Incorporate Expat 1.95.6.

Modules/expat/ascii.h

-/*
-Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
-See the file COPYING for copying permission.
+/* Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
+   See the file COPYING for copying permission.
 */
 
 #define ASCII_A 0x41

Modules/expat/internal.h

+/* internal.h
+
+   Internal definitions used by Expat.  This is not needed to compile
+   client code.
+
+   The following calling convention macros are defined for frequently
+   called functions:
+
+   FASTCALL    - Used for those internal functions that have a simple
+                 body and a low number of arguments and local variables.
+
+   PTRCALL     - Used for functions called though function pointers.
+
+   PTRFASTCALL - Like PTRCALL, but for low number of arguments.
+
+   inline      - Used for selected internal functions for which inlining
+                 may improve performance on some platforms.
+
+   Note: Use of these macros is based on judgement, not hard rules,
+         and therefore subject to change.
+*/
+
+#if defined(__GNUC__)
+/* Instability reported with egcs on a RedHat Linux 7.3.
+   Let's comment it out:
+   #define FASTCALL __attribute__((stdcall, regparm(3)))
+   and let's try this:
+*/
+#define FASTCALL __attribute__((regparm(3)))
+#define PTRCALL
+#define PTRFASTCALL __attribute__((regparm(3)))
+
+#elif defined(WIN32)
+/* Using __fastcall seems to have an unexpected negative effect under
+   MS VC++, especially for function pointers, so we won't use it for
+   now on that platform. It may be reconsidered for a future release
+   if it can be made more effective.
+   Likely reason: __fastcall on Windows is like stdcall, therefore
+   the compiler cannot perform stack optimizations for call clusters.
+*/
+#define FASTCALL
+#define PTRCALL
+#define PTRFASTCALL
+
+#endif
+
+#ifndef FASTCALL
+#define FASTCALL
+#endif
+
+#ifndef PTRCALL
+#define PTRCALL
+#endif
+
+#ifndef PTRFASTCALL
+#define PTRFASTCALL
+#endif
+
+#ifndef XML_MIN_SIZE
+#if !defined(__cplusplus) && !defined(inline)
+#ifdef __GNUC__
+#define inline __inline
+#endif /* __GNUC__ */
+#endif
+#endif /* XML_MIN_SIZE */
+
+#ifdef __cplusplus
+#define inline inline
+#else
+#ifndef inline
+#define inline
+#endif
+#endif

Modules/expat/xmlrole.h

-/*
-Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
-See the file COPYING for copying permission.
+/* Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
+   See the file COPYING for copying permission.
 */
 
 #ifndef XmlRole_INCLUDED
 #define XmlRole_INCLUDED 1
 
+#ifdef __VMS
+/*      0        1         2         3      0        1         2         3
+        1234567890123456789012345678901     1234567890123456789012345678901 */
+#define XmlPrologStateInitExternalEntity    XmlPrologStateInitExternalEnt
+#endif
+
 #include "xmltok.h"
 
