"
 	"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
@@ -31,7 +32,7 @@ import (
 )
 
 var (
-	bfExtensions = blackfriday.CommonExtensions | blackfriday.HardLineBreak | blackfriday.Footnotes
+	bfExtensions = blackfriday.NoIntraEmphasis | blackfriday.FencedCode | blackfriday.Autolink | blackfriday.Strikethrough | blackfriday.SpaceHeadings | blackfriday.HardLineBreak
 	m            *minify.M
 )
 
@@ -54,8 +55,7 @@ func (r *renderer) RenderNode(w io.Writer, node *blackfriday.Node, entering bool
 		html = r.f.ReplaceMentions(r.ctx, html, r.mentions)
 
 		// we don't have much recourse if this fails
-		_, err := io.WriteString(w, html)
+		if _, err := io.WriteString(w, html); err != nil {
-		if err != nil {
 			log.Errorf("error outputting markdown text: %s", err)
 		}
 		return status
@@ -63,7 +63,7 @@ func (r *renderer) RenderNode(w io.Writer, node *blackfriday.Node, entering bool
 	return r.HTMLRenderer.RenderNode(w, node, entering)
 }
 
-func (f *formatter) FromMarkdown(ctx context.Context, md string, mentions []*gtsmodel.Mention, tags []*gtsmodel.Tag) string {
+func (f *formatter) FromMarkdown(ctx context.Context, markdownText string, mentions []*gtsmodel.Mention, tags []*gtsmodel.Tag, emojis []*gtsmodel.Emoji) string {
 
 	renderer := &renderer{
 		f:        f,
@@ -75,11 +75,28 @@ func (f *formatter) FromMarkdown(ctx context.Context, md string, mentions []*gts
 		}),
 	}
 
-	// parse markdown, use custom renderer to add hashtag/mention links
+	// Temporarily replace all found emoji shortcodes in the markdown text with
-	contentBytes := blackfriday.Run([]byte(md), blackfriday.WithExtensions(bfExtensions), blackfriday.WithRenderer(renderer))
+	// their ID so that they're not parsed as anything by the markdown parser -
+	// this fixes cases where emojis with some underscores in them are parsed as
+	// words with emphasis, eg `:_some_emoji:` becomes `:<em>some</em>emoji:`
+	//
+	// Since the IDs of the emojis are just uppercase letters + numbers they should
+	// be safe to pass through the markdown parser without unexpected effects.
+	for _, e := range emojis {
+		markdownText = strings.ReplaceAll(markdownText, ":"+e.Shortcode+":", ":"+e.ID+":")
+	}
 
-	// clean anything dangerous out of it
+	// parse markdown text into html, using custom renderer to add hashtag/mention links
-	content := SanitizeHTML(string(contentBytes))
+	htmlContentBytes := blackfriday.Run([]byte(markdownText), blackfriday.WithExtensions(bfExtensions), blackfriday.WithRenderer(renderer))
+	htmlContent := string(htmlContentBytes)
+
+	// Replace emoji IDs in the parsed html content with their shortcodes again
+	for _, e := range emojis {
+		htmlContent = strings.ReplaceAll(htmlContent, ":"+e.ID+":", ":"+e.Shortcode+":")
+	}
+
+	// clean anything dangerous out of the html
+	htmlContent = SanitizeHTML(htmlContent)
 
 	if m == nil {
 		m = minify.New()
@@ -89,7 +106,7 @@ func (f *formatter) FromMarkdown(ctx context.Context, md string, mentions []*gts
 		})
 	}
 
-	minified, err := m.String("text/html", content)
+	minified, err := m.String("text/html", htmlContent)
 	if err != nil {
 		log.Errorf("error minifying markdown text: %s", err)
 	}

internal/text/markdown_test.go

@@ -62,7 +62,7 @@ const (
 	mdCodeBlockWithNewlines         = "some code coming up\n\n```\n\n\n\n```\nthat was some code"
 	mdCodeBlockWithNewlinesExpected = "<p>some code coming up</p><pre><code>\n\n\n</code></pre><p>that was some code</p>"
 	mdWithFootnote                  = "fox mulder,fbi.[^1]\n\n[^1]: federated bureau of investigation"
-	mdWithFootnoteExpected          = "<p>fox mulder,fbi.<sup id=\"fnref:1\"><a href=\"#fn:1\" rel=\"nofollow noreferrer\">1</a></sup></p><div><hr><ol><li id=\"fn:1\">federated bureau of investigation<br></li></ol></div>"
+	mdWithFootnoteExpected          = "<p>fox mulder,fbi.[^1]</p><p>[^1]: federated bureau of investigation</p>"
 	mdWithBlockQuote                = "get ready, there's a block quote coming:\n\n>line1\n>line2\n>\n>line3\n\n"
 	mdWithBlockQuoteExpected        = "<p>get ready, there’s a block quote coming:</p><blockquote><p>line1<br>line2</p><p>line3</p></blockquote>"
 	mdHashtagAndCodeBlock           = "#Hashtag\n\n```\n#Hashtag\n```"
@@ -76,22 +76,22 @@ type MarkdownTestSuite struct {
 }
 
