Alexhuszagh · October 4, 2016 19:29
diff --git a/samsung_to_silence.py b/samsung_to_silence.py
 #!/usr/bin/env python
 '''
    samsung_to_silence.py
    ---------------------

    Convert Samsung SMS plaintext backups (XML) to the silence format.

    :warning: This has only been tested with the Samsung Galaxy S6, other
    systems may differ.

    :author: Alex Huszagh
    :license: MIT

    :example:
        python samsung_to_silence.py --samsung sms.xml
 '''

 from __future__ import print_function
 import argparse
 from collections import OrderedDict
 import xml.etree.cElementTree as cET

 try:
    from urllib.parse import unquote
 except ImportError:
    # Python 2.x
    from urllib2 import unquote

 # ARGS
 # ----

 PARSER = argparse.ArgumentParser()
 PARSER.add_argument("--samsung",
    type=str,
    required=True,
    help="Path to Samsung SMS XML backup.")
 PARSER.add_argument("--silence",
    type=str,
    default="SilencePlaintextBackup.xml",
    help="Path to SIlence export.")
 ARGS = PARSER.parse_args()


 # CONSTANTS
 # ---------

 SILENCE_TEMPLATE = OrderedDict([
    ("protocol", "0"),
    ("address", "null"),
    ("date", "null"),
    ("type", "null"),
    ("subject", "null"),
    ("body", "null"),
    ("toa", "null"),
    ("sc_toa", "null"),
    ("service_center", "null"),
    ("read", "null"),
    ("status", "-1"),
 ])


 # OBJECTS
 # -------


 class SamsungParser(object):
    '''
    XML parser for the Samsung plaintext backup format.

    Example:
        <file ver="2">
        <thread n="94">
        <message type="SMS">
            <address>18001234567</address>
            <body>Message</body>
            <date>1475094117922</date>
            <read>1</read>
            <type>1</type>
            <locked>0</locked>
        </message>
        </thread>

    Args:
        path                Path to XML file
    '''

    def __init__(self, path):
        super(SamsungParser, self).__init__()

        self.parser = cET.iterparse(path, events=('end',))

    def __iter__(self):
        '''High-level wrapper for XML parser'''

        try:
            return self.parse()
        except cET.ParseError:
            raise IOError("Non-XML file found")
        except TypeError:
            raise IOError("XML file not open")

    def parse_message(self, element):
        '''Extract next SMS from XML backup'''

        message = SILENCE_TEMPLATE.copy()
        for name in ["address", "body", "date", "read", "type", "locked"]:
            message[name] = self.read_content(element, name)

        return message

    def parse(self):
        '''Generator for each SMS item'''

        for event, element in self.parser:
            if element.tag == "message" and element.attrib["type"] == "SMS":
                yield self.parse_message(element)

    def unescape(self, message):
        '''Unescape URL-encoded text from Samsung'''

        items = message.split("+")
        unescaped = []
        for item in items:
            unescaped.append(unquote(item))

        return ' '.join(unescaped)

    def read_content(self, element, name):
        '''Read context from child XML node with given name'''

        text = next(element.find(name).itertext())
        unescaped = self.unescape(text)
        return unescaped.decode("UTF-8")


 class SilenceWriter(object):
    '''
    XML writer for the Silence plaintext backup format.

    Example:
        <sms address="18001234567" body="Message" date="1475094117922" locked="0" protocol="0" read="1" sc_toa="null" service_center="null" status="-1" subject="null" toa="null" type="1" />

    Args:
        path                Path to output XML file
    '''

    declaration = b'<?xml version="1.0" encoding="UTF-8" standalone="yes"?>\n'
    comment = b'<!-- File Created By SamsungToSilence -->\n'

    def __init__(self, path):
        super(SilenceWriter, self).__init__()

        self.path = path
        self.root = cET.Element("smes")
        self.tree = cET.ElementTree(self.root)
        self.counter = 0

    def write(self, message):
        '''Write message to XML'''

        self.counter += 1
        element = cET.SubElement(self.root, "sms")
        for key, value in message.items():
            element.set(key, value)

    def close(self):
        '''Close tree and write out to file'''

        self.root.set("count", str(self.counter))
        with open(self.path, 'wb') as file:
            file.write(self.declaration)
            file.write(self.comment)
            self.tree.write(file, encoding="UTF-8", xml_declaration=False)


 # FUNCTIONS
 # ---------


 def main():
    '''Executable entry point'''

    parser = SamsungParser(ARGS.samsung)
    writer = SilenceWriter(ARGS.silence)
    for message in parser:
        writer.write(message)
    writer.close()


 if __name__ == '__main__':
    main()
	#!/usr/bin/env python
	'''
	samsung_to_silence.py
	---------------------

	Convert Samsung SMS plaintext backups (XML) to the silence format.

