Update to 2.13.5

NEWS

@@ -1,5 +1,21 @@
 NEWS file for libxml2
 
+v2.13.5: Nov 12 2024
+
+### Regressions
+
+- xmlIO: Fix reading from non-regular files like pipes (Nick Wellnhofer)
+- xmlreader: Fix return value of xmlTextReaderReadString (Nick Wellnhofer)
+- parser: Fix loading of parameter entities in external DTDs (Nick Wellnhofer)
+- parser: Fix downstream code that swaps DTDs (Nick Wellnhofer)
+- parser: Fix detection of duplicate attributes (Nick Wellnhofer)
+- string: Fix va_copy fallback (Nick Wellnhofer)
+
+### Bug fixes
+
+- xpath: Fix parsing of non-ASCII names (Nick Wellnhofer)
+
+
 v2.13.4: Sep 18 2024
 
 ### Regressions

README.md

@@ -2,7 +2,7 @@
 
 libxml2 Windows build with Visual Studio.
 
-This version is libxml2-2.13.4.
+This version is libxml2-2.13.5.
 
 Note that LZMA support is only available for VS2013 or later.
 

distfiles/download.url

@@ -1 +1 @@
-https://download.gnome.org/sources/libxml2/2.13/libxml2-2.13.4.tar.xz
+https://download.gnome.org/sources/libxml2/2.13/libxml2-2.13.5.tar.xz

distfiles/libxml2-2.13.4-import.md5 → distfiles/libxml2-2.13.5-import.md5

@@ -2,7 +2,7 @@ f437ed9058e8e5135e47c01e973376ba  Copyright
 799d00dfaca59a88a1ddcb7e564e6476  HTMLparser.c
 24211653eb3b5d6752f5eb91d6ecc3ef  HTMLtree.c
 dd63184811cb2ff705c3e466364d3773  INSTALL
-3b128a9995eb5addcc86f1c5984b2e83  NEWS
+7a926a25d2fe5ccd55e217e00ec1cb86  NEWS
 5f32c16a4eccf442197b65fa65f3c7b8  README.libxml2.md
 7283878f36a935a3c00df077cf45af54  SAX.c
 f2edb7da1ddb9093de9841519b68f576  SAX2.c
@@ -86,7 +86,7 @@ cdbcf52ea11b6ee99454e3b9a3adeaac  legacy.c
 62f33a8621e3442770fd15a540a7eba0  list.c
 040942573dbd47e7188991ab3c9c9a99  nanoftp.c
 4c676ca8672af9c242eab69fb9e2056f  nanohttp.c
-6f4d03099f4c60600f6640aea424356c  parser.c
+1a9b6424fcc1dd7a5d1d3b8df25894f5  parser.c
 47ca05f0faf204f7d1a308cd24a80111  parserInternals.c
 fc88d174a7b70de62c609c32ce3f55f8  pattern.c
 06b7f056c759cff032979e0075c5b318  relaxng.c
@@ -101,7 +101,7 @@ c48cacc169fbe69e961ceafbfb92ee71  testModule.c
 8efef0b6535d6c069678e9f6750d3742  testdict.c
 6a3e7cbf9864c04639a1a8ac0c388ea2  testdso.c
 e7f8098f4a9e147624c3cf7d652a70c0  testlimits.c
-b6224e455a78abefae0a5a1d128ed789  testparser.c
+5c573ad30e2cf82094729769850c4366  testparser.c
 71ea68a83739869caba574c1725fba96  testrecurse.c
 d746403de87ca28dbb43a4a76e63a3d6  threads.c
 1bcb15667ab695cdd2cc8d5b1bc05169  timsort.h
@@ -112,19 +112,19 @@ ff757d7c377b585c520e7ef78551ed87  valid.c
 9ca0965eeabe09b4f8d9a1c6c5d8c3b0  win32/libxml2.rc
 98e142c0d1c661f7604934d31f6cc0ec  xinclude.c
 0a034450d155e35ec8ba99ee2005b695  xlink.c
-44b28c907e7824182b409165130cf971  xmlIO.c
+38e6fc4ba7ac7f5cfb9ab7b3dbffc517  xmlIO.c
 7a41f0136e0179fafba0d6e9a4514d73  xmlcatalog.c
 7eeb0736114d53f08e4be4d611e579dd  xmllint.c
 9a191c58eb3f035c795898caea8e689b  xmlmemory.c
 9d8d0df11b80a714e97da726359977dc  xmlmodule.c
-e8fa4ddf7a04cf784396eccbe8381224  xmlreader.c
+1f17f7c808f45a7d24233f21a4a087e5  xmlreader.c
 fda6796d6766ba055664f2993eca5ed7  xmlregexp.c
 d877f026dd7130ed114ef4715be479bf  xmlsave.c
 7ad15c4a469cee12b90af8115dd76b5e  xmlschemas.c
 991c4949ca7a25c99a7beff8e7516915  xmlschemastypes.c
-82c2b7394f560d8af467b779026d3972  xmlstring.c
+b7ff570f5e58afc8fc939282e6707108  xmlstring.c
 c9097be7e143f99684f19d937a775144  xmlunicode.c
-a8ab2ac1eb6a00dccdae814c1cfcea68  xmlwriter.c
-be75156569c7a3cfcca092ed459db66c  xpath.c
+9bc7467b012af0f01626c8eb38e59eed  xmlwriter.c
+fa4329f88bf47c174874486f6093a0b5  xpath.c
 dcf0c2434204059b5d6f03346dec1004  xpointer.c
 8798ea780f20ac559991cf1710c29087  xzlib.c

