Bump version

No plan on implementing this feature

.github/ISSUE_TEMPLATE/bug_report.md

@@ -8,7 +8,8 @@ assignees: ''
 ---
 
 # Must have
-- WhatsApp version: [WhatsApp version] - [Android/iOS]
+- WhatsApp version: [WhatsApp version]
+- OS: [Android/iOS] - [version]
 - Platform: [Linux/Windows/MacOS]
 - Branch and version: [main/dev] - [exporter version]
 

.github/workflows/compile-binary.yml

@@ -12,22 +12,22 @@ jobs:
   linux:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@v3
+    - uses: actions/checkout@v4
     - name: Set up Python
-      uses: actions/setup-python@v4
+      uses: actions/setup-python@v5
       with:
-        python-version: '3.11'
+        python-version: '3.12'
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
-        pip install pycryptodome javaobj-py3 ordered-set zstandard nuitka
+        pip install pycryptodome javaobj-py3 ordered-set zstandard nuitka==2.3
         pip install .
     - name: Build binary with Nuitka
       run: |
         python -m nuitka --no-deployment-flag=self-execution --onefile --include-data-file=./Whatsapp_Chat_Exporter/whatsapp.html=./Whatsapp_Chat_Exporter/whatsapp.html --follow-imports Whatsapp_Chat_Exporter/__main__.py 
         cp __main__.bin wtsexporter_linux_x64
         sha256sum wtsexporter_linux_x64
-    - uses: actions/upload-artifact@v3
+    - uses: actions/upload-artifact@v4
       with:
         name: binary-linux
         path: |
@@ -36,22 +36,22 @@ jobs:
   windows:
     runs-on: windows-latest
     steps:
-    - uses: actions/checkout@v3
+    - uses: actions/checkout@v4
     - name: Set up Python
-      uses: actions/setup-python@v4
+      uses: actions/setup-python@v5
       with:
-        python-version: '3.11'
+        python-version: '3.12'
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
-        pip install pycryptodome javaobj-py3 ordered-set zstandard nuitka
+        pip install pycryptodome javaobj-py3 ordered-set zstandard nuitka==2.3
         pip install .
     - name: Build binary with Nuitka
       run: |
         python -m nuitka --no-deployment-flag=self-execution --onefile --include-data-file=./Whatsapp_Chat_Exporter/whatsapp.html=./Whatsapp_Chat_Exporter/whatsapp.html --assume-yes-for-downloads --follow-imports Whatsapp_Chat_Exporter\__main__.py 
         copy __main__.exe wtsexporter_x64.exe
         Get-FileHash wtsexporter_x64.exe
-    - uses: actions/upload-artifact@v3
+    - uses: actions/upload-artifact@v4
       with:
         name: binary-windows
         path: |
@@ -60,22 +60,22 @@ jobs:
   macos:
     runs-on: macos-latest
     steps:
-    - uses: actions/checkout@v3
+    - uses: actions/checkout@v4
     - name: Set up Python
-      uses: actions/setup-python@v4
+      uses: actions/setup-python@v5
       with:
-        python-version: '3.10'
+        python-version: '3.12'
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
-        pip install pycryptodome javaobj-py3 ordered-set zstandard nuitka
+        pip install pycryptodome javaobj-py3 ordered-set zstandard nuitka==2.3
         pip install .
     - name: Build binary with Nuitka
       run: |
         python -m nuitka --no-deployment-flag=self-execution --onefile --include-data-file=./Whatsapp_Chat_Exporter/whatsapp.html=./Whatsapp_Chat_Exporter/whatsapp.html --follow-imports Whatsapp_Chat_Exporter/__main__.py 
         cp __main__.bin wtsexporter_macos_x64
         shasum -a 256 wtsexporter_macos_x64
-    - uses: actions/upload-artifact@v3
+    - uses: actions/upload-artifact@v4
       with:
         name: binary-macos
         path: |

.github/workflows/python-publish.yml

@@ -11,6 +11,10 @@ name: Upload Python Package
 on:
   release:
     types: [published]
+  workflow_dispatch:
+
+permissions:
+  id-token: write
 
 jobs:
   deploy:

.gitignore

@@ -134,3 +134,14 @@ dmypy.json
 *.onefile-build/
 *.exe
 __main__
+
+
+# Dev time intermidiates & temp files
+result/
+WhatsApp/
+/*.db
+/*.db-*
+/myout
+/msgstore.db
+/myout-json
+.vscode/

README.md

@@ -4,12 +4,17 @@
 [![Python](https://img.shields.io/pypi/pyversions/Whatsapp-Chat-Exporter)](https://pypi.org/project/Whatsapp-Chat-Exporter/)
 
 A customizable Android and iPhone Whatsapp database parser that will give you the history of your Whatsapp conversations in HTML and JSON. Inspired by [Telegram Chat Export Tool](https://telegram.org/blog/export-and-more).  
-**If you plan to uninstall WhatsApp or delete your WhatsApp account, please make a backup of your WhatsApp database. You may want to use this exporter again on the same database in the future as the exporter develops**
+> [!TIP]
+> If you plan to uninstall WhatsApp or delete your WhatsApp account, please make a backup of your WhatsApp database. You may want to use this exporter again on the same database in the future as the exporter develops.
+
+If you would like to support this project, all you need to do is to contribute or share this project! If you think otherwise and want to make a donation, please refer to the [Donation Guide](https://blog.knugi.com/DONATE.html).
 
 # Usage
-**Usage in README may be removed in the future. Check the usage in [Wiki](https://github.com/KnugiHK/Whatsapp-Chat-Exporter/wiki)**.
 
-**If you want to use the old release (< 0.5) of the exporter, please follow the [old usage guide](https://github.com/KnugiHK/Whatsapp-Chat-Exporter/wiki/Old-Usage#usage)**.
+> [!NOTE]
+> Usage in README may be removed in the future. Check the usage in [Wiki](https://github.com/KnugiHK/Whatsapp-Chat-Exporter/wiki)
+> 
+> If you want to use the old release (< 0.5) of the exporter, please follow the [old usage guide](https://github.com/KnugiHK/Whatsapp-Chat-Exporter/wiki/Old-Usage#usage).
 
 First, install the exporter by:
 ```shell
@@ -35,6 +40,13 @@ Simply invoke the following command from shell.
 ```sh
 wtsexporter -a
 ```
+#### Enriching Contact from vCard
+Usually, the default WhatsApp contact database extracted from your phone will contains the contact names and the exporter will use it to map your chats. However, some reported cases showed that the database could has never been populated. 
+In this case, you can export your contacts to a vCard file from your phone or a cloud provider like Google Contacts. Then, install the necessary dependency and run the following command from the shell:
+```sh
+pip install whatsapp-chat-exporter["vcards"]
+wtsexporter -a --enrich-from-vcard contacts.vcf --default-country-code 852
+```
 
 ### Encrypted Android WhatsApp Backup
 In order to support the decryption, install pycryptodome if it is not installed
@@ -42,7 +54,10 @@ In order to support the decryption, install pycryptodome if it is not installed
 pip install pycryptodome # Or 
 pip install whatsapp-chat-exporter["android_backup"] # install along with this software
 ```
-### Crypt15 is now the easiest way to decrypt a backup. If you have the 32 bytes hex key generated when you enable End-to-End encrypted backup, you can use it to decrypt the backup. If you do not have the 32 bytes hex key, you can still use the key file extracted just like extacting key file for Crypt12 and Crypt14 to decrypt the backup.
+
+> [!TIP]
+> Crypt15 is now the easiest way to decrypt a backup. If you have the 32 bytes hex key generated when you enable End-to-End encrypted backup, you can use it to decrypt the backup. If you do not have the 32 bytes hex key, you can still use the key file extracted just like extacting key file for Crypt12 and Crypt14 to decrypt the backup.
+
 #### Crypt12 or Crypt14
 You will need the decryption key file from your phone. If you have root access, you can find it as `/data/data/com.whatsapp/files/key`. Otherwise, if you used WhatsApp-Key-DB-Extractor before, it will appear in the WhatsApp backup directory as `WhatsApp/Databases/.nomedia`.
 
