GNU bug report logs - #10777
24.0.93; Enhancement: Add support for libxml2 on MS-Windows

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 10777 in the body.
You can then email your comments to 10777 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: bug-gnu-emacs <at> gnu.org
Subject: 24.0.93; Enhancement: Add support for libxml2 on MS-Windows
Date: Fri, 10 Feb 2012 12:21:55 +0200

Package: emacs
Severity: wishlist
Tags: patch

See the subject.  The patch below accomplishes that, but in

  http://lists.gnu.org/archive/html/emacs-devel/2012-02/msg00288.html

it was decided that it should be applied after Emacs 24.1 is released.


=== modified file 'etc/NEWS'
--- etc/NEWS	2012-02-10 08:51:37 +0000
+++ etc/NEWS	2012-02-10 09:27:36 +0000
@@ -1539,6 +1539,10 @@ and also when HOME is set to C:\ by defa
 *** --without-gnutls disables automatic GnuTLS detection.
 
 +++
+*** --without-libxml2 omits support for libxml2, even if its presence
+is detected.
+
++++
 *** --lib for general library linkage, works with the USER_LIBS build variable.
 
 +++

=== modified file 'lib-src/ChangeLog'
--- lib-src/ChangeLog	2012-02-09 07:48:22 +0000
+++ lib-src/ChangeLog	2012-02-10 09:27:36 +0000
@@ -1,3 +1,7 @@
+2012-02-05  Eli Zaretskii  <eliz <at> gnu.org>
+
+	* makefile.w32-in (obj): Add xml.o.
+
 2012-01-05  Glenn Morris  <rgm <at> gnu.org>
 
 	* ebrowse.c (version) <emacs_copyright>:

=== modified file 'lib-src/makefile.w32-in'
--- lib-src/makefile.w32-in	2012-02-09 07:48:22 +0000
+++ lib-src/makefile.w32-in	2012-02-10 09:27:36 +0000
@@ -140,7 +140,7 @@ obj =   dosfns.o msdos.o \
 	process.o callproc.o unexw32.o \
 	region-cache.o sound.o atimer.o \
 	doprnt.o intervals.o textprop.o composite.o \
-	gnutls.o
+	gnutls.o xml.o
 
 #
 # These are the lisp files that are loaded up in loadup.el

=== modified file 'lisp/ChangeLog'
--- lisp/ChangeLog	2012-02-10 08:23:33 +0000
+++ lisp/ChangeLog	2012-02-10 09:27:36 +0000
@@ -1,3 +1,7 @@
+2012-02-04  Eli Zaretskii  <eliz <at> gnu.org>
+
+	* term/w32-win.el (dynamic-library-alist): Add libxml2 DLLs.
+
 2012-02-10  Glenn Morris  <rgm <at> gnu.org>
 
 	* textmodes/bibtex.el:

=== modified file 'lisp/term/w32-win.el'
--- lisp/term/w32-win.el	2012-01-19 07:21:25 +0000
+++ lisp/term/w32-win.el	2012-02-10 09:27:36 +0000
@@ -210,7 +210,8 @@ See the documentation of `create-fontset
        '(gdk-pixbuf "libgdk_pixbuf-2.0-0.dll")
        '(glib "libglib-2.0-0.dll")
        '(gobject "libgobject-2.0-0.dll")
-       '(gnutls "libgnutls-28.dll" "libgnutls-26.dll")))
+       '(gnutls "libgnutls-28.dll" "libgnutls-26.dll")
+       '(libxml2 "libxml2-2.dll" "libxml2.dll")))
 
 ;;; multi-tty support
 (defvar w32-initialized nil

=== modified file 'nt/ChangeLog'
--- nt/ChangeLog	2012-02-09 07:48:22 +0000
+++ nt/ChangeLog	2012-02-10 10:11:05 +0000
@@ -1,3 +1,10 @@
+2012-02-04  Eli Zaretskii  <eliz <at> gnu.org>
+
+	* configure.bat: Support building with libxml2.
+
+	* INSTALL:
+	* README.W32: Add information about libxml2.
+
 2012-02-06  Christoph Scholtes  <cschol2112 <at> googlemail.com>
 
 	* README.W32: Removed specific version information for libXpm

=== modified file 'nt/INSTALL'
--- nt/INSTALL	2012-02-04 21:34:36 +0000
+++ nt/INSTALL	2012-02-10 09:58:48 +0000
@@ -367,6 +367,40 @@
   You can get pre-built binaries (including any required DLL and the
   header files) at http://sourceforge.net/projects/ezwinports/files/.
 
