add RFC103 layer->encoding implementation (#4758)

mapcontext.c

@@ -1333,10 +1333,7 @@ int msWriteMapContext(mapObj *map, FILE *stream)
   }
 
   /* file header */
-  msOWSPrintEncodeMetadata(stream, &(map->web.metadata),
-                           NULL, "wms_encoding", OWS_NOERR,
-                           "<?xml version='1.0' encoding=\"%s\" standalone=\"no\" ?>\n",
-                           "ISO-8859-1");
+  msIO_fprintf( stream, "<?xml version='1.0' encoding=\"UTF-8\" standalone=\"no\" ?>\n");
 
   /* set the WMS_Viewer_Context information */
   pszEncodedVal = msEncodeHTMLEntities(version);

mapcopy.c

@@ -934,6 +934,7 @@ int msCopyLayer(layerObj *dst, layerObj *src)
   MS_COPYSTRING(dst->name, src->name);
   MS_COPYSTRING(dst->group, src->group);
   MS_COPYSTRING(dst->data, src->data);
+  MS_COPYSTRING(dst->encoding, src->encoding);
 
   MS_COPYSTELEM(status);
   MS_COPYSTELEM(type);

mapfile.c

@@ -3918,6 +3918,8 @@ int initLayer(layerObj *layer, mapObj *map)
   initExpression(&(layer->_geomtransform));
   layer->_geomtransform.type = MS_GEOMTRANSFORM_NONE;
 
+  layer->encoding = NULL;
+
   return(0);
 }
 
@@ -3958,6 +3960,7 @@ int freeLayer(layerObj *layer)
     msLayerClose(layer);
 
   msFree(layer->name);
+  msFree(layer->encoding);
   msFree(layer->group);
   msFree(layer->data);
   msFree(layer->classitem);
@@ -4236,6 +4239,9 @@ int loadLayer(layerObj *layer, mapObj *map)
         msSetError(MS_EOFERR, NULL, "loadLayer()");
         return(-1);
         break;
+      case(ENCODING):
+        if(getString(&layer->encoding) == MS_FAILURE) return(-1);
+        break;
       case(END):
         if(layer->type == -1) {
           msSetError(MS_MISCERR, "Layer type not set.", "loadLayer()");
@@ -4638,6 +4644,7 @@ static void writeLayer(FILE *stream, int indent, layerObj *layer)
   writeKeyword(stream, indent, "CONNECTIONTYPE", layer->connectiontype, 10, MS_SDE, "SDE", MS_OGR, "OGR", MS_POSTGIS, "POSTGIS", MS_WMS, "WMS", MS_ORACLESPATIAL, "ORACLESPATIAL", MS_WFS, "WFS", MS_PLUGIN, "PLUGIN", MS_UNION, "UNION", MS_UVRASTER, "UVRASTER", MS_CONTOUR, "CONTOUR");
   writeString(stream, indent, "DATA", NULL, layer->data);
   writeNumber(stream, indent, "DEBUG", 0, layer->debug); /* is this right? see loadLayer() */
+  writeString(stream, indent, "ENCODING", NULL, layer->encoding);
   writeExtent(stream, indent, "EXTENT", layer->extent);
   writeExpression(stream, indent, "FILTER", &(layer->filter));
   writeString(stream, indent, "FILTERITEM", NULL, layer->filteritem);

mapgml.c

@@ -1151,9 +1151,7 @@ int msGMLWriteQuery(mapObj *map, char *filename, const char *namespaces)
     }
   }
 
-  /* charset encoding: lookup "gml_encoding" metadata first, then  */
-  /* "wms_encoding", and if not found then use "ISO-8859-1" as default. */
-  msOWSPrintEncodeMetadata(stream, &(map->web.metadata), namespaces, "encoding", OWS_NOERR, "<?xml version=\"1.0\" encoding=\"%s\"?>\n\n", "ISO-8859-1");
+  msIO_fprintf(stream, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n\n");
   msOWSPrintValidateMetadata(stream, &(map->web.metadata), namespaces, "rootname", OWS_NOERR, "<%s ", "msGMLOutput");
 
   msOWSPrintEncodeMetadata(stream, &(map->web.metadata), namespaces, "uri", OWS_NOERR, "xmlns=\"%s\"", NULL);

maplayer.c

@@ -279,10 +279,20 @@ int msLayerNextShape(layerObj *layer, shapeObj *shape)
   /* We need to leverage the iteminfo (I think) at this point */
 
   rv = layer->vtable->LayerNextShape(layer, shape);
+  if(rv != MS_SUCCESS)
+    return rv;
 
