Update libxml to 2.7.7.

In doing so, I recreated all the patches and organized them under
a patches/ subdirectory.

BUG=32197

Review URL: http://codereview.chromium.org/2951008

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@52725 0039d316-1c4b-4281-b951-d872f2087c98

7 files changed, 637 insertions, 0 deletions

diff --git a/third_party/libxml/patches/LoadLibraryA b/third_party/libxml/patches/LoadLibraryA
new file mode 100644
index 0000000..89fff15
--- /dev/null
+++ b/third_party/libxml/patches/LoadLibraryA
@@ -0,0 +1,16 @@
+Change 'LoadLibrary' to 'LoadLibraryA' (used with 'const char*' as an
+argument)
+
+Index: libxml/xmlmodule.c
+===================================================================
+--- libxml.orig/xmlmodule.c	2010-07-09 14:17:46.959288280 -0700
++++ libxml/xmlmodule.c	2010-07-09 14:17:55.419051003 -0700
+@@ -300,7 +300,7 @@
+ static void *
+ xmlModulePlatformOpen(const char *name)
+ {
+-    return LoadLibrary(name);
++    return LoadLibraryA(name);
+ }
+ 
+ /*
diff --git a/third_party/libxml/patches/icu b/third_party/libxml/patches/icu
new file mode 100644
index 0000000..4503f92
--- /dev/null
+++ b/third_party/libxml/patches/icu
@@ -0,0 +1,453 @@
+Add code support for ICU.
+
+diff --git a/third_party/libxml/encoding.c b/third_party/libxml/encoding.c
+index b86a547..0f41df9 100644
+--- a/third_party/libxml/encoding.c
++++ b/third_party/libxml/encoding.c
+@@ -58,7 +58,7 @@ static xmlCharEncodingAliasPtr xmlCharEncodingAliases = NULL;
+ static int xmlCharEncodingAliasesNb = 0;
+ static int xmlCharEncodingAliasesMax = 0;
+ 
+-#ifdef LIBXML_ICONV_ENABLED
++#if defined(LIBXML_ICONV_ENABLED) || defined(LIBXML_ICU_ENABLED)
+ #if 0
+ #define DEBUG_ENCODING  /* Define this to get encoding traces */
+ #endif
+@@ -97,6 +97,54 @@ xmlEncodingErr(xmlParserErrors error, const char *msg, const char *val)
+                     NULL, 0, val, NULL, NULL, 0, 0, msg, val);
+ }
+ 
++#ifdef LIBXML_ICU_ENABLED
++static uconv_t* 
++openIcuConverter(const char* name, int toUnicode)
++{
++  UErrorCode status = U_ZERO_ERROR;
++  uconv_t *conv = (uconv_t *) xmlMalloc(sizeof(uconv_t));
++  if (conv == NULL)
++    return NULL;
++
++  conv->uconv = ucnv_open(name, &status);
++  if (U_FAILURE(status))
++    goto error;
++
++  status = U_ZERO_ERROR;
++  if (toUnicode) {
++    ucnv_setToUCallBack(conv->uconv, UCNV_TO_U_CALLBACK_STOP, 
++                        NULL, NULL, NULL, &status);
++  }
++  else {
++    ucnv_setFromUCallBack(conv->uconv, UCNV_FROM_U_CALLBACK_STOP, 
++                        NULL, NULL, NULL, &status);
++  }
++  if (U_FAILURE(status))
++    goto error;
++
++  status = U_ZERO_ERROR;
++  conv->utf8 = ucnv_open("UTF-8", &status);
++  if (U_SUCCESS(status))
++    return conv;
++
++error:
++  if (conv->uconv) 
++    ucnv_close(conv->uconv);
++  xmlFree(conv);
++  return NULL;
++}
++
++static void
++closeIcuConverter(uconv_t *conv)
++{
++  if (conv != NULL) {
++    ucnv_close(conv->uconv);
++    ucnv_close(conv->utf8);
++    xmlFree(conv);
++  }
++}
++#endif /* LIBXML_ICU_ENABLED */
++
+ /************************************************************************
+  *									*
+  *		Conversions To/From UTF8 encoding			*
+@@ -1306,7 +1354,11 @@ xmlNewCharEncodingHandler(const char *name,
+ #ifdef LIBXML_ICONV_ENABLED
+     handler->iconv_in = NULL;
+     handler->iconv_out = NULL;
+-#endif /* LIBXML_ICONV_ENABLED */
++#endif
++#ifdef LIBXML_ICU_ENABLED
++    handler->uconv_in = NULL;
++    handler->uconv_out = NULL;
++#endif
+ 
+     /*
+      * registers and returns the handler.