@@ -80,12 +95,13 @@ wtsexporter -a -k 432435053b5204b08e5c3823423399aa30ff061435ab89bc4e6713969cdaa5
 ```
 
 ## Working with iOS/iPadOS (iPhone or iPad)
-Do an iPhone/iPad Backup with iTunes first.
+Do an iPhone/iPad Backup with iTunes/Finder first.
 * iPhone backup on Mac: https://support.apple.com/HT211229
 * iPhone backup on Windows: https://support.apple.com/HT212156
 * iPad backup: https://support.apple.com/guide/ipad/ipad9a74df05xx/ipados
 ### Encrypted iOS/iPadOS Backup
-**If you are working on unencrypted iOS/iPadOS backup, skip this**
+> [!NOTE]
+> If you are working on unencrypted iOS/iPadOS backup, skip this.
 
 If you want to work on an encrypted iOS/iPadOS Backup, you should install iphone_backup_decrypt from [KnugiHK/iphone_backup_decrypt](https://github.com/KnugiHK/iphone_backup_decrypt) before you run the extract_iphone_media.py.
 ```sh
@@ -114,14 +130,12 @@ After extracting, you will get these:
 Invoke the wtsexporter with --help option will show you all options available.
 ```sh
 > wtsexporter --help
-usage: wtsexporter [-h] [-a] [-i] [-e EXPORTED] [-w WA] [-m MEDIA] [-b BACKUP] [-o OUTPUT] [-j [JSON]] [-d DB]
-                   [-k KEY] [-t TEMPLATE] [-s] [-c] [--offline OFFLINE] [--size [SIZE]] [--no-html] [--check-update]
-                   [--assume-first-as-me] [--no-avatar] [--import] [--business] [--preserve-timestamp] [--wab WAB]
-                   [--time-offset {-12 to 14}] [--date DATE] [--date-format FORMAT] [--include [phone number ...]]
-                   [--exclude [phone number ...]] [--create-separated-media]
+usage: wtsexporter [-h] [-a] [-i] [-e EXPORTED] [-w WA] [-m MEDIA] [-b BACKUP] [-o OUTPUT] [-j [JSON]] [--avoid-encoding-json] [--pretty-print-json [PRETTY_PRINT_JSON]] [-d DB] [-k KEY] [-t TEMPLATE] [-s]
+                   [-c] [--offline OFFLINE] [--size [SIZE]] [--no-html] [--check-update] [--assume-first-as-me] [--no-avatar] [--import] [--business] [--wab WAB] [--time-offset {-12 to 14}] [--date DATE]
+                   [--date-format FORMAT] [--include [phone number ...]] [--exclude [phone number ...]] [--dont-filter-empty] [--per-chat] [--create-separated-media] [--decrypt-chunk-size DECRYPT_CHUNK_SIZE]
+                   [--enrich-from-vcards ENRICH_FROM_VCARDS] [--default-country-code DEFAULT_CONTRY_CODE] [--txt [TEXT_FORMAT]]
 
-A customizable Android and iPhone WhatsApp database parser that will give you the history of your WhatsApp
-conversations in HTML and JSON. Android Backup Crypt12, Crypt14 and Crypt15 supported.
+A customizable Android and iOS/iPadOS WhatsApp database parser that will give you the history of your WhatsApp conversations in HTML and JSON. Android Backup Crypt12, Crypt14 and Crypt15 supported.
 
 options:
   -h, --help            show this help message and exit
@@ -133,11 +147,15 @@ options:
   -m MEDIA, --media MEDIA
                         Path to WhatsApp media folder (default: WhatsApp)
   -b BACKUP, --backup BACKUP
-                        Path to Android (must be used together with -k)/iPhone WhatsApp backup
+                        Path to Android (must be used together with -k)/iOS WhatsApp backup
   -o OUTPUT, --output OUTPUT
                         Output to specific directory (default: result)
   -j [JSON], --json [JSON]
                         Save the result to a single JSON file (default if present: result.json)
+  --avoid-encoding-json
+                        Don't encode non-ascii characters in the output JSON files
+  --pretty-print-json [PRETTY_PRINT_JSON]
+                        Pretty print the output JSON.
   -d DB, --db DB        Path to database file (default: msgstore.db/7c7fba66680ef796b916b067077cc246adacf01d)
   -k KEY, --key KEY     Path to key file
   -t TEMPLATE, --template TEMPLATE
@@ -153,7 +171,6 @@ options:
   --no-avatar           Do not render avatar in HTML output
   --import              Import JSON file and convert to HTML output
   --business            Use Whatsapp Business default files (iOS only)
-  --preserve-timestamp  Preserve the modification timestamp of the extracted files (iOS only)
   --wab WAB, --wa-backup WAB
                         Path to contact database in crypt15 format
   --time-offset {-12 to 14}
@@ -164,11 +181,20 @@ options:
                         Include chats that match the supplied phone number
   --exclude [phone number ...]
                         Exclude chats that match the supplied phone number
+  --dont-filter-empty   By default, the exporter will not render chats with no valid message. Setting this flag will cause the exporter to render those.
+  --per-chat            Output the JSON file per chat
   --create-separated-media
-                        Create a copy of the media seperated per chat in <MEDIA>/separated/ directory 
-                        (Android only)
+                        Create a copy of the media seperated per chat in <MEDIA>/separated/ directory
+  --decrypt-chunk-size DECRYPT_CHUNK_SIZE
+                        Specify the chunk size for decrypting iOS backup, which may affect the decryption speed.
+  --enrich-from-vcards ENRICH_FROM_VCARDS
+                        Path to an exported vcf file from Google contacts export. Add names missing from WhatsApp's default database
+  --default-country-code DEFAULT_CONTRY_CODE
+                        Use with --enrich-from-vcards. When numbers in the vcf file does not have a country code, this will be used. 1 is for US, 66 for Thailand etc. Most likely use the number of your own
+                        country
+  --txt [TEXT_FORMAT]   Export chats in text format similar to what WhatsApp officially provided (default if present: result/)
 
-WhatsApp Chat Exporter: 0.10.0 Licensed with MIT
+WhatsApp Chat Exporter: 0.10.5 Licensed with MIT. See https://wts.knugi.dev/docs?dest=osl for all open source licenses.
 ```
 
 # To do
@@ -179,6 +205,8 @@ This is a MIT licensed project.
 
 The Telegram Desktop's export is the reference for whatsapp.html in this repo.
 
-bplist.py was released by Vladimir "Farcaller" Pouzanov under MIT license.
+`bplist.py` was released by Vladimir "Farcaller" Pouzanov under MIT license.
+
+Please also refer to any files prefixed with `LICENSE` to obtain copies of the various licenses.
 
 WhatsApp Chat Exporter is not affiliated, associated, authorized, endorsed by, or in any way officially connected with the WhatsApp LLC, or any of its subsidiaries or its affiliates. The official WhatsApp LLC website can be found at https://www.whatsapp.com/.

Whatsapp_Chat_Exporter/__init__.py

@@ -1,3 +1,3 @@
 #!/usr/bin/python3
 
-__version__ = "0.10.0"
+__version__ = "0.10.5"

Whatsapp_Chat_Exporter/__main__.py

@@ -7,11 +7,18 @@ import shutil
 import json
 import string
 import glob
+try:
+    import vobject
+except ModuleNotFoundError:
+    vcards_deps_installed = False
+else:
+    from Whatsapp_Chat_Exporter.vcards_contacts import ContactsFromVCards
+    vcards_deps_installed = True
 from Whatsapp_Chat_Exporter import exported_handler, android_handler
 from Whatsapp_Chat_Exporter import ios_handler, ios_media_handler
 from Whatsapp_Chat_Exporter.data_model import ChatStore
-from Whatsapp_Chat_Exporter.utility import APPLE_TIME, Crypt, DbType
-from Whatsapp_Chat_Exporter.utility import check_update, import_from_json
+from Whatsapp_Chat_Exporter.utility import APPLE_TIME, Crypt, DbType, chat_is_empty, readable_to_bytes
+from Whatsapp_Chat_Exporter.utility import check_update, import_from_json, sanitize_filename, bytes_to_readable
 from argparse import ArgumentParser, SUPPRESS
 from datetime import datetime
 from sys import exit
@@ -26,7 +33,7 @@ def main():
         description = 'A customizable Android and iOS/iPadOS WhatsApp database parser that '
                       'will give you the history of your WhatsApp conversations in HTML '
                       'and JSON. Android Backup Crypt12, Crypt14 and Crypt15 supported.',
-        epilog = f'WhatsApp Chat Exporter: {__version__} Licensed with MIT. See'
+        epilog = f'WhatsApp Chat Exporter: {__version__} Licensed with MIT. See '
                   'https://wts.knugi.dev/docs?dest=osl for all open source licenses.'
     )
     parser.add_argument(
@@ -85,6 +92,20 @@ def main():
         type=str,
         const="result.json",
         help="Save the result to a single JSON file (default if present: result.json)")
+    parser.add_argument(
+        '--avoid-encoding-json',
+        dest='avoid_encoding_json',
+        default=False,
+        action='store_true',
+        help="Don't encode non-ascii characters in the output JSON files")
+    parser.add_argument(
+        '--pretty-print-json',
+        dest='pretty_print_json',
+        default=None,
+        nargs='?',
+        const=2,
+        type=int,
+        help="Pretty print the output JSON.")
     parser.add_argument(
         '-d',
         '--db',
@@ -141,7 +162,6 @@ def main():
         "--split",
         dest="size",
         nargs='?',
-        type=int,
         const=0,
         default=None,
         help="Maximum (rough) size of a single output file in bytes, 0 for auto"
@@ -188,13 +208,6 @@ def main():
         action='store_true',
         help="Use Whatsapp Business default files (iOS only)"
     )
-    parser.add_argument(
-        "--preserve-timestamp",
-        dest="preserve_timestamp",
-        default=False,
-        action='store_true',
-        help="Preserve the modification timestamp of the extracted files (iOS only)"
-    )
     parser.add_argument(
         "--wab",
         "--wa-backup",
@@ -239,6 +252,13 @@ def main():
         metavar="phone number",
         help="Exclude chats that match the supplied phone number"
     )
+    parser.add_argument(
+        "--dont-filter-empty",
+        dest="filter_empty",
+        default=True,
+        action='store_false',
+        help="By default, the exporter will not render chats with no valid message. Setting this flag will cause the exporter to render those."
+    )
     parser.add_argument(
         "--per-chat",
         dest="json_per_chat",
@@ -253,6 +273,34 @@ def main():
         action='store_true',
         help="Create a copy of the media seperated per chat in <MEDIA>/separated/ directory"
     )
+    parser.add_argument(
+        "--decrypt-chunk-size",
+        dest="decrypt_chunk_size",
+        default=1 * 1024 * 1024,
+        type=int,
+        help="Specify the chunk size for decrypting iOS backup, which may affect the decryption speed."
+    )
+    parser.add_argument(
+        "--enrich-from-vcards",
+        dest="enrich_from_vcards",
+        default=None,
+        help="Path to an exported vcf file from Google contacts export. Add names missing from WhatsApp's default database"
+    )
+    parser.add_argument(
+        "--default-country-code",
+        dest="default_contry_code",
+        default=None,
+        help="Use with --enrich-from-vcards. When numbers in the vcf file does not have a country code, this will be used. 1 is for US, 66 for Thailand etc. Most likely use the number of your own country"
+    )
+    parser.add_argument(
+        "--txt",
+        dest="text_format",
+        nargs='?',
+        default=None,
+        type=str,
+        const="result",
+        help="Export chats in text format similar to what WhatsApp officially provided (default if present: result/)")
+
     args = parser.parse_args()
 