   /* RFC89 Apply Layer GeomTransform */
   if(layer->_geomtransform.type != MS_GEOMTRANSFORM_NONE && rv == MS_SUCCESS) {
     rv = msGeomTransformShape(layer->map, layer, shape);      
+    if(rv != MS_SUCCESS)
+      return rv;
+  }
+
+  if(layer->encoding) {
+    rv = msLayerEncodeShapeAttributes(layer,shape);
+    if(rv != MS_SUCCESS)
+      return rv;
   }
 
   return rv;
@@ -323,10 +333,20 @@ int msLayerGetShape(layerObj *layer, shapeObj *shape, resultObj *record)
   */
 
   rv = layer->vtable->LayerGetShape(layer, shape, record);
+  if(rv != MS_SUCCESS)
+    return rv;
 
   /* RFC89 Apply Layer GeomTransform */
   if(layer->_geomtransform.type != MS_GEOMTRANSFORM_NONE && rv == MS_SUCCESS) {
     rv = msGeomTransformShape(layer->map, layer, shape); 
+    if(rv != MS_SUCCESS)
+      return rv;
+  }
+
+  if(layer->encoding) {
+    rv = msLayerEncodeShapeAttributes(layer,shape);
+    if(rv != MS_SUCCESS)
+      return rv;
   }
 
   return rv;

mapogcsos.c

@@ -82,7 +82,6 @@ static int msSOSException(mapObj *map, char *locator, char *exceptionCode)
   char *errorString     = NULL;
   char *errorMessage    = NULL;
   char *schemasLocation = NULL;
-  const char *encoding;
 
   xmlDocPtr  psDoc      = NULL;
   xmlNodePtr psRootNode = NULL;
@@ -91,7 +90,6 @@ static int msSOSException(mapObj *map, char *locator, char *exceptionCode)
 
   psNsOws = xmlNewNs(NULL, BAD_CAST "http://www.opengis.net/ows/1.1", BAD_CAST "ows");
 
-  encoding = msOWSLookupMetadata(&(map->web.metadata), "SO", "encoding");
   errorString = msGetErrorString("\n");
   errorMessage = msEncodeHTMLEntities(errorString);
   schemasLocation = msEncodeHTMLEntities(msOWSGetSchemasLocation(map));
@@ -104,13 +102,10 @@ static int msSOSException(mapObj *map, char *locator, char *exceptionCode)
 
   xmlNewNs(psRootNode, BAD_CAST "http://www.opengis.net/ows/1.1", BAD_CAST "ows");
 
-  if (encoding)
-    msIO_setHeader("Content-Type","text/xml; charset=%s", encoding);
-  else
-    msIO_setHeader("Content-Type","text/xml");
+  msIO_setHeader("Content-Type","text/xml; charset=UTF-8");
   msIO_sendHeaders();
 
-  xmlDocDumpFormatMemoryEnc(psDoc, &buffer, &size, (encoding ? encoding : "ISO-8859-1"), 1);
+  xmlDocDumpFormatMemoryEnc(psDoc, &buffer, &size, ("UTF-8"), 1);
 
   msIO_printf("%s", buffer);
 
@@ -1109,7 +1104,6 @@ int msSOSGetCapabilities(mapObj *map, sosParamsObj *sosparams, cgiRequestObj *re
   char *xsi_schemaLocation = NULL;
   char *script_url=NULL;
   const char *updatesequence=NULL;
-  const char *encoding;
 
   int i,j,k;
   layerObj *lp = NULL, *lpTmp = NULL;
@@ -1185,7 +1179,6 @@ int msSOSGetCapabilities(mapObj *map, sosParamsObj *sosparams, cgiRequestObj *re
 