+* Optional libxml2 support
+
+  If configure.bat finds the libxml/HTMLparser.h file in the include path,
+  Emacs is built with libxml2 support by default; to avoid that you can
+  pass the argument --without-libxml2.
+
+  In order to support libxml2 at runtime, a libxml2-enabled Emacs must
+  be able to find the relevant DLLs during startup; failure to do so
+  is not an error, but libxml2 features won't be available to the
+  running session.
+
+  One place where you can get pre-built Windows binaries of libxml2
+  (including any required DLL and the header files) is here:
+
+     http://sourceforge.net/projects/ezwinports/files/
+
+  To compile Emacs with libxml2 from that site, you will need to pass
+  the "--cflags -I/path/to/include/libxml2" option to configure.bat,
+  because libxml2 header files are installed in the include/libxml2
+  subdirectory of the directory where you unzip the binary
+  distribution.  Other binary distributions might use other
+  directories, although include/libxml2 is the canonical place where
+  libxml2 headers are installed on Posix platforms.
+
+  You will also need to install the libiconv "development" tarball,
+  because the libiconv headers need to be available to the compiler
+  when you compile with libxml2 support.  A MinGW port of libiconv can
+  be found on the MinGW site:
+
+   http://sourceforge.net/projects/mingw/files/MinGW/Base/libiconv/
+
+  You need the libiconv-X.Y.Z-N-mingw32-dev.tar.lzma tarball from that
+  site.
+
 * Experimental SVG support
 
   SVG support is currently experimental, and not built by default.

=== modified file 'nt/README.W32'
--- nt/README.W32	2012-02-09 07:48:22 +0000
+++ nt/README.W32	2012-02-10 09:27:36 +0000
@@ -172,6 +172,16 @@ See the end of the file for license cond
   You can get pre-built binaries (including any required DLL and the
   header files) at http://sourceforge.net/projects/ezwinports/files/.
 
+* libxml2 support
+
+  In order to support libxml2 at runtime, a libxml2-enabled Emacs must
+  be able to find the relevant DLLs during startup; failure to do so
+  is not an error, but libxml2 features won't be available to the
+  running session.
+
+  You can get pre-built binaries (including any required DLL and the
+  header files) at http://sourceforge.net/projects/ezwinports/files/.
+
 * Uninstalling Emacs
 
   If you should need to uninstall Emacs, simply delete all the files

=== modified file 'nt/configure.bat'
--- nt/configure.bat	2012-01-19 07:21:25 +0000
+++ nt/configure.bat	2012-02-10 09:27:36 +0000
@@ -131,6 +131,7 @@ if "%1" == "--without-jpeg" goto without
 if "%1" == "--without-gif" goto withoutgif
 if "%1" == "--without-tiff" goto withouttiff
 if "%1" == "--without-gnutls" goto withoutgnutls
+if "%1" == "--without-libxml2" goto withoutlibxml2
 if "%1" == "--without-xpm" goto withoutxpm
 if "%1" == "--with-svg" goto withsvg
 if "%1" == "--distfiles" goto distfiles
@@ -156,6 +157,7 @@ echo.   --without-gif           do not u
 echo.   --without-tiff          do not use TIFF library even if it is installed
 echo.   --without-xpm           do not use XPM library even if it is installed
 echo.   --without-gnutls        do not use GnuTLS library even if it is installed
+echo.   --without-libxml2       do not use libxml2 library even if it is installed
 echo.   --with-svg              use the RSVG library (experimental)
 echo.   --distfiles             path to files for make dist, e.g. libXpm.dll
 if "%use_extensions%" == "0" goto end
@@ -317,6 +319,14 @@ goto again
 
 rem ----------------------------------------------------------------------
 
+:withoutlibxml2
+set libxml2support=N
+set HAVE_LIBXML2=
+shift
+goto again
+
+rem ----------------------------------------------------------------------
+
 :withouttiff
 set tiffsupport=N
 set HAVE_TIFF=
@@ -569,6 +579,28 @@ set HAVE_GNUTLS=1
 :tlsDone
 rm -f junk.c junk.obj
 