 #ifdef __cplusplus
@@ -17,6 +22,7 @@ enum {
   XML_ROLE_NONE = 0,
   XML_ROLE_XML_DECL,
   XML_ROLE_INSTANCE_START,
+  XML_ROLE_DOCTYPE_NONE,
   XML_ROLE_DOCTYPE_NAME,
   XML_ROLE_DOCTYPE_SYSTEM_ID,
   XML_ROLE_DOCTYPE_PUBLIC_ID,
@@ -24,11 +30,13 @@ enum {
   XML_ROLE_DOCTYPE_CLOSE,
   XML_ROLE_GENERAL_ENTITY_NAME,
   XML_ROLE_PARAM_ENTITY_NAME,
+  XML_ROLE_ENTITY_NONE,
   XML_ROLE_ENTITY_VALUE,
   XML_ROLE_ENTITY_SYSTEM_ID,
   XML_ROLE_ENTITY_PUBLIC_ID,
   XML_ROLE_ENTITY_COMPLETE,
   XML_ROLE_ENTITY_NOTATION_NAME,
+  XML_ROLE_NOTATION_NONE,
   XML_ROLE_NOTATION_NAME,
   XML_ROLE_NOTATION_SYSTEM_ID,
   XML_ROLE_NOTATION_NO_SYSTEM_ID,
@@ -44,11 +52,13 @@ enum {
   XML_ROLE_ATTRIBUTE_TYPE_NMTOKENS,
   XML_ROLE_ATTRIBUTE_ENUM_VALUE,
   XML_ROLE_ATTRIBUTE_NOTATION_VALUE,
+  XML_ROLE_ATTLIST_NONE,
   XML_ROLE_ATTLIST_ELEMENT_NAME,
   XML_ROLE_IMPLIED_ATTRIBUTE_VALUE,
   XML_ROLE_REQUIRED_ATTRIBUTE_VALUE,
   XML_ROLE_DEFAULT_ATTRIBUTE_VALUE,
   XML_ROLE_FIXED_ATTRIBUTE_VALUE,
+  XML_ROLE_ELEMENT_NONE,
   XML_ROLE_ELEMENT_NAME,
   XML_ROLE_CONTENT_ANY,
   XML_ROLE_CONTENT_EMPTY,
@@ -64,6 +74,8 @@ enum {
   XML_ROLE_CONTENT_ELEMENT_REP,
   XML_ROLE_CONTENT_ELEMENT_OPT,
   XML_ROLE_CONTENT_ELEMENT_PLUS,
+  XML_ROLE_PI,
+  XML_ROLE_COMMENT,
 #ifdef XML_DTD
   XML_ROLE_TEXT_DECL,
   XML_ROLE_IGNORE_SECT,
@@ -73,15 +85,17 @@ enum {
 };
 
 typedef struct prolog_state {
-  int (*handler)(struct prolog_state *state,
+  int (PTRCALL *handler) (struct prolog_state *state,
                           int tok,
                           const char *ptr,
                           const char *end,
                           const ENCODING *enc);
   unsigned level;
+  int role_none;
 #ifdef XML_DTD
   unsigned includeLevel;
   int documentEntity;
+  int inEntityValue;
 #endif /* XML_DTD */
 } PROLOG_STATE;
 

Modules/expat/xmltok.h

-/*
-Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
-See the file COPYING for copying permission.
+/* Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
+   See the file COPYING for copying permission.
 */
 
 #ifndef XmlTok_INCLUDED
@@ -11,9 +10,11 @@ extern "C" {
 #endif
 
 /* The following token may be returned by XmlContentTok */
-#define XML_TOK_TRAILING_RSQB -5 /* ] or ]] at the end of the scan; might be start of
-                                    illegal ]]> sequence */
-/* The following tokens may be returned by both XmlPrologTok and XmlContentTok */
+#define XML_TOK_TRAILING_RSQB -5 /* ] or ]] at the end of the scan; might be
+                                    start of illegal ]]> sequence */
+/* The following tokens may be returned by both XmlPrologTok and
+   XmlContentTok.
+*/
 #define XML_TOK_NONE -4          /* The string to be scanned is empty */
 #define XML_TOK_TRAILING_CR -3   /* A CR at the end of the scan;
                                     might be part of CRLF sequence */
@@ -22,8 +23,8 @@ extern "C" {
 #define XML_TOK_INVALID 0
 
 /* The following tokens are returned by XmlContentTok; some are also
-  returned by XmlAttributeValueTok, XmlEntityTok, XmlCdataSectionTok */
-
+   returned by XmlAttributeValueTok, XmlEntityTok, XmlCdataSectionTok.
+*/
 #define XML_TOK_START_TAG_WITH_ATTS 1
 #define XML_TOK_START_TAG_NO_ATTS 2
 #define XML_TOK_EMPTY_ELEMENT_WITH_ATTS 3 /* empty element tag <e/> */
@@ -35,7 +36,9 @@ extern "C" {
 #define XML_TOK_ENTITY_REF 9
 #define XML_TOK_CHAR_REF 10               /* numeric character reference */
 
