mas01mc@354: #!/usr/bin/python
mas01mc@354: 
mas01mc@354: # Python code/library to query the audioDB via the SOAP web interface.
mas01mc@354: # by Malcolm Slaney, August/September 2008
mas01mc@354: # malcolm@ieee.org
mas01mc@354: 
mas01mc@354: import sys, socket
mas01mc@354: from xml.dom.minidom import parseString
mas01mc@354: 
mas01mc@354: global debug
mas01mc@354: debug = False
mas01mc@354: global dbName
mas01mc@354: dbName = 'tutorial.adb'
mas01mc@354: 
mas01mc@354: # From: http://www.informit.com/articles/article.aspx?p=686162&seqNum=2
mas01mc@354: #serverHost = 'research-hm3.corp.sk1.yahoo.com'
mas01mc@354: serverHost = 'localhost'
mas01mc@354: serverPort = 14475
mas01mc@354: 
mas01mc@354: # Start the server on serverHost with
mas01mc@354: #	./audioDB -s 14475
mas01mc@354: 
mas01mc@354: # Here are the templates used for the different kinds of queries.  We'll fill in the
mas01mc@354: # desired parameters are we go.
mas01mc@354: LIST_TEMPLATE = """
mas01mc@354: <?xml version="1.0" encoding="UTF-8"?>
mas01mc@354: <SOAP-ENV:Envelope
mas01mc@354:  xmlns:SOAP-ENV="http://schemas.xmlsoap.org/soap/envelope/"
mas01mc@354:  xmlns:SOAP-ENC="http://schemas.xmlsoap.org/soap/encoding/"
mas01mc@354:  xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
mas01mc@354:  xmlns:xsd="http://www.w3.org/2001/XMLSchema"
mas01mc@354:  xmlns:adb="http://tempuri.org/adb.xsd">
mas01mc@354:  <SOAP-ENV:Body>
mas01mc@354:   <adb:liszt>
mas01mc@354:    <dbName>%s</dbName>
mas01mc@354:    <lisztOffset>0</lisztOffset>
mas01mc@354:    <lisztLength>100000</lisztLength>
mas01mc@354:   </adb:liszt>
mas01mc@354:  </SOAP-ENV:Body>
mas01mc@354: </SOAP-ENV:Envelope>
mas01mc@354: """
mas01mc@354: 
mas01mc@354: SEQUENCE_TEMPLATE = """
mas01mc@354: <?xml version="1.0" encoding="UTF-8"?>
mas01mc@354: <SOAP-ENV:Envelope
mas01mc@354: xmlns:SOAP-ENV="http://schemas.xmlsoap.org/soap/envelope/"
mas01mc@354: xmlns:SOAP-ENC="http://schemas.xmlsoap.org/soap/encoding/"
mas01mc@354: xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
mas01mc@354: xmlns:xsd="http://www.w3.org/2001/XMLSchema"
mas01mc@354: xmlns:adb="http://tempuri.org/adb.xsd">
mas01mc@354: <SOAP-ENV:Body>
mas01mc@354:  <adb:sequenceQueryByKey>
mas01mc@354:   <dbName>%s</dbName>
mas01mc@354:   <featureFileName>%s</featureFileName>
mas01mc@354:   <queryType>%s</queryType>
mas01mc@354:   <trackFileName></trackFileName>
mas01mc@354:   <timesFileName></timesFileName>
mas01mc@354:   <queryPoint>%s</queryPoint>
mas01mc@354:   <pointNN>%s</pointNN>			<!-- Number of nearest neighbors to retrieve -->
mas01mc@354:   <trackNN>%s</trackNN>			<!-- Number of tracks to retrieve -->
mas01mc@354:   <sequenceLength>%s</sequenceLength>	<!-- Number of frames in a shingle -->
mas01mc@354:   <radius>%s</radius>			<!-- Distance radius to search -->