+@@ -1371,7 +1423,7 @@ xmlInitCharEncodingHandlers(void) {
+     xmlNewCharEncodingHandler("ASCII", asciiToUTF8, NULL);
+     xmlNewCharEncodingHandler("US-ASCII", asciiToUTF8, NULL);
+ #endif /* LIBXML_OUTPUT_ENABLED */
+-#ifndef LIBXML_ICONV_ENABLED
++#if !defined(LIBXML_ICONV_ENABLED) && !defined(LIBXML_ICU_ENABLED)
+ #ifdef LIBXML_ISO8859X_ENABLED
+     xmlRegisterCharEncodingHandlersISO8859x ();
+ #endif
+@@ -1578,6 +1630,10 @@ xmlFindCharEncodingHandler(const char *name) {
+     xmlCharEncodingHandlerPtr enc;
+     iconv_t icv_in, icv_out;
+ #endif /* LIBXML_ICONV_ENABLED */
++#ifdef LIBXML_ICU_ENABLED
++    xmlCharEncodingHandlerPtr enc;
++    uconv_t *ucv_in, *ucv_out;
++#endif /* LIBXML_ICU_ENABLED */
+     char upper[100];
+     int i;
+ 
+@@ -1647,6 +1703,35 @@ xmlFindCharEncodingHandler(const char *name) {
+ 		    "iconv : problems with filters for '%s'\n", name);
+     }
+ #endif /* LIBXML_ICONV_ENABLED */
++#ifdef LIBXML_ICU_ENABLED
++    /* check whether icu can handle this */
++    ucv_in = openIcuConverter(name, 1);
++    ucv_out = openIcuConverter(name, 0);
++    if (ucv_in != NULL && ucv_out != NULL) {
++	    enc = (xmlCharEncodingHandlerPtr)
++	          xmlMalloc(sizeof(xmlCharEncodingHandler));
++	    if (enc == NULL) {
++                closeIcuConverter(ucv_in);
++                closeIcuConverter(ucv_out);
++		return(NULL);
++	    }
++	    enc->name = xmlMemStrdup(name);
++	    enc->input = NULL;
++	    enc->output = NULL;
++	    enc->uconv_in = ucv_in;
++	    enc->uconv_out = ucv_out;
++#ifdef DEBUG_ENCODING
++            xmlGenericError(xmlGenericErrorContext,
++		    "Found ICU converter handler for encoding %s\n", name);
++#endif
++	    return enc;
++    } else if (ucv_in != NULL || ucv_out != NULL) {
++            closeIcuConverter(ucv_in);
++            closeIcuConverter(ucv_out);
++	    xmlEncodingErr(XML_ERR_INTERNAL_ERROR,
++		    "ICU converter : problems with filters for '%s'\n", name);
++    }
++#endif /* LIBXML_ICU_ENABLED */
+ 
+ #ifdef DEBUG_ENCODING
+     xmlGenericError(xmlGenericErrorContext,
+@@ -1737,6 +1822,75 @@ xmlIconvWrapper(iconv_t cd, unsigned char *out, int *outlen,
+ 
+ /************************************************************************
+  *									*
++ *		ICU based generic conversion functions	         	*
++ *									*
++ ************************************************************************/
++
++#ifdef LIBXML_ICU_ENABLED
++/**
++ * xmlUconvWrapper:
++ * @cd: ICU uconverter data structure
++ * @toUnicode : non-zero if toUnicode. 0 otherwise.
