Python script to dump all fields in KAPE targets and modules, including documentation

kape_info.py

import csv
import yaml
import argparse
import os
import sys

filenames = []


def main():

    if args.target:
        fieldnames = ['Disabled', 'Local', 'Target', 'Description', 'Author', 'Version', 'Id',
                      'RecreateDirectories', 'Name', 'Category', 'Path',
                      'Recursive', 'FileMask', 'SaveAsFileName',
                      'AlwaysAddToQueue', 'Comment', 'MinSize', 'MaxSize',
                      'Documentation'
                      ]
        output = 'Targets.csv'

    else:
        fieldnames = ['Disabled', 'Local', 'Module', 'Description', 'Category', 'Author',
                      'Version', 'Id', 'BinaryUrl', 'ExportFormat',
                      'WaitTimeout', 'FileMask', 'Executable', 'CommandLine',
                      'ExportFormat', 'ExportFile', 'Append', 'Documentation'
                      ]
        output = 'Modules.csv'

    with open(output, 'w', newline='') as f_output:
        csv_output = csv.DictWriter(f_output, fieldnames=fieldnames)
        csv_output.writeheader()

        for filename in filenames:
            print(f'Parsing {filename}')
            with open(filename) as f_input:
                try:
                    data = yaml.safe_load(f_input)
                except Exception as e:
                    print(f'Unable to parse {f_input}: {e}')
                    continue
                doc = False
                f_input.seek(0)
                Documentation = []
                while True:
                    line = f_input.readline()
                    if not line:
                        break
                    if doc:
                        Documentation.append(line)
                    if '# Documentation' in line:
                        doc = True

            if args.target:
                Disabled = 'False'
                Local = 'False'
                Target = os.path.basename(f_input.name).split('.')[0]
                if '!Disabled' in f_input.name:
                    Disabled = 'True'
                if '!Local' in f_input.name:
                    Local = 'True'
                Description = data['Description']
                Author = data['Author']
                Version = data['Version']
                Id = data['Id']
                RecreateDirectories = data['RecreateDirectories']

                for dic in data['Targets']:
                    row = {'Disabled': Disabled, 'Local': Local, 'Target': Target, 'Description': Description,
                           'Author': Author, 'Version': Version, 'Id': Id,
                           'RecreateDirectories': RecreateDirectories
                           }

                    for get in ['Name', 'Category', 'Path', 'Recursive',
                                'FileMask', 'SaveAsFileName',
                                'AlwaysAddToQueue', 'Comment', 'MinSize',
                                'MaxSize'
                                ]:
                        try:
                            row[get] = dic[get]
                        except KeyError:
                            pass

                    row['Documentation'] = ''.join(Documentation).replace('# ', '')

                    csv_output.writerow(row)
            else:
                Disabled = 'False'
                Local = 'False'
                Module = os.path.basename(f_input.name).split('.')[0]
                if '!Disabled' in f_input.name:
                    Disabled = 'True'
                if '!Local' in f_input.name:
                    Local = 'True'
                Description = data['Description']
                Category = data['Category']
                Author = data['Author']
                Version = data['Version']
                Id = data['Id']
                try:
                    BinaryUrl = data['BinaryUrl']
                except KeyError:
                    BinaryUrl = ''
                ExportFormat = data['ExportFormat']
                try:
                    WaitTimeout = data['WaitTimeout']
                except KeyError:
                    WaitTimeout = ''
                try:
                    FileMask = data['FileMask']
                except KeyError:
                    FileMask = ''

                for dic in data['Processors']:
                    row = {'Disabled': Disabled, 'Local': Local, 'Module': Module, 'Description': Description,
                           'Category': Category, 'Author': Author,
                           'Version': Version, 'Id': Id, 'BinaryUrl': BinaryUrl,
                           'ExportFormat': ExportFormat,
                           'WaitTimeout': WaitTimeout, 'FileMask': FileMask
                           }

                    for get in ['Executable', 'CommandLine', 'ExportFormat',
                                'ExportFile', 'Append'
                                ]:
                        try:
                            row[get] = dic[get]
                        except KeyError:
                            pass

                    row['Documentation'] = ''.join(Documentation).replace('# ', '')

                    csv_output.writerow(row)


parser = argparse.ArgumentParser()
parser.add_argument("-d", "--dir", help="Directory to be parsed")
parser.add_argument("--target", help="Directory to be parsed", action='store_true')
parser.add_argument("--module", help="Directory to be parsed", action='store_true')

if len(sys.argv) == 1:
    parser.print_help()
    parser.exit()

args = parser.parse_args()

for path, subdirs, files in os.walk(args.dir):
    if 'bin' not in path:

        for name in files:
            filenames.append(os.path.join(path, name))

if __name__ == "__main__":
    main()

Requires PyYAML
pip install PyYAML