Added exception for invalid XML

.gitignore

@@ -16,4 +16,6 @@ dist
 build
 setup_script-001.iss
 setup_script-002.iss
-Output
+Output
+invalid_eads
+harvard-schematron.xml

as_xtf_GUI.py

@@ -443,9 +443,9 @@ def get_aspace_log(defaults, xtf_checkbox=True):
     correct_creds = False
     close_program = False
     while correct_creds is False:
-        asplog_col1 = [[sg.Text("Enter your ArchivesSpace username:", font=("Roboto", 12))],
-                       [sg.Text("Enter your ArchivesSpace password:", font=("Roboto", 12))],
-                       [sg.Text("Enter your ArchivesSpace API URL:", font=("Roboto", 12))]]
+        asplog_col1 = [[sg.Text("Enter your ArchivesSpace username:", font=("Roboto", 11))],
+                       [sg.Text("Enter your ArchivesSpace password:", font=("Roboto", 11))],
+                       [sg.Text("Enter your ArchivesSpace API URL:", font=("Roboto", 11))]]
         asplog_col2 = [[sg.InputText(focus=True, key="_ASPACE_UNAME_")],
                        [sg.InputText(password_char='*', key="_ASPACE_PWORD_")],
                        [sg.InputText(defaults["as_api"], key="_ASPACE_API_")]]
@@ -519,11 +519,11 @@ def get_xtf_log(defaults):
     correct_creds = False
     close_program = False
     while correct_creds is False:
-        xtflog_col1 = [[sg.Text("Enter your XTF username:", font=("Roboto", 12))],
-                       [sg.Text("Enter your XTF password:", font=("Roboto", 12))],
-                       [sg.Text("Enter XTF Hostname:", font=("Roboto", 12))],
-                       [sg.Text("Enter XTF Remote Path:", font=("Roboto", 12))],
-                       [sg.Text("Enter XTF Indexer Path:", font=("Roboto", 12))]]
+        xtflog_col1 = [[sg.Text("Enter your XTF username:", font=("Roboto", 11))],
+                       [sg.Text("Enter your XTF password:", font=("Roboto", 11))],
+                       [sg.Text("Enter XTF Hostname:", font=("Roboto", 11))],
+                       [sg.Text("Enter XTF Remote Path:", font=("Roboto", 11))],
+                       [sg.Text("Enter XTF Indexer Path:", font=("Roboto", 11))]]
         xtflog_col2 = [[sg.InputText(focus=True, key="_XTF_UNAME_")],
                        [sg.InputText(password_char='*', key="_XTF_PWORD_")],
                        [sg.InputText(defaults["xtf_default"]["xtf_host"], key="_XTF_HOSTNAME_")],
@@ -614,16 +614,23 @@ def get_eads(input_ids, defaults, cleanup_options, repositories, client, values_
                 if defaults["ead_export_default"]["_CLEAN_EADS_"] is True:
                     if defaults["ead_export_default"]["_KEEP_RAW_"] is True:
                         print("Cleaning up EAD record...", end='', flush=True)
-                        results = clean.cleanup_eads(resource_export.filepath, cleanup_options,
-                                                     defaults["ead_export_default"]["_OUTPUT_DIR_"],
-                                                     keep_raw_exports=True)
-                        print("Done")
+                        valid, results = clean.cleanup_eads(resource_export.filepath, cleanup_options,
+                                                            defaults["ead_export_default"]["_OUTPUT_DIR_"],
+                                                            keep_raw_exports=True)
+                        if valid:
+                            print("Done")
+                            print(results)
+                        else:
+                            print("XML validation error\n" + results)
                     else:
                         print("Cleaning up EAD record...", end='', flush=True)
-                        results = clean.cleanup_eads(resource_export.filepath, cleanup_options,
-                                                     defaults["ead_export_default"]["_OUTPUT_DIR_"])
-                        print("Done")
-                    print(results)
+                        valid, results = clean.cleanup_eads(resource_export.filepath, cleanup_options,
+                                                            defaults["ead_export_default"]["_OUTPUT_DIR_"])
+                        if valid:
+                            print("Done")
+                            print(results)
+                        else:
+                            print("XML validation error\n" + results)
             else:
                 print(resource_export.error + "\n")
         else:

cleanup.py

@@ -358,16 +358,22 @@ def cleanup_eads(filepath, custom_clean, output_dir="clean_eads", keep_raw_expor
         custom_clean (list): strings as passed from as_xtf_GUI.py that determines what methods will be run against the
         lxml element when running the clean_suite() method. The user can specify what they want cleaned in
         as_xtf_GUI.py, so this is how those specifications are passed.
-        output_dir (str, optional): filepath of where the EAD record should be sent after cleaning, as specified by the user ("clean_eads" is default)
-        keep_raw_exports (bool, optional): if a user in as_xtf_GUI.py specifies to keep the exports that come from
+        output_dir (str): filepath of where the EAD record should be sent after cleaning, as specified by the user ("clean_eads" is default)
+        keep_raw_exports (bool): if a user in as_xtf_GUI.py specifies to keep the exports that come from
         as_export.py, this parameter will prevent the function from deleting those files in source_eads.
 
     Returns:
         results (str): filled with result information when methods are performed
     """
     filename = Path(filepath).name  # get file name + extension
+    valid_err = ""
     parser = etree.XMLParser(remove_blank_text=True, ns_clean=True)  # clean up redundant namespace declarations
-    tree = etree.parse(filepath, parser=parser)
+    try:
+        tree = etree.parse(filepath, parser=parser)
+    except:
+        valid_err += "There was an error with the xml data. The file is saved in {}\n\n".format(Path(filepath).parent)
+        valid_err += "-" * 135
+        return False, valid_err
     ead_root = tree.getroot()
     ead = EADRecord(ead_root)
     clean_ead, results = ead.clean_suite(ead, custom_clean)
@@ -383,7 +389,7 @@ def cleanup_eads(filepath, custom_clean, output_dir="clean_eads", keep_raw_expor
         os.remove(filepath)
     if keep_raw_exports is False:  # prevents program from rerunning cleanup on cleaned files
         os.remove(filepath)
-        return results
+        return True, results
     else:
         results += "\nKeeping raw ASpace exports in {}\n".format(output_dir)
-        return results
+        return True, results

requirements.txt

@@ -1,5 +1,5 @@
 loguru==0.5.0
 lxml==4.5.1
 paramiko==2.7.1
-PySimpleGUI==4.19.0.2
+PySimpleGUI==4.20.0
 scp==0.13.2