include/libxml/xmlversion.h

@@ -15,21 +15,21 @@
  *
  * the version string like "1.2.3"
  */
-#define LIBXML_DOTTED_VERSION "2.13.4"
+#define LIBXML_DOTTED_VERSION "2.13.5"
 
 /**
  * LIBXML_VERSION:
  *
  * the version number: 1.2.3 value is 10203
  */
-#define LIBXML_VERSION 21304
+#define LIBXML_VERSION 21305
 
 /**
  * LIBXML_VERSION_STRING:
  *
  * the version number string, 1.2.3 value is "10203"
  */
-#define LIBXML_VERSION_STRING "21304"
+#define LIBXML_VERSION_STRING "21305"
 
 /**
  * LIBXML_VERSION_EXTRA:
@@ -44,7 +44,7 @@
  * Macro to check that the libxml version in use is compatible with
  * the version the software has been compiled against
  */
-#define LIBXML_TEST_VERSION xmlCheckVersion(21304);
+#define LIBXML_TEST_VERSION xmlCheckVersion(21305);
 
 /**
  * LIBXML_THREAD_ENABLED:

parser.c

@@ -8912,6 +8912,35 @@ xmlAttrHashInsert(xmlParserCtxtPtr ctxt, unsigned size, const xmlChar *name,
     return(INT_MAX);
 }
 
+static int
+xmlAttrHashInsertQName(xmlParserCtxtPtr ctxt, unsigned size,
+                       const xmlChar *name, const xmlChar *prefix,
+                       unsigned hashValue, int aindex) {
+    xmlAttrHashBucket *table = ctxt->attrHash;
+    xmlAttrHashBucket *bucket;
+    unsigned hindex;
+
+    hindex = hashValue & (size - 1);
+    bucket = &table[hindex];
+
+    while (bucket->index >= 0) {
+        const xmlChar **atts = &ctxt->atts[bucket->index];
+
+        if ((name == atts[0]) && (prefix == atts[1]))
+            return(bucket->index);
+
+        hindex++;
+        bucket++;
+        if (hindex >= size) {
+            hindex = 0;
+            bucket = table;
+        }
+    }
+
+    bucket->index = aindex;
+
+    return(INT_MAX);
+}
 /**
  * xmlParseStartTag2:
  * @ctxt:  an XML parser context
@@ -8960,6 +8989,8 @@ xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref,
     int nratts, nbatts, nbdef;
     int i, j, nbNs, nbTotalDef, attval, nsIndex, maxAtts;
     int alloc = 0;
+    int numNsErr = 0;
+    int numDupErr = 0;
 
     if (RAW != '<') return(NULL);
     NEXT1;
@@ -9338,10 +9369,12 @@ xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref,
             if (res < INT_MAX) {
                 if (aprefix == atts[res+1]) {
                     xmlErrAttributeDup(ctxt, aprefix, attname);
+                    numDupErr += 1;
                 } else {
                     xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED,
                              "Namespaced Attribute %s in '%s' redefined\n",
                              attname, nsuri, NULL);
+                    numNsErr += 1;
                 }
             }
         }
@@ -9440,6 +9473,43 @@ xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref,
 	}
     }
 
