opex-manifest-generator 1.1.11__py3-none-any.whl → 1.1.13__py3-none-any.whl

opex_manifest_generator/__init__.py

@@ -1,16 +1,16 @@
-"""
-opex_manifest_generator package definitions
-
-Author: Christopher Prince
-license: Apache License 2.0"
-"""
-
-from .opex_manifest import OpexManifestGenerator,OpexDir,OpexFile
-from .hash import HashGenerator
-from .common import *
-from .cli import parse_args,run_cli
-import importlib.metadata
-
-__author__ = "Christopher Prince (c.pj.prince@gmail.com)"
-__license__ = "Apache License Version 2.0"
+"""
+opex_manifest_generator package definitions
+
+Author: Christopher Prince
+license: Apache License 2.0"
+"""
+
+from .opex_manifest import OpexManifestGenerator,OpexDir,OpexFile
+from .hash import HashGenerator
+from .common import *
+from .cli import parse_args,run_cli
+import importlib.metadata
+
+__author__ = "Christopher Prince (c.pj.prince@gmail.com)"
+__license__ = "Apache License Version 2.0"
 __version__ = importlib.metadata.version("opex_manifest_generator")

opex_manifest_generator/cli.py

@@ -1,160 +1,161 @@
-"""
-Cli interaction.
-
-author: Christopher Prince
-license: Apache License 2.0"
-"""
-
-import argparse
-import os
-import time
-from opex_manifest_generator.opex_manifest import OpexManifestGenerator
-import importlib.metadata
-
-def parse_args():
-    parser = argparse.ArgumentParser(description = "OPEX Manifest Generator for Preservica Uploads")
-    parser.add_argument('root', default = os.getcwd(), help = "The root path to generate Opexes for")
-    parser.add_argument("-c", "--autoclass", required = False,
-                        choices = ['catalog', 'c', 'accession', 'a', 'both', 'b', 'generic', 'g', 'catalog-generic', 'cg', "accession-generic", "ag", "both-generic", "bg"],
-                        type = str.lower,
-                        help="""Toggles whether to utilise the auto_classification_generator 
-                        to generate an on the fly Reference listing.
-                        
-                        There are several options, {catalog} will generate
-                        a Archival Reference following an ISAD(G) sturcutre.
-                        {accession} will create a running number of files.
-                        {both} will do both at the same time!
-                        {generic} will populate the title and description fields with the folder/file's name,
-                        if used in conjunction with one of the above options:
-                        {generic-catalog,generic-accession, generic-both} it will do both simultaneously.
-                        """)
-    parser.add_argument("-p", "--prefix", required = False, nargs = '+',
-                        help= """Assign a prefix when utilising the --autoclass option. Prefix will append any text before all generated text.
-                        When utilising the {both} option fill in like: [catalog-prefix, accession-prefix] without square brackets.                        
-                        """)
-    parser.add_argument("-fx", "--fixity", required = False, const = "SHA-1", default = None,
-                        nargs = '?', choices = ['NONE', 'SHA-1', 'MD5', 'SHA-256', 'SHA-512'], type = str.upper,
-                        help="Generates a hash for each file and adds it to the opex, can select the algorithm to utilise.")
-    parser.add_argument("-rme", "--remove-empty", required = False, action = 'store_true', default = False,
-                        help = "Remove and log empty directories from root. Log will be exported to 'meta' / output folder.")
-    parser.add_argument("-o", "--output", required = False, nargs = 1,
-                        help = "Sets the output to send any generated files to. Will not affect creation of a meta dir.")
