URI/FQDN extraction

app/Entities.py

@@ -178,3 +178,27 @@ class Attachment:
     if 'href' in at_ref:
       self.href = at_ref['href']
 
+class URIException(Exception):
+  message = None
+  def __init__(self,message):
+    self.message = message
+
+class URI:
+  id = None
+  uri = None
+  fqdn = None
+  href = None
+
+  def __init__(self,uri_ref):
+    if 'id' not in uri_ref:
+      raise URIException("'id' is mandatory!")
+    self.id = uri_ref['id']
+    if 'uri' not in uri_ref:
+      raise URIException("'uri' is mandatory!")
+    self.uri = uri_ref['uri']
+    if 'fqdn' not in uri_ref:
+      raise URIException("'fqdn' is mandatory!")
+    self.fqdn = uri_ref['fqdn']
+    if 'href' in uri_ref:
+      self.href = uri_ref['href']
+    

app/Gulag.py

@@ -1,8 +1,8 @@
-import json,sys,os,logging
+import json,sys,os,logging,re
 import email,email.header,email.message
 from GulagDB import GulagDB,GulagDBException
 from GulagMailbox import IMAPmailbox,IMAPmailboxException
-from GulagUtils import whoami
+from GulagUtils import whoami,extract_uris,extract_fqdn
 
 class GulagException(Exception):
   message = None
@@ -78,6 +78,7 @@ class Gulag:
       for unseen in imap_mb.get_unseen_messages():
         quarmail_ids = []
         attachments = []
+        uris = {}
         uid = unseen['imap_uid']
         msg = email.message_from_bytes(unseen['msg'])
         msg_size = len(msg)
@@ -154,12 +155,30 @@ class Gulag:
             })
             attachments.append(attach_id)
           # Ende if part.get_filename()
+          # get all URIs
+          ctype = part.get_content_type()
+          if(ctype == 'text/plain' or ctype == 'text/html'):
+            curis = {}
+            curis = extract_uris(part.get_payload(decode=True).decode("utf-8"))
+            if(len(curis) > 0):
+              uris = {**uris, **curis}
         # Ende for msg.walk()
         # QuarMail und Attachments verknüpfen
         if(len(attachments) > 0):
           for quarmail_id in quarmail_ids:
             for attachment_id in attachments:
               self.db.quarmail2attachment(str(quarmail_id), str(attachment_id))
+        if(len(uris) > 0):
+          for quarmail_id in quarmail_ids:
+            for uri in uris:
+              try:
+                uri_id = self.db.add_uri({
+                  "uri": uri,
+                  "fqdn": extract_fqdn(uri)
+                })
+                self.db.quarmail2uri(str(quarmail_id), str(uri_id))
+              except GulagDBException as e:
+                logging.error(whoami(self) + e.message)
       # Ende for(unseen)
       imap_mb.close()
     # Ende for get_mailboxes
@@ -256,10 +275,38 @@ class Gulag:
     if 'data' not in args:
       return at_db
 
-  def get_uris(self):
-    # https://stackoverflow.com/questions/1792366/extract-urls-out-of-email-in-python
-    return True
-
+  def get_quarmail_uris(self,args):
+    if('from_rfc822_message' not in args):
+      try:
+        return self.db.get_quarmail_uris(args['quarmail_id'])
+      except GulagDBException as e:
+        raise GulagException(whoami(self) + e.message) from e
+    qm_db = None
+    try:
+      qm_db = self.db.get_quarmail({"id": args['quarmail_id']})
+    except GulagDBException as e:
+      logging.warning(whoami(self) + e.message)
+      raise GulagException(whoami(self) + e.message) from e
+    mailbox = None
+    try:
+      mailbox = self.db.get_mailbox(qm_db['mailbox_id'])
+    except GulagDBException as e:
+      logging.warning(whoami(self) + e.message)
+      raise GulagException(whoami(self) + e.message) from e 
+    imap_mb = None
+    try:
+      imap_mb = IMAPmailbox(mailbox)
+      mparts = imap_mb.get_main_parts(qm_db['imap_uid'])
+      uris = []
+      uri_pattern = r'(https?:\/\/[^\s<>"]+)'
+      for part in mparts:
+        for m in re.finditer(uri_pattern, part.decode("utf-8")):
+          uris.append(m.group(0))
+      return uris
+    except IMAPmailboxException as e:
+      logging.warning(whoami(self) + e.message)
+      raise GulagException(whoami(self) + e.message) from e
+    
   def rspamd_http2imap(self,args):
     mailbox = None
     try:
@@ -302,7 +349,7 @@ class Gulag:
         )
         logging.error(err)
         raise GulagException(err)
-      if('rfc822_message' not in args['rfc822_message']):
+      if('rfc822_message' not in args):
         err = str(whoami(self) 
           + "Missing rfc822_message!"
         )

app/GulagDB.py

@@ -2,7 +2,7 @@ import mysql.connector as mariadb
 from Entities import(
   Mailbox,MailboxException,QuarMail,
   QuarMailException,Attachment,
-  AttachmentException
+  AttachmentException,URI,URIException
 )
 from GulagUtils import whoami
 