++ * @out:  a pointer to an array of bytes to store the result
++ * @outlen:  the length of @out
++ * @in:  a pointer to an array of ISO Latin 1 chars
++ * @inlen:  the length of @in
++ *
++ * Returns 0 if success, or 
++ *     -1 by lack of space, or
++ *     -2 if the transcoding fails (for *in is not valid utf8 string or
++ *        the result of transformation can't fit into the encoding we want), or
++ *     -3 if there the last byte can't form a single output char.
++ *     
++ * The value of @inlen after return is the number of octets consumed
++ *     as the return value is positive, else unpredictable.
++ * The value of @outlen after return is the number of ocetes consumed.
++ */
++static int
++xmlUconvWrapper(uconv_t *cd, int toUnicode, unsigned char *out, int *outlen,
++                const unsigned char *in, int *inlen) {
++    const char *ucv_in = (const char *) in;
++    char *ucv_out = (char *) out;
++    UErrorCode err = U_ZERO_ERROR;
++
++    if ((out == NULL) || (outlen == NULL) || (inlen == NULL) || (in == NULL)) {
++        if (outlen != NULL) *outlen = 0;
++        return(-1);
++    }
++
++    /* 
++     * TODO(jungshik)
++     * 1. is ucnv_convert(To|From)Algorithmic better?
++     * 2. had we better use an explicit pivot buffer?
++     * 3. error returned comes from 'fromUnicode' only even
++     *    when toUnicode is true !
++     */
++    if (toUnicode) {
++        /* encoding => UTF-16 => UTF-8 */
++        ucnv_convertEx(cd->utf8, cd->uconv, &ucv_out, ucv_out + *outlen,
++                       &ucv_in, ucv_in + *inlen, NULL, NULL, NULL, NULL,
++                       0, TRUE, &err);
++    } else {
++        /* UTF-8 => UTF-16 => encoding */
++        ucnv_convertEx(cd->uconv, cd->utf8, &ucv_out, ucv_out + *outlen,
++                       &ucv_in, ucv_in + *inlen, NULL, NULL, NULL, NULL,
++                       0, TRUE, &err);
++    }
++    *inlen = ucv_in - (const char*) in; 
++    *outlen = ucv_out - (char *) out;
++    if (U_SUCCESS(err))
++        return 0;
++    if (err == U_BUFFER_OVERFLOW_ERROR)
++        return -1;
++    if (err == U_INVALID_CHAR_FOUND || err == U_ILLEGAL_CHAR_FOUND)
++        return -2;
++    /* if (err == U_TRUNCATED_CHAR_FOUND) */
++    return -3;
++}
++#endif /* LIBXML_ICU_ENABLED */
++
++/************************************************************************
++ *									*
+  *		The real API used by libxml for on-the-fly conversion	*
+  *									*
+  ************************************************************************/
+@@ -1810,6 +1964,16 @@ xmlCharEncFirstLineInt(xmlCharEncodingHandler *handler, xmlBufferPtr out,
+ 	if (ret == -1) ret = -3;
+     }
+ #endif /* LIBXML_ICONV_ENABLED */
++#ifdef LIBXML_ICU_ENABLED
++    else if (handler->uconv_in != NULL) {
++	ret = xmlUconvWrapper(handler->uconv_in, 1, &out->content[out->use],
++	                      &written, in->content, &toconv);
++	xmlBufferShrink(in, toconv);
++	out->use += written;
++	out->content[out->use] = 0;
++	if (ret == -1) ret = -3;
++    }
++#endif /* LIBXML_ICU_ENABLED */
+ #ifdef DEBUG_ENCODING
+     switch (ret) {
+         case 0:
+@@ -1915,6 +2079,17 @@ xmlCharEncInFunc(xmlCharEncodingHandler * handler, xmlBufferPtr out,
+             ret = -3;
+     }
+ #endif /* LIBXML_ICONV_ENABLED */
++#ifdef LIBXML_ICU_ENABLED
++    else if (handler->uconv_in != NULL) {
++        ret = xmlUconvWrapper(handler->uconv_in, 1, &out->content[out->use],
++                              &written, in->content, &toconv);
++        xmlBufferShrink(in, toconv);
++        out->use += written;
++        out->content[out->use] = 0;