+if (%libxml2support%) == (N) goto xml2Done
+
+echo Checking for libxml2....
+echo #include "libxml/HTMLparser.h" >junk.c
+echo main(){} >>junk.c
+echo %COMPILER% %usercflags% %mingwflag% -c junk.c -o junk.obj >>config.log
+%COMPILER% %usercflags% %mingwflag% -c junk.c -o junk.obj >junk.out 2>>config.log
+if exist junk.obj goto havelibxml2
+
+echo ...libxml/HTMLparser.h not found, building without libxml2 support
+echo The failed program was: >>config.log
+type junk.c >>config.log
+set HAVE_LIBXML2=
+goto xml2Done
+
+:havelibxml2
+echo ...libxml2 header available, building with libxml2 support
+set HAVE_LIBXML2=1
+
+:xml2Done
+rm -f junk.c junk.obj
+
 if (%jpegsupport%) == (N) goto jpegDone
 
 echo Checking for jpeg-6b...
@@ -757,6 +789,7 @@ if (%doldflags%) == (Y) echo #define USE
 if (%profile%) == (Y) echo #define PROFILING 1 >>config.tmp
 if not "(%HAVE_PNG%)" == "()" echo #define HAVE_PNG 1 >>config.tmp
 if not "(%HAVE_GNUTLS%)" == "()" echo #define HAVE_GNUTLS 1 >>config.tmp
+if not "(%HAVE_LIBXML2%)" == "()" echo #define HAVE_LIBXML2 1 >>config.tmp
 if not "(%HAVE_JPEG%)" == "()" echo #define HAVE_JPEG 1 >>config.tmp
 if not "(%HAVE_GIF%)" == "()" echo #define HAVE_GIF 1 >>config.tmp
 if not "(%HAVE_TIFF%)" == "()" echo #define HAVE_TIFF 1 >>config.tmp
@@ -896,6 +929,7 @@ set HAVE_DISTFILES=
 set distFilesOk=
 set pngsupport=
 set tlssupport=
+set libxml2support=
 set jpegsupport=
 set gifsupport=
 set tiffsupport=

=== modified file 'src/ChangeLog'
--- src/ChangeLog	2012-02-10 08:38:22 +0000
+++ src/ChangeLog	2012-02-10 09:34:47 +0000
@@ -1,3 +1,31 @@
+2012-02-10  Eli Zaretskii  <eliz <at> gnu.org>
+
+	Support building on MS-Windows with libxml2.
+
+	* makefile.w32-in (OBJ2): Add xml.$(O).
+	(GLOBAL_SOURCES): Add xml.c.
+	($(BLD)/xml.$(O)): New dependency list.
+
+	* xml.c (DEF_XML2_FN, LOAD_XML2_FN) [WINDOWSNT]: New macros.
+	(fn_htmlReadMemory, fn_xmlReadMemory, fn_xmlDocGetRootElement)
+	(fn_xmlFreeDoc, fn_xmlCleanupParser, fn_xmlCheckVersion)
+	[!WINDOWSNT]: New macros.
+	(init_libxml2_functions, libxml2_loaded_p): New functions.
+	(parse_region): Call fn_xmlCheckVersion instead of using the macro
+	LIBXML_TEST_VERSION.  Call libxml2 functions via the fn_* macros.
+	(xml_cleanup_parser): New function, export for fn_xmlCleanupParser.
+	Calls xmlCleanupParser only if libxml2 was loaded (or statically
+	linked in).
+	(Flibxml_parse_html_region, Flibxml_parse_xml_region): Call
+	init_libxml2_functions before calling libxml2 functions.
+	(syms_of_xml) <Qlibxml2_dll>: DEFSYM it.
+
+	* emacs.c: Don't include libxml/parser.h.
+	(shut_down_emacs): Call xml_cleanup_parser, instead of calling
+	xmlCleanupParser directly.
+
+	* lisp.h [HAVE_LIBXML2]: Add prototype for xml_cleanup_parser.
+
 2012-02-10  Glenn Morris  <rgm <at> gnu.org>
 
 	* fns.c (Fsecure_hash): Doc fix.