 func (suite *MarkdownTestSuite) TestParseSimple() {
-	s := suite.formatter.FromMarkdown(context.Background(), simpleMarkdown, nil, nil)
+	s := suite.formatter.FromMarkdown(context.Background(), simpleMarkdown, nil, nil, nil)
 	suite.Equal(simpleMarkdownExpected, s)
 }
 
 func (suite *MarkdownTestSuite) TestParseWithCodeBlock() {
-	s := suite.formatter.FromMarkdown(context.Background(), withCodeBlock, nil, nil)
+	s := suite.formatter.FromMarkdown(context.Background(), withCodeBlock, nil, nil, nil)
 	suite.Equal(withCodeBlockExpected, s)
 }
 
 func (suite *MarkdownTestSuite) TestParseWithInlineCode() {
-	s := suite.formatter.FromMarkdown(context.Background(), withInlineCode, nil, nil)
+	s := suite.formatter.FromMarkdown(context.Background(), withInlineCode, nil, nil, nil)
 	suite.Equal(withInlineCodeExpected, s)
 }
 
 func (suite *MarkdownTestSuite) TestParseWithInlineCode2() {
-	s := suite.formatter.FromMarkdown(context.Background(), withInlineCode2, nil, nil)
+	s := suite.formatter.FromMarkdown(context.Background(), withInlineCode2, nil, nil, nil)
 	suite.Equal(withInlineCode2Expected, s)
 }
 
@@ -100,17 +100,17 @@ func (suite *MarkdownTestSuite) TestParseWithHashtag() {
 		suite.testTags["Hashtag"],
 	}
 
-	s := suite.formatter.FromMarkdown(context.Background(), withHashtag, nil, foundTags)
+	s := suite.formatter.FromMarkdown(context.Background(), withHashtag, nil, foundTags, nil)
 	suite.Equal(withHashtagExpected, s)
 }
 
 func (suite *MarkdownTestSuite) TestParseWithHTML() {
-	s := suite.formatter.FromMarkdown(context.Background(), mdWithHTML, nil, nil)
+	s := suite.formatter.FromMarkdown(context.Background(), mdWithHTML, nil, nil, nil)
 	suite.Equal(mdWithHTMLExpected, s)
 }
 
 func (suite *MarkdownTestSuite) TestParseWithCheekyHTML() {
-	s := suite.formatter.FromMarkdown(context.Background(), mdWithCheekyHTML, nil, nil)
+	s := suite.formatter.FromMarkdown(context.Background(), mdWithCheekyHTML, nil, nil, nil)
 	suite.Equal(mdWithCheekyHTMLExpected, s)
 }
 
@@ -118,36 +118,36 @@ func (suite *MarkdownTestSuite) TestParseWithHashtagInitial() {
 	s := suite.formatter.FromMarkdown(context.Background(), mdWithHashtagInitial, nil, []*gtsmodel.Tag{
 		suite.testTags["Hashtag"],
 		suite.testTags["welcome"],
-	})
+	}, nil)
 	suite.Equal(mdWithHashtagInitialExpected, s)
 }
 
 func (suite *MarkdownTestSuite) TestParseCodeBlockWithNewlines() {
-	s := suite.formatter.FromMarkdown(context.Background(), mdCodeBlockWithNewlines, nil, nil)
+	s := suite.formatter.FromMarkdown(context.Background(), mdCodeBlockWithNewlines, nil, nil, nil)
 	suite.Equal(mdCodeBlockWithNewlinesExpected, s)
 }
 
 func (suite *MarkdownTestSuite) TestParseWithFootnote() {
-	s := suite.formatter.FromMarkdown(context.Background(), mdWithFootnote, nil, nil)
+	s := suite.formatter.FromMarkdown(context.Background(), mdWithFootnote, nil, nil, nil)
 	suite.Equal(mdWithFootnoteExpected, s)
 }
 
 func (suite *MarkdownTestSuite) TestParseWithBlockquote() {
-	s := suite.formatter.FromMarkdown(context.Background(), mdWithBlockQuote, nil, nil)
+	s := suite.formatter.FromMarkdown(context.Background(), mdWithBlockQuote, nil, nil, nil)
 	suite.Equal(mdWithBlockQuoteExpected, s)
 }
 
 func (suite *MarkdownTestSuite) TestParseHashtagWithCodeBlock() {
 	s := suite.formatter.FromMarkdown(context.Background(), mdHashtagAndCodeBlock, nil, []*gtsmodel.Tag{
 		suite.testTags["Hashtag"],
-	})
+	}, nil)
 	suite.Equal(mdHashtagAndCodeBlockExpected, s)
 }
 
 func (suite *MarkdownTestSuite) TestParseMentionWithCodeBlock() {
 	s := suite.formatter.FromMarkdown(context.Background(), mdMentionAndCodeBlock, []*gtsmodel.Mention{
 		suite.testMentions["local_user_2_mention_zork"],
-	}, nil)
+	}, nil, nil)
 	suite.Equal(mdMentionAndCodeBlockExpected, s)
 }