Codespace silver spork 69j5r7wjg5jqf4qrq

.vscode/settings.json

@@ -0,0 +1,11 @@
+{
+    "python.testing.unittestArgs": [
+        "-v",
+        "-s",
+        ".",
+        "-p",
+        "*test.py"
+    ],
+    "python.testing.pytestEnabled": false,
+    "python.testing.unittestEnabled": true
+}

package/bin/android-sms-extractor

@@ -36,6 +36,14 @@ parser.add_argument('--datetime-format', action='store', dest='dt_format',
 	help="Set datetime (strftime) format for text output (default: %(default)s)",
 	default="%Y-%m-%d %H:%M:%S")
 
+parser.add_argument('--destination', action='store', dest='destination',
+	help="Extract only conversations with this normalized destination"
+	)
+
+parser.add_argument('--show-uris', action='store_true', dest='show_uris',
+	help="Append attachment URI to placeholder text when available"
+	)
+
 #===============================================================================
 # Parse arguments
 #===============================================================================
@@ -58,9 +66,18 @@ c = db_conn.cursor()
 #===============================================================================
 # Query and loop through all SMS conversations
 #===============================================================================
-c.execute("SELECT conversations._id, conversations.name, \
+query = "SELECT conversations._id, conversations.name, \
 	conversations.participant_normalized_destination \
-	FROM conversations WHERE latest_message_id NOT NULL ORDER BY name")
+	FROM conversations WHERE latest_message_id NOT NULL"
+params = []
+
+if args.destination:
+	query += " AND conversations.participant_normalized_destination = ?"
+	params.append(args.destination)
+
+query += " ORDER BY name"
+
+c.execute(query, params)
 
 for conversation in c.fetchall():
 	data_id = conversation[0]
@@ -71,29 +88,81 @@ for conversation in c.fetchall():
 	print("{} [{}]".format(data_name, data_dest))
 	print("==================================================")
 
+	# Detect available columns in the parts table for improved attachment handling
+	c.execute("PRAGMA table_info(parts)")
+	part_columns = [row[1] for row in c.fetchall()]
+	optional_part_columns = [
+		"content_type",
+		"content_uri",
+		"display_name",
+		"name",
+		"filename",
+	]
+	available_columns = ["parts.text", "parts.timestamp"] + [
+		"parts." + name for name in optional_part_columns if name in part_columns
+	]
+	select_columns = ", ".join(available_columns)
+
 	# Query and loop through all messages in this conversation
-	c.execute("SELECT messages._id, parts.text, parts.timestamp, \
-		participants.sim_slot_id FROM messages \
-		LEFT JOIN parts ON messages._id = parts.message_id \
-		LEFT JOIN participants ON messages.sender_id = participants._id \
-		WHERE messages.conversation_id = ? ORDER BY parts.timestamp", (data_id,))
+	c.execute("SELECT messages._id, {} , participants.sim_slot_id FROM messages "
+		"LEFT JOIN parts ON messages._id = parts.message_id " \
+		"LEFT JOIN participants ON messages.sender_id = participants._id "
+		"WHERE messages.conversation_id = ? ORDER BY parts.timestamp".format(select_columns), (data_id,))
 
 	for result in c.fetchall():
 		msg_id = result[0]
 		msg_text = result[1]
 		msg_time = result[2]
 
-		# Remove the last 3 digits (milliseconds) from timestamp
-		msg_time = int(str(msg_time)[:-3])
-		msg_time_formatted = datetime.fromtimestamp(msg_time).strftime(args.dt_format)
-
-		sim_slot_id = result[3]
-
-		# A sim_slot_id equal to 0 indicates that the SMS was sent from the device
-		if sim_slot_id == 0:
-			msg_direction = "===>"
+	attachment_type = None
+	attachment_uri = None
+	attachment_name = None
+	attachment_filename = None
+
+	if len(result) > 3:
+		attachment_type = result[3]
+	if len(result) > 4:
+		attachment_uri = result[4]
+	if len(result) > 5:
+		attachment_name = result[5]
+	if len(result) > 6:
+		attachment_filename = result[6]
+
+	# Treat all-black block placeholders as absent text for media parts
+	if msg_text is not None:
+		trimmed = msg_text.strip()
+		if trimmed and all(ch in {"\u2588", "\u25A0", "\u2592", "\u2593"} for ch in trimmed):
+			msg_text = None
+
+	# Remove the last 3 digits (milliseconds) from timestamp
+	msg_time = int(str(msg_time)[:-3])
+	msg_time_formatted = datetime.fromtimestamp(msg_time).strftime(args.dt_format)
+
+	if sim_slot_id == 0:
+		msg_direction = "===>"
+	else:
+		msg_direction = "<---"
+
+	if msg_text is None or msg_text == "":
+		if attachment_type:
+			if attachment_type.startswith("image/"):
+				msg_text = "[Image attachment]"
+			elif attachment_type.startswith("audio/"):
+				msg_text = "[Audio attachment]"
+			elif attachment_type.startswith("video/"):
+				msg_text = "[Video attachment]"
+			else:
+				msg_text = "[{} attachment]".format(attachment_type)
+			if attachment_name:
+				msg_text += " {}".format(attachment_name)
+			elif attachment_filename:
+				msg_text += " {}".format(attachment_filename)
+			if args.show_uris and attachment_uri:
+				msg_text += " {}".format(attachment_uri)
+			elif not attachment_name and not attachment_filename and attachment_uri:
+				msg_text += " {}".format(attachment_uri)
 		else:
-			msg_direction = "<---"
+			msg_text = ""
 
-		print("[{} {}]:\n{}\n".format(
-			msg_time_formatted, msg_direction, msg_text))
+	print("[{} {}]:\n{}\n".format(
+		msg_time_formatted, msg_direction, msg_text))

readme.md

@@ -11,4 +11,7 @@ With this Python script you can extract all SMS messages from the SQLite databas
 5. Pass the path to the extracted database file to this script:  
    `$ android-sms-extractor bugle_db`
 
+You can also append attachment information for media parts:
+- `$ android-sms-extractor --show-uris bugle_db`
+
 **IMPORTANT:** This script has been written in 2019 when I needed a quick solution to export all SMS from my Android device (*OnePlus X*) running *LineageOS 14.1*. This script was later in 2022 also successfully tested with the SMS database of `com.android.messaging` in *LineageOS 17.1*. If it also works for you: Good! If not: I'm sorry!