Text/status parsing fixes (#141)

* aaaaaa

* vendor minify

* update + test markdown parsing

1 files changed, 22 insertions, 11 deletions

diff --git a/internal/text/common.go b/internal/text/common.go
index 4f0bad9dc..f6a5ca5f5 100644
--- a/internal/text/common.go
+++ b/internal/text/common.go
@@ -20,6 +20,7 @@ package text
 
 import (
 	"fmt"
+	"html"
 	"strings"
 
 	"github.com/superseriousbusiness/gotosocial/internal/gtsmodel"
@@ -29,23 +30,33 @@ import (
 // preformat contains some common logic for making a string ready for formatting, which should be used for all user-input text.
 func preformat(in string) string {
 	// do some preformatting of the text
-	// 1. Trim all the whitespace
-	s := strings.TrimSpace(in)
+
+	// 1. unescape everything that might be html escaped
+	s := html.UnescapeString(in)
+
+	// 2. trim leading or trailing whitespace
+	s = strings.TrimSpace(s)
 	return s
 }
 
 // postformat contains some common logic for html sanitization of text, wrapping elements, and trimming newlines and whitespace
 func postformat(in string) string {
 	// do some postformatting of the text
-	// 1. sanitize html to remove any dodgy scripts or other disallowed elements
-	s := SanitizeOutgoing(in)
-	// 2. wrap the whole thing in a paragraph
-	s = fmt.Sprintf(`<p>%s</p>`, s)
-	// 3. remove any cheeky newlines
-	s = strings.ReplaceAll(s, "\n", "")
-	// 4. remove any whitespace added as a result of the formatting
-	s = strings.TrimSpace(s)
-	return s
+
+	// 1. sanitize html to remove potentially dangerous elements
+	s := SanitizeHTML(in)
+
+	// 2. the sanitize step tends to escape characters inside codeblocks, which is behavior we don't want, so unescape everything again
+	s = html.UnescapeString(s)
+
+	// 3. minify html to remove any trailing newlines, spaces, unnecessary elements, etc etc
+	mini, err := minifyHTML(s)
+	if err != nil {
+		// if the minify failed, just return what we have
+		return s
+	}
+	// return minified version of the html
+	return mini
 }
 
 func (f *formatter) ReplaceTags(in string, tags []*gtsmodel.Tag) string {