def createRow(email_id, mail, attach, msg_body, body_type, categories):
one = lambda arr : head(arr) if arr else ''
originating_ips = mail.get_all('x-originating-ip', [])
forensic_bcc = mail.get_all('x-libpst-forensic-bcc', [])
msgid = [clean_string(s, [(r'\n', ''), EXPR_OPTS['fix_utf8'], EXPR_OPTS['fix_tab'] ]) for s in mail.get_all('message-id', [])]
inreplyto = [clean_string(s, [(r'\n', ''), EXPR_OPTS['fix_utf8'], EXPR_OPTS['fix_tab'] ]) for s in mail.get_all('in-reply-to', [])]
references = [clean_string(s, [(r'\n', ' '), EXPR_OPTS['fix_utf8'], EXPR_OPTS['fix_tab'] ]) for s in mail.get_all('references', [])]
mail_date= mail.get_all('date', None)
subject = mail.get_all('subject', [])
subject = convert_encoded(one(subject))
senders, senders_line = addrs(mail.get_all('from', []))
# Parse and convert all addresses to utf-8
# results are tuple (addr, addr_line) where the _line variables contain the full email header and
# the first value is a list of only the address portion
tos, tos_line = addrs(mail.get_all('to', []))
delivered_to, delivered_to_line = addrs(mail.get_all('Delivered-To', []))
# Sometime tos is the string "Undisclosed-Recipient" in which case we can try the delivery address
if not tos and delivered_to:
tos = delivered_to
ccs, ccs_line = addrs(mail.get_all('cc', []))
bccs, bccs_line = addrs(mail.get_all('bcc', []))
doc = { "id": email_id,
"datetime": dateToUTCstr(head(mail_date)) if mail_date else None,
"originating_ips" : originating_ips,
"categories" : categories,
"senders": senders,
"senders_line": senders_line,
"tos": tos,
"tos_line": tos_line,
"delivered_to": delivered_to,
"delivered_to_line": delivered_to_line,
"ccs": ccs,
"ccs_line" : ccs_line,
"bccs": bccs,
"bccs_line" : bccs_line,
"forensic-bcc" : forensic_bcc,
"attachments": attach,
"messageid": msgid,
"inreplyto": inreplyto,
"references": references,
"subject": subject,
"body": msg_body,
"body_as_html": body_type == 'html'
}
return doc
def createRow(email_id, mail, attach, msg_body, body_type, categories):
one = lambda arr: head(arr) if arr else ''
originating_ips = mail.get_all('x-originating-ip', [])
forensic_bcc = mail.get_all('x-libpst-forensic-bcc', [])
msgid = [
clean_string(
s, [(r'\n', ''), EXPR_OPTS['fix_utf8'], EXPR_OPTS['fix_tab']])
for s in mail.get_all('message-id', [])
]
inreplyto = [
clean_string(
s, [(r'\n', ''), EXPR_OPTS['fix_utf8'], EXPR_OPTS['fix_tab']])
for s in mail.get_all('in-reply-to', [])
]
references = [
clean_string(
s, [(r'\n', ' '), EXPR_OPTS['fix_utf8'], EXPR_OPTS['fix_tab']])
for s in mail.get_all('references', [])
]
mail_date = mail.get_all('date', None)
subject = mail.get_all('subject', [])
subject = convert_encoded(one(subject))
senders, senders_line = addrs(mail.get_all('from', []))
# Parse and convert all addresses to utf-8
# results are tuple (addr, addr_line) where the _line variables contain the full email header and
# the first value is a list of only the address portion
tos, tos_line = addrs(mail.get_all('to', []))
delivered_to, delivered_to_line = addrs(mail.get_all('Delivered-To', []))
# Sometime tos is the string "Undisclosed-Recipient" in which case we can try the delivery address
if not tos and delivered_to:
tos = delivered_to
ccs, ccs_line = addrs(mail.get_all('cc', []))
bccs, bccs_line = addrs(mail.get_all('bcc', []))
doc = {
"id": email_id,
"datetime": dateToUTCstr(head(mail_date)) if mail_date else None,
"originating_ips": originating_ips,
"categories": categories,
"senders": senders,
"senders_line": senders_line,
"tos": tos,
"tos_line": tos_line,
"delivered_to": delivered_to,
"delivered_to_line": delivered_to_line,
"ccs": ccs,
"ccs_line": ccs_line,
"bccs": bccs,
"bccs_line": bccs_line,
"forensic-bcc": forensic_bcc,
"attachments": attach,
"messageid": msgid,
"inreplyto": inreplyto,
"references": references,
"subject": subject,
"body": msg_body,
"body_as_html": body_type == 'html'
}
return doc
def createRow(email_id, mail, attach, msg_body, categories):
#addr_tostr = lambda arr : ";".join(arr)
#addrs = lambda arr : [clean_string(addr.lower(), [(r'\'', '')]) for
#name, addr in getaddresses(arr)]
#csv_sep = lambda arr : ",".join(arr) if arr else ''
#scolon_sep = lambda arr : ";".join(arr) if arr else ''
##
## return tuple (extracted emails array, unprocessed parts as array)
##
def addrs(arr):
items = []
arr = [
clean_string(s.lower(),
[EXPR_OPTS['fix_utf8'], (r'\t', ';'), (r'\n', ';')])
for s in arr
]
for name, addr in getaddresses(arr):
if '@' in addr:
items.append(addr)
elif '@' in name:
items.append(name)
return ([clean_string(s.lower(), [(r'\'', '')]) for s in items], arr)
one = lambda arr: head(arr) if arr else ''
originating_ips = mail.get_all('x-originating-ip', [])
forensic_bcc = mail.get_all('x-libpst-forensic-bcc', [])
msgid = [
clean_string(
s, [(r'\n', ''), EXPR_OPTS['fix_utf8'], EXPR_OPTS['fix_tab']])
for s in mail.get_all('message-id', [])
]
inreplyto = [
clean_string(
s, [(r'\n', ''), EXPR_OPTS['fix_utf8'], EXPR_OPTS['fix_tab']])
for s in mail.get_all('in-reply-to', [])
]
references = [
clean_string(
s, [(r'\n', ' '), EXPR_OPTS['fix_utf8'], EXPR_OPTS['fix_tab']])
for s in mail.get_all('references', [])
]
mail_date = mail.get_all('date', None)
subject = mail.get_all('subject', [])
#importance ??