     # Check for updates
@@ -264,10 +312,10 @@ def main():
         parser.error("You must define only one device type.")
     if not args.android and not args.ios and not args.exported and not args.import_json:
         parser.error("You must define the device type.")
-    if args.no_html and not args.json:
-        parser.error("You must either specify a JSON output file or enable HTML output.")
+    if args.no_html and not args.json and not args.text_format:
+        parser.error("You must either specify a JSON output file, text file output directory or enable HTML output.")
     if args.import_json and (args.android or args.ios or args.exported or args.no_html):
-        parser.error("You can only use --import with -j and without --no-html.")
+        parser.error("You can only use --import with -j and without --no-html, -a, -i, -e.")
     elif args.import_json and not os.path.isfile(args.json):
         parser.error("JSON file not found.")
     if args.android and args.business:
@@ -277,6 +325,13 @@ def main():
         (args.json[-5:] == ".json" and os.path.isfile(args.json[:-5]))
     ):
         parser.error("When --per-chat is enabled, the destination of --json must be a directory.")
+    if args.enrich_from_vcards is not None and args.default_contry_code is None:
+        parser.error("When --enrich-from-vcards is provided, you must also set --default-country-code")
+    if args.size is not None and not isinstance(args.size, int) and not args.size.isnumeric():
+        try:
+            args.size = readable_to_bytes(args.size)
+        except ValueError:
+            parser.error("The value for --split must be ended in pure bytes or with a proper unit (e.g., 1048576 or 1MB)")
     if args.filter_date is not None:
         if " - " in args.filter_date:
             start, end = args.filter_date.split(" - ")
@@ -320,6 +375,16 @@ def main():
 
     data = {}
 
+    if args.enrich_from_vcards is not None:
+        if not vcards_deps_installed:
+            parser.error(
+                "You don't have the dependency to enrich contacts with vCard.\n"
+                "Read more on how to deal with enriching contacts:\n"
+                "https://github.com/KnugiHK/Whatsapp-Chat-Exporter/blob/main/README.md#usage"
+            )
+        contact_store = ContactsFromVCards()
+        contact_store.load_vcf_file(args.enrich_from_vcards, args.default_contry_code)
+
     if args.android:
         contacts = android_handler.contacts
         messages = android_handler.messages
@@ -403,7 +468,7 @@ def main():
             args.media = identifiers.DOMAIN
         if args.backup is not None:
             if not os.path.isdir(args.media):
-                ios_media_handler.extract_media(args.backup, identifiers, args.preserve_timestamp)
+                ios_media_handler.extract_media(args.backup, identifiers, args.decrypt_chunk_size)
             else:
                 print("WhatsApp directory already exists, skipping WhatsApp file extraction.")
         if args.db is None:
@@ -429,6 +494,9 @@ def main():
                 if args.android:
                     android_handler.calls(db, data, args.timezone_offset, filter_chat)
             if not args.no_html:
+                if args.enrich_from_vcards is not None and not contact_store.is_empty():
+                    contact_store.enrich_from_vcards(data)
+
                 create_html(
                     data,
                     args.output,
@@ -436,7 +504,8 @@ def main():
                     args.embedded,
                     args.offline,
                     args.size,
-                    args.no_avatar
+                    args.no_avatar,
+                    args.filter_empty
                 )
         else:
             print(
@@ -471,7 +540,9 @@ def main():
                 args.template,
                 args.embedded,
                 args.offline,
-                args.size
+                args.size,
+                args.no_avatar,
+                args.filter_empty
             )
         for file in glob.glob(r'*.*'):
             shutil.copy(file, args.output)
@@ -483,16 +554,33 @@ def main():
             args.template,
             args.embedded,
             args.offline,
-            args.size
+            args.size,
+            args.no_avatar,
+            args.filter_empty
         )
 
+    if args.text_format:
+        print("Writing text file...")
+        android_handler.create_txt(data, args.text_format)
+
     if args.json and not args.import_json:
+        if args.filter_empty:
+            data = {k: v for k, v in data.items() if not chat_is_empty(v)}
+
+        if args.enrich_from_vcards is not None and not contact_store.is_empty():
+            contact_store.enrich_from_vcards(data)
+
         if isinstance(data[next(iter(data))], ChatStore):
             data = {jik: chat.to_json() for jik, chat in data.items()}
+            
         if not args.json_per_chat:
             with open(args.json, "w") as f:
-                data = json.dumps(data)
-                print(f"\nWriting JSON file...({int(len(data)/1024/1024)}MB)")
+                data = json.dumps(
+                    data,
+                    ensure_ascii=not args.avoid_encoding_json,
+                    indent=args.pretty_print_json
+                )
+                print(f"\nWriting JSON file...({bytes_to_readable(len(data))})")
                 f.write(data)
         else:
             if args.json[-5:] == ".json":
@@ -505,8 +593,9 @@ def main():
                     contact = data[jik]["name"].replace('/', '')
                 else:
                     contact = jik.replace('+', '')
-                with open(f"{args.json}/{contact}.json", "w") as f:
-                    f.write(json.dumps(data[jik]))
+                with open(f"{args.json}/{sanitize_filename(contact)}.json", "w") as f:
+                    file_content_to_write = json.dumps({jik: data[jik]}, ensure_ascii=not args.avoid_encoding_json, indent=2 if args.pretty_print_json else None)
+                    f.write(file_content_to_write)
                     print(f"Writing JSON file...({index + 1}/{total})", end="\r")
             print()
     else:

Whatsapp_Chat_Exporter/android_handler.py

@@ -10,11 +10,12 @@ from mimetypes import MimeTypes
 from markupsafe import escape as htmle
 from hashlib import sha256
 from base64 import b64decode, b64encode
+from datetime import datetime
 from Whatsapp_Chat_Exporter.data_model import ChatStore, Message
-from Whatsapp_Chat_Exporter.utility import MAX_SIZE, ROW_SIZE, DbType, determine_metadata, JidType
-from Whatsapp_Chat_Exporter.utility import rendering, Crypt, Device, get_file_name, setup_template
+from Whatsapp_Chat_Exporter.utility import MAX_SIZE, ROW_SIZE, DbType, convert_time_unit, determine_metadata
+from Whatsapp_Chat_Exporter.utility import rendering, Crypt, Device, get_file_name, setup_template, JidType
 from Whatsapp_Chat_Exporter.utility import brute_force_offset, CRYPT14_OFFSETS, get_status_location
-from Whatsapp_Chat_Exporter.utility import get_chat_condition, slugify
+from Whatsapp_Chat_Exporter.utility import get_chat_condition, slugify, bytes_to_readable, chat_is_empty
 
 try:
     import zlib
@@ -157,7 +158,11 @@ def contacts(db, data):
     c = db.cursor()
     c.execute("""SELECT count() FROM wa_contacts""")
     total_row_number = c.fetchone()[0]
-    print(f"Processing contacts...({total_row_number})")
+    if total_row_number == 0:
+        print("No contacts profiles found in the default database, consider using --enrich-from-vcards for adopting names from exported contacts from Google")
+        return False
+    else:
+        print(f"Processing contacts...({total_row_number})")
 
     c.execute("""SELECT jid, COALESCE(display_name, wa_name) as display_name, status FROM wa_contacts; """)
     row = c.fetchone()
@@ -174,10 +179,12 @@ def messages(db, data, media_folder, timezone_offset, filter_date, filter_chat):
     try:
         c.execute(f"""SELECT count()
                       FROM messages
+                        INNER JOIN jid
+                            ON messages.key_remote_jid = jid.raw_string
                       WHERE 1=1
                         {f'AND timestamp {filter_date}' if filter_date is not None else ''}
-                        {get_chat_condition(filter_chat[0], True, "messages.key_remote_jid")}
-                        {get_chat_condition(filter_chat[1], False, "messages.key_remote_jid")}""")
+                        {get_chat_condition(filter_chat[0], True, ["messages.key_remote_jid", "messages.remote_resource"], "jid", "android")}
+                        {get_chat_condition(filter_chat[1], False, ["messages.key_remote_jid", "messages.remote_resource"], "jid", "android")}""")
 
     except sqlite3.OperationalError:
         c.execute(f"""SELECT count()
@@ -186,10 +193,12 @@ def messages(db, data, media_folder, timezone_offset, filter_date, filter_chat):
                             ON chat._id = message.chat_row_id
                         INNER JOIN jid
                             ON jid._id = chat.jid_row_id
+                        LEFT JOIN jid jid_group
+                            ON jid_group._id = message.sender_jid_row_id
                       WHERE 1=1
                         {f'AND timestamp {filter_date}' if filter_date is not None else ''}
-                        {get_chat_condition(filter_chat[0], True, "jid.raw_string")}
-                        {get_chat_condition(filter_chat[1], False, "jid.raw_string")}""")
+                        {get_chat_condition(filter_chat[0], True, ["jid.raw_string", "jid_group.raw_string"], "jid", "android")}
+                        {get_chat_condition(filter_chat[1], False, ["jid.raw_string", "jid_group.raw_string"], "jid", "android")}""")
     total_row_number = c.fetchone()[0]
     print(f"Processing messages...(0/{total_row_number})", end="\r")
 
