calvin.erfmann/WhatsApp-Chat-Exporter
1
0
Fork 0
mirror of https://github.com/KnugiHK/WhatsApp-Chat-Exporter.git synced 2026-04-27 08:21:33 +00:00
Code Issues Packages Projects Releases Wiki Activity

PEP8

Browse Source
This commit is contained in:
KnugiHK
2025-05-05 17:13:43 +08:00
parent 3220ed2d3f
commit a58dd78be8
4 changed files with 224 additions and 171 deletions
Download Patch File Download Diff File Expand all files Collapse all files

203
Whatsapp_Chat_Exporter/__main__.py
View File

@@ -34,12 +34,12 @@ def setup_argument_parser() -> ArgumentParser:
"""Set up and return the argument parser with all options."""
parser = ArgumentParser(
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.',
'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: {importlib.metadata.version("whatsapp_chat_exporter")} Licensed with MIT. See '
'https://wts.knugi.dev/docs?dest=osl for all open source licenses.'
'https://wts.knugi.dev/docs?dest=osl for all open source licenses.'
)
# Device type arguments
device_group = parser.add_argument_group('Device Type')
device_group.add_argument(
@@ -54,7 +54,7 @@ def setup_argument_parser() -> ArgumentParser:
"-e", "--exported", dest="exported", default=None,
help="Define the target as exported chat file and specify the path to the file"
)
# Input file paths
input_group = parser.add_argument_group('Input Files')
input_group.add_argument(
@@ -86,7 +86,7 @@ def setup_argument_parser() -> ArgumentParser:
"--wab", "--wa-backup", dest="wab", default=None,
help="Path to contact database in crypt15 format"
)
# Output options
output_group = parser.add_argument_group('Output Options')
output_group.add_argument(
@@ -109,7 +109,7 @@ def setup_argument_parser() -> ArgumentParser:
"--size", "--output-size", "--split", dest="size", nargs='?', const=0, default=None,
help="Maximum (rough) size of a single output file in bytes, 0 for auto"
)
# JSON formatting options
json_group = parser.add_argument_group('JSON Options')
json_group.add_argument(
@@ -128,7 +128,7 @@ def setup_argument_parser() -> ArgumentParser:
"--import", dest="import_json", default=False, action='store_true',
help="Import JSON file and convert to HTML output"
)
# HTML options
html_group = parser.add_argument_group('HTML Options')
html_group.add_argument(
@@ -155,7 +155,7 @@ def setup_argument_parser() -> ArgumentParser:
"--headline", dest="headline", default="Chat history with ??",
help="The custom headline for the HTML output. Use '??' as a placeholder for the chat name"
)
# Media handling
media_group = parser.add_argument_group('Media Handling')
media_group.add_argument(
@@ -166,7 +166,7 @@ def setup_argument_parser() -> ArgumentParser:
"--create-separated-media", dest="separate_media", default=False, action='store_true',
help="Create a copy of the media seperated per chat in <MEDIA>/separated/ directory"
)
# Filtering options
filter_group = parser.add_argument_group('Filtering Options')
filter_group.add_argument(
@@ -195,7 +195,7 @@ def setup_argument_parser() -> ArgumentParser:
"Setting this flag will cause the exporter to render those. "
"This is useful if chat(s) are missing from the output")
)
# Contact enrichment
contact_group = parser.add_argument_group('Contact Enrichment')
contact_group.add_argument(
@@ -219,7 +219,7 @@ def setup_argument_parser() -> ArgumentParser:
"The chats (JSON files only) and media from the source directory will be merged into the target directory. "
"No chat messages or media will be deleted from the target directory; only new chat messages and media will be added to it. "
"This enables chat messages and media to be deleted from the device to free up space, while ensuring they are preserved in the exported backups."
)
)
)
inc_merging_group.add_argument(
"--source-dir",
@@ -233,7 +233,7 @@ def setup_argument_parser() -> ArgumentParser:
default=None,
help="Sets the target directory. Used for performing incremental merges."
)
# Miscellaneous
misc_group = parser.add_argument_group('Miscellaneous')
misc_group.add_argument(
@@ -260,7 +260,7 @@ def setup_argument_parser() -> ArgumentParser:
"--max-bruteforce-worker", dest="max_bruteforce_worker", default=10, type=int,
help="Specify the maximum number of worker for bruteforce decryption."
