From 9d17f25521d675d356ecde4a383d82ae28070df9 Mon Sep 17 00:00:00 2001
From: Radim Blazek <radim.blazek@gmail.com>
Date: Tue, 3 Dec 2013 22:26:30 +0100
Subject: [PATCH] Convert WMS GetFeatureInfo GML to UTF-8 supported by Expat,
 fixes #9082

---
 src/core/qgsgml.h                    |  5 ++++-
 src/core/qgsgmlschema.h              |  1 +
 src/providers/wms/qgswmsprovider.cpp | 18 +++++++++++++++---
 3 files changed, 20 insertions(+), 4 deletions(-)

diff --git a/src/core/qgsgml.h b/src/core/qgsgml.h
index 2f82ea50796a..496d7abb9cbc 100644
--- a/src/core/qgsgml.h
+++ b/src/core/qgsgml.h
@@ -50,6 +50,7 @@ class CORE_EXPORT QgsGml : public QObject
     ~QgsGml();
 
     /** Does the Http GET request to the wfs server
+     *  Supports only UTF-8, UTF-16, ISO-8859-1, ISO-8859-1 XML encodings.
      *  @param uri GML URL
      *  @param wkbType wkbType to retrieve
      *  @param extent retrieved extents
@@ -57,7 +58,9 @@ class CORE_EXPORT QgsGml : public QObject
      */
     int getFeatures( const QString& uri, QGis::WkbType* wkbType, QgsRectangle* extent = 0 );
 
-    /** Read from GML data. Constructor uri param is ignored */
+    /** Read from GML data. Constructor uri param is ignored
+     *  Supports only UTF-8, UTF-16, ISO-8859-1, ISO-8859-1 XML encodings.
+     */
     int getFeatures( const QByteArray &data, QGis::WkbType* wkbType, QgsRectangle* extent = 0 );
 
     /** Get parsed features for given type name */
diff --git a/src/core/qgsgmlschema.h b/src/core/qgsgmlschema.h
index cefe71107471..4fa35d0ad21b 100644
--- a/src/core/qgsgmlschema.h
+++ b/src/core/qgsgmlschema.h
@@ -85,6 +85,7 @@ class CORE_EXPORT QgsGmlSchema: public QObject
 
     /** Guess GML schema from data if XSD does not exist.
       * Currently only recognizes UMN Mapserver GetFeatureInfo GML response.
+      * Supports only UTF-8, UTF-16, ISO-8859-1, US-ASCII XML encodings.
       * @param data GML data
       * @return true in case of success */
     bool guessSchema( const QByteArray &data );
diff --git a/src/providers/wms/qgswmsprovider.cpp b/src/providers/wms/qgswmsprovider.cpp
index 111bed0e9c01..33667526a772 100644
--- a/src/providers/wms/qgswmsprovider.cpp
+++ b/src/providers/wms/qgswmsprovider.cpp
@@ -61,6 +61,7 @@
 #include <QSettings>
 #include <QEventLoop>
 #include <QCoreApplication>
+#include <QTextCodec>
 #include <QTime>
 
 #ifdef QGISDEBUG
@@ -4376,7 +4377,18 @@ QgsRasterIdentifyResult QgsWmsProvider::identify( const QgsPoint & thePoint, Qgs
         }
       }
 
-      QgsDebugMsg( "GML (first 2000 bytes):\n" + QString::fromUtf8( mIdentifyResultBodies.value( gmlPart ).left( 2000 ) ) );
+      QByteArray gmlByteArray = mIdentifyResultBodies.value( gmlPart );
+      QgsDebugMsg( "GML (first 2000 bytes):\n" + gmlByteArray.left( 2000 ) );
+
+      // QgsGmlSchema.guessSchema() and QgsGml::getFeatures() are using Expat
+      // which only accepts UTF-8, UTF-16, ISO-8859-1
+      // http://sourceforge.net/p/expat/bugs/498/
+      QDomDocument dom;
+      dom.setContent( gmlByteArray ); // gets XML encoding
+      QTextStream stream( &gmlByteArray );
+      stream.setCodec( QTextCodec::codecForName( "UTF-8" ) );
+      dom.save( stream, 4, QDomNode::EncodingFromTextStream );
+
       QGis::WkbType wkbType;
       QgsGmlSchema gmlSchema;
 
@@ -4418,7 +4430,7 @@ QgsRasterIdentifyResult QgsWmsProvider::identify( const QgsPoint & thePoint, Qgs
       else
       {
         // guess from GML
-        bool ok = gmlSchema.guessSchema( mIdentifyResultBodies.value( gmlPart ) );
+        bool ok = gmlSchema.guessSchema( gmlByteArray );
         if ( ! ok )
         {
           QgsError err = gmlSchema.error();
@@ -4454,7 +4466,7 @@ QgsRasterIdentifyResult QgsWmsProvider::identify( const QgsPoint & thePoint, Qgs
         }
         QgsGml gml( featureTypeName, geometryAttribute, fields );
         // TODO: avoid converting to string and back
-        int ret = gml.getFeatures( mIdentifyResultBodies.value( gmlPart ), &wkbType );
+        int ret = gml.getFeatures( gmlByteArray, &wkbType );
 #ifdef QGISDEBUG
         QgsDebugMsg( QString( "parsing result = %1" ).arg( ret ) );
 #else