From 99513481649665af16f389c2995bf77653a4c35b Mon Sep 17 00:00:00 2001
From: Jon Michael Aanes <jonjmaa@gmail.com>
Date: Thu, 31 Oct 2024 22:37:50 +0100
Subject: [PATCH] Automatically select period key

---
 libpurple_to_markdown/__main__.py | 61 +++++++++++++++++++++++--------
 1 file changed, 46 insertions(+), 15 deletions(-)

diff --git a/libpurple_to_markdown/__main__.py b/libpurple_to_markdown/__main__.py
index d8c3829..3c7f9dc 100644
--- a/libpurple_to_markdown/__main__.py
+++ b/libpurple_to_markdown/__main__.py
@@ -1,7 +1,8 @@
 import argparse
 import logging
 from pathlib import Path
-from collections.abc import Iterable
+from collections.abc import Iterable, Iterator
+import dataclasses
 
 from . import (
     filter_useless_messages,
@@ -10,32 +11,60 @@ from . import (
     merge_adjacent_messages,
     synctech_sms,
 )
-from .data import Message
+from .data import Message, MYSELF
 
 logger = logging.getLogger(__name__)
 
+def group_messages(messages: Iterable[Message], key) -> dict[str, list[Message]]:
+    by_key: dict[str, list[Message]] = {}
+    for msg in messages:
+        by_key.setdefault(key(msg), []).append(msg)
+        del msg
+    return by_key
+
 
 def group_messages_by_chat_id(messages: Iterable[Message]) -> dict[str, list[Message]]:
-    by_period: dict[str, list[Message]] = {}
-    for msg in messages:
-        by_period.setdefault(msg.chat_id, []).append(msg)
-        del msg
-    return by_period
+    return group_messages(messages, key=lambda msg: msg.chat_id)
+
+def year_and_month_period_key(msg: Message):
+    return f'{msg.sent_at.year}-{msg.sent_at.month:02}'
+
+def year_period_key(msg: Message):
+    return f'{msg.sent_at.year}'
+
+def year_quarter_period_key(msg: Message):
+    quarter = int((msg.sent_at.month-1)/3)+1
+    return f'{msg.sent_at.year}-Q{quarter:01}'
+
+MAX_AVERAGE_MESSAGES_PER_PERIOD = 120
 
 def group_messages_by_period(messages: Iterable[Message]) -> dict[str, list[Message]]:
-    by_period: dict[str, list[Message]] = {}
-    for msg in messages:
-        period_key = f'{msg.sent_at.year}-{msg.sent_at.month:02}'
-        by_period.setdefault(period_key, []).append(msg)
-        del msg
-    return by_period
 
+    possible_period_keys = [(lambda msg: 'Full History'), year_period_key, year_quarter_period_key, year_and_month_period_key]
+
+    for period_key in possible_period_keys:
+        grouped = group_messages(messages, key=period_key)
+        average_num_messages = sum(len(grouped[k]) for k in grouped) / len(grouped)
+        if average_num_messages <= MAX_AVERAGE_MESSAGES_PER_PERIOD:
+            break
+
+        del period_key, average_num_messages
+
+    return grouped
+
+def replace_myself(messages: Iterable[Message], myself: str) -> Iterator[Message]:
+    for msg in messages:
+        if msg.sender == MYSELF:
+            yield dataclasses.replace(msg, sender = myself)
+        else:
+            yield msg
 
 def parse_args():
     parser = argparse.ArgumentParser()
     parser.add_argument('--purple', type=Path, dest='purple_folder')
     parser.add_argument('--synctech', type=Path, dest='synctech_sms_backup_file')
     parser.add_argument('--output', type=Path)
+    parser.add_argument('--myself', type=str, default='Myself')
     return parser.parse_args()
 
 
@@ -52,6 +81,7 @@ def main():
         logger.fatal('No input file given!')
         return
 
+    all_messages = replace_myself(all_messages, myself=args.myself)
     all_messages = list(all_messages)
     logger.info('%d messages after loading', len(all_messages))
 
@@ -64,11 +94,12 @@ def main():
 
     for chat_id, messages_in_chat_original in messages_by_chat_id.items():
         messages_in_chat = merge_adjacent_messages(messages_in_chat_original )
-
         messages_by_period = group_messages_by_period(messages_in_chat)
+        logger.info('  "%s": %d messages, %d periods (%d msg/period avg)', chat_id, len(messages_in_chat_original), len(messages_by_period), len(messages_in_chat_original)/ len(messages_by_period))
 
         for period_key, messages in messages_by_period.items():
-            output_file = args.output / f'{chat_id} - {period_key}.md'
+            output_file = args.output / chat_id / f'{period_key}.md'
+            output_file.parent.mkdir(exist_ok=True)
             logger.info('Writing % 5d messages to %s', len(messages), output_file)
             with open(output_file, 'w') as f:
                 f.write(