Add .eml support with nested attachments parsing

[gautamp8]

No description provided.

[satwikkansal]

this looks suspicious, gotta check what it is doing in bigger context

[Copilot]

Pull Request Overview

Add support for processing .eml email files—including header/body extraction and nested attachment handling—into the existing attachment pipeline.

• Introduce new methods for parsing .eml files, extracting/saving attachments, and converting HTML bodies to text
• Integrate .eml support in _process_document and extend forward to queue nested attachments
• Update summary generation to report email files and nested attachments

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
mxtoai/tools/attachment_processing_tool.py	Added .eml parsing methods, integrated .eml handling into document processing and forwarding, and enhanced summary reporting for nested attachments
mxtoai/agents/email_agent.py	Updated imports to include Tool alongside ToolCallingAgent

Comments suppressed due to low confidence (2)

mxtoai/tools/attachment_processing_tool.py:639

• The images counter in _create_attachment_summary is never incremented when handling image attachments, so the final status report will never include images. Consider adding images += 1 in the block where content.get("type") == "image".

summary_parts.append(f"Image {att['filename']}: Requires visual processing")

mxtoai/tools/attachment_processing_tool.py:132

• The new .eml parsing functionality (_process_eml_file, attachment extraction, body extraction) lacks unit tests. Consider adding tests to verify email parsing, nested attachment saving, and HTML-to-text conversion.

def _process_eml_file(self, file_path: Path, extract_attachments: bool = True) -> tuple[str, list[dict[str, Any]]]:

[Copilot]

[nitpick] Using a regex to strip HTML tags is brittle and may fail on complex markup. Consider using a dedicated HTML-to-text parser (e.g., html.parser or a third-party library) for more reliable results.

Suggested change

	def _html_to_text(self, html: str) -> str:
	"""
	Convert HTML to plain text using simple regex.
	Args:
	html: HTML content.
	Returns:
	str: Plain text version of the HTML.
	"""
	# Remove HTML tags
	text = re.sub(r"<[^>]+>", " ", html)
	# Normalize whitespace
	return re.sub(r"\s+", " ", text).strip()
	class _HTMLToTextParser(HTMLParser):
	"""
	Custom HTML parser to extract plain text from HTML content.
	"""
	def __init__(self):
	super().__init__()
	self.text_parts = []
	def handle_data(self, data):
	self.text_parts.append(data)
	def get_text(self) -> str:
	return "".join(self.text_parts)
	def _html_to_text(self, html: str) -> str:
	"""
	Convert HTML to plain text using a dedicated HTML parser.
	Args:
	html: HTML content.
	Returns:
	str: Plain text version of the HTML.
	"""
	parser = self._HTMLToTextParser()
	parser.feed(html)
	parser.close()
	# Normalize whitespace
	return " ".join(parser.get_text().split())

[Copilot]

The Tool import is not used in this file and can be removed to clean up unused imports.

Suggested change

	from smolagents import Tool, ToolCallingAgent
	from smolagents import ToolCallingAgent

[satwikkansal]

this looks weird, what is / doing here

[satwikkansal]

did we verify that the cleanup is happening for these, post processing of the task

[satwikkansal]

are we passing attachments through llm and generating summaries? How will this work for excel files and others?