=== modified file 'src/emacs.c'
--- src/emacs.c	2012-01-19 07:21:25 +0000
+++ src/emacs.c	2012-02-10 09:27:36 +0000
@@ -82,10 +82,6 @@ along with GNU Emacs.  If not, see <http
 #include <sys/personality.h>
 #endif
 
-#ifdef HAVE_LIBXML2
-#include <libxml/parser.h>
-#endif
-
 #ifndef O_RDWR
 #define O_RDWR 2
 #endif
@@ -2124,7 +2120,7 @@ shut_down_emacs (int sig, int no_x, Lisp
 #endif
 
 #ifdef HAVE_LIBXML2
-  xmlCleanupParser ();
+  xml_cleanup_parser ();
 #endif
 }
 

=== modified file 'src/lisp.h'
--- src/lisp.h	2012-01-19 07:21:25 +0000
+++ src/lisp.h	2012-02-10 09:27:36 +0000
@@ -3542,6 +3542,7 @@ EXFUN (Fmsdos_downcase_filename, 1);
 #ifdef HAVE_LIBXML2
 /* Defined in xml.c */
 extern void syms_of_xml (void);
+extern void xml_cleanup_parser (void);
 #endif
 
 #ifdef HAVE_MENUS

=== modified file 'src/makefile.w32-in'
--- src/makefile.w32-in	2012-01-19 07:21:25 +0000
+++ src/makefile.w32-in	2012-02-10 09:27:36 +0000
@@ -128,6 +128,7 @@ OBJ2 =  $(BLD)/sysdep.$(O)		\
 	$(BLD)/image.$(O)		\
 	$(BLD)/terminal.$(O)            \
 	$(BLD)/menu.$(O)		\
+	$(BLD)/xml.$(O)			\
 	$(BLD)/w32term.$(O)		\
 	$(BLD)/w32xfns.$(O)		\
 	$(BLD)/w32fns.$(O)		\
@@ -225,7 +226,7 @@ GLOBAL_SOURCES =   dosfns.c msdos.c \
 	process.c callproc.c unexw32.c \
 	region-cache.c sound.c atimer.c \
 	doprnt.c intervals.c textprop.c composite.c \
-	gnutls.c
+	gnutls.c xml.c
 SOME_MACHINE_OBJECTS = dosfns.o msdos.o \
 	xterm.o xfns.o xmenu.o xselect.o xrdb.o xsmfns.o dbusbind.o
 obj = $(GLOBAL_SOURCES:.c=.o)
@@ -940,6 +941,13 @@ $(BLD)/gnutls.$(O) : \
 	$(LISP_H) \
 	$(PROCESS_H)
 
+$(BLD)/xml.$(O) : \
+	$(SRC)/xml.c \
+	$(SRC)/w32.h \
+	$(CONFIG_H) \
+	$(LISP_H) \
+	$(SRC)/buffer.h
+
 $(BLD)/image.$(O) : \
 	$(SRC)/image.c \
 	$(SRC)/epaths.h \

=== modified file 'src/xml.c'
--- src/xml.c	2012-01-05 09:46:05 +0000
+++ src/xml.c	2012-02-10 09:41:14 +0000
@@ -28,6 +28,99 @@ along with GNU Emacs.  If not, see <http
 #include "lisp.h"
 #include "buffer.h"
 