)
return parser
@@ -272,52 +272,59 @@ def validate_args(parser: ArgumentParser, args) -> None:
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 and not args.text_format:
parser.error("You must either specify a JSON output file, text file output directory or enable HTML output.")
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, -a, -i, -e.")
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.incremental_merge and (args.source_dir is None or args.target_dir is None):
parser.error("You must specify both --source-dir and --target-dir for incremental merge.")
parser.error(
"You must specify both --source-dir and --target-dir for incremental merge.")
if args.android and args.business:
parser.error("WhatsApp Business is only available on iOS for now.")
if "??" not in args.headline:
parser.error("--headline must contain '??' for replacement.")
# JSON validation
if args.json_per_chat and args.json and (
(args.json.endswith(".json") and os.path.isfile(args.json)) or
(args.json.endswith(".json") and os.path.isfile(args.json)) or
(not args.json.endswith(".json") and os.path.isfile(args.json))
):
parser.error("When --per-chat is enabled, the destination of --json must be a directory.")
parser.error(
"When --per-chat is enabled, the destination of --json must be a directory.")
# vCards validation
if args.enrich_from_vcards is not None and args.default_country_code is None:
parser.error("When --enrich-from-vcards is provided, you must also set --default-country-code")
parser.error(
"When --enrich-from-vcards is provided, you must also set --default-country-code")
# Size validation
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)")
parser.error(
"The value for --split must be ended in pure bytes or with a proper unit (e.g., 1048576 or 1MB)")
# Date filter validation and processing
if args.filter_date is not None:
process_date_filter(parser, args)
# Crypt15 key validation
if args.key is None and args.backup is not None and args.backup.endswith("crypt15"):
args.key = getpass("Enter your encryption key: ")
# Theme validation
if args.whatsapp_theme:
args.template = "whatsapp_new.html"
# Chat filter validation
if args.filter_chat_include is not None and args.filter_chat_exclude is not None:
parser.error("Chat inclusion and exclusion filters cannot be used together.")
parser.error(
"Chat inclusion and exclusion filters cannot be used together.")
validate_chat_filters(parser, args.filter_chat_include)
validate_chat_filters(parser, args.filter_chat_exclude)
@@ -327,21 +334,24 @@ def validate_chat_filters(parser: ArgumentParser, chat_filter: Optional[List[str
if chat_filter is not None:
for chat in chat_filter:
if not chat.isnumeric():
parser.error("Enter a phone number in the chat filter. See https://wts.knugi.dev/docs?dest=chat")
parser.error(
"Enter a phone number in the chat filter. See https://wts.knugi.dev/docs?dest=chat")
def process_date_filter(parser: ArgumentParser, args) -> None:
"""Process and validate date filter arguments."""
if " - " in args.filter_date:
start, end = args.filter_date.split(" - ")
start = int(datetime.strptime(start, args.filter_date_format).timestamp())
start = int(datetime.strptime(
start, args.filter_date_format).timestamp())
end = int(datetime.strptime(end, args.filter_date_format).timestamp())
if start < 1009843200 or end < 1009843200:
parser.error("WhatsApp was first released in 2009...")
if start > end:
parser.error("The start date cannot be a moment after the end date.")
parser.error(
"The start date cannot be a moment after the end date.")
if args.android:
args.filter_date = f"BETWEEN {start}000 AND {end}000"
elif args.ios:
@@ -353,13 +363,15 @@ def process_date_filter(parser: ArgumentParser, args) -> None:
def process_single_date_filter(parser: ArgumentParser, args) -> None:
"""Process single date comparison filters."""
if len(args.filter_date) < 3:
parser.error("Unsupported date format. See https://wts.knugi.dev/docs?dest=date")
_timestamp = int(datetime.strptime(args.filter_date[2:], args.filter_date_format).timestamp())
parser.error(
"Unsupported date format. See https://wts.knugi.dev/docs?dest=date")
_timestamp = int(datetime.strptime(
args.filter_date[2:], args.filter_date_format).timestamp())
if _timestamp < 1009843200:
parser.error("WhatsApp was first released in 2009...")