	:warning: This has only been tested with the Samsung Galaxy S6, other
	systems may differ.

	:author: Alex Huszagh
	:license: MIT

	:example:
	python samsung_to_silence.py --samsung sms.xml
	'''

	from __future__ import print_function
	import argparse
	from collections import OrderedDict
	import xml.etree.cElementTree as cET

	try:
	from urllib.parse import unquote
	except ImportError:
	# Python 2.x
	from urllib2 import unquote

	# ARGS
	# ----

	PARSER = argparse.ArgumentParser()
	PARSER.add_argument("--samsung",
	type=str,
	required=True,
	help="Path to Samsung SMS XML backup.")
	PARSER.add_argument("--silence",
	type=str,
	default="SilencePlaintextBackup.xml",
	help="Path to SIlence export.")
	ARGS = PARSER.parse_args()


	# CONSTANTS
	# ---------

	SILENCE_TEMPLATE = OrderedDict([
	("protocol", "0"),
	("address", "null"),
	("date", "null"),
	("type", "null"),
	("subject", "null"),
	("body", "null"),
	("toa", "null"),
	("sc_toa", "null"),
	("service_center", "null"),
	("read", "null"),
	("status", "-1"),
	])


	# OBJECTS
	# -------


	class SamsungParser(object):
	'''
	XML parser for the Samsung plaintext backup format.

	Example:
	<file ver="2">
	<thread n="94">
	<message type="SMS">
	<address>18001234567</address>
	<body>Message</body>
	<date>1475094117922</date>
	<read>1</read>
	<type>1</type>
	<locked>0</locked>
	</message>
	</thread>

	Args:
	path Path to XML file
	'''

	def __init__(self, path):
	super(SamsungParser, self).__init__()

	self.parser = cET.iterparse(path, events=('end',))

	def __iter__(self):
	'''High-level wrapper for XML parser'''

	try:
	return self.parse()
	except cET.ParseError:
	raise IOError("Non-XML file found")
	except TypeError:
	raise IOError("XML file not open")

	def parse_message(self, element):
	'''Extract next SMS from XML backup'''

	message = SILENCE_TEMPLATE.copy()
	for name in ["address", "body", "date", "read", "type", "locked"]:
	message[name] = self.read_content(element, name)

	return message

	def parse(self):
	'''Generator for each SMS item'''

	for event, element in self.parser:
	if element.tag == "message" and element.attrib["type"] == "SMS":
	yield self.parse_message(element)

	def unescape(self, message):
	'''Unescape URL-encoded text from Samsung'''

	items = message.split("+")
	unescaped = []
	for item in items:
	unescaped.append(unquote(item))

	return ' '.join(unescaped)

	def read_content(self, element, name):
	'''Read context from child XML node with given name'''

	text = next(element.find(name).itertext())
	unescaped = self.unescape(text)
	return unescaped.decode("UTF-8")


	class SilenceWriter(object):
	'''
	XML writer for the Silence plaintext backup format.

	Example:
	<sms address="18001234567" body="Message" date="1475094117922" locked="0" protocol="0" read="1" sc_toa="null" service_center="null" status="-1" subject="null" toa="null" type="1" />

	Args:
	path Path to output XML file
	'''

	declaration = b'<?xml version="1.0" encoding="UTF-8" standalone="yes"?>\n'
	comment = b'<!-- File Created By SamsungToSilence -->\n'

	def __init__(self, path):
	super(SilenceWriter, self).__init__()

	self.path = path
	self.root = cET.Element("smes")
	self.tree = cET.ElementTree(self.root)
	self.counter = 0

	def write(self, message):
	'''Write message to XML'''

	self.counter += 1
	element = cET.SubElement(self.root, "sms")
	for key, value in message.items():
	element.set(key, value)

	def close(self):
	'''Close tree and write out to file'''

	self.root.set("count", str(self.counter))
	with open(self.path, 'wb') as file:
	file.write(self.declaration)
	file.write(self.comment)
	self.tree.write(file, encoding="UTF-8", xml_declaration=False)


	# FUNCTIONS
	# ---------


	def main():
	'''Executable entry point'''

	parser = SamsungParser(ARGS.samsung)
	writer = SilenceWriter(ARGS.silence)
	for message in parser:
	writer.write(message)
	writer.close()


	if __name__ == '__main__':
	main()