-    parser.add_argument("-dmd", "--disable-meta-dir", required = False, action = 'store_false',
-                        help = """Set whether to disable the creation of a 'meta' directory for generated files,
-                        default behaviour is to always generate this directory""")
-    parser.add_argument("-clr", "--clear-opex", required = False, action = 'store_true', default = False,
-                        help = """Clears existing opex files from a directory. If set with no further options will only clear opexes; 
-                        if multiple options are set will clear opexes and then run the program""")
-    parser.add_argument("-opt","--options-file", required = False, default=os.path.join(os.path.dirname(__file__),'options.properties'),
-                        help="Specify a custom Options file, changing the set presets for column headers (Title,Description,etc)")
-    parser.add_argument("-s", "--start-ref", required = False, nargs = '?', default = 1, 
-                        help="Set a custom Starting reference for the Auto Classification generator. The generated reference will")
-    parser.add_argument("-mdir","--metadata-dir", required=False, nargs= '?',
-                        default = os.path.join(os.path.dirname(os.path.realpath(__file__)), "metadata"),
-                        help="Specify the metadata directory to pull XML files from")
-    parser.add_argument("-m", "--metadata", required = False, const = 'e', default = 'none',
-                        nargs = '?', choices = ['none', 'n', 'exact', 'e', 'flat', 'f'], type = str.lower,
-                        help="Set whether to include xml metadata fields in the generation of the Opex")
-    parser.add_argument("-ex", "--export", required = False, action = 'store_true', default = False,
-                        help="Set whether to export the generated auto classification references to an AutoClass spreadsheet")
-    parser.add_argument("-i", "--input", required = False, nargs='?', 
-                        help="Set to utilise a CSV / XLSX spreadsheet to import data from")
-    parser.add_argument("-rm", "--remove", required = False, action = "store_true", default = False,
-                        help="Set whether to enable removals of files and folders from a directory. ***Currently in testing")
-    parser.add_argument("-z", "--zip", required = False, action = 'store_true',
-                        help="Set to zip files")
-    parser.add_argument("-fmt", "--output-format", required = False, default = "xlsx", choices = ['xlsx', 'csv'],
-                        help="Set whether to output to an xlsx or csv format")
-    parser.add_argument("-v", "--version", action = 'version', version = '%(prog)s {version}'.format(version = importlib.metadata.version("opex_manifest_generator")))
-    parser.add_argument("--accession-mode", required=False, choices=["file",'directory','both'],
-                        help="""Set the mode when utilising the Accession option in autoclass.
-                        file - only adds on files, folder - only adds on folders, both - adds on files and folders""")
-    parser.add_argument("--hidden", required = False, action = 'store_true', default = False,
-                        help="Set whether to include hidden files and folders")
-    parser.add_argument("--print-xmls", required = False, action = "store_true", default = False,
-                        help="Prints the elements from your xmls to the consoles")
-    args = parser.parse_args()
-    return args
-
-def run_cli():
-    args = parse_args()    
-    print(f"Running Opex Generation on: {args.root}")
-    if not args.output:
-        args.output = os.path.abspath(args.root)
-        print(f'Output path defaulting to root directory: {args.output}')
-    else:
-        args.output = os.path.abspath(args.output[0])
-        print(f'Output path set to: {args.output}')    
-    if args.input and args.autoclass:
-        print(f'Both Input and Auto-Class options have been selected, please use only one...')
-        time.sleep(5); raise SystemExit()
-    if not args.metadata in {'none', 'n'} and not args.input:
-        print(f'Warning: Metadata Flag has been given without Input. Metadata won\'t be generated.')
-        time.sleep(5)
-    if args.print_xmls:
-        OpexManifestGenerator.print_descriptive_xmls()
-    acc_prefix = None
-    if args.prefix:
-        if args.autoclass in {"both", "b", "both-generic", "bg"}:
-            if len(args.prefix) < 2 or len(args.prefix) > 2:
-                print('"Both" option is selected, please pass only two prefixes: [-p CATALOG_PREFIX ACCESSION_PREFIX]');
-                time.sleep(3); raise SystemExit
-            for n, a in enumerate(args.prefix):
-                if n == 0:
-                    args.prefix = str(a)
-                elif n == 1:
-                    acc_prefix = str(a)
-            print(f"Prefixes are set as: \t Catalog: {args.prefix} \t Acc: {acc_prefix}")
-        elif args.autoclass in {"accession", "a", "accession-generic", "ag"}:
-            for a in args.prefix:
-                acc_prefix = str(a)
-            print('Prefix is set as: ' + acc_prefix)                        
-        elif args.autoclass in {"catalog", "c", "catalog-generic", "cg"}:
-            acc_prefix = None
-            for a in args.prefix: 
-                args.prefix = str(a)
-            print('Prefix is set as: ' + args.prefix)
-        elif args.autoclass in {"generic", "g"}:
-            pass
-        else:
-            print('''An invalid option has been selected, please select a valid option: 
-                  {c, catalog
-                   a, accession
-                   b, both
-                   g, generic
-                   cg, catalog-generic
-                   ag, accesion-generic
-                   bg, both-generic}''')    
-            time.sleep(3)
-            raise SystemExit               
-    if args.fixity:
-        print(f'Fixity is activated, using {args.fixity} algorithm')
-    if args.remove:
-        i = "y"
-        # i = input(inspect.cleandoc("""You have enabled the remove functionality.
