Merge pull request #109 from demisto/fixed-obj-issue

Fixed an issue where some EML files were incorrectly decoded.

Author: moishce

CHANGELOG.md

@@ -1,5 +1,8 @@
 # Changelog
 
+v0.1.35
+* Fixed an issue where some EML files were incorrectly decoded.
+
 v0.1.34
 * Fixed an issue that prevented the headers from being parsed correctly for eml files (unknown-8bit encoding).
 * Fixed an issue that email with attachment was parsed incorrectly for eml files.

parse_emails/common.py

@@ -44,7 +44,7 @@ def convert_to_unicode(s, is_msg_header=True):
         return res.strip()
     except Exception:
         if s and 'unknown-8bit' in s:
-            s = email.header.make_header(email.header.decode_header(s))
+            s = str(email.header.make_header(email.header.decode_header(s)))
         else:
             for file_data in ENCODINGS_TYPES:
                 try:

parse_emails/tests/parse_emails_test.py

@@ -991,6 +991,7 @@ def test_handle_eml_unknown8bit():
     email_parser = EmailParser(file_path='parse_emails/tests/test_data/test-unknown-8bit.eml')
     results = email_parser.parse()
     assert results['From'] == '[email protected]'
+    assert results['HeadersMap']['From'] == '"test" <[email protected]>'
 
 
 def test_multipart_eml_with_eml_attachment_containing_html_body():

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "parse-emails"
-version = "0.1.34"
+version = "0.1.35"
 description = "Parses an email message file and extracts the data from it."
 authors = ["Demisto"]
 license = "MIT"