[bugfix] Markdown formatting updates (#743)

* add minify dependency specifically for markdown

* rearrange markdown formatting

* update markdown tests

3 files changed, 58 insertions, 19 deletions

diff --git a/internal/api/client/status/statuscreate_test.go b/internal/api/client/status/statuscreate_test.go
index a42654a42..2a82ff8a5 100644
--- a/internal/api/client/status/statuscreate_test.go
+++ b/internal/api/client/status/statuscreate_test.go
@@ -44,7 +44,7 @@ type StatusCreateTestSuite struct {
 const (
 	statusWithLinksAndTags = "#test alright, should be able to post #links with fragments in them now, let's see........\n\nhttps://docs.gotosocial.org/en/latest/user_guide/posts/#links\n\n#gotosocial\n\n(tobi remember to pull the docker image challenge)"
 	statusMarkdown         = "# Title\n\n## Smaller title\n\nThis is a post written in [markdown](https://www.markdownguide.org/)\n\n<img src=\"https://d33wubrfki0l68.cloudfront.net/f1f475a6fda1c2c4be4cac04033db5c3293032b4/513a4/assets/images/markdown-mark-white.svg\"/>"
-	statusMarkdownExpected = "<h1>Title</h1>\n\n<h2>Smaller title</h2>\n\n<p>This is a post written in <a href=\"https://www.markdownguide.org/\" rel=\"nofollow noreferrer noopener\" target=\"_blank\">markdown</a></p>\n\n<p><img src=\"https://d33wubrfki0l68.cloudfront.net/f1f475a6fda1c2c4be4cac04033db5c3293032b4/513a4/assets/images/markdown-mark-white.svg\" crossorigin=\"anonymous\"/></p>\n"
+	statusMarkdownExpected = "<h1>Title</h1><h2>Smaller title</h2><p>This is a post written in <a href=\"https://www.markdownguide.org/\" rel=\"nofollow noreferrer noopener\" target=\"_blank\">markdown</a></p><p><img src=\"https://d33wubrfki0l68.cloudfront.net/f1f475a6fda1c2c4be4cac04033db5c3293032b4/513a4/assets/images/markdown-mark-white.svg\" crossorigin=\"anonymous\"></p>"
 )
 
 // Post a new status with some custom visibility settings
diff --git a/internal/text/markdown.go b/internal/text/markdown.go
index a5c62f23f..8952b99d6 100644
--- a/internal/text/markdown.go
+++ b/internal/text/markdown.go
@@ -23,17 +23,39 @@ import (
 
 	"github.com/russross/blackfriday/v2"
 	"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
+	"github.com/superseriousbusiness/gotosocial/internal/log"
+	"github.com/tdewolff/minify/v2"
+	"github.com/tdewolff/minify/v2/html"
 )
 
-func (f *formatter) FromMarkdown(ctx context.Context, md string, mentions []*gtsmodel.Mention, tags []*gtsmodel.Tag) string {
-	// do the markdown parsing *first*
-	contentBytes := blackfriday.Run([]byte(md))
+var m *minify.M
 
+func (f *formatter) FromMarkdown(ctx context.Context, md string, mentions []*gtsmodel.Mention, tags []*gtsmodel.Tag) string {
 	// format tags nicely
-	content := f.ReplaceTags(ctx, string(contentBytes), tags)
+	content := f.ReplaceTags(ctx, md, tags)
 
 	// format mentions nicely
 	content = f.ReplaceMentions(ctx, content, mentions)
 