mas01mc@354:   <absolute-threshold>-4.0</absolute-threshold>
mas01mc@354:   <usingQueryPoint>1</usingQueryPoint>
mas01mc@354:   <lsh-exact>0</lsh-exact>
mas01mc@354:  </adb:sequenceQueryByKey>
mas01mc@354: </SOAP-ENV:Body>
mas01mc@354: </SOAP-ENV:Envelope>
mas01mc@354: """
mas01mc@354: 
mas01mc@354: STATUS_TEMPLATE = """
mas01mc@354: <?xml version="1.0" encoding="UTF-8"?>
mas01mc@354: <SOAP-ENV:Envelope
mas01mc@354:  xmlns:SOAP-ENV="http://schemas.xmlsoap.org/soap/envelope/"
mas01mc@354:  xmlns:SOAP-ENC="http://schemas.xmlsoap.org/soap/encoding/"
mas01mc@354:  xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
mas01mc@354:  xmlns:xsd="http://www.w3.org/2001/XMLSchema"
mas01mc@354:  xmlns:adb="http://tempuri.org/adb.xsd">
mas01mc@354:  <SOAP-ENV:Body>
mas01mc@354:   <adb:status>
mas01mc@354:    <dbName>%s</dbName>
mas01mc@354:   </adb:status>
mas01mc@354:  </SOAP-ENV:Body>
mas01mc@354: </SOAP-ENV:Envelope>
mas01mc@354: """
mas01mc@354: 
mas01mc@354: SHINGLE_QUERY_TEMPLATE = """
mas01mc@354: <?xml version="1.0" encoding="UTF-8"?>
mas01mc@354: <SOAP-ENV:Envelope
mas01mc@354:  xmlns:SOAP-ENV="http://schemas.xmlsoap.org/soap/envelope/"
mas01mc@354:  xmlns:SOAP-ENC="http://schemas.xmlsoap.org/soap/encoding/"
mas01mc@354:  xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
mas01mc@354:  xmlns:xsd="http://www.w3.org/2001/XMLSchema"
mas01mc@354:  xmlns:adb="http://tempuri.org/adb.xsd">
mas01mc@354:  <SOAP-ENV:Body>
mas01mc@354:   <adb:shingleQuery>
mas01mc@354:    <dbName>%s</dbName>
mas01mc@354:    <qVector>
mas01mc@354:    <dim>%s</dim>
mas01mc@354:       %s
mas01mc@354:       %s
mas01mc@354:    </qVector>
mas01mc@354:    <keyList></keyList>
mas01mc@354:    <timesFileName></timesFileName>
mas01mc@354:    <queryType>32</queryType>
mas01mc@354:    <queryPos>%s</queryPos>
mas01mc@354:    <pointNN>%s</pointNN>
mas01mc@354:    <trackNN>%s</trackNN>
mas01mc@354:    <sequenceLength>1</sequenceLength>
mas01mc@354:    <radius>%s</radius>
mas01mc@354:    <absolute-threshold>%s</absolute-threshold>
mas01mc@354:    <relative-threshold>%s</relative-threshold>
mas01mc@354:    <exhaustive>%s</exhaustive>
mas01mc@354:    <lsh-exact>%s</lsh-exact>
mas01mc@354:   </adb:shingleQuery>
mas01mc@354:  </SOAP-ENV:Body>
mas01mc@354: </SOAP-ENV:Envelope>
mas01mc@354: """
mas01mc@354: 
mas01mc@354: ###############  List Query - Show the files in the database ###########
mas01mc@354: # Return a list of (key identifier, frame length) pairs.
mas01mc@354: def RunListQuery():
mas01mc@354: 	global debug, dbName
mas01mc@354: 	message = LIST_TEMPLATE%(dbName)
mas01mc@354: 	
mas01mc@354: 	response = SendXMLCommand(message)
mas01mc@354: 	return response
mas01mc@354: 
mas01mc@354: # Construct a list from the two DOM entries passed.  Used with the map routine to 
mas01mc@354: # assemble the output.