+    /*
+     * Using a single hash table for nsUri/localName pairs cannot
+     * detect duplicate QNames reliably. The following example will
+     * only result in two namespace errors.
+     *
+     * <doc xmlns:a="a" xmlns:b="a">
+     *   <elem a:a="" b:a="" b:a=""/>
+     * </doc>
+     *
+     * If we saw more than one namespace error but no duplicate QNames
+     * were found, we have to scan for duplicate QNames.
+     */
+    if ((numDupErr == 0) && (numNsErr > 1)) {
+        memset(ctxt->attrHash, -1,
+               attrHashSize * sizeof(ctxt->attrHash[0]));
+
+        for (i = 0, j = 0; j < nratts; i += 5, j++) {
+            unsigned hashValue, nameHashValue, prefixHashValue;
+            int res;
+
+            aprefix = atts[i+1];
+            if (aprefix == NULL)
+                continue;
+
+            attname = atts[i];
+            /* Hash values always have bit 31 set, see dict.c */
+            nameHashValue = ctxt->attallocs[j] | 0x80000000;
+            prefixHashValue = xmlDictComputeHash(ctxt->dict, aprefix);
+
+            hashValue = xmlDictCombineHash(nameHashValue, prefixHashValue);
+            res = xmlAttrHashInsertQName(ctxt, attrHashSize, attname,
+                                         aprefix, hashValue, i);
+            if (res < INT_MAX)
+                xmlErrAttributeDup(ctxt, aprefix, attname);
+        }
+    }
+
     /*
      * Reconstruct attribute pointers
      */
@@ -11763,6 +11833,7 @@ xmlIOParseDTD(xmlSAXHandlerPtr sax, xmlParserInputBufferPtr input,
         xmlFreeParserInputBuffer(input);
 	return(NULL);
     }
+    xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD);
 
     /*
      * generate a parser input from the I/O handler
@@ -11852,6 +11923,7 @@ xmlSAXParseDTD(xmlSAXHandlerPtr sax, const xmlChar *ExternalID,
     if (ctxt == NULL) {
 	return(NULL);
     }
+    xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD);
 
     /*
      * Canonicalise the system ID
@@ -12143,6 +12215,15 @@ xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) {
 
         while (list != NULL) {
             list->parent = (xmlNodePtr) ent;
+
+            /*
+             * Downstream code like the nginx xslt module can set
+             * ctxt->myDoc->extSubset to a separate DTD, so the entity
+             * might have a different or a NULL document.
+             */
+            if (list->doc != ent->doc)
+                xmlSetTreeDoc(list, ent->doc);
+
             if (list->next == NULL)
                 ent->last = list;
             list = list->next;

testparser.c

@@ -125,6 +125,57 @@ testCFileIO(void) {
     return err;
 }
 
+#ifdef LIBXML_VALID_ENABLED
+static void
+testSwitchDtdExtSubset(void *vctxt, const xmlChar *name ATTRIBUTE_UNUSED,
+                       const xmlChar *externalId ATTRIBUTE_UNUSED,
+                       const xmlChar *systemId ATTRIBUTE_UNUSED) {
+    xmlParserCtxtPtr ctxt = vctxt;
+
+    ctxt->myDoc->extSubset = ctxt->_private;
+}
+
+static int
+testSwitchDtd(void) {
+    const char dtdContent[] =
+        "<!ENTITY test '<elem1/><elem2/>'>\n";
+    const char docContent[] =
+        "<!DOCTYPE doc SYSTEM 'entities.dtd'>\n"
+        "<doc>&test;</doc>\n";
+    xmlParserInputBufferPtr input;
+    xmlParserCtxtPtr ctxt;
+    xmlDtdPtr dtd;
+    xmlDocPtr doc;
+    xmlEntityPtr ent;
+    int err = 0;
+
+    input = xmlParserInputBufferCreateStatic(dtdContent,
+                                             sizeof(dtdContent) - 1,
+                                             XML_CHAR_ENCODING_NONE);
+    dtd = xmlIOParseDTD(NULL, input, XML_CHAR_ENCODING_NONE);
+
+    ctxt = xmlNewParserCtxt();
+    ctxt->_private = dtd;
+    ctxt->sax->externalSubset = testSwitchDtdExtSubset;
+    doc = xmlCtxtReadMemory(ctxt, docContent, sizeof(docContent) - 1, NULL,
+                            NULL, XML_PARSE_NOENT | XML_PARSE_DTDLOAD);
+    xmlFreeParserCtxt(ctxt);
+
+    ent = xmlGetDocEntity(doc, BAD_CAST "test");
+    if (ent->children->doc != NULL) {
+        fprintf(stderr, "Entity content should have NULL doc\n");
+        err = 1;
+    }
+
+    /* Free doc before DTD */
+    doc->extSubset = NULL;
+    xmlFreeDoc(doc);
+    xmlFreeDtd(dtd);
+
+    return err;
+}
+#endif /* LIBXML_VALID_ENABLED */
+
 #ifdef LIBXML_SAX1_ENABLED
 static int
 testBalancedChunk(void) {
@@ -571,6 +622,9 @@ main(void) {
     err |= testUnsupportedEncoding();
     err |= testNodeGetContent();
     err |= testCFileIO();
+#ifdef LIBXML_VALID_ENABLED
+    err |= testSwitchDtd();
+#endif
 #ifdef LIBXML_SAX1_ENABLED
     err |= testBalancedChunk();
 #endif