++        if (ret == -1)
++            ret = -3;
++    }
++#endif /* LIBXML_ICU_ENABLED */
+     switch (ret) {
+         case 0:
+ #ifdef DEBUG_ENCODING
+@@ -2015,6 +2190,15 @@ retry:
+ 	    out->content[out->use] = 0;
+ 	}
+ #endif /* LIBXML_ICONV_ENABLED */
++#ifdef LIBXML_ICU_ENABLED
++	else if (handler->uconv_out != NULL) {
++	    ret = xmlUconvWrapper(handler->uconv_out, 0,
++                              &out->content[out->use],
++ 				              &written, NULL, &toconv);
++	    out->use += written;
++	    out->content[out->use] = 0;
++	}
++#endif /* LIBXML_ICU_ENABLED */
+ #ifdef DEBUG_ENCODING
+ 	xmlGenericError(xmlGenericErrorContext,
+ 		"initialized encoder\n");
+@@ -2061,6 +2245,26 @@ retry:
+ 	}
+     }
+ #endif /* LIBXML_ICONV_ENABLED */
++#ifdef LIBXML_ICU_ENABLED
++    else if (handler->uconv_out != NULL) {
++	ret = xmlUconvWrapper(handler->uconv_out, 0,
++                              &out->content[out->use],
++	                      &written, in->content, &toconv);
++	xmlBufferShrink(in, toconv);
++	out->use += written;
++	writtentot += written;
++	out->content[out->use] = 0;
++	if (ret == -1) {
++	    if (written > 0) {
++		/*
++		 * Can be a limitation of iconv
++		 */
++		goto retry;
++	    }
++	    ret = -3;
++	}
++    }
++#endif /* LIBXML_ICU_ENABLED */
+     else {
+ 	xmlEncodingErr(XML_I18N_NO_OUTPUT,
+ 		       "xmlCharEncOutFunc: no output function !\n", NULL);
+@@ -2173,6 +2377,22 @@ xmlCharEncCloseFunc(xmlCharEncodingHandler *handler) {
+ 	xmlFree(handler);
+     }
+ #endif /* LIBXML_ICONV_ENABLED */
++#ifdef LIBXML_ICU_ENABLED
++    if ((handler->uconv_out != NULL) || (handler->uconv_in != NULL)) {
++	if (handler->name != NULL)
++	    xmlFree(handler->name);
++	handler->name = NULL;
++	if (handler->uconv_out != NULL) {
++	    closeIcuConverter(handler->uconv_out);
++	    handler->uconv_out = NULL;
++	}
++	if (handler->uconv_in != NULL) {
++	    closeIcuConverter(handler->uconv_in);
++	    handler->uconv_in = NULL;
++	}
++	xmlFree(handler);
++    }
++#endif
+ #ifdef DEBUG_ENCODING
+     if (ret)
+         xmlGenericError(xmlGenericErrorContext,
+@@ -2248,6 +2468,22 @@ xmlByteConsumed(xmlParserCtxtPtr ctxt) {
+ 		    cur += toconv;
+ 		} while (ret == -2);
+ #endif
++#ifdef LIBXML_ICU_ENABLED
++	    } else if (handler->uconv_out != NULL) {
++	        do {
++		    toconv = in->end - cur;
++		    written = 32000;
++		    ret = xmlUconvWrapper(handler->uconv_out, 0, &convbuf[0],
++	                      &written, cur, &toconv);
++		    if (ret < 0) {
++		        if (written > 0)
++			    ret = -2;
++			else
++			    return(-1);
++		    }
++		    unused += written;
++		    cur += toconv;
++		} while (ret == -2);
+             } else {
+ 	        /* could not find a converter */
+ 	        return(-1);
+@@ -2259,8 +2495,9 @@ xmlByteConsumed(xmlParserCtxtPtr ctxt) {
+     }
+     return(in->consumed + (in->cur - in->base));
+ }
++#endif
+ 
+-#ifndef LIBXML_ICONV_ENABLED
++#if !defined(LIBXML_ICONV_ENABLED) && !defined(LIBXML_ICU_ENABLED)
+ #ifdef LIBXML_ISO8859X_ENABLED
+ 
+ /**
+diff --git a/third_party/libxml/include/libxml/encoding.h b/third_party/libxml/include/libxml/encoding.h
+index c74b25f..b5f8b48 100644
+--- a/third_party/libxml/include/libxml/encoding.h
++++ b/third_party/libxml/include/libxml/encoding.h
+@@ -26,6 +26,24 @@
+ 
+ #ifdef LIBXML_ICONV_ENABLED
+ #include <iconv.h>
++#else 
++#ifdef LIBXML_ICU_ENABLED
++#include <unicode/ucnv.h>
++#if 0
++/* Forward-declare UConverter here rather than pulling in <unicode/ucnv.h>
++ * to prevent unwanted ICU symbols being exposed to users of libxml2.