mas01mc@354: def ParseListConstruct(f,l):
mas01mc@354: 	el = [f.firstChild.data.encode('latin-1'),l.firstChild.data.encode('latin-1')]
mas01mc@354: 	return el
mas01mc@354: 	
mas01mc@354: def ParseListXML(response):
mas01mc@354: 	dom = parseString(response)
mas01mc@354: 	fileElements = dom.getElementsByTagName('Rkey')
mas01mc@354: 	# print fileElements
mas01mc@354: 	lenElements = dom.getElementsByTagName('Rlen')
mas01mc@354: 	# print lenElements
mas01mc@354: 	return map(ParseListConstruct, fileElements, lenElements)
mas01mc@354: 
mas01mc@354: ###############  Status Query - Show the status of the database ###########
mas01mc@354: # Return a dictionary with the status fields
mas01mc@354: def GetDomElement(dom, field):
mas01mc@354: 	els = dom.getElementsByTagName(field)
mas01mc@354: 	if len(els) > 0:
mas01mc@354: 		return els[0].firstChild.data.encode('latin-1')
mas01mc@354: 	else:
mas01mc@354: 		return ""
mas01mc@354: 
mas01mc@354: def RunStatusQuery():
mas01mc@354: 	global debug, dbName
mas01mc@354: 	message = STATUS_TEMPLATE%(dbName)
mas01mc@354: 
mas01mc@354: 	response = SendXMLCommand(message)
mas01mc@354: 	# print response
mas01mc@354: 	dom = parseString(response)
mas01mc@354: 	status = {}
mas01mc@354: 	status['numFiles'] = GetDomElement(dom, 'numFiles')
mas01mc@354: 	status['dim'] = GetDomElement(dom, 'dim')
mas01mc@354: 	status['length'] = GetDomElement(dom, 'length')
mas01mc@354: 	status['dudCount'] = GetDomElement(dom, 'dudCount')
mas01mc@354: 	status['nullCount'] = GetDomElement(dom, 'nullCount')
mas01mc@354: 	return status
mas01mc@354: 
mas01mc@354: 
mas01mc@354: 
mas01mc@354: ###############  Shingle/Matrix Query - Show the data closest to shingle range ###########
mas01mc@354: #
mas01mc@354: # Encode features as a matrix with dim columns, and (optionally) powers as a matrix with one column
mas01mc@354: # dim   # number of columns
mas01mc@354: # f1,1 f1,2...f1,dim # first row
mas01mc@354: # f2,1 f2,2...f2,dim
mas01mc@354: # ...
mas01mc@354: # fN,1 fN,2...fN,dim # last row
mas01mc@354: # p1,1         # first row's power
mas01mc@354: # p2,1
mas01mc@354: # ...
mas01mc@354: # pN,1         # last row's power
mas01mc@354: #
mas01mc@354: def RunShingleQuery():
mas01mc@354: 	global debug, dbName
mas01mc@354: 	featureDim = '3'
mas01mc@354: 	queryVector='<v>1.0</v><v>0.5</v><v>0.25</v><v>0.5</v><v>1.5</v><v>1.0</v>'  # two rows of features
mas01mc@354: 	powerVector='<p>-1.0</p><p>-1.0</p>'   # one power feature per row
mas01mc@354: 	queryPos = '0'  # where in the feature sequence to start the shingle query
mas01mc@354: 	pointNN = '10'  # how many near points to return per track
mas01mc@354: 	trackNN = '10'  # how many near tracks to return
mas01mc@354: 	radius = '1.0'  # search radius
mas01mc@354: 	absoluteThreshold = '-4.5'  # absolute silence threshold in Bels (query and database shingles)
mas01mc@354: 	relativeThreshold = '0'     # relative silence threshold in Bels between features, 0 = ignore
mas01mc@354: 	exhaustive = '0'            # 1 = perform query using all subsequences of features of length sequenceLength
mas01mc@354: 	lshExact = '0'              # if using an index then compute exact distances after LSH retrieval
mas01mc@354: 	message = SHINGLE_QUERY_TEMPLATE
mas01mc@354: 	message = SHINGLE_QUERY_TEMPLATE%(dbName, featureDim, queryVector, powerVector, queryPos, pointNN, trackNN, radius, absoluteThreshold, relativeThreshold, exhaustive, lshExact);
mas01mc@354: 	# print message
mas01mc@354: 	print message
