dmarc_to_database.py

   1 #!/usr/bin/python3
   2
   3 import configparser
   4 import imaplib
   5 import email
   6 import io
   7 import zipfile
   8 import xml.etree.ElementTree
   9 import psycopg2
  10 import re
  11 import datetime
  12
  13 def fetch_msg(num):
  14     return mailbox.uid('FETCH', num, '(RFC822)')[1][0][1]
  15
  16 def xml_of_part(part):
  17     with zipfile.ZipFile(io.BytesIO(part.get_payload(decode=True))) as zf:
  18         fn = zf.infolist()[0].filename
  19         contents = zf.read(fn).decode('utf-8')
  20         return xml.etree.ElementTree.fromstring(contents)
  21
  22
  23 def xml_of(message):
  24     reports = []
  25     if message.is_multipart():
  26         for p in message.get_payload():
  27             if 'zip' in p.get_content_type():
  28                 reports += [xml_of_part(p)]
  29     else:
  30         reports = [xml_of_part(message)]
  31     return reports
  32
  33 def extract_report(msg):
  34     pmsg = email.message_from_bytes(msg)
  35     return xml_of(pmsg)
  36
  37 def maybe_strip(text):
  38     if text:
  39         return text.strip()
  40     else:
  41         return ''
  42
  43 field_maps = {
  44     './policy_published/adkim':
  45         {'pg_field_name': 'policy_published_adkim',
  46          'pg_table': 'reports', 'pg_type': 'varchar'},
  47     './policy_published/aspf':
  48         {'pg_field_name': 'policy_published_aspf',
  49          'pg_table': 'reports', 'pg_type': 'varchar'},
  50      './policy_published/domain':
  51         {'pg_field_name': 'policy_published_domain',
  52          'pg_table': 'reports', 'pg_type': 'varchar'},
  53      './policy_published/p':
  54         {'pg_field_name': 'policy_published_p',
  55          'pg_table': 'reports', 'pg_type': 'varchar'},
  56      './policy_published/pct':
  57         {'pg_field_name': 'policy_published_pct',
  58          'pg_table': 'reports', 'pg_type': 'int'},
  59      './record[{}]/auth_results/dkim/domain':
  60         {'pg_field_name': 'auth_results_dkim_domain',
  61          'pg_table': 'report_items', 'pg_type': 'varchar'},
  62      './record[{}]/auth_results/dkim/result':
  63         {'pg_field_name': 'auth_results_dkim_result',
  64          'pg_table': 'report_items', 'pg_type': 'varchar'},
  65      './record[{}]/auth_results/spf/domain':
  66         {'pg_field_name': 'auth_results_spf_domain',
  67          'pg_table': 'report_items', 'pg_type': 'varchar'},
  68      './record[{}]/auth_results/spf/result':
  69         {'pg_field_name': 'auth_results_spf_result',
  70          'pg_table': 'report_items', 'pg_type': 'varchar'},
  71      './record[{}]/identifiers/header_from':
  72         {'pg_field_name': 'identifiers_header_from',
  73          'pg_table': 'report_items', 'pg_type': 'varchar'},
  74      './record[{}]/row/count':
  75         {'pg_field_name': 'count',
  76          'pg_table': 'report_items', 'pg_type': 'int'},
  77      './record[{}]/row/policy_evaluated/disposition':
  78         {'pg_field_name': 'policy_evaluated_disposition',
  79          'pg_table': 'report_items', 'pg_type': 'varchar'},
  80      './record[{}]/row/policy_evaluated/dkim':
  81         {'pg_field_name': 'policy_evaluated_dkim',
  82          'pg_table': 'report_items', 'pg_type': 'varchar'},
  83      './record[{}]/row/policy_evaluated/spf':
  84         {'pg_field_name': 'policy_evaluated_spf',
  85          'pg_table': 'report_items', 'pg_type': 'varchar'},
  86      './record[{}]/row/source_ip':
  87         {'pg_field_name': 'source_ip',
  88          'pg_table': 'report_items', 'pg_type': 'inet'},
  89      './report_metadata/date_range/begin':
  90         {'pg_field_name': 'report_metadata_date_range_begin',
  91          'pg_table': 'reports', 'pg_type': 'timestamp'},
  92      './report_metadata/date_range/end':
  93         {'pg_field_name': 'report_metadata_date_range_end',
  94          'pg_table': 'reports', 'pg_type': 'timestamp'},
  95      './report_metadata/email':
  96         {'pg_field_name': 'report_metadata_email',
  97          'pg_table': 'reports', 'pg_type': 'varchar'},
  98      './report_metadata/org_name':
  99         {'pg_field_name': 'report_metadata_org_name',
 100          'pg_table': 'reports', 'pg_type': 'varchar'},
 101      './report_metadata/report_id':