@@ -245,8 +254,8 @@ def messages(db, data, media_folder, timezone_offset, filter_date, filter_chat):
                             ON receipt_user.message_row_id = messages._id
                     WHERE messages.key_remote_jid <> '-1'
                         {f'AND messages.timestamp {filter_date}' if filter_date is not None else ''}
-                        {get_chat_condition(filter_chat[0], True, "messages.key_remote_jid")}
-                        {get_chat_condition(filter_chat[1], False, "messages.key_remote_jid")}
+                        {get_chat_condition(filter_chat[0], True, ["messages.key_remote_jid", "messages.remote_resource"], "jid_global", "android")}
+                        {get_chat_condition(filter_chat[1], False, ["messages.key_remote_jid", "messages.remote_resource"], "jid_global", "android")}
                     GROUP BY messages._id
                     ORDER BY messages.timestamp ASC;"""
         )
@@ -313,8 +322,8 @@ def messages(db, data, media_folder, timezone_offset, filter_date, filter_chat):
                             ON receipt_user.message_row_id = message._id
                     WHERE key_remote_jid <> '-1'
                         {f'AND message.timestamp {filter_date}' if filter_date is not None else ''}
-                        {get_chat_condition(filter_chat[0], True, "key_remote_jid")}
-                        {get_chat_condition(filter_chat[1], False, "key_remote_jid")}
+                        {get_chat_condition(filter_chat[0], True, ["key_remote_jid", "jid_group.raw_string"], "jid_global", "android")}
+                        {get_chat_condition(filter_chat[1], False, ["key_remote_jid", "jid_group.raw_string"], "jid_global", "android")}
                     GROUP BY message._id;"""
             )
         except Exception as e:
@@ -445,9 +454,9 @@ def messages(db, data, media_folder, timezone_offset, filter_date, filter_chat):
                         msg = content["data"]
                         if msg is not None:
                             if "\r\n" in msg:
-                                msg = msg.replace("\r\n", "<br>")
+                                msg = msg.replace("\r\n", " <br>")
                             if "\n" in msg:
-                                msg = msg.replace("\n", "<br>")
+                                msg = msg.replace("\n", " <br>")
             else:
                 if content["status"] == 0 and content["edit_version"] == 7 or table_message and content["media_wa_type"] == 15:
                     msg = "Message deleted"
@@ -460,9 +469,9 @@ def messages(db, data, media_folder, timezone_offset, filter_date, filter_chat):
                         msg = content["data"]
                         if msg is not None:
                             if "\r\n" in msg:
-                                msg = msg.replace("\r\n", "<br>")
+                                msg = msg.replace("\r\n", " <br>")
                             if "\n" in msg:
-                                msg = msg.replace("\n", "<br>")
+                                msg = msg.replace("\n", " <br>")
             message.data = msg
 
         data[content["key_remote_jid"]].add_message(content["_id"], message)
@@ -487,10 +496,12 @@ def media(db, data, media_folder, filter_date, filter_chat, separate_media=True)
                     FROM message_media
                         INNER JOIN messages
                             ON message_media.message_row_id = messages._id
+                        INNER JOIN jid
+                            ON messages.key_remote_jid = jid.raw_string
                     WHERE 1=1  
                         {f'AND messages.timestamp {filter_date}' if filter_date is not None else ''}
-                        {get_chat_condition(filter_chat[0], True, "messages.key_remote_jid")}
-                        {get_chat_condition(filter_chat[1], False, "messages.key_remote_jid")}""")
+                        {get_chat_condition(filter_chat[0], True, ["messages.key_remote_jid", "remote_resource"], "jid", "android")}
+                        {get_chat_condition(filter_chat[1], False, ["messages.key_remote_jid", "remote_resource"], "jid", "android")}""")
     except sqlite3.OperationalError:
         c.execute(f"""SELECT count()
                     FROM message_media
@@ -500,10 +511,12 @@ def media(db, data, media_folder, filter_date, filter_chat, separate_media=True)
                             ON chat._id = message.chat_row_id
                         INNER JOIN jid
                             ON jid._id = chat.jid_row_id
+                        LEFT JOIN jid jid_group
+                            ON jid_group._id = message.sender_jid_row_id
                     WHERE 1=1    
                         {f'AND message.timestamp {filter_date}' if filter_date is not None else ''}
-                        {get_chat_condition(filter_chat[0], True, "jid.raw_string")}
-                        {get_chat_condition(filter_chat[1], False, "jid.raw_string")}""")
+                        {get_chat_condition(filter_chat[0], True, ["jid.raw_string", "jid_group.raw_string"], "jid", "android")}
+                        {get_chat_condition(filter_chat[1], False, ["jid.raw_string", "jid_group.raw_string"], "jid", "android")}""")
     total_row_number = c.fetchone()[0]
     print(f"\nProcessing media...(0/{total_row_number})", end="\r")
     i = 0
@@ -525,8 +538,8 @@ def media(db, data, media_folder, filter_date, filter_chat, separate_media=True)
                         ON messages.key_remote_jid = jid.raw_string
                 WHERE jid.type <> 7
                     {f'AND messages.timestamp {filter_date}' if filter_date is not None else ''}
-                    {get_chat_condition(filter_chat[0], True, "messages.key_remote_jid")}
-                    {get_chat_condition(filter_chat[1], False, "messages.key_remote_jid")}
+                    {get_chat_condition(filter_chat[0], True, ["messages.key_remote_jid", "remote_resource"], "jid", "android")}
+                    {get_chat_condition(filter_chat[1], False, ["messages.key_remote_jid", "remote_resource"], "jid", "android")}
                 ORDER BY messages.key_remote_jid ASC"""
         )
     except sqlite3.OperationalError:
@@ -547,10 +560,12 @@ def media(db, data, media_folder, filter_date, filter_chat, separate_media=True)
                         ON jid._id = chat.jid_row_id
                     LEFT JOIN media_hash_thumbnail
 						ON message_media.file_hash = media_hash_thumbnail.media_hash
+                    LEFT JOIN jid jid_group
+                        ON jid_group._id = message.sender_jid_row_id
                 WHERE jid.type <> 7
                     {f'AND message.timestamp {filter_date}' if filter_date is not None else ''}
-                    {get_chat_condition(filter_chat[0], True, "key_remote_jid")}
-                    {get_chat_condition(filter_chat[1], False, "key_remote_jid")}
+                    {get_chat_condition(filter_chat[0], True, ["key_remote_jid", "jid_group.raw_string"], "jid", "android")}
+                    {get_chat_condition(filter_chat[1], False, ["key_remote_jid", "jid_group.raw_string"], "jid", "android")}
                 ORDER BY jid.raw_string ASC"""
         )
     content = c.fetchone()
@@ -581,17 +596,6 @@ def media(db, data, media_folder, filter_date, filter_chat, separate_media=True)
                 shutil.copy2(file_path, new_path)
                 message.data = new_path
         else:
-            if False: # Block execution
-                try:
-                    r = requests.get(content["message_url"])
-                    if r.status_code != 200:
-                        raise RuntimeError()
-                except:
-                    message.data = "The media is missing"
-                    message.mime = "media"
-                    message.meta = True
-                else:
-                    ...
             message.data = "The media is missing"
             message.mime = "media"
             message.meta = True
@@ -619,10 +623,12 @@ def vcard(db, data, media_folder, filter_date, filter_chat):
                  FROM messages_vcards
                     INNER JOIN messages
                         ON messages_vcards.message_row_id = messages._id
+                    INNER JOIN jid
+                        ON messages.key_remote_jid = jid.raw_string
                  WHERE 1=1
                     {f'AND messages.timestamp {filter_date}' if filter_date is not None else ''}
-                    {get_chat_condition(filter_chat[0], True, "messages.key_remote_jid")}
-                    {get_chat_condition(filter_chat[1], False, "messages.key_remote_jid")}
+                    {get_chat_condition(filter_chat[0], True, ["messages.key_remote_jid", "remote_resource"], "jid", "android")}
+                    {get_chat_condition(filter_chat[1], False, ["messages.key_remote_jid", "remote_resource"], "jid", "android")}
                  ORDER BY messages.key_remote_jid ASC;"""
         )
     except sqlite3.OperationalError:
@@ -637,10 +643,12 @@ def vcard(db, data, media_folder, filter_date, filter_chat):
                         ON chat._id = message.chat_row_id
                     INNER JOIN jid
                         ON jid._id = chat.jid_row_id
+                    LEFT JOIN jid jid_group
+                        ON jid_group._id = message.sender_jid_row_id
                 WHERE 1=1
                     {f'AND message.timestamp {filter_date}' if filter_date is not None else ''}
-                    {get_chat_condition(filter_chat[0], True, "key_remote_jid")}
-                    {get_chat_condition(filter_chat[1], False, "key_remote_jid")}
+                    {get_chat_condition(filter_chat[0], True, ["key_remote_jid", "jid_group.raw_string"], "jid", "android")}
+                    {get_chat_condition(filter_chat[1], False, ["key_remote_jid", "jid_group.raw_string"], "jid", "android")}
                  ORDER BY message.chat_row_id ASC;"""
         )
 
