From 90c4b5d63f2b50cbc3a228ea30ac2762781fa55d Mon Sep 17 00:00:00 2001
From: Chris Smowton <smowton@github.com>
Date: Fri, 16 Apr 2021 15:58:51 +0100
Subject: [PATCH] Switch to using HTML entities for escaping

---
 extractor/extractor.go   |  6 +++---
 extractor/trap/labels.go |  4 ++--
 extractor/util/util.go   | 13 ++++++++-----
 3 files changed, 13 insertions(+), 10 deletions(-)

diff --git a/extractor/extractor.go b/extractor/extractor.go
index dd2346728b9..6413ef6a1b6 100644
--- a/extractor/extractor.go
+++ b/extractor/extractor.go
@@ -139,7 +139,7 @@ func ExtractWithFlags(buildFlags []string, patterns []string) error {
 
 		scope := extractPackageScope(tw, pkg)
 		tw.ForEachObject(extractObjectType)
-		lbl := tw.Labeler.GlobalID(util.EscapeCurlyBraces(pkg.PkgPath) + ";pkg")
+		lbl := tw.Labeler.GlobalID(util.EscapeTrapSpecialChars(pkg.PkgPath) + ";pkg")
 		dbscheme.PackagesTable.Emit(tw, lbl, pkg.Name, pkg.PkgPath, scope)
 
 		if len(pkg.Errors) != 0 {
@@ -624,7 +624,7 @@ func (extraction *Extraction) extractFileInfo(tw *trap.Writer, file string) {
 			extraction.Lock.Unlock()
 			break
 		}
-		lbl := tw.Labeler.GlobalID(util.EscapeCurlyBraces(path) + ";folder")
+		lbl := tw.Labeler.GlobalID(util.EscapeTrapSpecialChars(path) + ";folder")
 		dbscheme.FoldersTable.Emit(tw, lbl, path, component)
 		if i > 0 {
 			dbscheme.ContainerParentTable.Emit(tw, parentLbl, lbl)
@@ -1496,7 +1496,7 @@ func getTypeLabel(tw *trap.Writer, tp types.Type) (trap.Label, bool) {
 				if field.Embedded() {
 					name = ""
 				}
-				fmt.Fprintf(&b, "%s,{%s},%s", name, fieldTypeLbl, util.EscapeCurlyBraces(tp.Tag(i)))
+				fmt.Fprintf(&b, "%s,{%s},%s", name, fieldTypeLbl, util.EscapeTrapSpecialChars(tp.Tag(i)))
 			}
 			lbl = tw.Labeler.GlobalID(fmt.Sprintf("%s;structtype", b.String()))
 		case *types.Pointer:
diff --git a/extractor/trap/labels.go b/extractor/trap/labels.go
index c6fbfa1d160..2176cdf929c 100644
--- a/extractor/trap/labels.go
+++ b/extractor/trap/labels.go
@@ -74,7 +74,7 @@ func (l *Labeler) FileLabel() Label {
 
 // FileLabelFor returns the label for the file for which the trap writer `tw` is associated
 func (l *Labeler) FileLabelFor(path string) Label {
-	return l.GlobalID(util.EscapeCurlyBraces(path) + ";sourcefile")
+	return l.GlobalID(util.EscapeTrapSpecialChars(path) + ";sourcefile")
 }
 
 // LocalID associates a label with the given AST node `nd` and returns it
@@ -103,7 +103,7 @@ func (l *Labeler) ScopeID(scope *types.Scope, pkg *types.Package) Label {
 		} else {
 			if pkg != nil && pkg.Scope() == scope {
 				// if this scope is the package scope
-				pkgLabel := l.GlobalID(util.EscapeCurlyBraces(pkg.Path()) + ";package")
+				pkgLabel := l.GlobalID(util.EscapeTrapSpecialChars(pkg.Path()) + ";package")
 				label = l.GlobalID("{" + pkgLabel.String() + "};scope")
 			} else {
 				label = l.FreshID()
diff --git a/extractor/util/util.go b/extractor/util/util.go
index 54aee83b8b0..5725c03d5b6 100644
--- a/extractor/util/util.go
+++ b/extractor/util/util.go
@@ -198,10 +198,13 @@ func GetExtractorPath() (string, error) {
 	return extractorPath, nil
 }
 
-func EscapeCurlyBraces(s string) string {
-	// Replace carets with ^caret, then curly braces with ^lcbrace and ^rcbrace.
-	s = strings.ReplaceAll(s, "^", "^caret")
-	s = strings.ReplaceAll(s, "{", "^lcbrace")
-	s = strings.ReplaceAll(s, "}", "^rcbrace")
+func EscapeTrapSpecialChars(s string) string {
+	// Replace TRAP special characters with their HTML entities, as well as '&' to avoid ambiguity.
+	s = strings.ReplaceAll(s, "&", "&amp;")
+	s = strings.ReplaceAll(s, "{", "&lbrace;")
+	s = strings.ReplaceAll(s, "}", "&rbrace;")
+	s = strings.ReplaceAll(s, "\"", "&quot;")
+	s = strings.ReplaceAll(s, "@", "&commat;")
+	s = strings.ReplaceAll(s, "#", "&num;")
 	return s
 }