iMessage loader: implement message content extraction from attributed… (#13634)

- **Description:** We are adding functionality to extract message
content from the `attributedBody` field of the database, in case the
content is not in the `text` field.
  - **Issue:** Closes #13326 and #10680 
  - **Dependencies:** None.
  - **Tag maintainer:** @eyurtsev, @hwchase17

---------

Co-authored-by: onotate <johnp.pham@mail.utoronto.ca>

libs/langchain/langchain/chat_loaders/imessage.py

@@ -46,6 +46,36 @@ class IMessageChatLoader(BaseChatLoader):
                 "Please install it with `pip install pysqlite3`"
             ) from e
 
+    def _parse_attributedBody(self, attributedBody: bytes) -> str:
+        """
+        Parse the attributedBody field of the message table
+        for the text content of the message.
+
+        The attributedBody field is a binary blob that contains
+        the message content after the byte string b"NSString":
+
+                              5 bytes      1-3 bytes    `len` bytes
+        ... | b"NSString" |   preamble   |   `len`   |    contents    | ...
+
+        The 5 preamble bytes are always b"\x01\x94\x84\x01+"
+
+        The size of `len` is either 1 byte or 3 bytes:
+        - If the first byte in `len` is b"\x81" then `len` is 3 bytes long.
+          So the message length is the 2 bytes after, in little Endian.
+        - Otherwise, the size of `len` is 1 byte, and the message length is
+          that byte.
+
+        Args:
+            attributedBody (bytes): attributedBody field of the message table.
+        Return:
+            str: Text content of the message.
+        """
+        content = attributedBody.split(b"NSString")[1][5:]
+        length, start = content[0], 1
+        if content[0] == 129:
+            length, start = int.from_bytes(content[1:3], "little"), 3
+        return content[start : start + length].decode("utf-8", errors="ignore")
+
     def _load_single_chat_session(
         self, cursor: "sqlite3.Cursor", chat_id: int
     ) -> ChatSession:
@@ -62,7 +92,7 @@ class IMessageChatLoader(BaseChatLoader):
         results: List[HumanMessage] = []
 
         query = """
-        SELECT message.date, handle.id, message.text
+        SELECT message.date, handle.id, message.text, message.attributedBody
         FROM message
         JOIN chat_message_join ON message.ROWID = chat_message_join.message_id
         JOIN handle ON message.handle_id = handle.ROWID
@@ -72,18 +102,24 @@ class IMessageChatLoader(BaseChatLoader):
         cursor.execute(query, (chat_id,))
         messages = cursor.fetchall()
 
-        for date, sender, text in messages:
-            if text:  # Skip empty messages
-                results.append(
-                    HumanMessage(
-                        role=sender,
-                        content=text,
-                        additional_kwargs={
-                            "message_time": date,
-                            "sender": sender,
-                        },
-                    )
+        for date, sender, text, attributedBody in messages:
+            if text:
+                content = text
+            elif attributedBody:
+                content = self._parse_attributedBody(attributedBody)
+            else:  # Skip messages with no content
+                continue
+
+            results.append(
+                HumanMessage(
+                    role=sender,
+                    content=content,
+                    additional_kwargs={
+                        "message_time": date,
+                        "sender": sender,
+                    },
                 )
+            )
 
         return ChatSession(messages=results)
 

libs/langchain/tests/unit_tests/chat_loaders/test_imessage.py

@@ -0,0 +1,28 @@
+import pathlib
+
+from langchain.chat_loaders import imessage, utils
+
+
+def test_imessage_chat_loader() -> None:
+    chat_path = pathlib.Path(__file__).parent / "data" / "imessage_chat.db"
+    loader = imessage.IMessageChatLoader(str(chat_path))
+
+    chat_sessions = list(
+        utils.map_ai_messages(loader.lazy_load(), sender="testemail@gmail.com")
+    )
+    assert chat_sessions, "Chat sessions should not be empty"
+
+    assert chat_sessions[0]["messages"], "Chat messages should not be empty"
+
+    # message content in text field
+    assert "Yeh" in chat_sessions[0]["messages"][0].content, "Chat content mismatch"
+
+    # short message content in attributedBody field
+    assert (
+        "John is the almighty" in chat_sessions[0]["messages"][16].content
+    ), "Chat content mismatch"
+
+    # long message content in attributedBody field
+    long_msg = "aaaaabbbbbaaaaabbbbbaaaaabbbbbaaaaabbbbbaaaaabbbbbaaaaabbbbbaaaaabbbbba"
+    "aaaabbbbbaaaaabbbbbaaaaabbbbbaaaaabbbbbaaaaabbbbbaaaaabbb"
+    assert long_msg in chat_sessions[0]["messages"][18].content, "Chat content mismatch"