Add support for telegram JSON file format

Add the --telegram command line argument that, combined with a JSON
output, generates a Telegram compatible JSON file [1].

The JSON is per-chat, so the --telegram argument implies the
--json-per-chat setting.

I took a few shortcuts:
* Contact and Ids are inferred from the chat id or phone numbers
* All text is marked as plain (e.g. no markup or different types)
* Only personal chats and private groups supported
* Private groups are defined if the chat has a name
* Various ids try to match the ones in WA but may require bulk edits

[1] - https://core.telegram.org/import-export

Fixes: https://github.com/KnugiHK/WhatsApp-Chat-Exporter/issues/152

Whatsapp_Chat_Exporter/__main__.py

@@ -15,6 +15,7 @@ from Whatsapp_Chat_Exporter.data_model import ChatCollection, ChatStore
 from Whatsapp_Chat_Exporter.utility import APPLE_TIME, CLEAR_LINE, Crypt, check_update
 from Whatsapp_Chat_Exporter.utility import readable_to_bytes, safe_name, bytes_to_readable
 from Whatsapp_Chat_Exporter.utility import import_from_json, incremental_merge, DbType
+from Whatsapp_Chat_Exporter.utility import telegram_json_format
 from argparse import ArgumentParser, SUPPRESS
 from datetime import datetime
 from getpass import getpass
@@ -148,6 +149,10 @@ def setup_argument_parser() -> ArgumentParser:
         '--pretty-print-json', dest='pretty_print_json', default=None, nargs='?', const=2, type=int,
         help="Pretty print the output JSON."
     )
+    json_group.add_argument(
+        "--telegram", dest="telegram", default=False, action='store_true',
+        help="Output the JSON in a format compatible with Telegram export (implies json-per-chat)"
+    )
     json_group.add_argument(
         "--per-chat", dest="json_per_chat", default=False, action='store_true',
         help="Output the JSON file per chat"
@@ -648,7 +653,7 @@ def export_json(args, data: ChatCollection, contact_store=None) -> None:
         data = {jik: chat.to_json() for jik, chat in data.items()}
 
     # Export as a single file or per chat
-    if not args.json_per_chat:
+    if not args.json_per_chat and not args.telegram:
         export_single_json(args, data)
     else:
         export_multiple_json(args, data)
@@ -684,9 +689,13 @@ def export_multiple_json(args, data: Dict) -> None:
         else:
             contact = jik.replace('+', '')
 
+        if args.telegram:
+            obj = telegram_json_format(jik, data[jik])
+        else:
+            obj = {jik: data[jik]}
         with open(f"{json_path}/{safe_name(contact)}.json", "w") as f:
             file_content = json.dumps(
-                {jik: data[jik]},
+                obj,
                 ensure_ascii=not args.avoid_encoding_json,
                 indent=args.pretty_print_json
             )

Whatsapp_Chat_Exporter/utility.py

@@ -627,6 +627,72 @@ def safe_name(text: Union[str, bytes]) -> str:
     return "-".join(''.join(safe_chars).split())
 
 
+def get_from_string(msg: Dict, chat_id: str) -> str:
+    """Return the number or name for the sender"""
+    if msg["from_me"]:
+        return "Me"
+    if msg["sender"]:
+        return str(msg["sender"])
+    return str(chat_id)
+
+
+def get_from_id(msg: Dict, chat_id: str) -> str:
+    """Return the user id for the sender"""
+    if msg["from_me"]:
+        return "user00000"
+    if msg["sender"]:
+        return "user" + msg["sender"]
+    return f"user{chat_id}"
+
+
+def get_reply_id(data: Dict, reply_key: str) -> Optional[str]:
+    """Get the id of the message corresponding to the reply"""
+    if not reply_key:
+        return None
+    for msg_id, msg in data["messages"].items():
+        if msg["key_id"] == reply_key:
+            return int(msg_id)
+    return None
+
+
+def telegram_json_format(jik: str, data: Dict) -> Dict:
+    """Convert the data to the Telegram export format"""
+    try:
+        chat_id = int(''.join([c for c in jik if c.isdigit()]))
+    except ValueError:
+        # not a real chat: e.g. statusbroadcast
+        chat_id = 0
+    obj = {
+            "name": data["name"] if data["name"] else jik,
+            # TODO can we do better than this?
+            "type": "private_group" if data["name"] else "personal_chat",
+            "id": chat_id,
+            "messages": [ {
+                "id": int(msgId),
+                "type": "message",
+                "date": datetime.fromtimestamp(msg["timestamp"]).isoformat().split(".")[0],
+                "date_unixtime": int(msg["timestamp"]),
+                "from": get_from_string(msg, chat_id),
+                "from_id": get_from_id(msg, chat_id),
+                "reply_to_message_id": get_reply_id(data, msg["reply"]),
+                "text": msg["data"],
+                "text_entities": [
+                    {
+                        # TODO this will lose formatting and different types
+                        "type": "plain",
+                        "text": msg["data"],
+                        }
+                    ],
+                } for msgId, msg in data["messages"].items()]
+            }
+    # remove empty messages and replies
+    for msg_id, msg in enumerate(obj["messages"]):
+        if not msg["reply_to_message_id"]:
+            del obj["messages"][msg_id]["reply_to_message_id"]
+    obj["messages"] = [m for m in obj["messages"] if m["text"]]
+    return obj
+
+
 class WhatsAppIdentifier(StrEnum):
     # AppDomainGroup-group.net.whatsapp.WhatsApp.shared-ChatStorage.sqlite
     MESSAGE = "7c7fba66680ef796b916b067077cc246adacf01d"