-        #                         This action will remove files listed for removal, it is irreversible.
-        #                         Please type Y to confirm, otherwise program will close: """))
-        if not i.lower() == "y":
-            print("Closing program..."); time.sleep(3); raise SystemExit()
-    time.sleep(3)
-    OpexManifestGenerator(root = args.root, 
-                          output_path = args.output, 
-                          autoclass_flag = args.autoclass, 
-                          prefix = args.prefix, 
-                          acc_prefix = acc_prefix, 
-                          empty_flag = args.remove_empty, 
-                          remove_flag = args.remove, 
-                          clear_opex_flag = args.clear_opex, 
-                          algorithm = args.fixity, 
-                          startref = args.start_ref, 
-                          export_flag = args.export, 
-                          meta_dir_flag = args.disable_meta_dir, 
-                          metadata_flag = args.metadata,
-                          metadata_dir = args.metadata_dir,
-                          hidden_flag= args.hidden,
-                          zip_flag = args.zip, 
-                          input = args.input, 
-                          output_format = args.output_format).main()
-    
-if __name__ == "__main__":
+"""
+Cli interaction.
+
+author: Christopher Prince
+license: Apache License 2.0"
+"""
+
+import argparse
+import os
+import time
+from opex_manifest_generator.opex_manifest import OpexManifestGenerator
+import importlib.metadata
+
+def parse_args():
+    parser = argparse.ArgumentParser(description = "OPEX Manifest Generator for Preservica Uploads")
+    parser.add_argument('root', default = os.getcwd(), help = "The root path to generate Opexes for")
+    parser.add_argument("-c", "--autoclass", required = False,
+                        choices = ['catalog', 'c', 'accession', 'a', 'both', 'b', 'generic', 'g', 'catalog-generic', 'cg', "accession-generic", "ag", "both-generic", "bg"],
+                        type = str.lower,
+                        help="""Toggles whether to utilise the auto_classification_generator 
+                        to generate an on the fly Reference listing.
+                        
+                        There are several options, {catalog} will generate
+                        a Archival Reference following an ISAD(G) sturcutre.
+                        {accession} will create a running number of files.
+                        {both} will do both at the same time!
+                        {generic} will populate the title and description fields with the folder/file's name,
+                        if used in conjunction with one of the above options:
+                        {generic-catalog,generic-accession, generic-both} it will do both simultaneously.
+                        """)
+    parser.add_argument("-p", "--prefix", required = False, nargs = '+',
+                        help= """Assign a prefix when utilising the --autoclass option. Prefix will append any text before all generated text.
+                        When utilising the {both} option fill in like: [catalog-prefix, accession-prefix] without square brackets.                        
+                        """)
+    parser.add_argument("-fx", "--fixity", required = False, const = "SHA-1", default = None,
+                        nargs = '?', choices = ['NONE', 'SHA-1', 'MD5', 'SHA-256', 'SHA-512'], type = str.upper,
+                        help="Generates a hash for each file and adds it to the opex, can select the algorithm to utilise.")
+    parser.add_argument("-rme", "--remove-empty", required = False, action = 'store_true', default = False,
+                        help = "Remove and log empty directories from root. Log will be exported to 'meta' / output folder.")
+    parser.add_argument("-o", "--output", required = False, nargs = 1,
+                        help = "Sets the output to send any generated files to. Will not affect creation of a meta dir.")
