fix: ignore unparsed entity logic for std xml entities

PR-URL: #266 Credit: @SethFalco Close: #266 Reviewed-by: @HeikoTheissen
isaacs · May 27, 2024 · 93a63dd · 93a63dd
1 parent b577600
commit 93a63dd
Show file tree

Hide file tree

Showing 4 changed files with 63 additions and 3 deletions.
diff --git a/lib/sax.js b/lib/sax.js
@@ -1483,13 +1483,13 @@
           }
 
           if (c === ';') {
-            if (parser.opt.unparsedEntities) {
-              var parsedEntity = parseEntity(parser)
+            var parsedEntity = parseEntity(parser)
+            if (parser.opt.unparsedEntities && !Object.values(sax.XML_ENTITIES).includes(parsedEntity)) {
               parser.entity = ''
               parser.state = returnState
               parser.write(parsedEntity)
             } else {
-              parser[buffer] += parseEntity(parser)
+              parser[buffer] += parsedEntity
               parser.entity = ''
               parser.state = returnState
             }

diff --git a/test/unparsed-entities-amp.js b/test/unparsed-entities-amp.js
@@ -0,0 +1,17 @@
+require(__dirname).test({
+  opt: {unparsedEntities: true},
+  xml: '<svg>' +
+    '<text>' +
+    '&amp;' +
+    '</text>' +
+    '</svg>',
+  expect: [
+    ['opentagstart', {'name': 'SVG', attributes: {}}],
+    ['opentag', {'name': 'SVG', attributes: {}, isSelfClosing: false}],
+    ['opentagstart', {'name': 'TEXT', attributes: {}}],
+    ['opentag', {'name': 'TEXT', attributes: {}, isSelfClosing: false}],
+    ['text', '&'],
+    ['closetag', 'TEXT'],
+    ['closetag', 'SVG']
+  ]
+})
diff --git a/test/unparsed-entities-predefind-in-custom.js b/test/unparsed-entities-predefind-in-custom.js
@@ -0,0 +1,32 @@
+var sax = require('../lib/sax');
+sax.ENTITIES.entity_reference = "<text>entity reference</text>";
+sax.ENTITIES.escaped_entity_reference = "&lt;text&gt;escaped entity reference&lt;/text&gt;";
+require(__dirname).test({
+  opt: {unparsedEntities: true},
+  xml: '<svg>' +
+    '<text>' +
+    '&lt;text&gt;escaped literal&lt;/text&gt;' +
+    '</text>' +
+    '&entity_reference;' +
+    '<text>' +
+    '&escaped_entity_reference;' +
+    '</text>' +
+    '</svg>',
+  expect: [
+    ['opentagstart', {'name': 'SVG', attributes: {}}],
+    ['opentag', {'name': 'SVG', attributes: {}, isSelfClosing: false}],
+    ['opentagstart', {'name': 'TEXT', attributes: {}}],
+    ['opentag', {'name': 'TEXT', attributes: {}, isSelfClosing: false}],
+    ['text', '<text>escaped literal</text>'],
+    ['closetag', 'TEXT'],
+    ['opentagstart', {'name': 'TEXT', attributes: {}}],
+    ['opentag', {'name': 'TEXT', attributes: {}, isSelfClosing: false}],
+    ['text', 'entity reference'],
+    ['closetag', 'TEXT'],
+    ['opentagstart', {'name': 'TEXT', attributes: {}}],
+    ['opentag', {'name': 'TEXT', attributes: {}, isSelfClosing: false}],
+    ['text', '<text>escaped entity reference</text>'],
+    ['closetag', 'TEXT'],
+    ['closetag', 'SVG']
+  ]
+})
diff --git a/test/unparsed-entities-quote-in-attr.js b/test/unparsed-entities-quote-in-attr.js
@@ -0,0 +1,11 @@
+require(__dirname).test({
+  opt: {unparsedEntities: true},
+  xml: '<doc a="&#34;">' +
+    '</doc>',
+  expect: [
+    ['opentagstart', {'name': 'DOC', attributes: {}}],
+    ['attribute', { name: 'A', value: '"'} ],
+    ['opentag', {'name': 'DOC', attributes: {A: '"'}, isSelfClosing: false}],
+    ['closetag', 'DOC']
+  ]
+})