@@ -676,8 +684,8 @@ def calls(db, data, timezone_offset, filter_chat):
                     LEFT JOIN chat
                         ON call_log.jid_row_id = chat.jid_row_id
                 WHERE 1=1
-                    {get_chat_condition(filter_chat[0], True, "jid.raw_string")}
-                    {get_chat_condition(filter_chat[1], False, "jid.raw_string")}""")
+                    {get_chat_condition(filter_chat[0], True, ["jid.raw_string"])}
+                    {get_chat_condition(filter_chat[1], False, ["jid.raw_string"])}""")
     total_row_number = c.fetchone()[0]
     if total_row_number == 0:
         return
@@ -698,8 +706,8 @@ def calls(db, data, timezone_offset, filter_chat):
                     LEFT JOIN chat
                         ON call_log.jid_row_id = chat.jid_row_id
                 WHERE 1=1
-                    {get_chat_condition(filter_chat[0], True, "jid.raw_string")}
-                    {get_chat_condition(filter_chat[1], False, "jid.raw_string")}"""
+                    {get_chat_condition(filter_chat[0], True, ["jid.raw_string"])}
+                    {get_chat_condition(filter_chat[1], False, ["jid.raw_string"])}"""
     )
     chat = ChatStore(Device.ANDROID, "WhatsApp Calls")
     content = c.fetchone()
@@ -724,15 +732,21 @@ def calls(db, data, timezone_offset, filter_chat):
             f"call {'to' if call.from_me else 'from'} "
             f"{call.sender} was "
         )
-        if content['call_result'] == 2:
+        if content['call_result'] in (0, 4, 7):
+            call.data += "cancelled." if call.from_me else "missed."
+        elif content['call_result'] == 2:
             call.data += "not answered." if call.from_me else "missed."
         elif content['call_result'] == 3:
             call.data += "unavailable."
         elif content['call_result'] == 5:
+            call_time = convert_time_unit(content['duration'])
+            call_bytes = bytes_to_readable(content['bytes_transferred'])
             call.data += (
-                f"initiated and lasted for {content['duration']} second(s) "
-                f"with {content['bytes_transferred']} bytes transferred."
+                f"initiated and lasted for {call_time} "
+                f"with {call_bytes} data transferred."
             )
+        else:
+            call.data += "in an unknown state."
         chat.add_message(content["_id"], call)
         content = c.fetchone()
     data["000000000000000"] = chat
@@ -745,7 +759,8 @@ def create_html(
         embedded=False,
         offline_static=False,
         maximum_size=None,
-        no_avatar=False
+        no_avatar=False,
+        filter_empty=True
     ):
     template = setup_template(template, no_avatar)
 
@@ -759,7 +774,7 @@ def create_html(
 
     for current, contact in enumerate(data):
         chat = data[contact]
-        if len(chat.messages) == 0:
+        if filter_empty and chat_is_empty(chat):
             continue
         safe_file_name, name = get_file_name(contact, chat)
 
@@ -791,6 +806,7 @@ def create_html(
                     current_size = 0
                     current_page += 1
                 else:
+                    render_box.append(message)
                     if message.key_id == last_msg:
                         if current_page == 1:
                             output_file_name = f"{output_folder}/{safe_file_name}.html"
@@ -806,8 +822,6 @@ def create_html(
                             False,
                             chat
                         )
-                    else:
-                        render_box.append(message)
         else:
             output_file_name = f"{output_folder}/{safe_file_name}.html"
             rendering(
@@ -824,3 +838,38 @@ def create_html(
             print(f"Generating chats...({current}/{total_row_number})", end="\r")
 
     print(f"Generating chats...({total_row_number}/{total_row_number})", end="\r")
+
+
+def create_txt(data, output):
+    os.makedirs(output, exist_ok=True)
+    for jik, chat in data.items():
+        if chat.name is not None:
+            contact = chat.name.replace('/', '')
+        else:
+            contact = jik.replace('+', '')
+        output_file = os.path.join(output, f"{contact}.txt")
+        with open(output_file, "w", encoding="utf8") as f:
+            for message in chat.messages.values():
+                date = datetime.fromtimestamp(message.timestamp).date()
+                if message.meta and message.mime != "media":
+                    continue  # Skip any metadata in text format
+                if message.from_me:
+                    name = "You"
+                else:
+                    name = message.sender if message.sender else contact
+                prefix = f"[{date} {message.time}] {name}: "
+                prefix_length = len(prefix)
+                if message.media and ("/" in message.mime or message.mime == "media"):
+                    if message.data == "The media is missing":
+                        message_text = "<The media is missing>"
+                    else:
+                        message_text = f"<media file in {message.data}>"
+                else:
+                    if message.data is None:
+                        message_text = ""
+                    else:
+                        message_text = message.data.replace('<br>', f'\n{" " * prefix_length}')
+                if message.caption is not None:
+                    message_text += "\n" + ' ' * len(prefix) + message.caption.replace('<br>', f'\n{" " * prefix_length}')
+                f.write(f"{prefix}{message_text}\n")
+

Whatsapp_Chat_Exporter/data_model.py

@@ -34,6 +34,7 @@ class ChatStore():
         self.their_avatar = None
         self.their_avatar_thumb = None
         self.status = None
+        self.media_base = ""
     
     def add_message(self, id, message):
         if not isinstance(message, Message):

Whatsapp_Chat_Exporter/ios_handler.py

@@ -29,24 +29,41 @@ def contacts(db, data):
 def messages(db, data, media_folder, timezone_offset, filter_date, filter_chat):
     c = db.cursor()
     # Get contacts
-    c.execute(f"""SELECT count()
-                FROM ZWACHATSESSION
-                WHERE 1=1
-                    {get_chat_condition(filter_chat[0], True, "ZWACHATSESSION.ZCONTACTJID")}
-                    {get_chat_condition(filter_chat[1], False, "ZWACHATSESSION.ZCONTACTJID")}""")
+    c.execute(
+        f"""SELECT count() 
+            FROM (SELECT DISTINCT ZCONTACTJID,
+                ZPARTNERNAME,
+                ZWAPROFILEPUSHNAME.ZPUSHNAME
+            FROM ZWACHATSESSION
+                INNER JOIN ZWAMESSAGE
+                    ON ZWAMESSAGE.ZCHATSESSION = ZWACHATSESSION.Z_PK
+                LEFT JOIN ZWAPROFILEPUSHNAME
+                    ON ZWACHATSESSION.ZCONTACTJID = ZWAPROFILEPUSHNAME.ZJID
+                LEFT JOIN ZWAGROUPMEMBER
+                        ON ZWAMESSAGE.ZGROUPMEMBER = ZWAGROUPMEMBER.Z_PK
+            WHERE 1=1
+                {get_chat_condition(filter_chat[0], True, ["ZWACHATSESSION.ZCONTACTJID", "ZMEMBERJID"], "ZGROUPINFO", "ios")}
+                {get_chat_condition(filter_chat[1], False, ["ZWACHATSESSION.ZCONTACTJID", "ZMEMBERJID"], "ZGROUPINFO", "ios")}
+            GROUP BY ZCONTACTJID);"""
+    )
     total_row_number = c.fetchone()[0]
     print(f"Processing contacts...({total_row_number})")
 
