46249649141088ccf310859496d7f36ac9a41326
8 import xml
.etree
.ElementTree
14 return mailbox
.uid('FETCH', num
, '(RFC822)')[1][0][1]
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
)
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
)]
30 reports
= [xml_of_part(message
)]
33 def extract_report(msg
):
34 pmsg
= email
.message_from_bytes(msg
)
37 def maybe_strip(text
):
43 field_maps
= {'./policy_published/adkim': {'pg_field_name': 'policy_published_adkim',
44 'pg_table': 'reports',
45 'pg_type': 'varchar'},
46 './policy_published/aspf': {'pg_field_name': 'policy_published_aspf',
47 'pg_table': 'reports',
48 'pg_type': 'varchar'},
49 './policy_published/domain': {'pg_field_name': 'policy_published_domain',
50 'pg_table': 'reports',
51 'pg_type': 'varchar'},
52 './policy_published/p': {'pg_field_name': 'policy_published_p',
53 'pg_table': 'reports',
54 'pg_type': 'varchar'},
55 './policy_published/pct': {'pg_field_name': 'policy_published_pct',
56 'pg_table': 'reports',
58 './record[{}]/auth_results/dkim/domain': {'pg_field_name': 'auth_results_dkim_domain',
59 'pg_table': 'report_items',
60 'pg_type': 'varchar'},
61 './record[{}]/auth_results/dkim/result': {'pg_field_name': 'auth_results_dkim_result',
62 'pg_table': 'report_items',
63 'pg_type': 'varchar'},
64 './record[{}]/auth_results/spf/domain': {'pg_field_name': 'auth_results_spf_domain',
65 'pg_table': 'report_items',
66 'pg_type': 'varchar'},
67 './record[{}]/auth_results/spf/result': {'pg_field_name': 'auth_results_spf_result',
68 'pg_table': 'report_items',
69 'pg_type': 'varchar'},
70 './record[{}]/identifiers/header_from': {'pg_field_name': 'identifiers_header_from',
71 'pg_table': 'report_items',
72 'pg_type': 'varchar'},
73 './record[{}]/row/count': {'pg_field_name': 'count',
74 'pg_table': 'report_items',
76 './record[{}]/row/policy_evaluated/disposition': {'pg_field_name': 'policy_evaluated_disposition',
77 'pg_table': 'report_items',
78 'pg_type': 'varchar'},
79 './record[{}]/row/policy_evaluated/dkim': {'pg_field_name': 'policy_evaluated_dkim',
80 'pg_table': 'report_items',
81 'pg_type': 'varchar'},
82 './record[{}]/row/policy_evaluated/spf': {'pg_field_name': 'policy_evaluated_spf',
83 'pg_table': 'report_items',
84 'pg_type': 'varchar'},
85 './record[{}]/row/source_ip': {'pg_field_name': 'source_ip',
86 'pg_table': 'report_items',
88 './report_metadata/date_range/begin': {'pg_field_name': 'report_metadata_date_range_begin',
89 'pg_table': 'reports',
90 'pg_type': 'timestamp'},
91 './report_metadata/date_range/end': {'pg_field_name': 'report_metadata_date_range_end',
92 'pg_table': 'reports',
93 'pg_type': 'timestamp'},
94 './report_metadata/email': {'pg_field_name': 'report_metadata_email',
95 'pg_table': 'reports',
96 'pg_type': 'varchar'},
97 './report_metadata/org_name': {'pg_field_name': 'report_metadata_org_name',
98 'pg_table': 'reports',
99 'pg_type': 'varchar'},
100 './report_metadata/report_id': {'pg_field_name': 'report_metadata_report_id',
101 'pg_table': 'reports',
102 'pg_type': 'varchar'}}
106 def build_insert_command(table_name
, report
, preamble_values
=None, i
=None):
109 values
= preamble_values
.copy()
112 for f
in [f
for f
in field_maps
if field_maps
[f
]['pg_table'] == table_name
]:
117 field_names
+= [field_maps
[f
]['pg_field_name']]
118 if field_maps
[f
]['pg_type'] == 'int':
119 values
[field_maps
[f
]['pg_field_name']] = int(report
.find(fp
).text
)
120 elif field_maps
[f
]['pg_type'] == 'timestamp':
121 values
[field_maps
[f
]['pg_field_name']] = datetime
.datetime
.utcfromtimestamp(int(report
.find(fp
).text
))
122 elif field_maps
[f
]['pg_type'] == 'inet':
123 values
[field_maps
[f
]['pg_field_name']] = maybe_strip(report
.find(fp
).text
)
125 values
[field_maps
[f
]['pg_field_name']] = maybe_strip(report
.find(fp
).text
)
126 insert_string
= 'insert into {} ('.format(table_name
)
128 insert_string
+= ', '.join(sorted(preamble_values
.keys())) + ', '
129 insert_string
+= ', '.join(field_names
) + ') '
130 insert_string
+= 'values ('
132 insert_string
+= ', '.join('%({})s'.format(fn
) for fn
in sorted(preamble_values
.keys())) + ', '
133 insert_string
+= ', '.join('%({})s'.format(f
) for f
in field_names
) + ');'
134 return insert_string
, values
137 def write_report(connection
, cursor
, report
):
138 insert_string
, values
= build_insert_command('reports', report
)
139 # print(insert_string, values)
140 cursor
.execute(insert_string
, values
)
142 for i
in range(1, len(report
.findall('./record'))+1):
144 cursor
.execute('select id, report_metadata_report_id from reports where report_metadata_report_id = %s;',
145 [report
.find('./report_metadata/report_id').text
])
146 results
= cursor
.fetchall()
147 if len(results
) != 1:
148 raise RuntimeError('Could not find report record for report item')
150 report_id
= results
[0][0]
151 insert_string
, values
= build_insert_command('report_items', report
, i
=i
,
152 preamble_values
={'report_id': report_id
})
153 # print(insert_string, values)
154 cursor
.execute(insert_string
, values
)
157 config
= configparser
.ConfigParser()
158 config
.read('dmarc.ini')
160 conn
= psycopg2
.connect(host
=config
['database']['server'],
161 database
=config
['database']['database'],
162 user
=config
['database']['username'],
163 password
=config
['database']['password'])
166 cur
.execute('select max(report_metadata_date_range_end) from reports')
167 results
= cur
.fetchall()
168 most_recent_date
= results
[0][0]
170 mailbox
= imaplib
.IMAP4(host
=config
['imap']['server'],
171 port
=config
['imap']['port'])
173 mailbox
.login(config
['imap']['username'], config
['imap']['password'])
174 mailbox
.select('INBOX', readonly
=True)
178 mails_from
= "SINCE " + (most_recent_date
- datetime
.timedelta(days
=2)).strftime("%d-%b-%Y")
181 resp
, nums
= mailbox
.uid('SEARCH', None, mails_from
)
184 dmarc_reports
= [report
for report_set
in [extract_report(fetch_msg(n
)) for n
in nums
[0].split()]
185 for report
in report_set
]
190 for report
in dmarc_reports
:
191 cur
.execute('select id, report_metadata_report_id from reports where report_metadata_report_id = %s;',
192 [report
.find('./report_metadata/report_id').text
])
193 results
= cur
.fetchall()
195 print('write', report
.find('./report_metadata/report_id').text
)
196 write_report(conn
, cur
, report
)