-/* The following tokens may be returned by both XmlPrologTok and XmlContentTok */
+/* The following tokens may be returned by both XmlPrologTok and
+   XmlContentTok.
+*/
 #define XML_TOK_PI 11                     /* processing instruction */
 #define XML_TOK_XML_DECL 12               /* XML decl or text decl */
 #define XML_TOK_COMMENT 13
@@ -75,8 +78,9 @@ extern "C" {
 /* The following token is returned only by XmlCdataSectionTok */
 #define XML_TOK_CDATA_SECT_CLOSE 40
 
-/* With namespace processing this is returned by XmlPrologTok
-   for a name with a colon. */
+/* With namespace processing this is returned by XmlPrologTok for a
+   name with a colon.
+*/
 #define XML_TOK_PREFIXED_NAME 41
 
 #ifdef XML_DTD
@@ -121,37 +125,45 @@ typedef struct {
 struct encoding;
 typedef struct encoding ENCODING;
 
-struct encoding {
-  int (*scanners[XML_N_STATES])(const ENCODING *,
+typedef int (PTRCALL *SCANNER)(const ENCODING *,
                                const char *,
                                const char *,
                                const char **);
-  int (*literalScanners[XML_N_LITERAL_TYPES])(const ENCODING *,
+
+struct encoding {
+  SCANNER scanners[XML_N_STATES];
+  SCANNER literalScanners[XML_N_LITERAL_TYPES];
+  int (PTRCALL *sameName)(const ENCODING *,
                           const char *,
+                          const char *);
+  int (PTRCALL *nameMatchesAscii)(const ENCODING *,
                                   const char *,
-					      const char **);
-  int (*sameName)(const ENCODING *,
-	          const char *, const char *);
-  int (*nameMatchesAscii)(const ENCODING *,
-			  const char *, const char *, const char *);
-  int (*nameLength)(const ENCODING *, const char *);
-  const char *(*skipS)(const ENCODING *, const char *);
-  int (*getAtts)(const ENCODING *enc, const char *ptr,
-	         int attsMax, ATTRIBUTE *atts);
-  int (*charRefNumber)(const ENCODING *enc, const char *ptr);
-  int (*predefinedEntityName)(const ENCODING *, const char *, const char *);
-  void (*updatePosition)(const ENCODING *,
+                                  const char *,
+                                  const char *);
+  int (PTRFASTCALL *nameLength)(const ENCODING *, const char *);
+  const char *(PTRFASTCALL *skipS)(const ENCODING *, const char *);
+  int (PTRCALL *getAtts)(const ENCODING *enc,
+                         const char *ptr,
+                         int attsMax,
+                         ATTRIBUTE *atts);
+  int (PTRFASTCALL *charRefNumber)(const ENCODING *enc, const char *ptr);
+  int (PTRCALL *predefinedEntityName)(const ENCODING *,
+                                      const char *,
+                                      const char *);
+  void (PTRCALL *updatePosition)(const ENCODING *,
                                  const char *ptr,
                                  const char *end,
                                  POSITION *);
-  int (*isPublicId)(const ENCODING *enc, const char *ptr, const char *end,
+  int (PTRCALL *isPublicId)(const ENCODING *enc,
+                            const char *ptr,
+                            const char *end,
                             const char **badPtr);
-  void (*utf8Convert)(const ENCODING *enc,
+  void (PTRCALL *utf8Convert)(const ENCODING *enc,
                               const char **fromP,
                               const char *fromLim,
                               char **toP,
                               const char *toLim);
-  void (*utf16Convert)(const ENCODING *enc,
+  void (PTRCALL *utf16Convert)(const ENCODING *enc,
                                const char **fromP,
                                const char *fromLim,
                                unsigned short **toP,
@@ -161,24 +173,25 @@ struct encoding {
   char isUtf16;
 };
 
-/*
-Scan the string starting at ptr until the end of the next complete token,
-but do not scan past eptr.  Return an integer giving the type of token.
+/* Scan the string starting at ptr until the end of the next complete
+   token, but do not scan past eptr.  Return an integer giving the
+   type of token.
 
-Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set.
+   Return XML_TOK_NONE when ptr == eptr; nextTokPtr will not be set.
 
