From 1f0c261fd2d3da250f62040236f7629f603f0b03 Mon Sep 17 00:00:00 2001 From: tobi Date: Wed, 7 May 2025 11:48:21 +0000 Subject: [PATCH] [bugfix] Fix indentation on multi-line alt text in web view (#4149) This pull request closes https://codeberg.org/superseriousbusiness/gotosocial/issues/3403 by searching for alt or title attributes in HTML, looking for the shortest indent (if any) at the start of multiline content of said attributes, and remove that shortest indent from each line. This is a bit more fiddly than the "easy" way of doing it, but it has the advantage that it preserves user-added indents at the start of lines of alt text. Reviewed-on: https://codeberg.org/superseriousbusiness/gotosocial/pulls/4149 Co-authored-by: tobi Co-committed-by: tobi --- internal/router/template.go | 158 ++++++++++++++++++++++++------- internal/router/template_test.go | 46 +++++++-- web/template/page.tmpl | 2 +- 3 files changed, 167 insertions(+), 39 deletions(-) diff --git a/internal/router/template.go b/internal/router/template.go index 0bae96548..1e3b2e715 100644 --- a/internal/router/template.go +++ b/internal/router/template.go @@ -25,7 +25,9 @@ import ( "path/filepath" "reflect" "regexp" + "slices" "strings" + "sync" "unsafe" apimodel "code.superseriousbusiness.org/gotosocial/internal/api/model" @@ -134,25 +136,25 @@ func LoadTemplates(engine *gin.Engine) error { } var funcMap = template.FuncMap{ - "add": add, - "acctInstance": acctInstance, - "objectPosition": objectPosition, - "demojify": demojify, - "deref": deref, - "emojify": emojify, - "escape": escape, - "increment": increment, - "indent": indent, - "indentAttr": indentAttr, - "isNil": isNil, - "outdentPre": outdentPre, - "noescapeAttr": noescapeAttr, - "noescape": noescape, - "oddOrEven": oddOrEven, - "subtract": subtract, - "timestampPrecise": timestampPrecise, - "timestampVague": timestampVague, - "visibilityIcon": visibilityIcon, + "add": add, + "acctInstance": acctInstance, + "objectPosition": objectPosition, + "demojify": demojify, + "deref": deref, + "emojify": emojify, + "escape": escape, + "increment": increment, + "indent": indent, + "indentAttr": indentAttr, + "isNil": isNil, + "outdentPreformatted": outdentPreformatted, + "noescapeAttr": noescapeAttr, + "noescape": noescape, + "oddOrEven": oddOrEven, + "subtract": subtract, + "timestampPrecise": timestampPrecise, + "timestampVague": timestampVague, + "visibilityIcon": visibilityIcon, } func oddOrEven(n int) string { @@ -291,11 +293,31 @@ func subtract(n1 int, n2 int) int { } var ( - indentRegex = regexp.MustCompile(`(?m)^`) + // Find starts of lines to replace with indent. + indentRegex = regexp.MustCompile(`(?m)^`) + + // One indent level. indentStr = " " indentStrLen = len(indentStr) - indents = strings.Repeat(indentStr, 12) - indentPre = regexp.MustCompile(fmt.Sprintf(`(?Ums)^((?:%s)+)
.*
`, indentStr)) + + // Preformatted slice of indents. + indents = strings.Repeat(indentStr, 12) + + // Measure indent at the start of a line. + indentDepthStr = fmt.Sprintf(`^((?:%s)+)`, indentStr) + indentDepth = regexp.MustCompile(`(?m)` + indentDepthStr) + + // Find 
 tags and determine how indented they are.
+	indentPre = regexp.MustCompile(fmt.Sprintf(`(?Ums)%s
.*
`, indentDepthStr))
+	// Find content of alt or title attributes.
+	indentAltOrTitle = regexp.MustCompile(`(?Ums)\b(?:alt|title)="(.*)"(?:\b|>|$)`)
+
+	// Map of lazily-compiled replaceIndent
+	// regexes, keyed by the indent they
+	// replace, to avoid recompilation.
+	//
+	// At *most* 12 entries long.
+	replaceIndents = sync.Map{}
 )
 
 // indent appropriately indents the given html
@@ -318,32 +340,104 @@ func indentAttr(n int, html template.HTMLAttr) template.HTMLAttr {
 	return noescapeAttr(out)
 }
 
-// outdentPre outdents all `
` tags in the
-// given HTML so that they render correctly in code
-// blocks, even if they were indented before.
-func outdentPre(html template.HTML) template.HTML {
+// outdentPreformatted outdents all preformatted text in
+// the given HTML, ie., in `alt` and `title` attributes,
+// and between `
` tags, so that it renders correctly,
+// even if it was indented before.
+func outdentPreformatted(html template.HTML) template.HTML {
 	input := string(html)
 	output := regexes.ReplaceAllStringFunc(indentPre, input,
 		func(match string, buf *bytes.Buffer) string {
 			// Reuse the regex to pull out submatches.
 			matches := indentPre.FindAllStringSubmatch(match, -1)
+
+			// Ensure matches
+			// expected length.
 			if len(matches) != 1 {
 				return match
 			}
 
