Make parse take a list of messages

Author: emersion

emailthreads/quotes.py

@@ -2,7 +2,7 @@
 import sys
 from email.message import EmailMessage
 
-from util import *
+from .util import *
 
 def trim_empty_lines(block):
 	start = 0

emailthreads/threads.py

@@ -2,10 +2,12 @@
 import sys
 from email.message import EmailMessage
 
-from util import *
-from quotes import *
+from .util import *
+from .quotes import *
 
 def get_message_by_id(msgs, msg_id):
+	if msg_id is None or msg_id == "":
+		return None
 	# TODO: handle weird brackets stuff
 	for msg in msgs:
 		if msg["message-id"] == msg_id:
@@ -119,15 +121,27 @@ def __repr__(self):
 
 		return "\n".join(repr_lines)
 
-def parse(msg, refs=[]):
-	# For some reason Python strips "Re:" prefixes
-	subject = flatten_header_field(msg["subject"])
+def build_message_tree(messages):
+	heads = []
+	replies = []
+
+	for msg in messages:
+		in_reply_to = get_message_by_id(messages, msg['in-reply-to'])
+		if in_reply_to is None:
+			heads.append(msg)
+		else:
+			replies.append((msg, in_reply_to))
+
+	if len(heads) != 1:
+		raise Exception("expected exactly one head message, got " + str(len(heads)))
+	head = heads[0]
 
-	in_reply_to = get_message_by_id(refs, msg['in-reply-to'])
-	if in_reply_to is None or flatten_header_field(in_reply_to["subject"]) != subject:
-		text = get_text(msg)
-		text_lines = text.splitlines()
-		return Thread(text_lines, msg, (0, len(text_lines)))
+	replies = sorted(replies, key=lambda reply: reply[0]['date'])
+
+	return (head, replies)
+
+def parse_reply(msg, in_reply_to, thread):
+	subject = flatten_header_field(msg["subject"])
 
 	blocks = parse_blocks(msg)
 	blocks = trim_quotes_footer(blocks)
@@ -136,8 +150,6 @@ def parse(msg, refs=[]):
 	# print("\n".join([str(block) for block in blocks]))
 	blocks = merge_blocks(blocks)
 
-	thread = parse(in_reply_to, refs)
-
 	last_quote = None
 	for block in blocks:
 		if isinstance(block, Text):
@@ -162,3 +174,15 @@ def parse(msg, refs=[]):
 			last_quote = block
 
 	return thread
+
+def parse(messages):
+	(head, replies) = build_message_tree(messages)
+
+	text = get_text(head)
+	text_lines = text.splitlines()
+	thread = Thread(text_lines, head, (0, len(text_lines)))
+
+	for (msg, in_reply_to) in replies:
+		parse_reply(msg, in_reply_to, thread)
+
+	return thread

test/test_parse.py

@@ -31,7 +31,7 @@ def test_with_scissor(self):
 		patch = self._load_msg_from_file("scissor/patch.eml")
 		reply = self._load_msg_from_file("scissor/reply.eml")
 
-		thread = emailthreads.parse(reply, [patch])
+		thread = emailthreads.parse([patch, reply])
 
 		got = self._normalize(str(thread))
 		want = self._read_file("scissor/output.txt")
@@ -44,7 +44,7 @@ def test_with_multiple_replies(self):
 		reply2 = self._load_msg_from_file("multiple-replies/reply2.eml")
 		reply3 = self._load_msg_from_file("multiple-replies/reply3.eml")
 
-		thread = emailthreads.parse(reply3, [patch, reply1, reply2])
+		thread = emailthreads.parse([patch, reply1, reply2, reply3])
 
 		got = self._normalize(str(thread))
 		want = self._read_file("multiple-replies/output3.txt")