Add command 'contacts' for extracting contact information

.gitignore

@@ -11,3 +11,4 @@ data/*
 build
 imports
 backup*
+blocklist.txt

blocklist.example.txt

@@ -0,0 +1,2 @@
+[email protected]
+[email protected]

config.yml

@@ -11,3 +11,4 @@ output:
   invoice_file: invoices.pdf
   match_dir: build/matches
   rank_dir: build/rankings
+  contacts_dir: build/contacts

lib/inspector.py

@@ -6,6 +6,7 @@
 from operator import itemgetter
 from os.path import basename, join
 
+import pendulum
 from pandas import DataFrame
 from PyPDF2 import PdfFileReader, PdfFileMerger
 
@@ -217,3 +218,54 @@ def rank(self, year, quarter) -> None:
         ranking_file = join(self.config['rank_dir'], file_name + '.csv')
 
         self.export_csv(ranking, ranking_file)
+
+
+    def contacts(self, cutoff_date):
+        today = pendulum.today()
+
+        # Set default date
+        if cutoff_date is None:
+            cutoff_date = today.subtract(years=2).to_datetime_string()[:10]
+
+        # Select order files to be analyzed
+        order_files = build_path(self.config['order_dir'])
+
+        # Fetch their content
+        orders = load_json(order_files)
+
+        # Load blacklisted mail addresses
+        with open('blocklist.txt', 'r') as file:
+            blocklist = file.read().splitlines()
+
+        codes = set()
+        contacts  = []
+
+        for order in sorted(orders, key=itemgetter('Datum'), reverse=True):
+            mail_address = order['Email']
+
+            if mail_address in blocklist:
+                continue
+
+            # Throw out everything before cutoff date
+            if order['Datum'] < cutoff_date:
+                continue
+
+            # Prepare dictionary
+            contact = {}
+
+            contact['Anrede'] = order['Anrede']
+            contact['Vorname'] = order['Vorname']
+            contact['Nachname'] = order['Nachname']
+            contact['Name'] = order['Name']
+            contact['Email'] = order['Email']
+            contact['Letzte Bestelltung'] = order['Datum']
+
+            if mail_address not in codes:
+                codes.add(mail_address)
+                contacts.append(contact)
+
+        # Write ranking to CSV file
+        file_name = cutoff_date + '_' + today.to_datetime_string()[:10]
+        contacts_file = join(self.config['contacts_dir'], file_name + '.csv')
+
+        self.export_csv(contacts, contacts_file)

main.py

@@ -132,5 +132,23 @@ def rank(
     click.echo(' done!')
 
 
+@click.option('-c', '--config', help='Path to configuration file.')
+@click.option('-d', '--date', help='Cutoff date in ISO date format, eg \'YYYY-MM-DD\'. Default: today two years ago')
+@cli.command()
+def contacts(config: str, date: str = None):
+    """Generates mailmerge-ready contact list"""
+
+    # Load config
+    config = load_config(config)
+
+    # Initialize object
+    handler = Inspector(config)
+
+    # Load & match data sources
+    click.echo('Generating contact list ..', nl=False)
+    handler.contacts(date)
+    click.echo(' done!')
+
+
 if __name__ == '__main__':
     cli()

requirements.txt

@@ -1,8 +1,17 @@
-knv-paypal-matcher @ git+https://github.com/fundevogel/knv-pypal@96480054d480c932da3e890d87053c6fc2047ebd
+astroid==2.5.6
+click==7.1.2
+isort==5.8.0
+lazy-object-proxy==1.6.0
+mccabe==0.6.1
 numpy==1.20.2
 pandas==1.2.4
+pendulum==2.1.2
+pylint==2.8.2
 PyPDF2==1.26.0
 python-dateutil==2.8.1
 pytz==2021.1
+pytzdata==2020.1
 PyYAML==5.4.1
 six==1.15.0
+toml==0.10.2
+wrapt==1.12.1