++ * One particular case is Qt4 conflicting on UChar32.
++ */
++#include <stdint.h>
++struct UConverter;
++typedef struct UConverter UConverter;
++#ifdef _MSC_VER
++typedef wchar_t UChar;
++#else
++typedef uint16_t UChar;
++#endif
++#endif
++#endif
+ #endif
+ #ifdef __cplusplus
+ extern "C" {
+@@ -125,6 +143,13 @@ typedef int (* xmlCharEncodingOutputFunc)(unsigned char *out, int *outlen,
+  * Block defining the handlers for non UTF-8 encodings.
+  * If iconv is supported, there are two extra fields.
+  */
++#ifdef LIBXML_ICU_ENABLED
++struct _uconv_t {
++  UConverter *uconv; /* for conversion between an encoding and UTF-16 */
++  UConverter *utf8; /* for conversion between UTF-8 and UTF-16 */
++};
++typedef struct _uconv_t uconv_t;
++#endif
+ 
+ typedef struct _xmlCharEncodingHandler xmlCharEncodingHandler;
+ typedef xmlCharEncodingHandler *xmlCharEncodingHandlerPtr;
+@@ -136,6 +161,10 @@ struct _xmlCharEncodingHandler {
+     iconv_t                    iconv_in;
+     iconv_t                    iconv_out;
+ #endif /* LIBXML_ICONV_ENABLED */
++#ifdef LIBXML_ICU_ENABLED
++    uconv_t                    *uconv_in;
++    uconv_t                    *uconv_out;
++#endif /* LIBXML_ICU_ENABLED */
+ };
+ 
+ #ifdef __cplusplus
+diff --git a/third_party/libxml/include/libxml/parser.h b/third_party/libxml/include/libxml/parser.h
+index dd79c42..3580b63 100644
+--- a/third_party/libxml/include/libxml/parser.h
++++ b/third_party/libxml/include/libxml/parser.h
+@@ -1222,6 +1222,7 @@ typedef enum {
+     XML_WITH_DEBUG_MEM = 29,
+     XML_WITH_DEBUG_RUN = 30,
+     XML_WITH_ZLIB = 31,
++    XML_WITH_ICU = 32,
+     XML_WITH_NONE = 99999 /* just to be sure of allocation size */
+ } xmlFeature;
+ 
+diff --git a/third_party/libxml/include/libxml/xmlversion.h.in b/third_party/libxml/include/libxml/xmlversion.h.in
+index 4739f3a..de310ab 100644
+--- a/third_party/libxml/include/libxml/xmlversion.h.in
++++ b/third_party/libxml/include/libxml/xmlversion.h.in
+@@ -269,6 +269,15 @@ XMLPUBFUN void XMLCALL xmlCheckVersion(int version);
+ #endif
+ 
+ /**
++ * LIBXML_ICU_ENABLED:
++ *
++ * Whether icu support is available
++ */
++#if @WITH_ICU@
++#define LIBXML_ICU_ENABLED
++#endif
++
++/**
+  * LIBXML_ISO8859X_ENABLED:
+  *
+  * Whether ISO-8859-* support is made available in case iconv is not
+diff --git a/third_party/libxml/parser.c b/third_party/libxml/parser.c
+index 85e7599..3ba2a06 100644
+--- a/third_party/libxml/parser.c
++++ b/third_party/libxml/parser.c
+@@ -954,6 +954,12 @@ xmlHasFeature(xmlFeature feature)
+ #else
+             return(0);
+ #endif
++        case XML_WITH_ICU:
++#ifdef LIBXML_ICU_ENABLED
++            return(1);
++#else
++            return(0);
++#endif
