Don't replace & => & when followed by character that could be an …

…entity (some browsers allow entities without closing semicolon)
tdewolff · Dec 12, 2022 · 9190118 · 9190118
1 parent f2e7ed0
commit 9190118
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 8 deletions.
diff --git a/util.go b/util.go
@@ -280,12 +280,7 @@ func replaceEntities(b []byte, i int, entitiesMap map[string][]byte, revEntities
 			} else if r[0] == '&' {
 				// check if for example &amp; is followed by something that could potentially be an entity
 				k := j + 1
-				if k < len(b) && b[k] == '#' {
-					k++
-				}
-				for ; k < len(b) && k-j <= MaxEntityLength && (b[k] >= '0' && b[k] <= '9' || b[k] >= 'a' && b[k] <= 'z' || b[k] >= 'A' && b[k] <= 'Z'); k++ {
-				}
-				if k < len(b) && b[k] == ';' {
+				if k < len(b) && (b[k] >= '0' && b[k] <= '9' || b[k] >= 'a' && b[k] <= 'z' || b[k] >= 'A' && b[k] <= 'Z' || b[k] == '#') {
 					return b, k
 				}
 			}

diff --git a/util_test.go b/util_test.go
@@ -136,10 +136,11 @@ func TestReplaceEntities(t *testing.T) {
 		{"&#39;", `&#39;`},
 		{"&amp;amp;", `&amp;amp;`},
 		{"&amp;#34;", `&amp;#34;`},
-		{"&amp;a mp;", `&a mp;`},
+		//{"&amp;a mp;", `&a mp;`},
 		{"&amp;DiacriticalAcute;", `&amp;DiacriticalAcute;`},
 		{"&amp;CounterClockwiseContourIntegral;", `&amp;CounterClockwiseContourIntegral;`},
-		{"&amp;CounterClockwiseContourIntegralL;", `&CounterClockwiseContourIntegralL;`},
+		//{"&amp;CounterClockwiseContourIntegralL;", `&CounterClockwiseContourIntegralL;`},
+		{"&amp;parameterize", `&amp;parameterize`},
 		{"&varphi;", "&phiv;"},
 		{"&varpi;", "&piv;"},
 		{"&varnone;", "&varnone;"},