if args.filter_date[:2] == "> ":
if args.android:
args.filter_date = f">= {_timestamp}000"
@@ -371,7 +383,8 @@ def process_single_date_filter(parser: ArgumentParser, args) -> None:
elif args.ios:
args.filter_date = f"<= {_timestamp - APPLE_TIME}"
else:
parser.error("Unsupported date format. See https://wts.knugi.dev/docs?dest=date")
parser.error(
"Unsupported date format. See https://wts.knugi.dev/docs?dest=date")
def setup_contact_store(args) -> Optional['ContactsFromVCards']:
@@ -385,7 +398,8 @@ def setup_contact_store(args) -> Optional['ContactsFromVCards']:
)
exit(1)
contact_store = ContactsFromVCards()
contact_store.load_vcf_file(args.enrich_from_vcards, args.default_country_code)
contact_store.load_vcf_file(
args.enrich_from_vcards, args.default_country_code)
return contact_store
return None
@@ -395,9 +409,9 @@ def decrypt_android_backup(args) -> int:
if args.key is None or args.backup is None:
print("You must specify the backup file with -b and a key with -k")
return 1
print("Decryption key specified, decrypting WhatsApp backup...")
# Determine crypt type
if "crypt12" in args.backup:
crypt = Crypt.CRYPT12
@@ -408,7 +422,7 @@ def decrypt_android_backup(args) -> int:
else:
print("Unknown backup format. The backup file must be crypt12, crypt14 or crypt15.")
return 1
# Get key
keyfile_stream = False
if not os.path.isfile(args.key) and all(char in string.hexdigits for char in args.key.replace(" ", "")):
@@ -416,10 +430,10 @@ def decrypt_android_backup(args) -> int:
else:
key = open(args.key, "rb")
keyfile_stream = True
# Read backup
db = open(args.backup, "rb").read()
# Process WAB if provided
error_wa = 0
if args.wab:
@@ -436,7 +450,7 @@ def decrypt_android_backup(args) -> int:
)
if isinstance(key, io.IOBase):
key.seek(0)
# Decrypt message database
error_message = android_crypt.decrypt_backup(
db,
@@ -448,7 +462,7 @@ def decrypt_android_backup(args) -> int:
keyfile_stream=keyfile_stream,
max_worker=args.max_bruteforce_worker
)
# Handle errors
if error_wa != 0:
return error_wa
@@ -473,7 +487,7 @@ def handle_decrypt_error(error: int) -> None:
def process_contacts(args, data: ChatCollection, contact_store=None) -> None:
"""Process contacts from the database."""
contact_db = args.wa if args.wa else "wa.db" if args.android else "ContactsV2.sqlite"
if os.path.isfile(contact_db):
with sqlite3.connect(contact_db) as db:
db.row_factory = sqlite3.Row
@@ -486,42 +500,42 @@ def process_contacts(args, data: ChatCollection, contact_store=None) -> None:
def process_messages(args, data: ChatCollection) -> None:
"""Process messages, media and vcards from the database."""
msg_db = args.db if args.db else "msgstore.db" if args.android else args.identifiers.MESSAGE
if not os.path.isfile(msg_db):
print(
"The message database does not exist. You may specify the path "
"to database file with option -d or check your provided path."
)
exit(6)
filter_chat = (args.filter_chat_include, args.filter_chat_exclude)
with sqlite3.connect(msg_db) as db:
db.row_factory = sqlite3.Row
# Process messages
if args.android:
message_handler = android_handler
else:
message_handler = ios_handler
message_handler.messages(
db, data, args.media, args.timezone_offset,
db, data, args.media, args.timezone_offset,
args.filter_date, filter_chat, args.filter_empty
)
# Process media
message_handler.media(
db, data, args.media, args.filter_date,
db, data, args.media, args.filter_date,
filter_chat, args.filter_empty, args.separate_media
)
# Process vcards
message_handler.vcard(
db, data, args.media, args.filter_date,
db, data, args.media, args.filter_date,
filter_chat, args.filter_empty
)
# Process calls
process_calls(args, db, data, filter_chat)
@@ -540,9 +554,10 @@ def handle_media_directory(args) -> None:
"""Handle media directory copying or moving."""
if os.path.isdir(args.media):
media_path = os.path.join(args.output, args.media)
if os.path.isdir(media_path):
print("\nWhatsApp directory already exists in output directory. Skipping...", end="\n")
print(
"\nWhatsApp directory already exists in output directory. Skipping...", end="\n")
else:
if args.move_media:
try:
@@ -563,7 +578,7 @@ def create_output_files(args, data: ChatCollection, contact_store=None) -> None:
# Enrich from vcards if available
if contact_store and not contact_store.is_empty():
contact_store.enrich_from_vcards(data)
android_handler.create_html(
data,
args.output,
@@ -575,12 +590,12 @@ def create_output_files(args, data: ChatCollection, contact_store=None) -> None:
args.whatsapp_theme,
args.headline
)
# Create text files if requested
if args.text_format:
print("Writing text file...")