+    parser.add_argument("-dmd", "--disable-meta-dir", required = False, action = 'store_false',
+                        help = """Set whether to disable the creation of a 'meta' directory for generated files,
+                        default behaviour is to always generate this directory""")
+    parser.add_argument("-clr", "--clear-opex", required = False, action = 'store_true', default = False,
+                        help = """Clears existing opex files from a directory. If set with no further options will only clear opexes; 
+                        if multiple options are set will clear opexes and then run the program""")
+    parser.add_argument("-opt","--options-file", required = False, default=os.path.join(os.path.dirname(__file__),'options.properties'),
+                        help="Specify a custom Options file, changing the set presets for column headers (Title,Description,etc)")
+    parser.add_argument("-s", "--start-ref", required = False, nargs = '?', default = 1, 
+                        help="Set a custom Starting reference for the Auto Classification generator. The generated reference will")
+    parser.add_argument("-mdir","--metadata-dir", required=False, nargs= '?',
+                        default = os.path.join(os.path.dirname(os.path.realpath(__file__)), "metadata"),
+                        help="Specify the metadata directory to pull XML files from")
+    parser.add_argument("-m", "--metadata", required = False, const = 'e', default = 'none',
+                        nargs = '?', choices = ['none', 'n', 'exact', 'e', 'flat', 'f'], type = str.lower,
+                        help="Set whether to include xml metadata fields in the generation of the Opex")
+    parser.add_argument("-ex", "--export", required = False, action = 'store_true', default = False,
+                        help="Set whether to export the generated auto classification references to an AutoClass spreadsheet")
+    parser.add_argument("-i", "--input", required = False, nargs='?', 
+                        help="Set to utilise a CSV / XLSX spreadsheet to import data from")
+    parser.add_argument("-rm", "--remove", required = False, action = "store_true", default = False,
+                        help="Set whether to enable removals of files and folders from a directory. ***Currently in testing")
+    parser.add_argument("-z", "--zip", required = False, action = 'store_true',
+                        help="Set to zip files")
+    parser.add_argument("-fmt", "--output-format", required = False, default = "xlsx", choices = ['xlsx', 'csv'],
+                        help="Set whether to output to an xlsx or csv format")
+    parser.add_argument("-v", "--version", action = 'version', version = '%(prog)s {version}'.format(version = importlib.metadata.version("opex_manifest_generator")))
+    parser.add_argument("--accession-mode", required=False, choices=["file",'directory','both'],
+                        help="""Set the mode when utilising the Accession option in autoclass.
+                        file - only adds on files, folder - only adds on folders, both - adds on files and folders""")
+    parser.add_argument("--hidden", required = False, action = 'store_true', default = False,
+                        help="Set whether to include hidden files and folders")
+    parser.add_argument("--print-xmls", required = False, action = "store_true", default = False,
+                        help="Prints the elements from your xmls to the consoles")
+    args = parser.parse_args()
+    return args
+
+def run_cli():
+    args = parse_args()    
+    print(f"Running Opex Generation on: {args.root}")
+    if not args.output:
+        args.output = os.path.abspath(args.root)
+        print(f'Output path defaulting to root directory: {args.output}')
+    else:
+        args.output = os.path.abspath(args.output[0])
+        print(f'Output path set to: {args.output}')    
+    if args.input and args.autoclass:
+        print(f'Both Input and Auto-Class options have been selected, please use only one...')
+        time.sleep(5); raise SystemExit()
+    if not args.metadata in {'none', 'n'} and not args.input:
+        print(f'Warning: Metadata Flag has been given without Input. Metadata won\'t be generated.')