     c.execute(
-        f"""SELECT ZCONTACTJID,
+        f"""SELECT DISTINCT ZCONTACTJID,
                 ZPARTNERNAME,
-                ZPUSHNAME
-           FROM ZWACHATSESSION
+                ZWAPROFILEPUSHNAME.ZPUSHNAME
+            FROM ZWACHATSESSION
+                INNER JOIN ZWAMESSAGE
+                    ON ZWAMESSAGE.ZCHATSESSION = ZWACHATSESSION.Z_PK
                 LEFT JOIN ZWAPROFILEPUSHNAME
                     ON ZWACHATSESSION.ZCONTACTJID = ZWAPROFILEPUSHNAME.ZJID
-           WHERE 1=1
-                {get_chat_condition(filter_chat[0], True, "ZWACHATSESSION.ZCONTACTJID")}
-                {get_chat_condition(filter_chat[1], False, "ZWACHATSESSION.ZCONTACTJID")};"""
+                LEFT JOIN ZWAGROUPMEMBER
+                        ON ZWAMESSAGE.ZGROUPMEMBER = ZWAGROUPMEMBER.Z_PK
+            WHERE 1=1
+                {get_chat_condition(filter_chat[0], True, ["ZWACHATSESSION.ZCONTACTJID", "ZMEMBERJID"], "ZGROUPINFO", "ios")}
+                {get_chat_condition(filter_chat[1], False, ["ZWACHATSESSION.ZCONTACTJID", "ZMEMBERJID"], "ZGROUPINFO", "ios")}
+            GROUP BY ZCONTACTJID;"""
     )
     content = c.fetchone()
     while content is not None:
@@ -78,10 +95,12 @@ def messages(db, data, media_folder, timezone_offset, filter_date, filter_chat):
                   FROM ZWAMESSAGE
                         INNER JOIN ZWACHATSESSION
                             ON ZWAMESSAGE.ZCHATSESSION = ZWACHATSESSION.Z_PK
+                        LEFT JOIN ZWAGROUPMEMBER
+                            ON ZWAMESSAGE.ZGROUPMEMBER = ZWAGROUPMEMBER.Z_PK
                   WHERE 1=1
                     {f'AND ZMESSAGEDATE {filter_date}' if filter_date is not None else ''}
-                    {get_chat_condition(filter_chat[0], True, "ZWACHATSESSION.ZCONTACTJID")}
-                    {get_chat_condition(filter_chat[1], False, "ZWACHATSESSION.ZCONTACTJID")}""")
+                    {get_chat_condition(filter_chat[0], True, ["ZWACHATSESSION.ZCONTACTJID", "ZMEMBERJID"], "ZGROUPINFO", "ios")}
+                    {get_chat_condition(filter_chat[1], False, ["ZWACHATSESSION.ZCONTACTJID", "ZMEMBERJID"], "ZGROUPINFO", "ios")}""")
     total_row_number = c.fetchone()[0]
     print(f"Processing messages...(0/{total_row_number})", end="\r")
     c.execute(f"""SELECT ZCONTACTJID,
@@ -103,8 +122,8 @@ def messages(db, data, media_folder, timezone_offset, filter_date, filter_chat):
                         ON ZWAMESSAGE.ZCHATSESSION = ZWACHATSESSION.Z_PK
                  WHERE 1=1   
                     {f'AND ZMESSAGEDATE {filter_date}' if filter_date is not None else ''}
-                    {get_chat_condition(filter_chat[0], True, "ZCONTACTJID")}
-                    {get_chat_condition(filter_chat[1], False, "ZCONTACTJID")}
+                    {get_chat_condition(filter_chat[0], True, ["ZCONTACTJID", "ZMEMBERJID"], "ZGROUPINFO", "ios")}
+                    {get_chat_condition(filter_chat[1], False, ["ZCONTACTJID", "ZMEMBERJID"], "ZGROUPINFO", "ios")}
                  ORDER BY ZMESSAGEDATE ASC;""")
     i = 0
     content = c.fetchone()
@@ -216,10 +235,12 @@ def media(db, data, media_folder, filter_date, filter_chat, separate_media=False
                         ON ZWAMEDIAITEM.ZMESSAGE = ZWAMESSAGE.Z_PK
                     INNER JOIN ZWACHATSESSION
                         ON ZWAMESSAGE.ZCHATSESSION = ZWACHATSESSION.Z_PK
+                    LEFT JOIN ZWAGROUPMEMBER
+                        ON ZWAMESSAGE.ZGROUPMEMBER = ZWAGROUPMEMBER.Z_PK
                   WHERE 1=1
                     {f'AND ZMESSAGEDATE {filter_date}' if filter_date is not None else ''}
-                    {get_chat_condition(filter_chat[0], True, "ZWACHATSESSION.ZCONTACTJID")}
-                    {get_chat_condition(filter_chat[1], False, "ZWACHATSESSION.ZCONTACTJID")}
+                    {get_chat_condition(filter_chat[0], True, ["ZWACHATSESSION.ZCONTACTJID","ZMEMBERJID"], "ZGROUPINFO", "ios")}
+                    {get_chat_condition(filter_chat[1], False, ["ZWACHATSESSION.ZCONTACTJID", "ZMEMBERJID"], "ZGROUPINFO", "ios")}
                 """)
     total_row_number = c.fetchone()[0]
     print(f"\nProcessing media...(0/{total_row_number})", end="\r")
@@ -236,20 +257,25 @@ def media(db, data, media_folder, filter_date, filter_chat, separate_media=False
                         ON ZWAMEDIAITEM.ZMESSAGE = ZWAMESSAGE.Z_PK
                     INNER JOIN ZWACHATSESSION
                         ON ZWAMESSAGE.ZCHATSESSION = ZWACHATSESSION.Z_PK
+                    LEFT JOIN ZWAGROUPMEMBER
+                        ON ZWAMESSAGE.ZGROUPMEMBER = ZWAGROUPMEMBER.Z_PK
                  WHERE ZMEDIALOCALPATH IS NOT NULL
                     {f'AND ZWAMESSAGE.ZMESSAGEDATE {filter_date}' if filter_date is not None else ''}
-                    {get_chat_condition(filter_chat[0], True, "ZCONTACTJID")}
-                    {get_chat_condition(filter_chat[1], False, "ZCONTACTJID")}
+                    {get_chat_condition(filter_chat[0], True, ["ZCONTACTJID", "ZMEMBERJID"], "ZGROUPINFO", "ios")}
+                    {get_chat_condition(filter_chat[1], False, ["ZCONTACTJID", "ZMEMBERJID"], "ZGROUPINFO", "ios")}
                  ORDER BY ZCONTACTJID ASC""")
     content = c.fetchone()
     mime = MimeTypes()
     while content is not None:
         file_path = f"{media_folder}/Message/{content['ZMEDIALOCALPATH']}"
         ZMESSAGE = content["ZMESSAGE"]
-        message = data[content["ZCONTACTJID"]].messages[ZMESSAGE]
+        contact = data[content["ZCONTACTJID"]]
+        message = contact.messages[ZMESSAGE]
         message.media = True
+        if contact.media_base == "":
+            contact.media_base = media_folder + "/"
         if os.path.isfile(file_path):
-            message.data = file_path
+            message.data = '/'.join(file_path.split("/")[1:])
             if content["ZVCARDSTRING"] is None:
                 guess = mime.guess_type(file_path)[0]
                 if guess is not None:
@@ -259,7 +285,7 @@ def media(db, data, media_folder, filter_date, filter_chat, separate_media=False
             else:
                 message.mime = content["ZVCARDSTRING"]
             if separate_media:
-                chat_display_name = slugify(data[content["ZCONTACTJID"]].name or message.sender \
+                chat_display_name = slugify(contact.name or message.sender \
                                             or content["ZCONTACTJID"].split('@')[0], True)
                 current_filename = file_path.split("/")[-1]
                 new_folder = os.path.join(media_folder, "separated", chat_display_name)
@@ -268,17 +294,6 @@ def media(db, data, media_folder, filter_date, filter_chat, separate_media=False
                 shutil.copy2(file_path, new_path)
                 message.data = new_path
         else:
-            if False: # Block execution
-                try:
-                    r = requests.get(content["ZMEDIAURL"])
-                    if r.status_code != 200:
-                        raise RuntimeError()
-                except:
-                    message.data = "The media is missing"
-                    message.mime = "media"
-                    message.meta = True
-                else:
-                    ...
             message.data = "The media is missing"
             message.mime = "media"
             message.meta = True
@@ -306,10 +321,12 @@ def vcard(db, data, media_folder, filter_date, filter_chat):
                         ON ZWAMEDIAITEM.ZMESSAGE = ZWAMESSAGE.Z_PK
                     INNER JOIN ZWACHATSESSION
                         ON ZWAMESSAGE.ZCHATSESSION = ZWACHATSESSION.Z_PK
+                    LEFT JOIN ZWAGROUPMEMBER
+                        ON ZWAMESSAGE.ZGROUPMEMBER = ZWAGROUPMEMBER.Z_PK
                  WHERE 1=1
                     {f'AND ZWAMESSAGE.ZMESSAGEDATE {filter_date}' if filter_date is not None else ''}
-                    {get_chat_condition(filter_chat[0], True, "ZCONTACTJID")}
-                    {get_chat_condition(filter_chat[1], False, "ZCONTACTJID")};""")
+                    {get_chat_condition(filter_chat[0], True, ["ZCONTACTJID", "ZMEMBERJID"], "ZGROUPINFO", "ios")}
+                    {get_chat_condition(filter_chat[1], False, ["ZCONTACTJID", "ZMEMBERJID"], "ZGROUPINFO", "ios")};""")
     contents = c.fetchall()
     total_row_number = len(contents)
     print(f"\nProcessing vCards...(0/{total_row_number})", end="\r")

Whatsapp_Chat_Exporter/ios_media_handler.py

@@ -3,55 +3,59 @@
 import shutil
 import sqlite3
 import os
-import time
 import getpass
-import threading
 from Whatsapp_Chat_Exporter.utility import WhatsAppIdentifier
+from Whatsapp_Chat_Exporter.bplist import BPListReader
 try:
     from iphone_backup_decrypt import EncryptedBackup, RelativePath
-    from iphone_backup_decrypt import FailedToDecryptError
 except ModuleNotFoundError:
     support_encrypted = False
 else:
     support_encrypted = True
 
