From 21c61f7d027601ad4f1efde6717d88d858692d2c Mon Sep 17 00:00:00 2001
From: Jon Michael Aanes <jonjmaa@gmail.com>
Date: Sat, 26 Oct 2024 00:34:53 +0200
Subject: [PATCH] Initial commit for crazy log conversion project

---
 .gitignore                        | 20 +++++++
 libpurple_to_markdown/__init__.py |  0
 libpurple_to_markdown/__main__.py | 90 +++++++++++++++++++++++++++++++
 3 files changed, 110 insertions(+)
 create mode 100644 .gitignore
 create mode 100644 libpurple_to_markdown/__init__.py
 create mode 100644 libpurple_to_markdown/__main__.py

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..41180b8
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,20 @@
+# Program specific
+/output/
+/deps/
+/secrets
+/private_deps/
+/data/
+/config/
+
+# Python
+__pycache__/
+/build/
+/dist/
+*.egg-info/
+.mypy_cache/
+
+# Python, Testing
+/test/secrets.py
+/.coverage
+/.hypothesis/
+/htmlcov/
diff --git a/libpurple_to_markdown/__init__.py b/libpurple_to_markdown/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/libpurple_to_markdown/__main__.py b/libpurple_to_markdown/__main__.py
new file mode 100644
index 0000000..e5549c8
--- /dev/null
+++ b/libpurple_to_markdown/__main__.py
@@ -0,0 +1,90 @@
+
+import dataclasses
+import re
+import datetime
+from pathlib import Path
+import bs4
+
+@dataclasses.dataclass(frozen=True, order=True)
+class Message:
+    sent_at: datetime.datetime
+    sender: str
+    text: str
+
+    def __post_init__(self):
+        assert self.sent_at is not None
+        assert self.sender is not None
+        assert self.text is not None
+
+def datetime_sent(chat_start: datetime.datetime, message_sent: datetime.time) -> datetime.datetime:
+    naive = datetime.datetime.combine(chat_start.date(), message_sent, chat_start.tzinfo)
+    if chat_start.time() > message_sent:
+        naive = naive - datetime.timedelta(days=1)
+    return naive
+
+
+def parse_messages_in_chat_file(path: Path) -> list[Message]:
+    chat_start = datetime.datetime.fromisoformat(path.stem.removesuffix('CEST'))
+
+    with open(path) as f:
+        soup = bs4.BeautifulSoup(f)
+
+    messages = []
+
+    cur_sent_at: datetime.datetime | None = None
+    cur_sender: str | None = None
+    cur_text: str  | None = None
+
+    for c in soup.body.children:
+        if c.name == 'font':
+            # 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)
+
+            # Get sender
+            cur_sender = c.b.get_text().strip().removesuffix(':')
+
+        elif c.name is None:
+            cur_text = c.get_text()
+
+        elif c.name == 'a':
+            cur_text = '<' + c['href'] + '>'
+
+        elif c.name == 'br':
+            messages.append(Message(cur_sent_at, cur_sender, cur_text))
+            cur_sent_at = None
+            cur_sender = None
+            cur_text = None
+
+        elif c.name == 'h3':
+            pass # Ignore log header
+
+
+        else:
+            assert False, c
+
+    return messages
+
+
+
+
+def parse_messages_in_chat_folder(chat_folder_path: Path) -> list[Message]:
+    messages = []
+    for file_path in chat_folder_path.iterdir():
+        messages.extend(parse_messages_in_chat_file(file_path))
+
+    messages.sort()
+    return messages
+
+
+
+
+def main():
+    path = Path()
+
+    for message in parse_messages_in_chat_folder(path):
+        print(f'({message.sent_at}) {message.sender} : {message.text}')
+
+if __name__ == '__main__':
+    main()