Significantly improve message parsing performance

lib/xenu_nntp/message.py

@@ -4,6 +4,9 @@ import datetime
 import uuid
 
 import email.message
+import email.policy
+import email.parser
+import email.errors
 
 from email.utils  import parsedate_to_datetime
 from email.header import decode_header, Header
@@ -117,14 +120,10 @@ class MessageRange():
             obj.max = int(match[1])
             return obj
 
-class MessageState(enum.Enum):
-    EMPTY  = 0
-    HEADER = 1
-    BODY   = 2
-
 class Message(DatabaseTable):
     __slots__ = (
         '_cache',
+        '_parser',
         '_headers',
         '_headers_lc',
         '_body',
@@ -153,12 +152,12 @@ class Message(DatabaseTable):
         super().__init__()
 
         self._cache      = dict()
+        self._parser     = None
         self._headers    = None
         self._headers_lc = None
         self._body       = None
         self._key        = None
         self.id          = None
-        self.state       = MessageState.EMPTY
         self.line        = None
         self.content     = ''
 
@@ -304,49 +303,16 @@ class Message(DatabaseTable):
         elif value is not None:
             self._header_set('Subject', Header(value).encode())
 
-    def is_first_line(self):
-        return len(self.headers) == 1 and (self._body == '' or self._body is None)
-
     def readline(self, line: str):
-        if self.line is not None:
-            self.content += self.line
-
-        if self.state is MessageState.EMPTY:
-            self.state       = MessageState.HEADER
-            self._headers    = dict()
-            self._headers_lc = dict()
-
-        if self.state is MessageState.HEADER:
-            if line == '\n' or line == '\r\n':
-                self.state = MessageState.BODY
-            elif line[0] == ' ' or line[0] == '\t':
-                self._header_append(self._key, ' ' + line.strip())
+        if self._parser is None:
+            self.content = line
+            self._parser = email.parser.FeedParser()
         else:
-                match = self.RE_HEADER.match(line)
+            self.content += line
 
-                if match:
-                    self._key = match[1]
-
-                    self._header_append(self._key, match[2].rstrip())
-
-        elif self.state is MessageState.BODY:
-            if self._body is None:
-                self._body = ''
-            else:
-                self._body += self.line
-
-        self.line = line
-
-    def finish(self):
-        if self.line:
-            self.content += self.line
-            self._body   += self.line
-
-            self.line = None
-
-    def read(self, text: str):
-        obj = email.message_from_string(text)
+        self._parser.feed(line)
 
+    def read_email_obj(self, obj: email.message.Message):
         self._headers    = dict()
         self._headers_lc = dict()
 
@@ -354,7 +320,14 @@ class Message(DatabaseTable):
             self._header_set(key, obj.get(key))
 
         self._body = obj.get_payload()
-        self.line  = None
+
+    def finish(self):
+        if self._parser:
+            self.read_email_obj(self._parser.close())
+            self._parser = None
+
+    def read(self, text: str):
+        self.read_email_obj(email.message_from_string(text))
 
     def message_id_assign(self):
         sender = self.sender
@@ -401,12 +374,6 @@ class Message(DatabaseTable):
     @staticmethod
     def from_text(text: str):
         message = Message()
-
-        obj = email.message_from_string(text)
-
-        for key in obj.keys():
-            message._header_set(key, obj.get(key))
-
-        message._body = obj.get_payload()
+        message.read(text)
 
         return message