@@ -99,6 +99,11 @@ class GulagDB:
       cnt += 1
     return where_clause
 
+  def parse_filters(self,filters):
+    # TODO
+    # {"groupOp":"AND","rules":[{"field":"Customer","op":"eq","data":"eosp"}]}
+    return True
+
   def get_mailboxes(self):
     try:
       cursor = self.conn.cursor()
@@ -217,7 +222,7 @@ class GulagDB:
       data = cursor.fetchall()
       if not data:
         raise GulagDBException(whoami(self)
-          + "Quarmail with id '"+ args['id'] + "' does not exist!"
+          + "Quarmail with id '"+ str(args['id']) + "' does not exist!"
         )
       desc = cursor.description
       cursor.close()
@@ -380,3 +385,61 @@ class GulagDB:
     except mariadb.Error as e:
       raise GulagDBException(whoami(self) + str(e)) from e
 
+  def add_uri(self,args):
+    try:
+      cursor = self.conn.cursor()
+      cursor.execute("insert into URIs " +
+        "(uri, fqdn) values (%s,%s)",
+        (args['uri'], args['fqdn'])
+      )
+      return cursor.lastrowid
+    except mariadb.Error as e:
+      raise GulagDBException(whoami(self) + str(e)) from e
+
+  def del_uri(self,uri_id):
+    try:
+      cursor = self.conn.cursor()
+      cursor.execute(
+        "delete from URIs where uri_id=" +  uri_id + ";"
+      )
+      return cursor.lastrowid
+    except mariadb.Error as e:
+      raise GulagDBException(whoami(self) + str(e)) from e
+
+
+  def quarmail2uri(self,quarmail_id,uri_id):
+    try:
+      cursor = self.conn.cursor()
+      cursor.execute("insert into QuarMail2URI " +
+        "(quarmail_id, uri_id) values (%s,%s)",
+        (quarmail_id, uri_id)
+      )
+    except mariadb.Error as e:
+      raise GulagDBException(whoami(self) + str(e)) from e
+    
+  def get_quarmail_uris(self,quarmail_id):
+    try:
+      query = "select URIs.*"
+      query += " from QuarMail2URI"
+      query += " left join QuarMails ON QuarMails.id = QuarMail2URI.quarmail_id"
+      query += " left join URIs ON URIs.id = QuarMail2URI.uri_id"
+      query += " where QuarMails.id = " + str(quarmail_id) + ";"
+      cursor = self.conn.cursor()
+      cursor.execute(query)
+      results = []
+      data = cursor.fetchall()
+      if not data:
+        raise GulagDBException(whoami(self)
+          + "QuarMail("+ str(quarmail_id) +") has no uris!"
+        )
+      desc = cursor.description
+      for tuple in data:
+        dict = {}
+        for (name, value) in zip(desc, tuple):
+          dict[name[0]] = value
+        dict['href'] = self.uri_prefixes['quarmails'] + str(quarmail_id)
+        dict['href'] += "/uris/" + str(dict['id'])
+        results.append(URI(dict).__dict__)
+      return results
+    except mariadb.Error as e:
+      raise GulagDBException(whoami(self) + str(e)) from e

app/GulagMailbox.py

@@ -91,6 +91,19 @@ class IMAPmailbox:
       + str(self.email_address) + " not found!"
     )
 
+  def get_main_parts(self,imap_uid):
+    msg = email.message_from_bytes(self.get_message(imap_uid))
+    mparts = []
+    for part in msg.walk():
+      ctype = part.get_content_type()
+      if(ctype == 'text/plain' or ctype == 'text/html'):
+        mparts.append(part.get_payload(decode=True))
+    if(len(mparts) > 0):
+      return mparts
+    raise IMAPmailboxException(whoami(self) +
+      "IMAP_UID(" + str(imap_uid)+")@"+str(self.email_address)+" has no main parts!"
+    )
+
   def append_message(self,message):
     rv, data = self.mailbox.append(
       self.imap_mailbox,

app/GulagUtils.py

@@ -1,4 +1,4 @@
-import sys
+import sys,re
 from smtplib import SMTP
 
 def whoami(obj):
@@ -20,3 +20,17 @@ def send_mail(args):
   except TimeoutError as e:
     raise Exception('xyz') from e
 
+def extract_uris(string):
+  uris = {}
+  uri_pattern = r'(https?:\/\/[^\s<>"]+)'
+  for m in re.finditer(uri_pattern, string):
+    uris[m.group(0)] = {}
+  return uris
+
+def extract_fqdn(uri):
+  uri_pattern = r'(https?:\/\/[^\s<>"]+)'
+  if(re.match(uri_pattern,uri)):
+    m = re.match(r'https?:\/\/([^:\/]+)', uri)
+    return m.group(1)
+  else:
+    return None

app/Resources.py

@@ -82,6 +82,18 @@ class ResQuarMailAttachment(GulagResource):
     except GulagException as e:
       abort(400, message=e.message)
 
+class ResQuarMailURIs(GulagResource):
+  def get(self,quarmail_id): 
+    args = {
+      "quarmail_id": quarmail_id
+    }
+    if(request.args.get('from_rfc822_message')):
+        args['from_rfc822_message'] = True
+    try:
+      return self.gulag.get_quarmail_uris(args)
+    except GulagException as e:
+      abort(400, message=e.message)
+
 class ResAttachments(GulagResource):
   def get(self):
     return {"resource": "Attachments"}

app/gulag_server.py

@@ -7,7 +7,7 @@ from Gulag import Gulag,GulagException
 from Resources import (ResRoot,ResMailboxes,
   ResQuarMails,ResQuarMail,ResQuarMailAttachments,
   ResQuarMailAttachment,ResAttachments,ResAttachment,
-  ResRSPAMDImporter
+  ResRSPAMDImporter,ResQuarMailURIs
 )
 parser = argparse.ArgumentParser()
 parser.add_argument('--config', required=True, help="Path to config file")
@@ -44,6 +44,10 @@ try:
     '/api/v1/quarmails/<int:quarmail_id>/attachments/<int:attachment_id>',
     resource_class_kwargs={'gulag_object': gulag}
   )
+  api.add_resource(ResQuarMailURIs,
+    '/api/v1/quarmails/<int:quarmail_id>/uris',
+    resource_class_kwargs={'gulag_object': gulag}
+  )
   api.add_resource(ResAttachments,
     '/api/v1/attachments',
     resource_class_kwargs={'gulag_object': gulag}

db/gulag.sql

@@ -20,10 +20,6 @@ create table Mailboxes(
   smtp_pass varchar(2048) default null,
   comment varchar(256) default null
 )ENGINE = InnoDB;
-insert into Mailboxes (email_address,name,imap_user,imap_pass) 
-  values('quarantine-in@example.org','E-Mail inbound quarantine','quarantine-in','quarantine-in_secure_password');
-insert into Mailboxes (email_address,name,imap_user,imap_pass) 
-  values('quarantine-out@example.org','E-Mail outbound quarantine','quarantine-out','quarantine-out_secure_password');
 insert into Mailboxes (email_address,name,imap_user,imap_pass) 
   values('quarantine-sandbox@example.org','E-Mail sandbox quarantine','quarantine-sb','quarantine-sb_secure_password');
 
@@ -60,3 +56,15 @@ create table QuarMail2Attachment (
   foreign key (attachment_id) references Attachments (id) on delete cascade on update cascade
 )ENGINE = InnoDB;
 
+create table URIs (
+  id int unsigned auto_increment primary key,
+  uri varchar(2048),
+  fqdn varchar(512)
+)ENGINE = InnoDB;
+
+create table QuarMail2URI (
+  quarmail_id int unsigned,
+  uri_id int unsigned,
+  foreign key (quarmail_id) references QuarMails (id) on delete cascade on update cascade,
+  foreign key (uri_id) references URIs (id) on delete cascade on update cascade
+)ENGINE = InnoDB;

gulag-openapi-2.0.yaml

@@ -1,7 +1,7 @@
 swagger: '2.0'
 info:
   description: Gulag quarantine REST API
-  version: '18.12'
+  version: "1.0.0"
   title: Gulag quarantine REST API
   contact:
     email: info@dc-it-con.de
@@ -207,6 +207,35 @@ paths:
           description: bad input parameter
         500:
           description: server error
+          
+  /quarmails/{quarmail_id}/uris:
+    get:
+      summary: "retrieves all URIS from any main MIME part (text/plain,text/html)"
+      operationId: get_quarmail_uris
+      produces:
+      - application/json
+      parameters:
+      - in: path
+        name: quarmail_id
+        description: unique id of quarantined email
+        required: true
+        type: string
+      - in: query
+        name: from_rfc822_message
+        description: fetch all URIs from RFC822 message not from database
+        required: false
+        type: string
+      responses:
+        200:
+          description: array of URIs
+          schema:
+            type: array
+            items:
+              $ref: '#/definitions/URI'
+        400:
+          description: bad input parameter
+        500:
+          description: server error
 
 definitions:
   QuarMail:
@@ -321,3 +350,16 @@ definitions:
       data:
         type: string
         description: raw/encoded (see content_encoding) attachment payload
+  URI:
+    type: object
+    required:
+    - id
+    - uri
+    - fqdn
+    properties:
+      id:
+        type: integer
+      uri:
+        type: string
+      fqdn:
+        type: string