@nodable/flexible-xml-parser 1.6.0 → 1.7.0

package/src/XmlPartReader.js

@@ -2,7 +2,6 @@
 import { ParseError, ErrorCode } from './ParseError.js';
 import { collectRawAttributes } from './AttributeProcessor.js';
 import { isSpace } from "./util.js"
-import { name as isName, qName as isQName } from 'xml-naming';
 // Re-export flushAttributes so Xml2JsParser and XmlSpecialTagsReader can
 // continue to import it from here without changing their import lines.
 export { flushAttributes } from './AttributeProcessor.js';
@@ -52,35 +51,24 @@ export function readTagExp(parser) {
   // begins — captured before any reads so buildTagExpObj can compute each
   // attribute's absolute document position from its offset within attrsExp.
   const expStart = parser.source.startIndex;
-  let inSingleQuotes = false;
-  let inDoubleQuotes = false;
-  let i;
-  let EOE = false;
 
-  for (i = 0; parser.source.canRead(i); i++) {
-    const char = parser.source.readChAt(i);
-
-    if (char === "'" && !inDoubleQuotes) {
-      inSingleQuotes = !inSingleQuotes;
-    } else if (char === '"' && !inSingleQuotes) {
-      inDoubleQuotes = !inDoubleQuotes;
-    } else if (char === '>' && !inSingleQuotes && !inDoubleQuotes) {
-      EOE = true;
-      break;
-    }
-  }
-
-  if (!EOE) {
-    // Buffer exhausted before '>' — chunk boundary mid-tag. Throw UNEXPECTED_END
-    // so feed()/parseStream() rewinds to the level-0 outer mark and retries.
+  const relEnd = parser.source.scanTagExpEnd();
+
+  if (relEnd === -1) {
+    // Buffer exhausted before an unquoted '>' was found — chunk boundary
+    // mid-tag. Throw UNEXPECTED_END so feed()/parseStream() rewinds to the
+    // level-0 outer mark and retries. (Note: scanTagExpEnd() only returns a
+    // non-negative index once both quote flags are already balanced-closed —
+    // by construction, not by a separate post-scan check — so there is no
+    // longer a distinct "unclosed quote but '>' was found" case to detect;
+    // the old UNCLOSED_QUOTE branch here was checking the same two flags
+    // immediately after the only code path that requires them both false,
+    // making it permanently unreachable.)
     throw new ParseError("Unexpected closing of source waiting for '>'", ErrorCode.UNEXPECTED_END);
-  } else if (inSingleQuotes || inDoubleQuotes) {
-    // '>' found but a quote was never closed — real syntax error.
-    throw new ParseError("Invalid attribute expression. Quote is not properly closed", ErrorCode.UNCLOSED_QUOTE);
   }
 
-  const exp = parser.source.readStr(i);
-  parser.source.updateBufferBoundary(i + 1);
+  const exp = parser.source.readStr(relEnd);
+  parser.source.updateBufferBoundary(relEnd + 1);
   return buildTagExpObj(exp, parser, expStart);
 }
 
@@ -183,7 +171,7 @@ function buildTagExpObj(exp, parser, expStart, forceToReadAttrs = false) {
   tagExp.tagName = tagExp.tagName.trimEnd();
   tagExp._attrsExp = attrsExp;
 
-  if (!isQName(tagExp.tagName, parser.xmlDec.version)) {
+  if (!parser.getNameValidator('qName')(tagExp.tagName)) {
     throw new ParseError("Invalid tag name", ErrorCode.INVALID_TAG_NAME);
   }
 

package/src/XmlSpecialTagsReader.js

@@ -48,6 +48,19 @@ export function readPiTag(parser) {
     if (version === '1.1') {
       parser.xmlDec.version = 1.1;
     }
+    parser.xmlDec.encoding = tagExp.rawAttributes?.encoding;
+    parser.xmlDec.standalone = tagExp.rawAttributes?.standalone;
+
+    // BUG FIX: getNameValidator('qName') was already called (and memoized)
+    // above the moment this PI tag's own name ("xml") got validated — before
+    // xmlDec.version was known, so it was always cached with the '1.0'
+    // default. Every subsequent tag/attribute name in the document —
+    // including the root element — would silently be checked against XML
+    // 1.0 rules even for a document declaring version="1.1". Reset the
+    // cache now that the real version is known; this runs at most once per
+    // document (a <?xml?> declaration can only appear once), so the cost is
+    // negligible.
+    parser._nameValidators = Object.create(null);
   }
 
   // Flush attributes into the output builder's this.attributes accumulator
@@ -55,7 +68,7 @@ export function readPiTag(parser) {
   // does for regular tags. PI tags are not pushed onto the matcher, so no
   // updateCurrent() call is needed here.
   if (!skipOptions.attributes) {
-    flushAttributes(tagExp._attrsExp, parser, tagExp._attrsExpStart);
+    flushAttributes(tagExp._parsedAttrs, parser, tagExp._attrsExpStart, tagExp._rawAttrMatchCount);
   }
 
   if (tagExp.tagName === "xml") {

package/src/CharsSymbol.js

@@ -1,16 +0,0 @@
-export default {
-  "<" : "<", //tag start
-  ">" : ">", //tag end
-  "/" : "/", //close tag
-  "!" : "!", //comment or docttype
-  "!--" : "!--", //comment
-  "-->" : "-->", //comment end
-  "?" : "?", //pi
-  "?>" : "?>", //pi end
-  "?xml" : "?xml", //pi end
-  "![" : "![", //cdata
-  "]]>" : "]]>", //cdata end
-  "[" : "[",
-  "-" : "-",
-  "D" : "D",
-}