Revert "feat: improve RTF support"

Revert the change because of a performance issue

backend/factories/eml.py

@@ -12,7 +12,6 @@
 from returns.result import ResultE, safe
 
 from backend import schemas
-from backend.monkeypatch import with_monkey_patched_rtfparser
 from backend.outlookmsgfile import Message
 from backend.utils import parse_urls_from_body
 from backend.validator import is_eml_file
@@ -47,10 +46,8 @@ def to_eml(data: bytes) -> bytes:
 
  # assume data is a msg file
  file = BytesIO(data)
- with with_monkey_patched_rtfparser():
- message = Message(file)
- email = message.to_email()
-
+ message = Message(file)
+ email = message.to_email()
  return email.as_bytes()
 
 

backend/outlookmsgfile.py

@@ -17,7 +17,6 @@
 import email.message
 import email.parser
 import email.policy
-import io
 import os
 import re
 from email.message import EmailMessage
@@ -26,11 +25,8 @@
 from typing import BinaryIO
 
 import compressed_rtf
-import html2text
 from compoundfiles import CompoundFileEntity, CompoundFileReader
 from loguru import logger
-from rtfparse.parser import Rtf_Parser
-from rtfparse.renderers.de_encapsulate_html import De_encapsulate_HTML
 
 FALLBACK_ENCODING = "cp1252"
 
@@ -85,8 +81,11 @@ def load_message_stream( # noqa: C901
 
  if "SENDER_NAME" in props:
  if "SENT_REPRESENTING_NAME" in props:
- if props["SENDER_NAME"] != props["SENT_REPRESENTING_NAME"]:
- props["SENDER_NAME"] += " (" + props["SENT_REPRESENTING_NAME"] + ")"
+ if props["SENT_REPRESENTING_NAME"]: # noqa: SIM102
+ if props["SENDER_NAME"] != props["SENT_REPRESENTING_NAME"]:
+ props["SENDER_NAME"] += (
+ " (" + props["SENT_REPRESENTING_NAME"] + ")"
+ )
  del props["SENT_REPRESENTING_NAME"]
  if props["SENDER_NAME"]:
  msg["From"] = formataddr((props["SENDER_NAME"], ""))
@@ -112,61 +111,30 @@ def load_message_stream( # noqa: C901
  msg["Subject"] = props["SUBJECT"]
  del props["SUBJECT"]
 
- # Add a plain text body from the BODY field.
- has_body = False
+ # Add the plain-text body from the BODY field.
  if "BODY" in props:
  body = props["BODY"]
  if isinstance(body, str):
  msg.set_content(body, cte="quoted-printable")
  else:
  msg.set_content(body, maintype="text", subtype="plain", cte="8bit")
- has_body = True
 
- # Add a HTML body from the RTF_COMPRESSED field.
- if "RTF_COMPRESSED" in props:
- # Decompress the value to Rich Text Format.
+ # Plain-text is not availabe. Use the rich text version.
+ else:
+ doc.rtf_attachments += 1
+ fn = f"messagebody_{doc.rtf_attachments}.rtf"
 
+ msg.set_content(
+ f"<no plain text message body --- see attachment {fn}>",
+ cte="quoted-printable",
+ )
+
+ # Decompress the value to Rich Text Format.
  rtf = props["RTF_COMPRESSED"]
  rtf = compressed_rtf.decompress(rtf)
 
- # Try rtfparse to de-encapsulate HTML stored in a rich
- # text container.
- try:
- rtf_blob = io.BytesIO(rtf)
- parsed = Rtf_Parser(rtf_file=rtf_blob).parse_file()
- html_stream = io.StringIO()
- De_encapsulate_HTML().render(parsed, html_stream)
- html_body = html_stream.getvalue()
-
- if not has_body:
- # Try to convert that to plain/text if possible.
- text_body = html2text.html2text(html_body)
- msg.set_content(text_body, subtype="text", cte="quoted-printable")
- has_body = True
-
- if not has_body:
- msg.set_content(html_body, subtype="html", cte="quoted-printable")
- has_body = True
- else:
- msg.add_alternative(html_body, subtype="html", cte="quoted-printable")
-
- # If that fails, just attach the RTF file to the message.
- except Exception:
- doc.rtf_attachments += 1
- fn = f"messagebody_{doc.rtf_attachments}.rtf"
-
- if not has_body:
- msg.set_content(
- f"<no plain text message body --- see attachment {fn}>",
- cte="quoted-printable",
- )
- has_body = True
-
- # Add RTF file as an attachment.
- msg.add_attachment(rtf, maintype="text", subtype="rtf", filename=fn)
-
- if not has_body:
- msg.set_content("<no message body>", cte="quoted-printable")
+ # Add RTF file as an attachment.
+ msg.add_attachment(rtf, maintype="text", subtype="rtf", filename=fn)
 
  # # Copy over string values of remaining properties as headers
  # # so we don't lose any information.
@@ -177,11 +145,7 @@ def load_message_stream( # noqa: C901
  # Add attachments.
  for stream in entry:
  if stream.name.startswith("__attach_version1.0_#"):
- try:
- process_attachment(msg, stream, doc)
- except KeyError as e:
- logger.error(f"Error processing attachment {e!s} not found")
- continue
+ process_attachment(msg, stream, doc)
 
  return msg
 
@@ -247,6 +211,7 @@ def parse_properties( # noqa: C901
  # Read the entry.
  property_type = stream[i + 0 : i + 2]
  property_tag = stream[i + 2 : i + 4]
+ # flags = stream[i+4:i+8]
  value = stream[i + 8 : i + 16]
  i += 16
 
@@ -265,6 +230,8 @@ def parse_properties( # noqa: C901
 
  # Variable Length Properties.
  elif isinstance(tag_type, VariableLengthValueLoader):
+ # value_length = stream[i + 8 : i + 12] # not used
+
  # Look up the stream in the document that holds the value.
  streamname = "__substg1.0_{0:0{1}X}{2:0{3}X}".format(
  property_tag, 4, property_type, 4
@@ -351,7 +318,9 @@ def parse_properties( # noqa: C901
 
 
 class FixedLengthValueLoader:
- pass
+ @staticmethod
+ def load(value):
+ raise NotImplementedError()
 
 
 class NULL(FixedLengthValueLoader):
@@ -399,20 +368,17 @@ def load(value):
  value = reduce(
  lambda a, b: (a << 8) + b, reversed(value)
  ) # bytestring to integer
- try:
- value = datetime(1601, 1, 1) + timedelta(seconds=value / 10000000)
- except OverflowError:
- value = None
-
- return value
+ return datetime(1601, 1, 1) + timedelta(seconds=value / 10000000)
 
 
 # TODO: The other fixed-length data types:
 # "FLOAT", "DOUBLE", "CURRENCY", "APPTIME", "ERROR"
 
 
 class VariableLengthValueLoader:
- pass
+ @staticmethod
+ def load(value):
+ raise NotImplementedError()
 
 
 class BINARY(VariableLengthValueLoader):
@@ -934,10 +900,9 @@ def load(entry, doc, **kwargs):
  0x3F06: ("YPOS", "I4"),
  0x3F07: ("CONTROL_ID", "BINARY"),
  0x3F08: ("INITIAL_DETAILS_PANE", "I4"),
- 0x3FDE: ("PR_INTERNET_CPID", "I4"),
- 0x3FFD: ("PR_MESSAGE_CODEPAGE", "I4"),
 }
 
+
 code_pages = {
  # Microsoft code page id: python codec name
  437: "cp437",