-Return XML_TOK_PARTIAL when the string does not contain a complete token;
-nextTokPtr will not be set.
+   Return XML_TOK_PARTIAL when the string does not contain a complete
+   token; nextTokPtr will not be set.
 
-Return XML_TOK_INVALID when the string does not start a valid token; nextTokPtr
-will be set to point to the character which made the token invalid.
+   Return XML_TOK_INVALID when the string does not start a valid
+   token; nextTokPtr will be set to point to the character which made
+   the token invalid.
 
-Otherwise the string starts with a valid token; nextTokPtr will be set to point
-to the character following the end of that token.
+   Otherwise the string starts with a valid token; nextTokPtr will be
+   set to point to the character following the end of that token.
 
-Each data character counts as a single token, but adjacent data characters
-may be returned together.  Similarly for characters in the prolog outside
-literals, comments and processing instructions.
+   Each data character counts as a single token, but adjacent data
+   characters may be returned together.  Similarly for characters in
+   the prolog outside literals, comments and processing instructions.
 */
 
 
@@ -201,9 +214,9 @@ literals, comments and processing instructions.
 
 #endif /* XML_DTD */
 
-/* This is used for performing a 2nd-level tokenization on
-the content of a literal that has already been returned by XmlTok. */ 
-
+/* This is used for performing a 2nd-level tokenization on the content
+   of a literal that has already been returned by XmlTok.
+*/
 #define XmlLiteralTok(enc, literalType, ptr, end, nextTokPtr) \
   (((enc)->literalScanners[literalType])(enc, ptr, end, nextTokPtr))
 
@@ -264,14 +277,16 @@ int  XmlParseXmlDecl(int isGeneralTextEntity,
 int XmlInitEncoding(INIT_ENCODING *, const ENCODING **, const char *name);
 const ENCODING *XmlGetUtf8InternalEncoding(void);
 const ENCODING *XmlGetUtf16InternalEncoding(void);
-int  XmlUtf8Encode(int charNumber, char *buf);
-int  XmlUtf16Encode(int charNumber, unsigned short *buf);
-
+int FASTCALL XmlUtf8Encode(int charNumber, char *buf);
+int FASTCALL XmlUtf16Encode(int charNumber, unsigned short *buf);
 int XmlSizeOfUnknownEncoding(void);
+
+typedef int (*CONVERTER)(void *userData, const char *p);
+
 ENCODING *
 XmlInitUnknownEncoding(void *mem,
                        int *table,
-		       int (*conv)(void *userData, const char *p),
+                       CONVERTER convert,
                        void *userData);
 
 int XmlParseXmlDeclNS(int isGeneralTextEntity,
@@ -284,13 +299,14 @@ int  XmlParseXmlDeclNS(int isGeneralTextEntity,
                       const char **encodingNamePtr,
                       const ENCODING **namedEncodingPtr,
                       int *standalonePtr);
+
 int XmlInitEncodingNS(INIT_ENCODING *, const ENCODING **, const char *name);
 const ENCODING *XmlGetUtf8InternalEncodingNS(void);
 const ENCODING *XmlGetUtf16InternalEncodingNS(void);
 ENCODING *
 XmlInitUnknownEncodingNS(void *mem,
                          int *table,
-		         int (*conv)(void *userData, const char *p),
+                         CONVERTER convert,
                          void *userData);
 #ifdef __cplusplus
 }

Modules/expat/xmltok_ns.c

-const ENCODING *NS(XmlGetUtf8InternalEncoding)(void)
+const ENCODING *
+NS(XmlGetUtf8InternalEncoding)(void)
 {
   return &ns(internal_utf8_encoding).enc;
 }
 
-const ENCODING *NS(XmlGetUtf16InternalEncoding)(void)
+const ENCODING *
+NS(XmlGetUtf16InternalEncoding)(void)
 {
-#if XML_BYTE_ORDER == 12
+#if BYTEORDER == 1234
   return &ns(internal_little2_encoding).enc;
-#elif XML_BYTE_ORDER == 21
+#elif BYTEORDER == 4321
   return &ns(internal_big2_encoding).enc;
 #else
   const short n = 1;
-  return *(const char *)&n ? &ns(internal_little2_encoding).enc : &ns(internal_big2_encoding).enc;
+  return (*(const char *)&n
+          ? &ns(internal_little2_encoding).enc
+          : &ns(internal_big2_encoding).enc);
 #endif
 }
 
