Working on more robust methods

libpurple_to_markdown/__init__.py

@@ -24,13 +24,23 @@ def datetime_sent(chat_start: datetime.datetime, message_sent: datetime.time) ->
         naive = naive - datetime.timedelta(days=1)
     return naive
 
+def parse_timestamp(c) -> datetime.time:
+    timestamp_obj = c
+    if c.font is not None:
+        c = c.font
+    m = re.match(r'\((\d+):(\d+):(\d+)\)', timestamp_obj.get_text())
+    return datetime.time(int(m.group(1)), int(m.group(2)), int(m.group(3)))
 
 def parse_messages_in_chat_file(path: Path) -> list[Message]:
     logger.info('Parsing %s', path)
-    chat_start = datetime.datetime.fromisoformat(path.stem.removesuffix('CEST'))
+    chat_start = datetime.datetime.fromisoformat(path.stem.removesuffix('CEST').removesuffix('CET'))
 
     with open(path) as f:
-        soup = bs4.BeautifulSoup(f)
+        soup = bs4.BeautifulSoup(f, 'lxml')
+
+    if len(soup.contents) == 0:
+        logger.warning('File is empty?')
+        return []
 
     messages = []
 
@@ -38,21 +48,26 @@ def parse_messages_in_chat_file(path: Path) -> list[Message]:
     cur_sender: str = 'NOT DEFINED'
     cur_text: str = ''
 
-    for c in soup.body.children:
-        if c.name == 'font':
+    if soup.p:
+        logger.warning('File indicates error message?')
+        return [] # TODO
+    else:
+        loglines = soup.body.children
+
+    for c in loglines:
+        if c.name in {'font','span'} and cur_sent_at is None:
             # Get timestamp
-            m = re.match(r'\((\d+):(\d+):(\d+)\)', c.font.get_text())
-            time_sent = datetime.time(int(m.group(1)), int(m.group(2)), int(m.group(3)))
-            cur_sent_at = datetime_sent(chat_start, time_sent)
+            cur_sent_at = datetime_sent(chat_start, parse_timestamp(c))
 
             # Get sender
-            cur_sender = c.b.get_text().strip().removesuffix(':')
+            if c.b:
+                cur_sender = c.b.get_text().strip().removesuffix(':')
 
-        elif c.name is None:
-            cur_text = c.get_text()
+        elif c.name in {None,'span','font'}:
+            cur_text += c.get_text()
 
         elif c.name == 'a':
-            cur_text = cur_text + '<' + c['href'] + '>'
+            cur_text += cur_text + '<' + c['href'] + '>'
 
         elif c.name == 'br':
             messages.append(Message(cur_sent_at, cur_sender, cur_text.strip()))
@@ -60,7 +75,11 @@ def parse_messages_in_chat_file(path: Path) -> list[Message]:
             cur_sender = 'NOT DEFINED'
             cur_text = ''
 
-        elif c.name == 'h3':
+        elif c.name == 'b':
+            # Indicates system message. Ignore
+            pass
+
+        elif c.name in {'h1', 'h3'}:
             pass # Ignore log header
 
 

libpurple_to_markdown/__main__.py

@@ -1,4 +1,5 @@
 import argparse
+import logging
 from pathlib import Path
 
 from . import (parse_messages_in_chat_folder, merge_adjacent_messages,
@@ -10,6 +11,8 @@ def parse_args():
     return parser.parse_args()
 
 def main():
+    logging.basicConfig()
+    logging.getLogger().setLevel('INFO')
     args = parse_args()
 
     messages = parse_messages_in_chat_folder(args.path)