win32/rcVersion.h

@@ -1,4 +1,4 @@
 #define LIBXML_MAJOR_VERSION 2
 #define LIBXML_MINOR_VERSION 13
-#define LIBXML_MICRO_VERSION 4
-#define LIBXML_DOTTED_VERSION "2.13.4"
+#define LIBXML_MICRO_VERSION 5
+#define LIBXML_DOTTED_VERSION "2.13.5"

xmlIO.c

@@ -1228,7 +1228,13 @@ xmlInputDefaultOpen(xmlParserInputBufferPtr buf, const char *filename) {
         if (xzStream == NULL) {
             close(fd);
         } else {
-            if (__libxml2_xzcompressed(xzStream) > 0) {
+            /*
+             * Non-regular files like pipes can't be reopened.
+             * If a file isn't seekable, we pipe uncompressed
+             * input through xzlib.
+             */
+            if ((lseek(fd, 0, SEEK_CUR) < 0) ||
+                (__libxml2_xzcompressed(xzStream) > 0)) {
                 buf->context = xzStream;
                 buf->readcallback = xmlXzfileRead;
                 buf->closecallback = xmlXzfileClose;
@@ -1255,12 +1261,13 @@ xmlInputDefaultOpen(xmlParserInputBufferPtr buf, const char *filename) {
         if (gzStream == NULL) {
             close(fd);
         } else {
-            char buff4[4];
-
-            if ((gzread(gzStream, buff4, 4) > 0) &&
+            /*
+             * Non-regular files like pipes can't be reopened.
+             * If a file isn't seekable, we pipe uncompressed
+             * input through zlib.
+             */
+            if ((lseek(fd, 0, SEEK_CUR) < 0) ||
                 (gzdirect(gzStream) == 0)) {
-                gzrewind(gzStream);
-
                 buf->context = gzStream;
                 buf->readcallback = xmlGzfileRead;
                 buf->closecallback = xmlGzfileClose;

xmlreader.c

@@ -58,7 +58,7 @@
   #ifdef __va_copy
     #define va_copy(dest, src) __va_copy(dest, src)
   #else
-    #define va_copy(dest, src) memcpy(dest, src, sizeof(va_list))
+    #define va_copy(dest, src) memcpy(&(dest), &(src), sizeof(va_list))
   #endif
 #endif
 
@@ -1750,7 +1750,8 @@ xmlTextReaderReadString(xmlTextReaderPtr reader)
         case XML_CDATA_SECTION_NODE:
             break;
         case XML_ELEMENT_NODE:
-            if (xmlTextReaderDoExpand(reader) == -1)
+            if ((xmlTextReaderDoExpand(reader) == -1) ||
+                (node->children == NULL))
                 return(NULL);
             break;
         case XML_ATTRIBUTE_NODE:

xmlstring.c

@@ -30,7 +30,7 @@
   #ifdef __va_copy
     #define va_copy(dest, src) __va_copy(dest, src)
   #else
-    #define va_copy(dest, src) memcpy(dest, src, sizeof(va_list))
+    #define va_copy(dest, src) memcpy(&(dest), &(src), sizeof(va_list))
   #endif
 #endif
 

xmlwriter.c

@@ -36,7 +36,7 @@
   #ifdef __va_copy
     #define va_copy(dest, src) __va_copy(dest, src)
   #else
-    #define va_copy(dest, src) memcpy(dest, src, sizeof(va_list))
+    #define va_copy(dest, src) memcpy(&(dest), &(src), sizeof(va_list))
   #endif
 #endif