-static
-const ENCODING *NS(encodings)[] = {
+static const ENCODING *NS(encodings)[] = {
   &ns(latin1_encoding).enc,
   &ns(ascii_encoding).enc,
   &ns(utf8_encoding).enc,
@@ -26,21 +29,25 @@ const ENCODING *NS(encodings)[] = {
   &ns(utf8_encoding).enc /* NO_ENC */
 };
 
-static
-int NS(initScanProlog)(const ENCODING *enc, const char *ptr, const char *end,
+static int PTRCALL
+NS(initScanProlog)(const ENCODING *enc, const char *ptr, const char *end,
                    const char **nextTokPtr)
 {
-  return initScan(NS(encodings), (const INIT_ENCODING *)enc, XML_PROLOG_STATE, ptr, end, nextTokPtr);
+  return initScan(NS(encodings), (const INIT_ENCODING *)enc,
+                  XML_PROLOG_STATE, ptr, end, nextTokPtr);
 }
 
-static
-int NS(initScanContent)(const ENCODING *enc, const char *ptr, const char *end,
+static int PTRCALL
+NS(initScanContent)(const ENCODING *enc, const char *ptr, const char *end,
                     const char **nextTokPtr)
 {
-  return initScan(NS(encodings), (const INIT_ENCODING *)enc, XML_CONTENT_STATE, ptr, end, nextTokPtr);
+  return initScan(NS(encodings), (const INIT_ENCODING *)enc,
+                  XML_CONTENT_STATE, ptr, end, nextTokPtr);
 }
 
-int NS(XmlInitEncoding)(INIT_ENCODING *p, const ENCODING **encPtr, const char *name)
+int
+NS(XmlInitEncoding)(INIT_ENCODING *p, const ENCODING **encPtr,
+                    const char *name)
 {
   int i = getEncodingIndex(name);
   if (i == UNKNOWN_ENC)
@@ -54,8 +61,8 @@ int NS(XmlInitEncoding)(INIT_ENCODING *p, const ENCODING **encPtr, const char *n
   return 1;
 }
 
-static
-const ENCODING *NS(findEncoding)(const ENCODING *enc, const char *ptr, const char *end)
+static const ENCODING *
+NS(findEncoding)(const ENCODING *enc, const char *ptr, const char *end)
 {
 #define ENCODING_MAX 128
   char buf[ENCODING_MAX];
@@ -73,7 +80,8 @@ const ENCODING *NS(findEncoding)(const ENCODING *enc, const char *ptr, const cha
   return NS(encodings)[i];
 }
 
-int NS(XmlParseXmlDecl)(int isGeneralTextEntity,
+int
+NS(XmlParseXmlDecl)(int isGeneralTextEntity,
                     const ENCODING *enc,
                     const char *ptr,
                     const char *end,

setup.py

@@ -690,9 +690,9 @@ class PyBuildExt(build_ext):
         # More information on Expat can be found at www.libexpat.org.
         #
         if sys.byteorder == "little":
-            xmlbo = "12"
+            xmlbo = "1234"
         else:
-            xmlbo = "21"
+            xmlbo = "4321"
         expatinc = os.path.join(os.getcwd(), srcdir, 'Modules', 'expat')
         exts.append(Extension('pyexpat',
                               sources = [
@@ -702,10 +702,9 @@ class PyBuildExt(build_ext):
             'expat/xmltok.c',
             ],
                               define_macros = [
-            ('HAVE_EXPAT_H',None),
             ('XML_NS', '1'),
             ('XML_DTD', '1'),
-            ('XML_BYTE_ORDER', xmlbo),
+            ('BYTEORDER', xmlbo),
             ('XML_CONTEXT_BYTES','1024'),
             ],
                               include_dirs = [expatinc]