fix: just one error for text before the root, and text after

Previously the parser would raise an error for every... single... non-whitespace... character that appeared outside the root. That's too much. We now raise the error at most once before the root, and at most once after.
lddubeau · Jun 30, 2018 · 101ea50 · 101ea50
1 parent c03c7d0
commit 101ea50
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 25 deletions.
diff --git a/lib/saxes.js b/lib/saxes.js
@@ -55,7 +55,7 @@
     parser.closed = parser.closedRoot = parser.sawRoot = false
     parser.tag = parser.error = null
     parser.state = S.BEGIN
-    parser.ENTITIES = Object.create(sax.XML_ENTITIES);
+    parser.ENTITIES = Object.create(sax.XML_ENTITIES)
     parser.attribList = []
 
     // namespaces form a prototype chain.
@@ -323,7 +323,7 @@
     ATTRIB_VALUE_ENTITY_Q: S++, // <foo bar="&quot;"
     ATTRIB_VALUE_ENTITY_U: S++, // <foo bar=&quot
     CLOSE_TAG: S++, // </a
-    CLOSE_TAG_SAW_WHITE: S++, // </a   >
+    CLOSE_TAG_SAW_WHITE: S++ // </a   >
   }
 
   sax.XML_ENTITIES = {
@@ -713,8 +713,19 @@
             parser.state = S.OPEN_WAKA
             parser.startTagPosition = parser.position
           } else {
-            if (!isWhitespace(c) && (!parser.sawRoot || parser.closedRoot)) {
-              fail(parser, 'Text data outside of root node.')
+            if (!isWhitespace(c)) {
+              // We use the reportedTextBeforeRoot and
+              // reportedTextAfterRoot flags to avoid reporting errors
+              // for every single character that is out of place.
+              if (!parser.sawRoot && !parser.reportedTextBeforeRoot) {
+                fail(parser, 'Text data outside of root node.')
+                parser.reportedTextBeforeRoot = true
+              }
+
+              if (parser.closedRoot && !parser.reportedTextAfterRoot) {
+                fail(parser, 'Text data outside of root node.')
+                parser.reportedTextAfterRoot = true
+              }
             }
             if (c === '&') {
               parser.state = S.TEXT_ENTITY

diff --git a/test/issue-86.js b/test/issue-86.js
@@ -30,20 +30,12 @@ require(__dirname).test({
     ],
     [
       'text',
-      'd'
-    ],
-    [
-      'error',
-      'Text data outside of root node.\nLine: 0\nColumn: 18\nChar: e'
-    ],
-    [
-      'text',
-      'e'
+      'de'
     ],
     [
       'error',
       'Unexpected end\nLine: 0\nColumn: 20\nChar: '
-    ]
+    ],
   ],
   opt: {}
 })
diff --git a/test/trailing-non-whitespace.js b/test/trailing-non-whitespace.js
@@ -17,17 +17,7 @@ require(__dirname).test({
 Line: 0\n\
 Column: 23\n\
 Char: t'],
-    ['text', 't'],
-    ['error', 'Text data outside of root node.\n\
-Line: 0\n\
-Column: 24\n\
-Char: o'],
-    ['text', 'o '],
-    ['error', 'Text data outside of root node.\n\
-Line: 0\n\
-Column: 26\n\
-Char: m'],
-    ['text', 'm'],
+    ['text', 'to monkey land'],
     ['end'],
     ['ready']
   ],