 
-def extract_encrypted(base_dir, password, identifiers, bplist_reader=None):
-    backup = EncryptedBackup(backup_directory=base_dir, passphrase=password, cleanup=False, check_same_thread=False)
-    print("Decrypting WhatsApp database...", end="")
+def extract_encrypted(base_dir, password, identifiers, decrypt_chunk_size):
+    print("Trying to decrypt the iOS backup...", end="")
+    backup = EncryptedBackup(
+        backup_directory=base_dir,
+        passphrase=password,
+        cleanup=False,
+        check_same_thread=False,
+        decrypt_chunk_size=decrypt_chunk_size
+    )
+    print("Done\nDecrypting WhatsApp database...", end="")
     try:
         backup.extract_file(
             relative_path=RelativePath.WHATSAPP_MESSAGES,
-            domain=identifiers.DOMAIN,
+            domain_like=identifiers.DOMAIN,
             output_filename=identifiers.MESSAGE
         )
         backup.extract_file(
             relative_path=RelativePath.WHATSAPP_CONTACTS,
-            domain=identifiers.DOMAIN,
+            domain_like=identifiers.DOMAIN,
             output_filename=identifiers.CONTACT
         )
-    except FailedToDecryptError:
+    except ValueError:
         print("Failed to decrypt backup: incorrect password?")
-        exit()
+        exit(7)
+    except FileNotFoundError:
+        print("Essential WhatsApp files are missing from the iOS backup.")
+        exit(6)
     else:
         print("Done")
-    extract_thread = threading.Thread(
-        target=backup.extract_files_by_domain,
-        args=(identifiers.DOMAIN, identifiers.DOMAIN, bplist_reader)
+
+    def extract_progress_handler(file_id, domain, relative_path, n, total_files):
+        if n % 100 == 0:
+            print(f"Decrypting and extracting files...({n}/{total_files})", end="\r")   
+        return True
+
+    backup.extract_files(
+        domain_like=identifiers.DOMAIN,
+        output_folder=identifiers.DOMAIN,
+        preserve_folders=True,
+        filter_callback=extract_progress_handler
     )
-    extract_thread.daemon = True
-    extract_thread.start()
-    dot = 0
-    while extract_thread.is_alive():
-        print(f"Decrypting and extracting files{'.' * dot}{' ' * (3 - dot)}", end="\r")
-        if dot < 3:
-            dot += 1
-            time.sleep(0.5)
-        else:
-            dot = 0
-            time.sleep(0.4)
-    print(f"All required files decrypted and extracted.", end="\n")
-    extract_thread.handled = True
+    print(f"All required files are decrypted and extracted.          ", end="\n")
     return backup
 
 
@@ -70,10 +74,7 @@ def is_encrypted(base_dir):
             return False
 
 
-def extract_media(base_dir, identifiers, preserve_timestamp=False):
-    if preserve_timestamp:
-        from Whatsapp_Chat_Exporter.bplist import BPListReader 
-        preserve_timestamp = BPListReader
+def extract_media(base_dir, identifiers, decrypt_chunk_size):
     if is_encrypted(base_dir):
         if not support_encrypted:
             print("You don't have the dependencies to handle encrypted backup.")
@@ -82,7 +83,7 @@ def extract_media(base_dir, identifiers, preserve_timestamp=False):
             return False
         print("Encryption detected on the backup!")
         password = getpass.getpass("Enter the password for the backup:")
-        extract_encrypted(base_dir, password, identifiers, preserve_timestamp)
+        extract_encrypted(base_dir, password, identifiers, decrypt_chunk_size)
     else:
         wts_db = os.path.join(base_dir, identifiers.MESSAGE[:2], identifiers.MESSAGE)
         contact_db = os.path.join(base_dir, identifiers.CONTACT[:2], identifiers.CONTACT)
@@ -95,8 +96,7 @@ def extract_media(base_dir, identifiers, preserve_timestamp=False):
         else:
             shutil.copyfile(wts_db, identifiers.MESSAGE)
         if not os.path.isfile(contact_db):
-            print("Contact database not found.")
-            exit()
+            print("Contact database not found. Skipping...")
         else:
             shutil.copyfile(contact_db, identifiers.CONTACT)
         _wts_id = identifiers.DOMAIN
@@ -136,11 +136,10 @@ def extract_media(base_dir, identifiers, preserve_timestamp=False):
                         pass
                 elif flags == 1:
                     shutil.copyfile(os.path.join(base_dir, folder, hashes), destination)
-                    if preserve_timestamp:
-                        metadata = BPListReader(row["metadata"]).parse()
-                        creation = metadata["$objects"][1]["Birth"]
-                        modification = metadata["$objects"][1]["LastModified"]
-                        os.utime(destination, (modification, modification))
+                    metadata = BPListReader(row["metadata"]).parse()
+                    creation = metadata["$objects"][1]["Birth"]
+                    modification = metadata["$objects"][1]["LastModified"]
+                    os.utime(destination, (modification, modification))
                 if row["_index"] % 100 == 0:
                     print(f"Extracting WhatsApp files...({row['_index']}/{total_row_number})", end="\r")
                 row = c.fetchone()

Whatsapp_Chat_Exporter/utility.py

@@ -1,11 +1,12 @@
 import jinja2
 import json
 import os
-from bleach import clean as sanitize
-from markupsafe import Markup
 import unicodedata
 import re
-from datetime import datetime
+import math
+from bleach import clean as sanitize
+from markupsafe import Markup
+from datetime import datetime, timedelta
 from enum import IntEnum
 from Whatsapp_Chat_Exporter.data_model import ChatStore
 try:
@@ -23,6 +24,60 @@ MAX_SIZE = 4 * 1024 * 1024  # Default 4MB
 ROW_SIZE = 0x3D0
 
 
+def convert_time_unit(time_second: int):
+    time = str(timedelta(seconds=time_second))
+    if "day" not in time:
+        if time_second < 1:
+            time = "less than a second"
+        elif time_second == 1:
+            time = "a second"
+        elif time_second < 60:
+            time = time[5:][1 if time_second < 10 else 0:] + " seconds"
+        elif time_second == 60:
+            time = "a minute"
+        elif time_second < 3600:
+            time = time[2:] + " minutes"
+        elif time_second == 3600:
+            time = "an hour"
+        else:
+            time += " hour"
+    return time
+
+
+def bytes_to_readable(size_bytes: int):
+    """From https://stackoverflow.com/a/14822210/9478891 
+    Authors: james-sapam & other contributors
+    Licensed under CC BY-SA 3.0
+    See git commit logs for changes, if any.
+    """
+    if size_bytes == 0:
+       return "0B"
+    size_name = ("B", "KB", "MB", "GB", "TB", "PB", "EB", "ZB", "YB")
+    i = int(math.floor(math.log(size_bytes, 1024)))
+    p = math.pow(1024, i)
+    s = round(size_bytes / p, 2)
+    return "%s %s" % (s, size_name[i])
+
+
+def readable_to_bytes(size_str: str):
+    SIZE_UNITS = {
+        'B': 1,
+        'KB': 1024,
+        'MB': 1024**2,
+        'GB': 1024**3,
+        'TB': 1024**4,
+        'PB': 1024**5,
+        'EB': 1024**6,
+        'ZB': 1024**7, 
+        'YB': 1024**8
+    }
+    size_str = size_str.upper().strip()
+    number, unit = size_str[:-2].strip(), size_str[-2:].strip()
+    if unit not in SIZE_UNITS or not number.isnumeric():
+        raise ValueError("Invalid input for size_str. Example: 1024GB")
+    return int(number) * SIZE_UNITS[unit]
+
+
 def sanitize_except(html):
     return Markup(sanitize(html, tags=["br"]))
 
@@ -93,6 +148,7 @@ def rendering(
                 w3css=w3css,
                 next=next,
                 status=chat.status,
+                media_base=chat.media_base
             )
         )
 
@@ -138,11 +194,15 @@ def import_from_json(json_file, data):
         print(f"Importing chats from JSON...({index + 1}/{total_row_number})", end="\r")
 
 
+def sanitize_filename(file_name: str):
+    return "".join(x for x in file_name if x.isalnum() or x in "- ")
+
+
 def get_file_name(contact: str, chat: ChatStore):
     if "@" not in contact and contact not in ("000000000000000", "000000000000001", "ExportedChat"):
         raise ValueError("Unexpected contact format: " + contact)
     phone_number = contact.split('@')[0]
-    if "-" in contact:
+    if "-" in contact and chat.name is not None:
         file_name = ""
     else:
         file_name = phone_number
@@ -150,23 +210,45 @@ def get_file_name(contact: str, chat: ChatStore):
     if chat.name is not None:
         if file_name != "":
             file_name += "-"
-        file_name += chat.name.replace("/", "-")
+        file_name += chat.name.replace("/", "-").replace("\\", "-")
         name = chat.name
     else:
         name = phone_number
 
-    return "".join(x for x in file_name if x.isalnum() or x in "- "), name
+    return sanitize_filename(file_name), name
 
 
-def get_chat_condition(filter, include, column):
+def get_chat_condition(filter, include, columns, jid=None, platform=None):
     if filter is not None:
-        if include:
-            return f'''AND ({' OR '.join(f"{column} LIKE '%{chat}%'" for chat in filter)})'''
-        else:
-            return f'''AND ({' AND '.join(f"{column} NOT LIKE '%{chat}%'" for chat in filter)})'''
+        conditions = []
+        if len(columns) < 2 and jid is not None:
+            raise ValueError("There must be at least two elements in argument columns if jid is not None")
+        if jid is not None:
+            if platform == "android":
+                is_group = f"{jid}.type == 1"
+            elif platform == "ios":
+                is_group = f"{jid} IS NOT NULL"
+            else:
+                raise ValueError("Only android and ios are supported for argument platform if jid is not None")
+        for index, chat in enumerate(filter):
+            if include:
+                conditions.append(f"{' OR' if index > 0 else ''} {columns[0]} LIKE '%{chat}%'")
+                if len(columns) > 1:
+                    conditions.append(f" OR ({columns[1]} LIKE '%{chat}%' AND {is_group})")
+            else:
+                conditions.append(f"{' AND' if index > 0 else ''} {columns[0]} NOT LIKE '%{chat}%'")
+                if len(columns) > 1:
+                    conditions.append(f" AND ({columns[1]} NOT LIKE '%{chat}%' AND {is_group})")
+        return f"AND ({' '.join(conditions)})"
     else:
         return ""
 