+
+static Lisp_Object Qlibxml2_dll;
+
+#ifdef WINDOWSNT
+
+#include <windows.h>
+#include "w32.h"
+
+/* Macro for defining functions that will be loaded from the libxml2 DLL.  */
+#define DEF_XML2_FN(rettype,func,args) static rettype (FAR CDECL *fn_##func)args
+
+/* Macro for loading libxml2 functions from the library.  */
+#define LOAD_XML2_FN(lib,func) {					\
+    fn_##func = (void *) GetProcAddress (lib, #func);			\
+    if (!fn_##func) goto bad_library;					\
+  }
+
+DEF_XML2_FN (htmlDocPtr, htmlReadMemory,
+	     (const char *, int, const char *, const char *, int));
+DEF_XML2_FN (xmlDocPtr, xmlReadMemory,
+	     (const char *, int, const char *, const char *, int));
+DEF_XML2_FN (xmlNodePtr, xmlDocGetRootElement, (xmlDocPtr));
+DEF_XML2_FN (void, xmlFreeDoc, (xmlDocPtr));
+DEF_XML2_FN (void, xmlCleanupParser, (void));
+DEF_XML2_FN (void, xmlCheckVersion, (int));
+
+static int
+libxml2_loaded_p (void)
+{
+  Lisp_Object found = Fassq (Qlibxml2_dll, Vlibrary_cache);
+
+  if (CONSP (found))
+    return EQ (XCDR (found), Qt) ? 1 : 0;
+  return 0;
+}
+
+#else  /* !WINDOWSNT */
+
+#define fn_htmlReadMemory       htmlReadMemory
+#define fn_xmlReadMemory        xmlReadMemory
+#define fn_xmlDocGetRootElement xmlDocGetRootElement
+#define fn_xmlFreeDoc           xmlFreeDoc
+#define fn_xmlCleanupParser     xmlCleanupParser
+#define fn_xmlCheckVersion      xmlCheckVersion
+
+static inine int
+libxml2_loaded_p (void)
+{
+  return 1;
+}
+
+#endif	/* !WINDOWSNT */
+
+static int
+init_libxml2_functions (Lisp_Object libraries)
+{
+#ifdef WINDOWSNT
+  Lisp_Object found = Fassq (Qlibxml2_dll, Vlibrary_cache);
+
+  if (libxml2_loaded_p ())
+    return 1;
+  else
+    {
+      HMODULE library;
+
+      if (!(library = w32_delayed_load (libraries, Qlibxml2_dll)))
+	{
+	  message ("%s", "libxml2 library not found");
+	  return 0;
+	}
+
+      /* LOAD_XML2_FN jumps to bad_library if it fails to find the
+	 named function.  */
+      LOAD_XML2_FN (library, htmlReadMemory);
+      LOAD_XML2_FN (library, xmlReadMemory);
+      LOAD_XML2_FN (library, xmlDocGetRootElement);
+      LOAD_XML2_FN (library, xmlFreeDoc);
+      LOAD_XML2_FN (library, xmlCleanupParser);
+      LOAD_XML2_FN (library, xmlCheckVersion);
+
+      Vlibrary_cache = Fcons (Fcons (Qlibxml2_dll, Qt), Vlibrary_cache);
+      return 1;
+    }
+
+ bad_library:
+  Vlibrary_cache = Fcons (Fcons (Qlibxml2_dll, Qnil), Vlibrary_cache);
+
+  return 0;
+#else  /* !WINDOWSNT */
+  return 1;
+#endif	/* !WINDOWSNT */
+}
+
 static Lisp_Object
 make_dom (xmlNode *node)
 {
@@ -92,7 +185,7 @@ parse_region (Lisp_Object start, Lisp_Ob
   EMACS_INT bytes;
   EMACS_INT istart, iend;
 
-  LIBXML_TEST_VERSION;
+  fn_xmlCheckVersion (LIBXML_VERSION);
 
   validate_region (&start, &end);
 
@@ -111,16 +204,16 @@ parse_region (Lisp_Object start, Lisp_Ob
   bytes = CHAR_TO_BYTE (iend) - CHAR_TO_BYTE (istart);
 