+			// Ensure inner matches
+			// expected length.
+			innerMatches := matches[0]
+			if len(innerMatches) != 2 {
+				return match
+			}
+
 			var (
-				indented = matches[0][0]
-				indent   = matches[0][1]
+				indentedContent = innerMatches[0]
+				indent          = innerMatches[1]
 			)
 
-			// Outdent everything in the inner match, add
-			// a newline at the end to make it a bit neater.
-			outdented := strings.ReplaceAll(indented, indent, "")
+			// Outdent everything in the inner match.
+			outdented := strings.ReplaceAll(indentedContent, indent, "")
 
 			// Replace original match with the outdented version.
-			return strings.ReplaceAll(match, indented, outdented)
+			return strings.ReplaceAll(match, indentedContent, outdented)
 		},
 	)
+
+	output = regexes.ReplaceAllStringFunc(indentAltOrTitle, output,
+		func(match string, buf *bytes.Buffer) string {
+			// Reuse the regex to pull out submatches.
+			matches := indentAltOrTitle.FindAllStringSubmatch(match, -1)
+
+			// Ensure matches
+			// expected length.
+			if len(matches) != 1 {
+				return match
+			}
+
+			// Ensure inner matches
+			// expected length.
+			innerMatches := matches[0]
+			if len(innerMatches) != 2 {
+				return match
+			}
+
+			// The content of the alt or title
+			// attr inside quotation marks.
+			indentedContent := innerMatches[1]
+
+			// Find all indents in this text.
+			indents := indentDepth.FindAllString(indentedContent, -1)
+			if len(indents) == 0 {
+				// No indents in this text,
+				// it's probably just something
+				// inline like `alt="whatever"`.
+				return match
+			}
+
+			// Find the shortest indent as this
+			// is undoubtedly the one we added.
+			//
+			// By targeting the shortest one we
+			// avoid removing user-inserted
+			// whitespace at the start of lines
+			// of alt text (eg., in poetry etc).
+			slices.Sort(indents)
+			indent := indents[0]
+
+			// Load or create + store the
+			// regex to replace this indent,
+			// avoiding recompilation.
+			var replaceIndent *regexp.Regexp
+			if replaceIndentI, ok := replaceIndents.Load(indent); ok {
+				// Got regex for this indent.
+				replaceIndent = replaceIndentI.(*regexp.Regexp)
+			} else {
+				// No regex stored for
+				// this indent yet, store it.
+				replaceIndent = regexp.MustCompile(`(?m)^` + indent)
+				replaceIndents.Store(indent, replaceIndent)
+			}
+
+			// Remove all occurrences of the indent
+			// at the start of a line in the match.
+			return replaceIndent.ReplaceAllString(match, "")
+		},
+	)
+
 	return noescape(output)
 }
 
diff --git a/internal/router/template_test.go b/internal/router/template_test.go
index 19bf759e0..1c82d3ba4 100644
--- a/internal/router/template_test.go
+++ b/internal/router/template_test.go
@@ -22,10 +22,19 @@ import (
 	"testing"
 )
 
-func TestOutdentPre(t *testing.T) {
+func TestOutdentPreformatted(t *testing.T) {
 	const html = template.HTML(`
         

-            
                
+            

                 
Here's a bunch of HTML, read it and weep, weep then!

                 
<section class="about-user">
                     <div class="col-header">
@@ -67,7 +76,15 @@ func TestOutdentPre(t *testing.T) {
             

         

         

-            
                
+            

                 
Here's a bunch of HTML, read it and weep, weep then!

                 
<section class="about-user">
                     <div class="col-header">
@@ -112,7 +129,16 @@ func TestOutdentPre(t *testing.T) {
 
 	const expected = template.HTML(`
         

-            
                
+            

                 
Here's a bunch of HTML, read it and weep, weep then!

 
<section class="about-user">
     <div class="col-header">
@@ -154,7 +180,15 @@ func TestOutdentPre(t *testing.T) {
             

         

         

-            
                
+            

                 
Here's a bunch of HTML, read it and weep, weep then!

 
<section class="about-user">
     <div class="col-header">
@@ -197,7 +231,7 @@ func TestOutdentPre(t *testing.T) {
         

 `)
 
-	out := outdentPre(html)
+	out := outdentPreformatted(html)
 	if out != expected {
 		t.Fatalf("unexpected output:\n`%s`\n", out)
 	}
diff --git a/web/template/page.tmpl b/web/template/page.tmpl
index 0a54e74cb..4ea168300 100644
--- a/web/template/page.tmpl
+++ b/web/template/page.tmpl
@@ -79,7 +79,7 @@ image/webp
             {{- include "page_header.tmpl" . | indent 3 }}
         
         

-            {{- include .pageContent . | indent 3 | outdentPre }}
+            {{- include .pageContent . | indent 3 | outdentPreformatted }}