+def _is_message_empty(message):
+    return (message.data is None or message.data == "") and not message.media
+
+def chat_is_empty(chat: ChatStore):
+    return len(chat.messages) == 0 or all(_is_message_empty(message) for message in chat.messages.values())
+
 
 # Android Specific
 CRYPT14_OFFSETS = (

Whatsapp_Chat_Exporter/vcards_contacts.py

@@ -0,0 +1,82 @@
+import vobject
+from typing import List, TypedDict
+
+
+class ExportedContactNumbers(TypedDict):
+    full_name: str
+    numbers: List[str]
+
+
+class ContactsFromVCards:
+    def __init__(self) -> None:
+        self.contact_mapping = []
+
+    def is_empty(self):
+        return self.contact_mapping == []
+
+    def load_vcf_file(self, vcf_file_path: str, default_country_code: str):
+        self.contact_mapping = read_vcards_file(vcf_file_path, default_country_code)
+
+    def enrich_from_vcards(self, chats):
+        for number, name in self.contact_mapping:
+            # short number must be a bad contact, lets skip it
+            if len(number) <= 5:
+                continue
+
+            for chat in filter_chats_by_prefix(chats, number).values():
+                if not hasattr(chat, 'name') or (hasattr(chat, 'name') and chat.name is None):
+                    setattr(chat, 'name', name)
+
+
+def read_vcards_file(vcf_file_path, default_country_code: str):
+    contacts = []
+    with open(vcf_file_path, mode="r", encoding="utf-8") as f:
+        reader = vobject.readComponents(f)
+        for row in reader:
+            if hasattr(row, 'fn'):
+                name = str(row.fn.value)
+            elif hasattr(row, 'n'):
+                name = str(row.n.value)
+            else:
+                name = None
+            if not hasattr(row, 'tel') or name is None:
+                continue
+            contact: ExportedContactNumbers = {
+                "full_name": name,
+                "numbers": list(map(lambda tel: tel.value, row.tel_list)),
+            }
+            contacts.append(contact)
+
+    return map_number_to_name(contacts, default_country_code)
+
+
+def filter_chats_by_prefix(chats, prefix: str):
+    return {k: v for k, v in chats.items() if k.startswith(prefix)}
+
+
+def map_number_to_name(contacts, default_country_code: str):
+    mapping = []
+    for contact in contacts:
+        for index, num in enumerate(contact['numbers']):
+            normalized = normalize_number(num, default_country_code)
+            if len(contact['numbers']) > 1:
+                name = f"{contact['full_name']} ({index+1})"
+            else:
+                name = contact['full_name']
+            mapping.append((normalized, name))
+    return mapping
+
+
+def normalize_number(number: str, country_code: str):
+    # Clean the number
+    number = ''.join(c for c in number if c.isdigit() or c == "+")
+
+    # A number that starts with a + or 00 means it already have a country code
+    for starting_char in ('+', "00"):
+        if number.startswith(starting_char):
+            return number[len(starting_char):]
+
+    # leading zero should be removed
+    if starting_char == '0':
+        number = number[1:]
+    return country_code + number  # fall back

Whatsapp_Chat_Exporter/vcards_contacts_test.py

@@ -0,0 +1,20 @@
+# from contacts_names_from_vcards import readVCardsFile
+
+from Whatsapp_Chat_Exporter.vcards_contacts import normalize_number, read_vcards_file
+
+
+def test_readVCardsFile():
+    assert len(read_vcards_file("contacts.vcf", "973")) > 0
+
+def test_create_number_to_name_dicts():
+    pass
+
+def test_fuzzy_match_numbers():
+    pass
+
+def test_normalize_number():
+    assert normalize_number('0531234567', '1') == '1531234567'
+    assert normalize_number('001531234567', '2') == '1531234567'
+    assert normalize_number('+1531234567', '34') == '1531234567'
+    assert normalize_number('053(123)4567', '34') == '34531234567'
+    assert normalize_number('0531-234-567', '58') == '58531234567'

Whatsapp_Chat_Exporter/whatsapp.html

@@ -87,6 +87,7 @@
 			max-height: 100px !important;
 		}
 		</style>
+	<base href="{{ media_base }}" target="_blank">
 	</head>
 	<body>
 		<header class="w3-center w3-top">
@@ -137,13 +138,18 @@
 									<p>{{ msg.data or 'Not supported WhatsApp internal message' }}</p>
 									{% endif %}
 								</div>
+								{% if msg.caption is not none %}
+								<div class="w3-container">
+									{{ msg.caption | urlize(none, true, '_blank') }}
+								</div>
+								{% endif %}
 							{% else %}
 								{% if msg.media == false %}
 								{{ msg.data | sanitize_except() | urlize(none, true, '_blank') }}
 								{% else %}
 									{% if "image/" in msg.mime %}
 									<a href="{{ msg.data }}">
-										<img src="{{ msg.thumb if msg.thumb is not none else msg.data }}" {{ 'class="sticker"' | safe if msg.sticker }} />
+										<img src="{{ msg.thumb if msg.thumb is not none else msg.data }}" {{ 'class="sticker"' | safe if msg.sticker }} loading="lazy"/>
 									</a>
 									{% elif "audio/" in msg.mime %}
 									<audio controls="controls" autobuffer="autobuffer">
@@ -161,17 +167,18 @@
 										{% filter escape %}{{ msg.data }}{% endfilter %}
 									{% endif %}
 									{% if msg.caption is not none %}
-									<br>
-									{{ msg.caption | urlize(none, true, '_blank') }}
+									<div class="w3-container">
+										{{ msg.caption | urlize(none, true, '_blank') }}
+									</div>
 									{% endif %}
 								{% endif %}
-								{% endif %}
+							{% endif %}
 							</div>
 						</div>
 						{% if not no_avatar and my_avatar is not none %}
 						<div class="w3-col m2 l2 pad-left-10">
 							<a href="{{ my_avatar }}">
-								<img src="{{ my_avatar }}" onerror="this.style.display='none'" class="avatar">
+								<img src="{{ my_avatar }}" onerror="this.style.display='none'" class="avatar" loading="lazy">
 							</a>
 						</div>
 						{% endif %}
@@ -191,9 +198,9 @@
 						{% if not no_avatar %}
 						<div class="w3-col m2 l2">
 								{% if their_avatar is not none %}
-								<a href="{{ their_avatar }}"><img src="{{ their_avatar_thumb or '' }}" onerror="this.style.display='none'" class="avatar"></a>
+								<a href="{{ their_avatar }}"><img src="{{ their_avatar_thumb or '' }}" onerror="this.style.display='none'" class="avatar" loading="lazy"></a>
 								{% else %}
-								<img src="{{ their_avatar_thumb or '' }}" onerror="this.style.display='none'" class="avatar">
+								<img src="{{ their_avatar_thumb or '' }}" onerror="this.style.display='none'" class="avatar" loading="lazy">
 								{% endif %}
 						</div>
 						<div class="w3-col m10 l10">
@@ -221,13 +228,18 @@
 									<p>{{ msg.data or 'Not supported WhatsApp internal message' }}</p>
 									{% endif %}
 								</div>
+								{% if msg.caption is not none %}
+								<div class="w3-container">
+								{{ msg.caption | urlize(none, true, '_blank') }}
+								</div>
+								{% endif %}
 							{% else %}
 								{% if msg.media == false %}
 								{{ msg.data | sanitize_except() | urlize(none, true, '_blank') }}
 								{% else %}
 									{% if "image/" in msg.mime %}
 									<a href="{{ msg.data }}">
-										<img src="{{ msg.thumb if msg.thumb is not none else msg.data }}" {{ 'class="sticker"' | safe if msg.sticker }} />
+										<img src="{{ msg.thumb if msg.thumb is not none else msg.data }}" {{ 'class="sticker"' | safe if msg.sticker }} loading="lazy"/>
 									</a>
 									{% elif "audio/" in msg.mime %}
 									<audio controls="controls" autobuffer="autobuffer">
@@ -245,8 +257,9 @@
 										{% filter escape %}{{ msg.data }}{% endfilter %}
 									{% endif %}
 									{% if msg.caption is not none %}
-									<br>
-									{{ msg.caption | urlize(none, true, '_blank') }}
+									<div class="w3-container">
+										{{ msg.caption | urlize(none, true, '_blank') }}
+									</div>
 									{% endif %}
 								{% endif %}
 							{% endif %}

setup.py

@@ -55,9 +55,10 @@ setuptools.setup(
         'crypt12':  ["pycryptodome"],
         'crypt14':  ["pycryptodome"],
         'crypt15':  ["pycryptodome", "javaobj-py3"],
-        'all': ["pycryptodome", "javaobj-py3"],
-        'everything': ["pycryptodome", "javaobj-py3"],
-        'backup': ["pycryptodome", "javaobj-py3"]
+        'all': ["pycryptodome", "javaobj-py3", "vobject"],
+        'everything': ["pycryptodome", "javaobj-py3", "vobject"],
+        'backup': ["pycryptodome", "javaobj-py3"],
+        'vcards': ["vobject", "pycryptodome", "javaobj-py3"],
     },
     entry_points={
         "console_scripts": [