   /* updateSequence */
   updatesequence = msOWSLookupMetadata(&(map->web.metadata), "SO", "updatesequence");
-  encoding = msOWSLookupMetadata(&(map->web.metadata), "SO", "encoding");
 
   if (sosparams->pszUpdateSequence != NULL) {
     i = msOWSNegotiateUpdateSequence(sosparams->pszUpdateSequence, updatesequence);
@@ -1605,10 +1598,7 @@ int msSOSGetCapabilities(mapObj *map, sosParamsObj *sosparams, cgiRequestObj *re
   if ( msIO_needBinaryStdout() == MS_FAILURE )
     return MS_FAILURE;
 
-  if (encoding)
-    msIO_setHeader("Content-Type","text/xml; charset=%s", encoding);
-  else
-    msIO_setHeader("Content-Type","text/xml");
+  msIO_setHeader("Content-Type","text/xml; charset=UTF-8");
   msIO_sendHeaders();
 
   /*TODO* : check the encoding validity. Internally libxml2 uses UTF-8
@@ -1624,7 +1614,7 @@ int msSOSGetCapabilities(mapObj *map, sosParamsObj *sosparams, cgiRequestObj *re
 
   context = msIO_getHandler(stdout);
 
-  xmlDocDumpFormatMemoryEnc(psDoc, &buffer, &size, (encoding ? encoding : "ISO-8859-1"), 1);
+  xmlDocDumpFormatMemoryEnc(psDoc, &buffer, &size, ("UTF-8"), 1);
   msIO_contextWrite(context, buffer, size);
   xmlFree(buffer);
 
@@ -1718,7 +1708,6 @@ int msSOSGetObservation(mapObj *map, sosParamsObj *sosparams, cgiRequestObj *req
   xmlNodePtr psObservationNode = NULL, psResultNode=NULL;
   const char *pszProcedure = NULL;
   const char *pszBlockSep=NULL;
-  const char *encoding;
   char *pszResult=NULL;
   int nDiffrentProc = 0;
   SOSProcedureNode *paDiffrentProc = NULL;
@@ -1731,7 +1720,6 @@ int msSOSGetObservation(mapObj *map, sosParamsObj *sosparams, cgiRequestObj *req
 
   /* establish local namespace */
   pszTmp = msOWSLookupMetadata(&(map->web.metadata), "SFO", "namespace_uri");
-  encoding = msOWSLookupMetadata(&(map->web.metadata), "SO", "encoding");
 
   if(pszTmp) user_namespace_uri = pszTmp;
 
@@ -2458,14 +2446,11 @@ this request. Check sos/ows_enable_request settings.", "msSOSGetObservation()",
   }
 
   /* output results */
-  if (encoding)
-    msIO_setHeader("Content-Type","text/xml; charset=%s", encoding);
-  else
-    msIO_setHeader("Content-Type","text/xml");
+  msIO_setHeader("Content-Type","text/xml; charset=UTF-8");
   msIO_sendHeaders();
 
   context = msIO_getHandler(stdout);
-  xmlDocDumpFormatMemoryEnc(psDoc, &buffer, &size, (encoding ? encoding : "ISO-8859-1"), 1);
+  xmlDocDumpFormatMemoryEnc(psDoc, &buffer, &size, ("UTF-8"), 1);
   msIO_contextWrite(context, buffer, size);
   free(schemalocation);
   free(xsi_schemaLocation);

mapowscommon.c

@@ -615,21 +615,6 @@ int _validateNamespace(xmlNsPtr psNsOws)
     return MS_FAILURE;
 }
 