android_handler.create_txt(data, args.text_format)
# Create JSON files if requested
if args.json and not args.import_json:
export_json(args, data, contact_store)
@@ -591,11 +606,11 @@ def export_json(args, data: ChatCollection, contact_store=None) -> None:
# Enrich from vcards if available
if contact_store and not contact_store.is_empty():
contact_store.enrich_from_vcards(data)
# Convert ChatStore objects to JSON
if isinstance(data.get(next(iter(data), None)), ChatStore):
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:
export_single_json(args, data)
@@ -619,11 +634,11 @@ def export_multiple_json(args, data: Dict) -> None:
"""Export data to multiple JSON files, one per chat."""
# Adjust output path if needed
json_path = args.json[:-5] if args.json.endswith(".json") else args.json
# Create directory if it doesn't exist
if not os.path.isdir(json_path):
os.makedirs(json_path, exist_ok=True)
# Export each chat
total = len(data.keys())
for index, jik in enumerate(data.keys()):
@@ -631,11 +646,11 @@ def export_multiple_json(args, data: Dict) -> None:
contact = data[jik]["name"].replace('/', '')
else:
contact = jik.replace('+', '')
with open(f"{json_path}/{sanitize_filename(contact)}.json", "w") as f:
file_content = json.dumps(
{jik: data[jik]},
ensure_ascii=not args.avoid_encoding_json,
{jik: data[jik]},
ensure_ascii=not args.avoid_encoding_json,
indent=args.pretty_print_json
)
f.write(file_content)
@@ -646,7 +661,7 @@ def export_multiple_json(args, data: Dict) -> None:
def process_exported_chat(args, data: ChatCollection) -> None:
"""Process an exported chat file."""
exported_handler.messages(args.exported, data, args.assume_first_as_me)
if not args.no_html:
android_handler.create_html(
data,
@@ -659,7 +674,7 @@ def process_exported_chat(args, data: ChatCollection) -> None:
args.whatsapp_theme,
args.headline
)
# Copy files to output directory
for file in glob.glob(r'*.*'):
shutil.copy(file, args.output)
@@ -670,23 +685,23 @@ def main():
# Set up and parse arguments
parser = setup_argument_parser()
args = parser.parse_args()
# Check for updates
if args.check_update:
exit(check_update())
# Validate arguments
validate_args(parser, args)
# Create output directory if it doesn't exist
os.makedirs(args.output, exist_ok=True)
# Initialize data collection
data = ChatCollection()
# Set up contact store for vCard enrichment if needed
contact_store = setup_contact_store(args)
if args.import_json:
# Import from JSON
import_from_json(args.json, data)
@@ -710,13 +725,13 @@ def main():
# Set default media path if not provided
if args.media is None:
args.media = "WhatsApp"
# Set default DB paths if not provided
if args.db is None:
args.db = "msgstore.db"
if args.wa is None:
args.wa = "wa.db"
# Decrypt backup if needed
if args.key is not None:
error = decrypt_android_backup(args)
@@ -729,24 +744,26 @@ def main():
else:
from Whatsapp_Chat_Exporter.utility import WhatsAppIdentifier as identifiers
args.identifiers = identifiers
# Set default media path if not provided
if args.media is None:
args.media = identifiers.DOMAIN
# Extract media from backup if needed
if args.backup is not None:
if not os.path.isdir(args.media):
ios_media_handler.extract_media(args.backup, identifiers, args.decrypt_chunk_size)
ios_media_handler.extract_media(
args.backup, identifiers, args.decrypt_chunk_size)
else:
print("WhatsApp directory already exists, skipping WhatsApp file extraction.")
print(
"WhatsApp directory already exists, skipping WhatsApp file extraction.")
# Set default DB paths if not provided
if args.db is None:
args.db = identifiers.MESSAGE
if args.wa is None:
args.wa = "ContactsV2.sqlite"
if args.incremental_merge:
incremental_merge(
args.source_dir,
@@ -756,16 +773,16 @@ def main():
args.avoid_encoding_json
)
print("Incremental merge completed successfully.")
else:
else:
# Process contacts
process_contacts(args, data, contact_store)
# Process messages, media, and calls
process_messages(args, data)
# Create output files
create_output_files(args, data, contact_store)
# Handle media directory
handle_media_directory(args)