+        time.sleep(5)
+    if args.print_xmls:
+        OpexManifestGenerator.print_descriptive_xmls()
+    acc_prefix = None
+    if args.prefix:
+        if args.autoclass in {"both", "b", "both-generic", "bg"}:
+            if len(args.prefix) < 2 or len(args.prefix) > 2:
+                print('"Both" option is selected, please pass only two prefixes: [-p CATALOG_PREFIX ACCESSION_PREFIX]');
+                time.sleep(3); raise SystemExit
+            for n, a in enumerate(args.prefix):
+                if n == 0:
+                    args.prefix = str(a)
+                elif n == 1:
+                    acc_prefix = str(a)
+            print(f"Prefixes are set as: \t Catalog: {args.prefix} \t Acc: {acc_prefix}")
+        elif args.autoclass in {"accession", "a", "accession-generic", "ag"}:
+            for a in args.prefix:
+                acc_prefix = str(a)
+            print('Prefix is set as: ' + acc_prefix)                        
+        elif args.autoclass in {"catalog", "c", "catalog-generic", "cg"}:
+            acc_prefix = None
+            for a in args.prefix: 
+                args.prefix = str(a)
+            print('Prefix is set as: ' + args.prefix)
+        elif args.autoclass in {"generic", "g"}:
+            pass
+        else:
+            print('''An invalid option has been selected, please select a valid option: 
+                  {c, catalog
+                   a, accession
+                   b, both
+                   g, generic
+                   cg, catalog-generic
+                   ag, accesion-generic
+                   bg, both-generic}''')    
+            time.sleep(3)
+            raise SystemExit               
+    if args.fixity:
+        print(f'Fixity is activated, using {args.fixity} algorithm')
+    if args.remove:
+        i = "y"
+        # i = input(inspect.cleandoc("""You have enabled the remove functionality.
+        #                         This action will remove files listed for removal, it is irreversible.
+        #                         Please type Y to confirm, otherwise program will close: """))
+        if not i.lower() == "y":
+            print("Closing program..."); time.sleep(3); raise SystemExit()
+    time.sleep(3)
+    OpexManifestGenerator(root = args.root, 
+                          output_path = args.output, 
+                          autoclass_flag = args.autoclass, 
+                          prefix = args.prefix, 
+                          acc_prefix = acc_prefix, 
+                          empty_flag = args.remove_empty, 
+                          remove_flag = args.remove, 
+                          clear_opex_flag = args.clear_opex, 
+                          algorithm = args.fixity, 
+                          startref = args.start_ref, 
+                          export_flag = args.export, 
+                          meta_dir_flag = args.disable_meta_dir, 
+                          metadata_flag = args.metadata,
+                          metadata_dir = args.metadata_dir,
+                          hidden_flag= args.hidden,
+                          zip_flag = args.zip, 
+                          input = args.input, 
+                          output_format = args.output_format,
+                          options_file=args.options_file).main()
+    
+if __name__ == "__main__":
     run_cli()

opex_manifest_generator/common.py

@@ -1,58 +1,58 @@
-"""
-Common tools for use throghout program.
-
-author: Christopher Prince
-license: Apache License 2.0"
-"""
-
-import zipfile, os, sys, time, stat
-import datetime
-from lxml import etree
-
-def zip_opex(file_path,opex_path):
-    zip_file = f"{file_path}.zip"
-    if not os.path.exists(zip_file):
-        with zipfile.ZipFile(zip_file,'w') as z:
-            z.write(file_path,os.path.basename(file_path))
-            z.write(opex_path,os.path.basename(opex_path))
-    else: print(f'A zip file already exists for: {zip_file}')
-
-def win_256_check(path: str):
-    if len(path) > 255 and sys.platform == "win32":
-        if path.startswith(u'\\\\?\\'): path = path
-        else: path = u"\\\\?\\" + path
-    return path
-
-def filter_win_hidden(path: str):
-    if sys.platform =="win32":
-        if bool(os.stat(path).st_file_attribute & stat.FILE_ATTRIBUTE_HIDDEN) is True:
-            return True
-        else:
-            return False
-    else:
-        return False
-
-def check_nan(value):
-    if str(value).lower() in {"nan","nat"}:
-        value = None
-    return value
-
-def check_opex(opex_path:str):
-    opex_path = opex_path + ".opex" 
-    if os.path.exists(win_256_check(opex_path)):
-        return False
-    else: 
-        return True
-
-def write_opex(path: str, opexxml: etree.Element):
-    opex_path = win_256_check(str(path) + ".opex")
-    opex = etree.indent(opexxml, "  ")
-    opex = etree.tostring(opexxml, pretty_print=True, xml_declaration=True, encoding="UTF-8", standalone=True)
-    with open(f'{opex_path}', 'w', encoding="UTF-8") as writer:
-        writer.write(opex.decode('UTF-8'))
-        print('Saved Opex File to: ' + opex_path)
-    return opex_path
-
-def print_running_time(start_time):
-    print(f'Running time: {datetime.datetime.now() - start_time}')
-    time.sleep(5)
+"""
+Common tools for use throghout program.