-	return SanitizeHTML(content)
+	// parse markdown
+	contentBytes := blackfriday.Run([]byte(content))
+
+	// clean anything dangerous out of it
+	content = SanitizeHTML(string(contentBytes))
+
+	if m == nil {
+		m = minify.New()
+		m.Add("text/html", &html.Minifier{
+			KeepEndTags:    true,
+			KeepQuotes:     true,
+			KeepWhitespace: true,
+		})
+	}
+
+	minified, err := m.String("text/html", content)
+	if err != nil {
+		log.Errorf("error minifying markdown text: %s", err)
+	}
+
+	return minified
 }
diff --git a/internal/text/markdown_test.go b/internal/text/markdown_test.go
index 74a18a685..af4a4fef6 100644
--- a/internal/text/markdown_test.go
+++ b/internal/text/markdown_test.go
@@ -44,19 +44,23 @@ that was some JSON :)
 `
 
 const (
-	simpleMarkdown           = "# Title\n\nHere's a simple text in markdown.\n\nHere's a [link](https://example.org)."
-	simpleMarkdownExpected   = "<h1>Title</h1>\n\n<p>Here’s a simple text in markdown.</p>\n\n<p>Here’s a <a href=\"https://example.org\" rel=\"nofollow noreferrer noopener\" target=\"_blank\">link</a>.</p>\n"
-	withCodeBlockExpected    = "<h1>Title</h1>\n\n<p>Below is some JSON.</p>\n\n<pre><code class=\"language-json\">{\n  &#34;key&#34;: &#34;value&#34;,\n  &#34;another_key&#34;: [\n    &#34;value1&#34;,\n    &#34;value2&#34;\n  ]\n}\n</code></pre>\n\n<p>that was some JSON :)</p>\n"
-	withInlineCode           = "`Nobody tells you about the <code><del>SECRET CODE</del></code>, do they?`"
-	withInlineCodeExpected   = "<p><code>Nobody tells you about the &lt;code&gt;&lt;del&gt;SECRET CODE&lt;/del&gt;&lt;/code&gt;, do they?</code></p>\n"
-	withInlineCode2          = "`Nobody tells you about the </code><del>SECRET CODE</del><code>, do they?`"
-	withInlineCode2Expected  = "<p><code>Nobody tells you about the &lt;/code&gt;&lt;del&gt;SECRET CODE&lt;/del&gt;&lt;code&gt;, do they?</code></p>\n"
-	withHashtag              = "# Title\n\nhere's a simple status that uses hashtag #Hashtag!"
-	withHashtagExpected      = "<h1>Title</h1>\n\n<p>here’s a simple status that uses hashtag <a href=\"http://localhost:8080/tags/Hashtag\" class=\"mention hashtag\" rel=\"tag nofollow noreferrer noopener\" target=\"_blank\">#<span>Hashtag</span></a>!</p>\n"
-	mdWithHTML               = "# Title\n\nHere's a simple text in markdown.\n\nHere's a <a href=\"https://example.org\">link</a>.\n\nHere's an image: <img src=\"https://gts.superseriousbusiness.org/assets/logo.png\" alt=\"The GoToSocial sloth logo.\" width=\"500\" height=\"600\">"
-	mdWithHTMLExpected       = "<h1>Title</h1>\n\n<p>Here’s a simple text in markdown.</p>\n\n<p>Here’s a <a href=\"https://example.org\" rel=\"nofollow noreferrer noopener\" target=\"_blank\">link</a>.</p>\n\n<p>Here’s an image: <img src=\"https://gts.superseriousbusiness.org/assets/logo.png\" alt=\"The GoToSocial sloth logo.\" width=\"500\" height=\"600\" crossorigin=\"anonymous\"></p>\n"
-	mdWithCheekyHTML         = "# Title\n\nHere's a simple text in markdown.\n\nHere's a cheeky little script: <script>alert(ahhhh)</script>"
-	mdWithCheekyHTMLExpected = "<h1>Title</h1>\n\n<p>Here’s a simple text in markdown.</p>\n\n<p>Here’s a cheeky little script: </p>\n"
+	simpleMarkdown                  = "# Title\n\nHere's a simple text in markdown.\n\nHere's a [link](https://example.org)."
+	simpleMarkdownExpected          = "<h1>Title</h1><p>Here’s a simple text in markdown.</p><p>Here’s a <a href=\"https://example.org\" rel=\"nofollow noreferrer noopener\" target=\"_blank\">link</a>.</p>"
