commit 8696ebe182b9867cbded474e1a977664d3e0e144
author Nick Wellnhofer <wellnhofer@aevum.de> Tue Mar 11 14:32:35 2025 +0100
committer Nick Wellnhofer <wellnhofer@aevum.de> Tue Mar 11 16:34:30 2025 +0100
tree 53a2fe2e0770d7755136af55d35f53d2e5951fd8
parent d83ff954af4693fca75834fc56f27f635ce20e75

parser: Fix ignorableWhitespace callback

If ignorableWhitespace differs from the "characters" callback, we have
to check for blanks as well.

Regressed with 1f5b537.

SAX2.c

diff --git a/SAX2.c b/SAX2.c
index 436151e..b5c2e47 100644
--- a/SAX2.c
+++ b/SAX2.c
@@ -2736,7 +2736,7 @@
     hdlr->reference = xmlSAX2Reference;
     hdlr->characters = xmlSAX2Characters;
     hdlr->cdataBlock = xmlSAX2CDataBlock;
-    hdlr->ignorableWhitespace = xmlSAX2IgnorableWhitespace;
+    hdlr->ignorableWhitespace = xmlSAX2Characters;
     hdlr->processingInstruction = xmlSAX2ProcessingInstruction;
     hdlr->comment = xmlSAX2Comment;
     hdlr->warning = xmlParserWarning;

parser.c

diff --git a/parser.c b/parser.c
index 42684e0..8a26968 100644
--- a/parser.c
+++ b/parser.c
@@ -4779,18 +4779,23 @@
 
 static void
 xmlCharacters(xmlParserCtxtPtr ctxt, const xmlChar *buf, int size) {
+    int checkBlanks;
+
     if ((ctxt->sax == NULL) || (ctxt->disableSAX))
         return;
 
+    checkBlanks = (!ctxt->keepBlanks) ||
+                  (ctxt->sax->ignorableWhitespace != ctxt->sax->characters);
+
     /*
      * Calling areBlanks with only parts of a text node
      * is fundamentally broken, making the NOBLANKS option
      * essentially unusable.
      */
-    if ((!ctxt->keepBlanks) &&
-        (ctxt->sax->ignorableWhitespace != ctxt->sax->characters) &&
+    if ((checkBlanks) &&
         (areBlanks(ctxt, buf, size, 1))) {
-        if (ctxt->sax->ignorableWhitespace != NULL)
+        if ((ctxt->sax->ignorableWhitespace != NULL) &&
+            (ctxt->keepBlanks))
             ctxt->sax->ignorableWhitespace(ctxt->userData, buf, size);
     } else {
         if (ctxt->sax->characters != NULL)
@@ -4798,9 +4803,9 @@
 
         /*
          * The old code used to update this value for "complex" data
-         * even if keepBlanks was true. This was probably a bug.
+         * even if checkBlanks was false. This was probably a bug.
          */
-        if ((!ctxt->keepBlanks) && (*ctxt->space == -1))
+        if ((checkBlanks) && (*ctxt->space == -1))
             *ctxt->space = -2;
     }
 }