+
+author: Christopher Prince
+license: Apache License 2.0"
+"""
+
+import zipfile, os, sys, time, stat
+import datetime
+from lxml import etree
+
+def zip_opex(file_path,opex_path):
+    zip_file = f"{file_path}.zip"
+    if not os.path.exists(zip_file):
+        with zipfile.ZipFile(zip_file,'w') as z:
+            z.write(file_path,os.path.basename(file_path))
+            z.write(opex_path,os.path.basename(opex_path))
+    else: print(f'A zip file already exists for: {zip_file}')
+
+def win_256_check(path: str):
+    if len(path) > 255 and sys.platform == "win32":
+        if path.startswith(u'\\\\?\\'): path = path
+        else: path = u"\\\\?\\" + path
+    return path
+
+def filter_win_hidden(path: str):
+    if sys.platform =="win32":
+        if bool(os.stat(path).st_file_attributes & stat.FILE_ATTRIBUTE_HIDDEN) is True:
+            return True
+        else:
+            return False
+    else:
+        return False
+
+def check_nan(value):
+    if str(value).lower() in {"nan","nat"}:
+        value = None
+    return value
+
+def check_opex(opex_path:str):
+    opex_path = opex_path + ".opex" 
+    if os.path.exists(win_256_check(opex_path)):
+        return False
+    else: 
+        return True
+
+def write_opex(path: str, opexxml: etree.Element):
+    opex_path = win_256_check(str(path) + ".opex")
+    opex = etree.indent(opexxml, "  ")
+    opex = etree.tostring(opexxml, pretty_print=True, xml_declaration=True, encoding="UTF-8", standalone=True)
+    with open(f'{opex_path}', 'w', encoding="UTF-8") as writer:
+        writer.write(opex.decode('UTF-8'))
+        print('Saved Opex File to: ' + opex_path)
+    return opex_path
+
+def print_running_time(start_time):
+    print(f'Running time: {datetime.datetime.now() - start_time}')
+    time.sleep(5)

opex_manifest_generator/hash.py

@@ -1,34 +1,34 @@
-"""
-Hash Generator class for generating Fixities for files.
-
-author: Christopher Prince
-license: Apache License 2.0"
-"""
-
-import hashlib
-
-class HashGenerator():
-    def __init__(self,algorithm="SHA-1"):
-        self.algorithm = algorithm
-        self.buffer = 4096
-
-    def hash_generator(self,file_path: str):
-        if self.algorithm == "SHA-1":
-            hash = hashlib.sha1()
-        elif self.algorithm == "MD5":
-            hash = hashlib.md5()
-        elif self.algorithm == "SHA-256":
-            hash = hashlib.sha256()
-        elif self.algorithm == "SHA-512":
-            hash = hashlib.sha512()
-        else:
-            hash = hashlib.sha1()
-        print(f'Generating Fixity using {self.algorithm} for: {file_path}')
-        with open(file_path, "rb") as f:
-            while True:
-                buff = f.read(self.buffer)
-                if not buff:
-                    break
-                hash.update(buff)
-            f.close()
+"""
+Hash Generator class for generating Fixities for files.
+
+author: Christopher Prince
+license: Apache License 2.0"
+"""
+
+import hashlib
+
+class HashGenerator():
+    def __init__(self,algorithm="SHA-1"):
+        self.algorithm = algorithm
+        self.buffer = 4096
+
+    def hash_generator(self,file_path: str):
+        if self.algorithm == "SHA-1":
+            hash = hashlib.sha1()
+        elif self.algorithm == "MD5":
+            hash = hashlib.md5()
+        elif self.algorithm == "SHA-256":
+            hash = hashlib.sha256()
+        elif self.algorithm == "SHA-512":
+            hash = hashlib.sha512()
+        else:
+            hash = hashlib.sha1()
+        print(f'Generating Fixity using {self.algorithm} for: {file_path}')
+        with open(file_path, "rb") as f:
+            while True:
+                buff = f.read(self.buffer)
+                if not buff:
+                    break
+                hash.update(buff)
+            f.close()
         return hash.hexdigest().upper()