Improved formatting

libpurple_to_markdown/__init__.py

@@ -45,7 +45,7 @@ def parse_messages_in_chat_file(path: Path) -> list[Message]:
     messages = []
 
     cur_sent_at: datetime.datetime | None = None
-    cur_sender: str = 'NOT DEFINED'
+    cur_sender: str | None = None
     cur_text: str = ''
 
     if soup.p:
@@ -61,18 +61,19 @@ def parse_messages_in_chat_file(path: Path) -> list[Message]:
 
             # Get sender
             if c.b:
+                assert cur_sender is None
                 cur_sender = c.b.get_text().strip().removesuffix(':')
 
         elif c.name in {None,'span','font'}:
             cur_text += c.get_text()
 
         elif c.name == 'a':
-            cur_text += cur_text + '<' + c['href'] + '>'
+            cur_text += '<' + c['href'] + '>'
 
         elif c.name == 'br':
             messages.append(Message(cur_sent_at, cur_sender, cur_text.strip()))
             cur_sent_at = None
-            cur_sender = 'NOT DEFINED'
+            cur_sender = None
             cur_text = ''
 
         elif c.name == 'b':
@@ -100,14 +101,42 @@ def parse_messages_in_chat_folder(chat_folder_path: Path) -> list[Message]:
     return messages
 
 
+def format_message_as_citation(out, msg):
+    out.append(f'{msg.sent_at.date()} {msg.sent_at.time()} [[{msg.sender}]]:')
+    out.append('\n')
+    for line in msg.text.split('\n'):
+        out.append(f'> {line}\n')
+        del line
+    out.append('\n')
+
+def format_message_as_table(out, msg):
+    out.append(f'| {msg.sent_at} | [[{msg.sender}]] | ')
+    for line in msg.text.split('\n'):
+        out.append(f'{line}')
+        del line
+    out.append('|\n')
+
 def format_messages(messages: list[Message]) -> str:
     out = [
             '# Chat 2018'
             '\n\n'
     ]
 
-    for msg in messages:
+    as_table = False
-        out.append(f'[[{msg.sent_at.date()}]] {msg.sent_at.time()} [[{msg.sender}]]:\n> {msg.text}\n')
+
+
+    for msg_idx, msg in enumerate(messages):
+        if msg_idx == 0 or messages[msg_idx-1].sent_at.date() != msg.sent_at.date():
+            out.append('---\n')
+            out.append(f'## [[{msg.sent_at.date()}]]\n\n')
+            if as_table:
+                out.append('| sent at | sender | text |\n')
+                out.append('| ------- | ------ | ---- |\n')
+
+        if as_table:
+            format_message_as_table(out, msg)
+        else:
+            format_message_as_citation(out, msg)
         del msg
 
     return ''.join(out)
@@ -117,11 +146,16 @@ MSG_ADJACENTCY_DIST = datetime.timedelta(minutes=2)
 def is_adjacent_messages(first, second):
     return first.sender == second.sender and second.sent_at - first.sent_at <= MSG_ADJACENTCY_DIST
 
+def merge_texts(text1: str, text2: str) -> str:
+    punctuated = text1.endswith('.?!,:')
+    #return text1 + (' ' if punctuated else '. ') + text2
+    return text1 + (' ' if punctuated else '. ') + '\n' + text2
+
 def merge_adjacent_messages(messages: list[Message]) -> list[Message]:
     out = []
     for msg in messages:
         if out and is_adjacent_messages(out[-1], msg):
-            out[-1] = dataclasses.replace(out[-1], text=out[-1].text + '\n\n' + msg.text)
+            out[-1] = dataclasses.replace(out[-1], text=merge_texts(out[-1].text, msg.text))
         else:
             out.append(msg)
     return out