Normalize entity values in attr in xmlNodeGetContent

Fix https://gitlab.gnome.org/GNOME/libxml2/-/work_items/1118

runtest.c

@@ -5435,6 +5435,48 @@ done:
     return ret;
 }
 
+static int
+xmlEntityValueNormTest(const char *filename ATTRIBUTE_UNUSED,
+                       const char *result ATTRIBUTE_UNUSED,
+                       const char *err ATTRIBUTE_UNUSED,
+                       int options ATTRIBUTE_UNUSED) {
+    int ret = 0;
+    xmlAttrPtr prop = NULL;
+    const char *str = "<!DOCTYPE doc [\n"
+        "<!ENTITY d  '&#xD;'>\n"
+        "<!ENTITY a  '&#xA;'>\n"
+        "<!ENTITY da '&#xD;&#xA;'>\n"
+        "<!ENTITY b  '&#10;'>\n"
+        "]>\n"
+        "<doc a1='&d;' a2='&a;' a3='&da;'\n"
+        "a4='&d;&d;A&a;&#x20;&a;B&da;'\n"
+        "a5='&#xd;&#xd;A&#xa;&#xa;B&#xd;&#xa;'\n"
+        "a6='&b;'\n"
+        "a7='\n\n&b;xyz'/>";
+    xmlDoc *doc1 = xmlReadDoc(BAD_CAST str, NULL, NULL, 0);
+    xmlDoc *doc2 = xmlReadDoc(BAD_CAST str, NULL, NULL, XML_PARSE_NOENT);
+
+    prop = doc1->children->next->properties;
+    while (prop != NULL) {
+        xmlChar *content1 = xmlGetProp(doc1->children->next, BAD_CAST prop->name);
+        xmlChar *content2 = xmlGetProp(doc2->children->next, BAD_CAST prop->name);
+
+        if (!xmlStrEqual(content1, content2)) {
+            ret = 1;
+            fprintf(stderr, "entity resolution differs %s\n", prop->name);
+        }
+
+        xmlFree(content1);
+        xmlFree(content2);
+
+        prop = prop->next;
+    }
+
+    xmlFreeDoc(doc1);
+    xmlFreeDoc(doc2);
+
+    return ret;
+}
 
 /************************************************************************
  *									*
@@ -5684,6 +5726,7 @@ testDesc testDescriptions[] = {
       xmlTextReaderReadOuterXmlTest, NULL, NULL, NULL, NULL, 0 },
 #endif
     { "xmlCopyEntity children test" , xmlCopyEntityTest, NULL, NULL, NULL, NULL, 0 },
+    { "xml Entity values normalization" , xmlEntityValueNormTest, NULL, NULL, NULL, NULL, 0 },
 
     {NULL, NULL, NULL, NULL, NULL, NULL, 0}
 };

tree.c

@@ -77,7 +77,7 @@ xmlGetPropNodeInternal(const xmlNode *node, const xmlChar *name,
 static xmlChar* xmlGetPropNodeValueInternal(const xmlAttr *prop);
 
 static void
-xmlBufGetChildContent(xmlBufPtr buf, const xmlNode *tree);
+xmlBufGetChildContent(xmlBufPtr buf, const xmlNode *tree, int normalize);
 
 static void
 xmlUnlinkNodeInternal(xmlNodePtr cur);
@@ -5122,8 +5122,43 @@ xmlNodeBufGetContent(xmlBuffer *buffer, const xmlNode *cur)
     return(0);
 }
 
+/**
+ * TODO: move this to parserInternals?
+ * TODO: This function should also remove leading and trailing
+ *       whitespaces, and also group whitespaces together, but right
+ *       now the parse doesn't do that with XML_PARSE_NOENT, so this
+ *       replacement is consistent with the parser normalization
+ *
+ * Normalize attritube entity values
+ * Replaces any space character with 0x20
+ * https://www.w3.org/TR/REC-xml/#AVNormalize
+ */
+static xmlChar *
+xmlAttrNormalize(xmlChar *src)
+{
+    xmlChar *out = NULL;
+    xmlChar *dst = NULL;
+
+    if (src == NULL)
+        return(NULL);
+
+    out = src;
+    dst = out;
+    while (*src != 0) {
+        if (*src < 0x20) {
+            src++;
+            *dst++ = 0x20;
+        } else {
+            *dst++ = *src++;
+        }
+    }
+    *dst = 0;
+    return(out);
+}
+
+
 static void
-xmlBufGetEntityRefContent(xmlBufPtr buf, const xmlNode *ref) {
+xmlBufGetEntityRefContent(xmlBufPtr buf, const xmlNode *ref, int normalize) {
     xmlEntityPtr ent;
 
     if (ref->children != NULL) {
@@ -5149,23 +5184,26 @@ xmlBufGetEntityRefContent(xmlBufPtr buf, const xmlNode *ref) {
         return;
 
     ent->flags |= XML_ENT_EXPANDING;
-    xmlBufGetChildContent(buf, (xmlNodePtr) ent);
+    xmlBufGetChildContent(buf, (xmlNodePtr) ent, normalize);
     ent->flags &= ~XML_ENT_EXPANDING;
 }
 
 static void
-xmlBufGetChildContent(xmlBufPtr buf, const xmlNode *tree) {
+xmlBufGetChildContent(xmlBufPtr buf, const xmlNode *tree, int normalize) {
     const xmlNode *cur = tree->children;
 
     while (cur != NULL) {
         switch (cur->type) {
             case XML_TEXT_NODE:
             case XML_CDATA_SECTION_NODE:
-                xmlBufCat(buf, cur->content);
+                if (normalize)
+                    xmlBufCat(buf, xmlAttrNormalize(cur->content));
+                else
+                    xmlBufCat(buf, cur->content);
                 break;
 
             case XML_ENTITY_REF_NODE:
-                xmlBufGetEntityRefContent(buf, cur);
+                xmlBufGetEntityRefContent(buf, cur, normalize);
                 break;
 
             default:
@@ -5208,9 +5246,11 @@ xmlBufGetNodeContent(xmlBuf *buf, const xmlNode *cur)
         case XML_HTML_DOCUMENT_NODE:
         case XML_DOCUMENT_FRAG_NODE:
         case XML_ELEMENT_NODE:
-        case XML_ATTRIBUTE_NODE:
         case XML_ENTITY_DECL:
-            xmlBufGetChildContent(buf, cur);
+            xmlBufGetChildContent(buf, cur, 0);
+            break;
+        case XML_ATTRIBUTE_NODE:
+            xmlBufGetChildContent(buf, cur, 1);
             break;
 
         case XML_CDATA_SECTION_NODE:
@@ -5221,7 +5261,7 @@ xmlBufGetNodeContent(xmlBuf *buf, const xmlNode *cur)
             break;
 
         case XML_ENTITY_REF_NODE:
-            xmlBufGetEntityRefContent(buf, cur);
+            xmlBufGetEntityRefContent(buf, cur, 0);
             break;
 
         case XML_NAMESPACE_DECL: