From 7020c53b28deecd2546386bf8b18efb97f4d67da Mon Sep 17 00:00:00 2001
From: Thomas Oettli <spacefreak@noop.ch>
Date: Sat, 26 Oct 2019 13:04:19 +0200
Subject: [PATCH] Add option notification_email_parser_lib

---
 README.md                      |  2 ++
 docs/pyquarantine.conf.example |  9 ++++++++-
 pyquarantine/notifications.py  | 16 +++++++++++++---
 3 files changed, 23 insertions(+), 4 deletions(-)

diff --git a/README.md b/README.md
index c056ee4..016a019 100644
--- a/README.md
+++ b/README.md
@@ -114,6 +114,8 @@ The following configuration options are optional in each quarantine section:
     Enable to strip images from e-mails. This option superseeds notification_email_replacement_img.
   * **notification_email_replacement_img**  
     Path to an image to replace images in e-mails. It is hold in memory during runtime.
+  * **notification_email_parser_lib**  
+    HTML parser library used to parse text part of emails.
 
 
 ### Actions
diff --git a/docs/pyquarantine.conf.example b/docs/pyquarantine.conf.example
index 885927b..679fac7 100644
--- a/docs/pyquarantine.conf.example
+++ b/docs/pyquarantine.conf.example
@@ -119,7 +119,7 @@ notification_email_template = templates/notification.template
 
 # Option: notification_email_strip_images
 # Notes:  Optionally enable this option to strip img tags from emails.
-# Values: [ TRUE|ON|YES|FALSE|OFF|NO ]
+# Values: [ TRUE | ON | YES | FALSE | OFF | NO ]
 #
 notification_email_strip_images = False
 
@@ -138,6 +138,13 @@ notification_email_replacement_img = templates/removed.png
 #
 notification_email_embedded_imgs = templates/logo.png
 
+# Option: notification_email_parser_lib
+# Notes:  Optionally set the parser library used to parse
+#         the text part of emails.
+# Values: [ lxml | html.parser ]
+#
+notification_email_parser_lib = lxml
+
 # Option: whitelist_type
 # Notes:  Set the whitelist type.
 # Values: [ db | none ]
diff --git a/pyquarantine/notifications.py b/pyquarantine/notifications.py
index 8eed921..cbe1eb4 100644
--- a/pyquarantine/notifications.py
+++ b/pyquarantine/notifications.py
@@ -132,7 +132,8 @@ class EMailNotification(BaseNotification):
         # check if optional config options are present in config
         defaults = {
             "notification_email_replacement_img": "",
-            "notification_email_strip_images": "false"
+            "notification_email_strip_images": "false",
+            "notification_email_parser_lib": "lxml"
         }
         for option in defaults.keys():
             if option not in config.keys() and \
@@ -181,6 +182,10 @@ class EMailNotification(BaseNotification):
         else:
             raise RuntimeError("error parsing notification_email_strip_images: unknown value")
 
+        self.parser_lib = self.config["notification_email_parser_lib"].strip()
+        if self.parser_lib not in ["lxml", "html.parser"]:
+            raise RuntimeError("error parsing notification_email_parser_lib: unknown value")
+
         # read email replacement image if specified
         replacement_img = self.config["notification_email_replacement_img"].strip()
         if not strip_images and replacement_img:
@@ -229,8 +234,13 @@ class EMailNotification(BaseNotification):
                 "{}: content mimetype is {}".format(
                     queueid, mimetype))
         self.logger.debug(
-            "{}: creating BeatufilSoup object".format(queueid))
-        return BeautifulSoup(text, "lxml")
+            "{}: trying to create BeatufilSoup object with parser lib {}, "
+            "text length is {} bytes".format(
+                queueid, self.parser_lib, len(text)))
+        soup = BeautifulSoup(text, self.parser_lib)
+        self.logger.debug(
+            "{}: sucessfully created BeautifulSoup object".format(queueid))
+        return soup
 
     def get_text_multipart(self, queueid, msg, preferred=_html_text):
         "Get the mail text of a multipart email in html form."