From be9e790b12d747caa16ff2fc71dfb86d0bc08a62 Mon Sep 17 00:00:00 2001
From: KnugiHK <24708955+KnugiHK@users.noreply.github.com>
Date: Fri, 16 Jun 2023 01:25:51 +0800
Subject: [PATCH] Better handling of binary message #44

---
 Whatsapp_Chat_Exporter/data_model.py |  1 +
 Whatsapp_Chat_Exporter/extract.py    | 17 ++++++++++++-----
 Whatsapp_Chat_Exporter/whatsapp.html |  8 ++++++++
 3 files changed, 21 insertions(+), 5 deletions(-)

diff --git a/Whatsapp_Chat_Exporter/data_model.py b/Whatsapp_Chat_Exporter/data_model.py
index 36051e5..ac1f6f4 100644
--- a/Whatsapp_Chat_Exporter/data_model.py
+++ b/Whatsapp_Chat_Exporter/data_model.py
@@ -57,6 +57,7 @@ def __init__(self, from_me: Union[bool,int], timestamp: int, time: Union[int,flo
         self.meta = False
         self.data = None
         self.sender = None
+        self.safe = False
         # Extra
         self.reply = None
         self.quoted_data = None
diff --git a/Whatsapp_Chat_Exporter/extract.py b/Whatsapp_Chat_Exporter/extract.py
index 9415300..2e9dfea 100644
--- a/Whatsapp_Chat_Exporter/extract.py
+++ b/Whatsapp_Chat_Exporter/extract.py
@@ -11,7 +11,7 @@
 from pathlib import Path
 from mimetypes import MimeTypes
 from hashlib import sha256
-from base64 import b64decode
+from base64 import b64decode, b64encode
 from Whatsapp_Chat_Exporter.data_model import ChatStore, Message
 from Whatsapp_Chat_Exporter.utility import MAX_SIZE, ROW_SIZE, Device, rendering, sanitize_except, determine_day, Crypt
 from Whatsapp_Chat_Exporter.utility import brute_force_offset, CRYPT14_OFFSETS
@@ -252,8 +252,6 @@ def messages(db, data, media_folder):
         except sqlite3.OperationalError:
             continue
         else:
-            if content is not None and isinstance(content["data"], bytes):
-                continue
             break
     while content is not None:
         if content["key_remote_jid"] not in data:
@@ -266,6 +264,17 @@ def messages(db, data, media_folder):
             time=content["timestamp"],
             key_id=content["key_id"],
         )
+        if isinstance(content["data"], bytes):
+            message.data = ("The message is binary data and its base64 is "
+                '<a href="https://gchq.github.io/CyberChef/#recipe=From_Base64'
+                "('A-Za-z0-9%2B/%3D',true,false)Text_Encoding_Brute_Force"
+                f"""('Decode')&input={b64encode(b64encode(content["data"])).decode()}">""")
+            message.data += b64encode(content["data"]).decode("utf-8") + "</a>"
+            message.safe = message.meta = True
+            data[content["key_remote_jid"]].add_message(content["_id"], message)
+            i += 1
+            content = c.fetchone()
+            continue
         invalid = False
         if "-" in content["key_remote_jid"] and content["key_from_me"] == 0:
             name = None
@@ -407,8 +416,6 @@ def messages(db, data, media_folder):
             except sqlite3.OperationalError:
                 continue
             else:
-                if content is not None and isinstance(content["data"], bytes):
-                    continue
                 break
     print(f"Processing messages...({total_row_number}/{total_row_number})", end="\r")
 
diff --git a/Whatsapp_Chat_Exporter/whatsapp.html b/Whatsapp_Chat_Exporter/whatsapp.html
index 03bbfb9..c5425c9 100644
--- a/Whatsapp_Chat_Exporter/whatsapp.html
+++ b/Whatsapp_Chat_Exporter/whatsapp.html
@@ -125,7 +125,11 @@
 							{% endif %}
 							{% if msg.meta == true or msg.media == false and msg.data is none %}
 								<div class="w3-panel w3-border-blue w3-pale-blue w3-rightbar w3-leftbar w3-threequarter w3-center">
+									{% if msg.safe %}
+									<p>{{ msg.data | safe or 'Not supported WhatsApp internal message' }}</p>
+									{% else %}
 									<p>{{ msg.data or 'Not supported WhatsApp internal message' }}</p>
+									{% endif %}
 								</div>
 							{% else %}
 								{% if msg.media == false %}
@@ -205,7 +209,11 @@
 							{% endif %}
 							{% if msg.meta == true or msg.media == false and msg.data is none %}
 								<div class="w3-panel w3-border-blue w3-pale-blue w3-rightbar w3-leftbar w3-threequarter w3-center">
+									{% if msg.safe %}
+									<p>{{ msg.data | safe or 'Not supported WhatsApp internal message' }}</p>
+									{% else %}
 									<p>{{ msg.data or 'Not supported WhatsApp internal message' }}</p>
+									{% endif %}
 								</div>
 							{% else %}
 								{% if msg.media == false %}