#ip ??
senders, senders_line = addrs(mail.get_all('from', []))
#senders = [target_email if s == 'mailer-daemon' else s for s in senders]
tos, tos_line = addrs(mail.get_all('to', []))
ccs, ccs_line = addrs(mail.get_all('cc', []))
bccs, bccs_line = addrs(mail.get_all('bcc', []))
subject = clean_string(
quopri.decodestring(one(subject)),
[EXPR_OPTS['fix_utf8'], EXPR_OPTS['fix_tab'], EXPR_OPTS['fix_cr']])
body = clean_string(
quopri.decodestring(msg_body),
[EXPR_OPTS['fix_utf8'], EXPR_OPTS['fix_tab'], EXPR_OPTS['fix_cr']])
doc = {
"id": email_id,
"datetime": dateToUTCstr(head(mail_date)) if mail_date else None,
"originating_ips": originating_ips,
"categories": categories,
"senders": senders,
"senders_line": senders_line,
"tos": tos,
"tos_line": tos_line,
"ccs": ccs,
"ccs_line": ccs_line,
"bccs": bccs,
"bccs_line": bccs_line,
"forensic-bcc": forensic_bcc,
"attachments": attach,
"messageid": msgid,
"inreplyto": inreplyto,
"references": references,
"subject": subject,
"body": body
}
return json.dumps(doc)
def createRow(email_id, mail, attach, msg_body, categories):
#addr_tostr = lambda arr : ";".join(arr)
#addrs = lambda arr : [clean_string(addr.lower(), [(r'\'', '')]) for
#name, addr in getaddresses(arr)]
#csv_sep = lambda arr : ",".join(arr) if arr else ''
#scolon_sep = lambda arr : ";".join(arr) if arr else ''
##
## return tuple (extracted emails array, unprocessed parts as array)
##
def addrs(arr):
items = []
arr = [clean_string(s.lower(), [EXPR_OPTS['fix_utf8'], (r'\t', ';'), (r'\n', ';') ]) for s in arr]
for name, addr in getaddresses(arr):
if '@' in addr:
items.append(addr)
elif '@' in name:
items.append(name)
return ([clean_string(s.lower(), [(r'\'', '')]) for s in items], arr)
one = lambda arr : head(arr) if arr else ''
originating_ips = mail.get_all('x-originating-ip', [])
forensic_bcc = mail.get_all('x-libpst-forensic-bcc', [])
msgid = [clean_string(s, [(r'\n', ''), EXPR_OPTS['fix_utf8'], EXPR_OPTS['fix_tab'] ]) for s in mail.get_all('message-id', [])]
inreplyto = [clean_string(s, [(r'\n', ''), EXPR_OPTS['fix_utf8'], EXPR_OPTS['fix_tab'] ]) for s in mail.get_all('in-reply-to', [])]
references = [clean_string(s, [(r'\n', ' '), EXPR_OPTS['fix_utf8'], EXPR_OPTS['fix_tab'] ]) for s in mail.get_all('references', [])]
mail_date= mail.get_all('date', None)
subject = mail.get_all('subject', [])
#importance ??
#ip ??
senders, senders_line = addrs(mail.get_all('from', []))
#senders = [target_email if s == 'mailer-daemon' else s for s in senders]
tos, tos_line = addrs(mail.get_all('to', []))
ccs, ccs_line = addrs(mail.get_all('cc', []))
bccs, bccs_line = addrs(mail.get_all('bcc', []))
subject = clean_string(quopri.decodestring(one(subject)),
[
EXPR_OPTS['fix_utf8'],
EXPR_OPTS['fix_tab'],
EXPR_OPTS['fix_cr']])
body = clean_string(quopri.decodestring(msg_body),
[
EXPR_OPTS['fix_utf8'],
EXPR_OPTS['fix_tab'],
EXPR_OPTS['fix_cr']])
doc = { "id": email_id,
"datetime": dateToUTCstr(head(mail_date)) if mail_date else None,
"originating_ips" : originating_ips,
"categories" : categories,
"senders": senders,
"senders_line": senders_line,
"tos": tos,
"tos_line": tos_line,
"ccs": ccs,
"ccs_line" : ccs_line,
"bccs": bccs,
"bccs_line" : bccs_line,
"forensic-bcc" : forensic_bcc,
"attachments": attach,
"messageid": msgid,
"inreplyto": inreplyto,
"references": references,
"subject": subject,
"body": body
}
return json.dumps(doc)