+         default:
+ 	    break;
+      }
diff --git a/third_party/libxml/patches/icu-configure b/third_party/libxml/patches/icu-configure
new file mode 100644
index 0000000..f7e2395
--- /dev/null
+++ b/third_party/libxml/patches/icu-configure
@@ -0,0 +1,28 @@
+Modifications to configure.in:
+- set the WITH_ICU flag unconditionally
+- only output files we use
+
+Index: libxml/configure.in
+===================================================================
+--- libxml.orig/configure.in	2010-07-09 15:00:21.600113911 -0700
++++ libxml/configure.in	2010-07-09 15:02:50.299108047 -0700
+@@ -1229,6 +1229,9 @@
+ fi
+ AC_SUBST(WITH_OUTPUT)
+ 
++WITH_ICU=1
++AC_SUBST(WITH_ICU)
++
+ WITH_ICONV=0
+ if test "$with_iconv" = "no" ; then
+     echo Disabling ICONV support
+@@ -1456,7 +1459,7 @@
+ ln -s Copyright COPYING
+ 
+ # keep on one line for cygwin c.f. #130896
+-AC_OUTPUT(libxml2.spec:libxml.spec.in Makefile include/Makefile include/libxml/Makefile doc/Makefile doc/examples/Makefile doc/devhelp/Makefile example/Makefile python/Makefile python/tests/Makefile xstc/Makefile include/libxml/xmlversion.h xml2-config libxml-2.0.pc libxml-2.0-uninstalled.pc python/setup.py)
++AC_OUTPUT(include/libxml/xmlversion.h xml2-config)
+ 
+-chmod +x xml2-config python/setup.py
++chmod +x xml2-config
+ echo Done configuring
diff --git a/third_party/libxml/patches/icu-win32 b/third_party/libxml/patches/icu-win32
new file mode 100644
index 0000000..2ea8f21
--- /dev/null
+++ b/third_party/libxml/patches/icu-win32
@@ -0,0 +1,68 @@
+diff --git a/third_party/libxml/win32/Makefile.msvc b/third_party/libxml/win32/Makefile.msvc
+index 2409905..253c46e 100644
+--- a/third_party/libxml/win32/Makefile.msvc
++++ b/third_party/libxml/win32/Makefile.msvc
+@@ -71,6 +71,9 @@ LIBS = $(LIBS) wsock32.lib ws2_32.lib
+ !if "$(WITH_ICONV)" == "1"
+ LIBS = $(LIBS) iconv.lib
+ !endif 
++!if "$(WITH_ICU)" == "1"
++LIBS = $(LIBS) icu.lib
++!endif
+ !if "$(WITH_ZLIB)" == "1"
+ LIBS = $(LIBS) zdll.lib
+ !endif
+diff --git a/third_party/libxml/win32/configure.js b/third_party/libxml/win32/configure.js
+index e71d2af..75def3f 100644
+--- a/third_party/libxml/win32/configure.js
++++ b/third_party/libxml/win32/configure.js
+@@ -40,6 +40,7 @@ var withXpath = true;
+ var withXptr = true;
+ var withXinclude = true;
+ var withIconv = true;
++var withIcu = false;
+ var withIso8859x = false;
+ var withZlib = false;
+ var withDebug = true;
+@@ -124,6 +125,7 @@ function usage()
+ 	txt += "  xptr:       Enable XPointer support (" + (withXptr? "yes" : "no")  + ")\n";
+ 	txt += "  xinclude:   Enable XInclude support (" + (withXinclude? "yes" : "no")  + ")\n";
+ 	txt += "  iconv:      Enable iconv support (" + (withIconv? "yes" : "no")  + ")\n";