   if (htmlp)
-    doc = htmlReadMemory ((char *) BYTE_POS_ADDR (CHAR_TO_BYTE (istart)),
-			  bytes, burl, "utf-8",
-			  HTML_PARSE_RECOVER|HTML_PARSE_NONET|
-			  HTML_PARSE_NOWARNING|HTML_PARSE_NOERROR|
-			  HTML_PARSE_NOBLANKS);
+    doc = fn_htmlReadMemory ((char *) BYTE_POS_ADDR (CHAR_TO_BYTE (istart)),
+			     bytes, burl, "utf-8",
+			     HTML_PARSE_RECOVER|HTML_PARSE_NONET|
+			     HTML_PARSE_NOWARNING|HTML_PARSE_NOERROR|
+			     HTML_PARSE_NOBLANKS);
   else
-    doc = xmlReadMemory ((char *) BYTE_POS_ADDR (CHAR_TO_BYTE (istart)),
-			 bytes, burl, "utf-8",
-			 XML_PARSE_NONET|XML_PARSE_NOWARNING|
-			 XML_PARSE_NOBLANKS |XML_PARSE_NOERROR);
+    doc = fn_xmlReadMemory ((char *) BYTE_POS_ADDR (CHAR_TO_BYTE (istart)),
+			    bytes, burl, "utf-8",
+			    XML_PARSE_NONET|XML_PARSE_NOWARNING|
+			    XML_PARSE_NOBLANKS |XML_PARSE_NOERROR);
 
   if (doc != NULL)
     {
@@ -139,19 +232,26 @@ parse_region (Lisp_Object start, Lisp_Ob
       if (NILP (result)) {
 	/* The document isn't just comments, so get the tree the
 	   proper way. */
-	xmlNode *node = xmlDocGetRootElement (doc);
+	xmlNode *node = fn_xmlDocGetRootElement (doc);
 	if (node != NULL)
 	  result = make_dom (node);
       } else
 	result = Fcons (intern ("top"),
 			Fcons (Qnil, Fnreverse (Fcons (r, result))));
 
-      xmlFreeDoc (doc);
+      fn_xmlFreeDoc (doc);
     }
 
   return result;
 }
 
+void
+xml_cleanup_parser (void)
+{
+  if (libxml2_loaded_p ())
+    fn_xmlCleanupParser ();
+}
+
 DEFUN ("libxml-parse-html-region", Flibxml_parse_html_region,
        Slibxml_parse_html_region,
        2, 3, 0,
@@ -159,7 +259,9 @@ DEFUN ("libxml-parse-html-region", Flibx
 If BASE-URL is non-nil, it is used to expand relative URLs.  */)
   (Lisp_Object start, Lisp_Object end, Lisp_Object base_url)
 {
-  return parse_region (start, end, base_url, 1);
+  if (init_libxml2_functions (Vdynamic_library_alist))
+    return parse_region (start, end, base_url, 1);
+  return Qnil;
 }
 
 DEFUN ("libxml-parse-xml-region", Flibxml_parse_xml_region,
@@ -169,7 +271,9 @@ DEFUN ("libxml-parse-xml-region", Flibxm
 If BASE-URL is non-nil, it is used to expand relative URLs.  */)
   (Lisp_Object start, Lisp_Object end, Lisp_Object base_url)
 {
-  return parse_region (start, end, base_url, 0);
+  if (init_libxml2_functions (Vdynamic_library_alist))
+    return parse_region (start, end, base_url, 0);
+  return Qnil;
 }
 
 
@@ -181,6 +285,8 @@ syms_of_xml (void)
 {
   defsubr (&Slibxml_parse_html_region);
   defsubr (&Slibxml_parse_xml_region);
+
+  DEFSYM (Qlibxml2_dll, "libxml2");
 }
 
 #endif /* HAVE_LIBXML2 */

Message #10 received at 10777-done <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 10777-done <at> debbugs.gnu.org
Subject: Re: bug#10777: 24.0.93;
	Enhancement: Add support for libxml2 on MS-Windows
Date: Sat, 07 Apr 2012 17:01:56 +0300

> Date: Fri, 10 Feb 2012 12:21:55 +0200
> From: Eli Zaretskii <eliz <at> gnu.org>
> 
> Package: emacs
> Severity: wishlist
> Tags: patch
> 
> See the subject.  The patch below accomplishes that, but in
> 
>   http://lists.gnu.org/archive/html/emacs-devel/2012-02/msg00288.html
> 
> it was decided that it should be applied after Emacs 24.1 is released.

Changes installed as trunk revision 107786.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.