-xmlNodePtr msOWSCommonxmlNewChildEncoded( xmlNodePtr psParent, xmlNsPtr psNs, const char* name,
-    const char *content, const char *encoding)
-{
-  char *encoded = NULL;
-  xmlNodePtr psNode;
-
-  if (encoding && content) {
-    encoded = msGetEncodedString(content, encoding);
-    psNode =  xmlNewChild(psParent, psNs, BAD_CAST name, BAD_CAST encoded);
-    msFree(encoded);
-    return psNode;
-  } else
-    return xmlNewChild(psParent, psNs, BAD_CAST name, BAD_CAST content);
-}
-
 /*
  * Valid an xml string against an XML schema
  * Inpired from: http://xml.developpez.com/sources/?page=validation#validate_XSD_CppCLI_2

mapowscommon.h

@@ -116,9 +116,6 @@ xmlNodePtr msOWSCommonWGS84BoundingBox(xmlNsPtr psNsOws, int dimensions, double
 
 int _validateNamespace(xmlNsPtr psNsOws);
 
-xmlNodePtr msOWSCommonxmlNewChildEncoded( xmlNodePtr psParent, xmlNsPtr psNs, const char* name,
-    const char *content, const char *encoding);
-
 int msOWSSchemaValidation(const char* xml_schema, const char* xml);
 
 #endif /* defined(USE_LIBXML2) */

mapserver.h

@@ -1658,7 +1658,9 @@ extern "C" {
 
 #ifndef SWIG    
     expressionObj _geomtransform;
-#endif    
+#endif
+
+    char *encoding; /* for iconving shape attributes. ignored if NULL or "utf-8" */
   };
 
 
@@ -2312,6 +2314,7 @@ void msPopulateTextSymbolForLabelAndString(textSymbolObj *ts, labelObj *l, char
   MS_DLL_EXPORT char *msLayerGetProcessingKey( layerObj *layer, const char *);
   MS_DLL_EXPORT int msLayerClearProcessing( layerObj *layer );
   MS_DLL_EXPORT char* msLayerGetFilterString( layerObj *layer );
+  MS_DLL_EXPORT int msLayerEncodeShapeAttributes( layerObj *layer, shapeObj *shape);
 
   MS_DLL_EXPORT int msLayerSupportsCommonFilters(layerObj *layer);
   MS_DLL_EXPORT int msTokenizeExpression(expressionObj *expression, char **list, int *listsize);

mapstring.c

@@ -2074,3 +2074,59 @@ int msStringInArray( const char * pszString, char **array, int numelements)
   }
   return MS_FALSE;
 }
+
+int msLayerEncodeShapeAttributes( layerObj *layer, shapeObj *shape) {
+
+#ifdef USE_ICONV
+  iconv_t cd = NULL;
+  const char *inp;
+  char *outp, *out = NULL;
+  size_t len, bufsize, bufleft, iconv_status;
+  int i;
+#endif
+
+  if( !layer->encoding || !*layer->encoding || !strcasecmp(layer->encoding, "UTF-8"))
+    return MS_SUCCESS;
+
+  cd = iconv_open("UTF-8", layer->encoding);
+  if(cd == (iconv_t)-1) {
+    msSetError(MS_IDENTERR, "Encoding not supported by libiconv (%s).",
+               "msGetEncodedString()", layer->encoding);
+    return MS_FAILURE;
+  }
+
+#ifdef USE_ICONV
+  for(i=0;i <shape->numvalues; i++) {
+    if(!shape->values[i] || (len = strlen(shape->values[i]))==0) {
+      continue;    /* Nothing to do */
+    }
+
+    bufsize = len * 6 + 1; /* Each UTF-8 char can be up to 6 bytes */
+    inp = shape->values[i];
+    out = (char*) msSmallMalloc(bufsize);
+
+    strlcpy(out, shape->values[i], bufsize);
+    outp = out;
+
+    bufleft = bufsize;
+    iconv_status = -1;
+
+    while (len > 0) {
+      iconv_status = iconv(cd, (char**)&inp, &len, &outp, &bufleft);
+      if(iconv_status == -1) {
+        msFree(out);
+        continue; /* silently ignore failed conversions */
+      }
+    }
+    out[bufsize - bufleft] = '\0';
+    msFree(shape->values[i]);
+    shape->values[i] = out;
+  }
+  iconv_close(cd);
+
+  return MS_SUCCESS;
+#else
+  msSetError(MS_MISCERR, "Not implemented since Iconv is not enabled.", "msGetEncodedString()");
+  return MS_FAILURE;
+#endif
+}