++	txt += "  icu:        Enable icu support (" + (withIcu? "yes" : "no")  + ")\n";
+ 	txt += "  iso8859x:   Enable ISO8859X support (" + (withIso8859x? "yes" : "no")  + ")\n";
+ 	txt += "  zlib:       Enable zlib support (" + (withZlib? "yes" : "no")  + ")\n";
+ 	txt += "  xml_debug:  Enable XML debbugging module (" + (withDebug? "yes" : "no")  + ")\n";
+@@ -233,6 +235,7 @@ function discoverVersion()
+ 	vf.WriteLine("WITH_XPTR=" + (withXptr? "1" : "0"));
+ 	vf.WriteLine("WITH_XINCLUDE=" + (withXinclude? "1" : "0"));
+ 	vf.WriteLine("WITH_ICONV=" + (withIconv? "1" : "0"));
++	vf.WriteLine("WITH_ICU=" + (withIcu? "1" : "0"));
+ 	vf.WriteLine("WITH_ISO8859X=" + (withIso8859x? "1" : "0"));
+ 	vf.WriteLine("WITH_ZLIB=" + (withZlib? "1" : "0"));
+ 	vf.WriteLine("WITH_DEBUG=" + (withDebug? "1" : "0"));
+@@ -319,6 +322,8 @@ function configureLibxml()
+ 			of.WriteLine(s.replace(/\@WITH_XINCLUDE\@/, withXinclude? "1" : "0"));
+ 		} else if (s.search(/\@WITH_ICONV\@/) != -1) {
+ 			of.WriteLine(s.replace(/\@WITH_ICONV\@/, withIconv? "1" : "0"));
++		} else if (s.search(/\@WITH_ICU\@/) != -1) {
++			of.WriteLine(s.replace(/\@WITH_ICU\@/, withIcu? "1" : "0"));
+ 		} else if (s.search(/\@WITH_ISO8859X\@/) != -1) {
+ 			of.WriteLine(s.replace(/\@WITH_ISO8859X\@/, withIso8859x? "1" : "0"));
+ 		} else if (s.search(/\@WITH_ZLIB\@/) != -1) {
+@@ -462,6 +467,8 @@ for (i = 0; (i < WScript.Arguments.length) && (error == 0); i++) {
+ 			withXinclude = strToBool(arg.substring(opt.length + 1, arg.length));
+ 		else if (opt == "iconv")
+ 			withIconv = strToBool(arg.substring(opt.length + 1, arg.length));
++		else if (opt == "icu")
++			withIcu = strToBool(arg.substring(opt.length + 1, arg.length));
+ 		else if (opt == "iso8859x")
+ 			withIso8859x = strToBool(arg.substring(opt.length + 1, arg.length));
+ 		else if (opt == "zlib")
+@@ -646,6 +653,7 @@ txtOut += "     XPath support: " + boolToStr(withXpath) + "\n";
+ txtOut += "  XPointer support: " + boolToStr(withXptr) + "\n";
+ txtOut += "  XInclude support: " + boolToStr(withXinclude) + "\n";
+ txtOut += "     iconv support: " + boolToStr(withIconv) + "\n";
++txtOut += "     icu   support: " + boolToStr(withIcu) + "\n";
+ txtOut += "  iso8859x support: " + boolToStr(withIso8859x) + "\n";
+ txtOut += "      zlib support: " + boolToStr(withZlib) + "\n";
+ txtOut += "  Debugging module: " + boolToStr(withDebug) + "\n";
diff --git a/third_party/libxml/patches/libxml-disable-autodetect b/third_party/libxml/patches/libxml-disable-autodetect
new file mode 100644
index 0000000..c37d41a
--- /dev/null
+++ b/third_party/libxml/patches/libxml-disable-autodetect
@@ -0,0 +1,38 @@
+See comment below inline in the patch.
