FGDC1 now writes unicode text to xml, related to issue-80

Author: ethoms-usgs

Scripts/GeMS_FGDC1_Arc10.py

@@ -3,10 +3,11 @@
 import sys, os, os.path, arcpy
 from GeMS_utilityFunctions import *
 from xml.dom.minidom import *
+import codecs
 
 debug = False
 
-versionString = 'GeMS_FGDC1_Arc10.py, version of 20 July 2020'
+versionString = 'GeMS_FGDC1_Arc10.py, version of 5 October 2021'
 rawurl = 'https://raw.githubusercontent.com/usgs/gems-tools-arcmap/master/Scripts/GeMS_FGDC1_Arc10.py'
 checkVersion(versionString, rawurl, 'gems-tools-arcmap')
 
@@ -85,9 +86,10 @@ def writeDomToFile(workDir,dom,fileName):
     if debug:
         addMsgAndPrint(arcpy.env.workspace)
         addMsgAndPrint('fileName='+fileName)
-    outf = open(os.path.join(workDir,fileName),'w')
-    dom.writexml(outf)
-    outf.close()
+    outf = os.path.join(workDir,fileName)
+    
+    with codecs.open(outf, "w", encoding="utf-8", errors="xmlcharrefreplace") as out:
+        dom.writexml(out, addindent="")
 
 ###########################################
 inGdb = sys.argv[1]
@@ -113,7 +115,9 @@ def writeDomToFile(workDir,dom,fileName):
 
 # parse mrXML to DOM
 try:
-    domMR = xml.dom.minidom.parse(mrXML)
+    with open(mrXML) as xml:
+        domMR = parse(xml)
+    #domMR = xml.dom.minidom.parse(mrXML)
     addMsgAndPrint('  Master record parsed successfully')
 except:
     addMsgAndPrint(arcpy.GetMessages())

Scripts/GeMS_FGDC2_Arc10.py

@@ -427,6 +427,8 @@ def fixObjXML(objName,objType,objLoc,domMR, fdDataSourceValues=[]):
     arcXMLfile = wksp+'/'+objName+'.xml'
     testAndDelete(arcXMLfile)
     arcpy.ExportMetadata_conversion(objLoc,translator,arcXMLfile)
+    #with open(xml_file) as xml:
+    #    arcXML = parse(xml)
     arcXML = xml.dom.minidom.parse(arcXMLfile)
     dom = copy.deepcopy(domMR)
     # updateTableDom updates entity-attribute info, also returns dataSourceValues
@@ -502,29 +504,29 @@ def writeDomToFile(workDir,dom,fileName):
     objLoc = inGdb+'/'+aTable
     fixObjXML(objName,objType,objLoc,domMR)
 
-fcs = arcpy.ListFeatureClasses()
-for fc in fcs:
-    objName = fc
-    objType = 'Feature class'
-    objLoc = inGdb+'/'+fc
-    fixObjXML(objName,objType,objLoc,domMR)
-
-fds = arcpy.ListDatasets('','Feature')
-for fd in fds:
-    arcpy.env.workspace = inGdb+'/'+fd
-    fcs = arcpy.ListFeatureClasses()
-    arcpy.env.workspace = inGdb
-    fdDS = []  # inventory of all DataSource_IDs used in feature dataset
-    for fc in fcs:
-        objName = fc
-        objType = 'Feature class'
-        objLoc = inGdb+'/'+fd+'/'+fc
-        localDS = fixObjXML(objName,objType,objLoc,domMR)
-        for ds in localDS:
-            fdDS.append(ds)
-    objName = fd
-    objType = 'Feature dataset'
-    objLoc = inGdb+'/'+fd
-    fixObjXML(objName,objType,objLoc,domMR,set(fdDS))
+# fcs = arcpy.ListFeatureClasses()
+# for fc in fcs:
+    # objName = fc
+    # objType = 'Feature class'
+    # objLoc = inGdb+'/'+fc
+    # fixObjXML(objName,objType,objLoc,domMR)
+
+# fds = arcpy.ListDatasets('','Feature')
+# for fd in fds:
+    # arcpy.env.workspace = inGdb+'/'+fd
+    # fcs = arcpy.ListFeatureClasses()
+    # arcpy.env.workspace = inGdb
+    # fdDS = []  # inventory of all DataSource_IDs used in feature dataset
+    # for fc in fcs:
+        # objName = fc
+        # objType = 'Feature class'
+        # objLoc = inGdb+'/'+fd+'/'+fc
+        # localDS = fixObjXML(objName,objType,objLoc,domMR)
+        # for ds in localDS:
+            # fdDS.append(ds)
+    # objName = fd
+    # objType = 'Feature dataset'
+    # objLoc = inGdb+'/'+fd
+    # fixObjXML(objName,objType,objLoc,domMR,set(fdDS))
 
 logFile.close()