 102         {'pg_field_name': 'report_metadata_report_id',
 103          'pg_table': 'reports', 'pg_type': 'varchar'}}
 104
 105
 106
 107 def build_insert_command(table_name, report, preamble_values=None, i=None):
 108     field_names = []
 109     if preamble_values:
 110         values = preamble_values.copy()
 111     else:
 112         values = {}
 113     for f in [f for f in field_maps if field_maps[f]['pg_table'] == table_name]:
 114         if i:
 115             fp = f.format(i)
 116         else:
 117             fp = f
 118         field_names += [field_maps[f]['pg_field_name']]
 119         if field_maps[f]['pg_type'] == 'int':
 120             values[field_maps[f]['pg_field_name']] = int(report.find(fp).text)
 121         elif field_maps[f]['pg_type'] == 'timestamp':
 122             values[field_maps[f]['pg_field_name']] = datetime.datetime.utcfromtimestamp(int(report.find(fp).text))
 123         elif field_maps[f]['pg_type'] == 'inet':
 124             values[field_maps[f]['pg_field_name']] = maybe_strip(report.find(fp).text)
 125         else:
 126             values[field_maps[f]['pg_field_name']] = maybe_strip(report.find(fp).text)
 127     insert_string = 'insert into {} ('.format(table_name)
 128     if preamble_values:
 129         insert_string += ', '.join(sorted(preamble_values.keys())) + ', '
 130     insert_string += ', '.join(field_names) + ') '
 131     insert_string += 'values ('
 132     if preamble_values:
 133         insert_string += ', '.join('%({})s'.format(fn) for fn in sorted(preamble_values.keys())) + ', '
 134     insert_string += ', '.join('%({})s'.format(f) for f in field_names) + ');'
 135     return insert_string, values
 136
 137
 138 def write_report(connection, cursor, report):
 139     insert_string, values = build_insert_command('reports', report)
 140     # print(insert_string, values)
 141     cursor.execute(insert_string, values)
 142
 143     for i in range(1, len(report.findall('./record'))+1):
 144         field_names = []
 145         cursor.execute('select id, report_metadata_report_id from reports where report_metadata_report_id = %s;',
 146             [report.find('./report_metadata/report_id').text])
 147         results = cursor.fetchall()
 148         if len(results) != 1:
 149             raise RuntimeError('Could not find report record for report item')
 150         else:
 151             report_id = results[0][0]
 152         insert_string, values = build_insert_command('report_items', report, i=i,
 153                                                      preamble_values={'report_id': report_id})
 154         # print(insert_string, values)
 155         cursor.execute(insert_string, values)
 156     connection.commit()
 157
 158 config = configparser.ConfigParser()
 159 config.read('dmarc.ini')
 160
 161 conn = psycopg2.connect(host=config['database']['server'],
 162                         database=config['database']['database'],
 163                         user=config['database']['username'],
 164                         password=config['database']['password'])
 165
 166 cur = conn.cursor()
 167 cur.execute('select max(report_metadata_date_range_end) from reports')
 168 results = cur.fetchall()
 169 most_recent_date = results[0][0]
 170
 171 mailbox = imaplib.IMAP4(host=config['imap']['server'],
 172                       port=config['imap']['port'])
 173 mailbox.starttls()
 174 mailbox.login(config['imap']['username'], config['imap']['password'])
 175 mailbox.select('INBOX', readonly=True)
 176
 177
 178 if most_recent_date:
 179     mails_from = "SINCE " + (most_recent_date - datetime.timedelta(days=2)).strftime("%d-%b-%Y")
 180 else:
 181     mails_from = "ALL"
 182 resp, nums = mailbox.uid('SEARCH', None, mails_from)
 183
 184
 185 dmarc_reports = [report for report_set in [extract_report(fetch_msg(n)) for n in nums[0].split()]
 186                 for report in report_set]
 187
 188 mailbox.close()
 189 mailbox.logout()
 190
 191 for report in dmarc_reports:
 192     cur.execute('select id, report_metadata_report_id from reports where report_metadata_report_id = %s;',
 193         [report.find('./report_metadata/report_id').text])
 194     results = cur.fetchall()
 195     if not results:
 196         print('write', report.find('./report_metadata/report_id').text)
 197         write_report(conn, cur, report)
 198
 199 conn.close()