+	withCodeBlockExpected           = "<h1>Title</h1><p>Below is some JSON.</p><pre><code class=\"language-json\">{\n  &#34;key&#34;: &#34;value&#34;,\n  &#34;another_key&#34;: [\n    &#34;value1&#34;,\n    &#34;value2&#34;\n  ]\n}\n</code></pre><p>that was some JSON :)</p>"
+	withInlineCode                  = "`Nobody tells you about the <code><del>SECRET CODE</del></code>, do they?`"
+	withInlineCodeExpected          = "<p><code>Nobody tells you about the &lt;code>&lt;del>SECRET CODE&lt;/del>&lt;/code>, do they?</code></p>"
+	withInlineCode2                 = "`Nobody tells you about the </code><del>SECRET CODE</del><code>, do they?`"
+	withInlineCode2Expected         = "<p><code>Nobody tells you about the &lt;/code>&lt;del>SECRET CODE&lt;/del>&lt;code>, do they?</code></p>"
+	withHashtag                     = "# Title\n\nhere's a simple status that uses hashtag #Hashtag!"
+	withHashtagExpected             = "<h1>Title</h1><p>here’s a simple status that uses hashtag <a href=\"http://localhost:8080/tags/Hashtag\" class=\"mention hashtag\" rel=\"tag nofollow noreferrer noopener\" target=\"_blank\">#<span>Hashtag</span></a>!</p>"
+	mdWithHTML                      = "# Title\n\nHere's a simple text in markdown.\n\nHere's a <a href=\"https://example.org\">link</a>.\n\nHere's an image: <img src=\"https://gts.superseriousbusiness.org/assets/logo.png\" alt=\"The GoToSocial sloth logo.\" width=\"500\" height=\"600\">"
+	mdWithHTMLExpected              = "<h1>Title</h1><p>Here’s a simple text in markdown.</p><p>Here’s a <a href=\"https://example.org\" rel=\"nofollow noreferrer noopener\" target=\"_blank\">link</a>.</p><p>Here’s an image: <img src=\"https://gts.superseriousbusiness.org/assets/logo.png\" alt=\"The GoToSocial sloth logo.\" width=\"500\" height=\"600\" crossorigin=\"anonymous\"></p>"
+	mdWithCheekyHTML                = "# Title\n\nHere's a simple text in markdown.\n\nHere's a cheeky little script: <script>alert(ahhhh)</script>"
+	mdWithCheekyHTMLExpected        = "<h1>Title</h1><p>Here’s a simple text in markdown.</p><p>Here’s a cheeky little script: </p>"
+	mdWithHashtagInitial            = "#welcome #Hashtag"
+	mdWithHashtagInitialExpected    = "<p><a href=\"http://localhost:8080/tags/welcome\" class=\"mention hashtag\" rel=\"tag nofollow noreferrer noopener\" target=\"_blank\">#<span>welcome</span></a> <a href=\"http://localhost:8080/tags/Hashtag\" class=\"mention hashtag\" rel=\"tag nofollow noreferrer noopener\" target=\"_blank\">#<span>Hashtag</span></a></p>"
+	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>"
 )
 
 type MarkdownTestSuite struct {
@@ -102,6 +106,19 @@ func (suite *MarkdownTestSuite) TestParseWithCheekyHTML() {
 	suite.Equal(mdWithCheekyHTMLExpected, s)
 }
 
+func (suite *MarkdownTestSuite) TestParseWithHashtagInitial() {
+	s := suite.formatter.FromMarkdown(context.Background(), mdWithHashtagInitial, nil, []*gtsmodel.Tag{
+		suite.testTags["Hashtag"],
+		suite.testTags["welcome"],
+	})
+	suite.Equal(mdWithHashtagInitialExpected, s)
+}
+
+func (suite *MarkdownTestSuite) TestParseCodeBlockWithNewlines() {
+	s := suite.formatter.FromMarkdown(context.Background(), mdCodeBlockWithNewlines, nil, nil)
+	suite.Equal(mdCodeBlockWithNewlinesExpected, s)
+}
+
 func TestMarkdownTestSuite(t *testing.T) {
 	suite.Run(t, new(MarkdownTestSuite))
 }