+It'd be nice to fix this properly but that will require upstream WebKit
+changes; for now, we can just patch out some code to keep behavior as it
+was before.
+
+diff --git a/third_party/libxml/parser.c b/third_party/libxml/parser.c
+index 3ba2a06..4a3cde5 100644
+--- a/third_party/libxml/parser.c
++++ b/third_party/libxml/parser.c
+@@ -11545,6 +11545,20 @@ xmldecl_done:
+ 	int cur = ctxt->input->cur - ctxt->input->base;
+ 	int res;
+ 
++        /* Chromium note: commenting out the following block of code is a gory
++         * hack, meant to partially undo
++         *  http://git.gnome.org/browse/libxml2/commit/?id=a6c76a
++         *
++         * WebKit and libxml disagree about who is responsible for the
++         * document encoding.
++         *
++         * This bug:
++         *   https://bugs.webkit.org/show_bug.cgi?id=30508
++         * has links to test cases, libxml bug reports, and mailing list threads
++         * arguing about it; for now, though, we can just undo the change that
++         * caused libxml to regress.
++         */
++#if 0
+         /*
+          * Specific handling if we autodetected an encoding, we should not
+          * push more than the first line ... which depend on the encoding
+@@ -11580,6 +11594,7 @@ xmldecl_done:
+                 remain = 0;
+             }
+         }
++#endif
+ 	res =xmlParserInputBufferPush(ctxt->input->buf, size, chunk);
+ 	if (res < 0) {
+ 	    ctxt->errNo = XML_PARSER_EOF;
diff --git a/third_party/libxml/patches/win32-clobber-makefile b/third_party/libxml/patches/win32-clobber-makefile
new file mode 100644
index 0000000..2f7fbb4
--- /dev/null
+++ b/third_party/libxml/patches/win32-clobber-makefile
@@ -0,0 +1,19 @@
+Index: libxml/win32/configure.js
+===================================================================
+--- libxml.orig/win32/configure.js	2010-07-09 14:56:07.769093841 -0700
++++ libxml/win32/configure.js	2010-07-09 15:36:48.590268611 -0700
+@@ -611,7 +611,13 @@
+ 	makefile = ".\\Makefile.mingw";
+ else if (compiler == "bcb")
+ 	makefile = ".\\Makefile.bcb";
+-fso.CopyFile(makefile, ".\\Makefile", true);
++var new_makefile = ".\\Makefile";
++var f = fso.FileExists(new_makefile);
++if (f) {
++       var t = fso.GetFile(new_makefile);
++       t.Attributes = 0;
++}
++fso.CopyFile(makefile, new_makefile, true);
+ WScript.Echo("Created Makefile.");
+ // Create the config.h.
+ var confighsrc = "..\\include\\win32config.h";
diff --git a/third_party/libxml/patches/xmlregexp-bogus-cast b/third_party/libxml/patches/xmlregexp-bogus-cast
new file mode 100644
index 0000000..aaebfaa
--- /dev/null
+++ b/third_party/libxml/patches/xmlregexp-bogus-cast
@@ -0,0 +1,15 @@
+Change bogus '(unsigned long)' cast to '(unsigned short)'
+
+Index: libxml/xmlregexp.c
+===================================================================
+--- libxml.orig/xmlregexp.c	2010-07-09 14:16:36.990430641 -0700
++++ libxml/xmlregexp.c	2010-07-09 14:16:40.939742007 -0700
+@@ -6470,7 +6470,7 @@
+     if (name != NULL) {
+ 	value += 30 * (*name);
+ 	while ((ch = *name++) != 0) {
+-	    value = value ^ ((value << 5) + (value >> 3) + (unsigned long)ch);
++	    value = value ^ ((value << 5) + (value >> 3) + (unsigned short)ch);
+ 	}
+     }
+     return (value);