mas01mc@354: 	response = SendXMLCommand(message)
mas01mc@354: 	ParseShingleXML(response)
mas01mc@354: 		
mas01mc@354: 
mas01mc@354: ###############  Sequence Query - Show the data closest to one query ###########
mas01mc@354: def RunSequenceQuery(argv):
mas01mc@354: 	global debug, dbName	
mas01mc@354: 	if len(argv) > 2:
mas01mc@354: 		dbKey = argv[2]
mas01mc@354: 		qType = '32'			# nSequence
mas01mc@354: 		qPos = argv[3]
mas01mc@354: 		pointNN = '10'
mas01mc@354: 		trackNN = '5'
mas01mc@354: 		seqLen = argv[4]
mas01mc@354: 		queryRadius = '0.5'
mas01mc@354: 	else:
mas01mc@354: 		dbKey = 'tmp/3.chr'
mas01mc@354: 		qType = '32'			# nSequence
mas01mc@354: 		qPos = '110'
mas01mc@354: 		pointNN = '10'
mas01mc@354: 		trackNN = '5'
mas01mc@354: 		seqLen = '20'
mas01mc@354: 		queryRadius = '0.4'
mas01mc@354: 
mas01mc@354: 	message = SEQUENCE_TEMPLATE
mas01mc@354: 	message = SEQUENCE_TEMPLATE%(dbName, dbKey, qType, qPos, pointNN, trackNN, seqLen, queryRadius)
mas01mc@354: 	# print message
mas01mc@354: 	response = SendXMLCommand(message)
mas01mc@354: 	ParseShingleXML(response)
mas01mc@354: 
mas01mc@354: def ParseShingleXML(response):
mas01mc@354: 	# Grab all the responses
mas01mc@354: 	#	See http://diveintopython.org/xml_processing/parsing_xml.html
mas01mc@354: 	dom = parseString(response)
mas01mc@354: 	resultList = []
mas01mc@354: 	for node in dom.getElementsByTagName('Rlist'):
mas01mc@354: 		# print node.toxml()
mas01mc@354: 		resultList.append(node.firstChild.data.encode('latin-1'))
mas01mc@354: 
mas01mc@354: 	distanceList = []
mas01mc@354: 	for node in dom.getElementsByTagName('Dist'):
mas01mc@354: 		# print node.toxml()
mas01mc@354: 		distanceList.append(node.firstChild.data.encode('latin-1'))
mas01mc@354: 
mas01mc@354: 	positionList = []
mas01mc@354: 	for node in dom.getElementsByTagName('Spos'):
mas01mc@354: 		# print node.toxml()
mas01mc@354: 		positionList.append(node.firstChild.data.encode('latin-1'))
mas01mc@354: 
mas01mc@354: 	# print resultList
mas01mc@354: 	# print distanceList
mas01mc@354: 	# print positionList
mas01mc@354: 
mas01mc@354: 	# Print out a summary of the most similar results
mas01mc@354: 	for i in range(0,len(resultList)):
mas01mc@354: 		if i > 0 and resultList[i] != resultList[i-1]:
mas01mc@354: 			print
mas01mc@354: 		print positionList[i], distanceList[i], resultList[i]
mas01mc@354: 
mas01mc@354: 	dom.unlink()
mas01mc@354: 
mas01mc@354: ###############  XML and Network Utilities ###########
mas01mc@354: # Send one XML SOAP command to the server.  Get back the response.
mas01mc@354: 
mas01mc@354: def SendXMLCommand(message):
mas01mc@354: 	global debug
mas01mc@354: 	if debug:
mas01mc@354: 		print message
mas01mc@354: 		print
mas01mc@354: 
mas01mc@354: 	#Create a socket
mas01mc@354: 	sSock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
mas01mc@354: 
mas01mc@354: 	#Connect to server
mas01mc@354: 	sSock.connect((serverHost, serverPort))
mas01mc@354: 
mas01mc@354: 	#Send messages
mas01mc@354: 	sSock.send(message)
mas01mc@354: 	data = ""
mas01mc@354: 	# Now loop, while getting all the data we can get from the socket.
mas01mc@354: 	while True:
mas01mc@354: 		c = sSock.recv(4096)
mas01mc@354: 		if c == "":
mas01mc@354: 			break
mas01mc@354: 		data += c
mas01mc@354: 	if data == "":
mas01mc@354: 		print "No response from the audioDB server"
mas01mc@354: 		sys.exit(0)
mas01mc@354: 	# Split off the HTTP header and the data
mas01mc@354: 	header,response = data.split("\r\n\r\n", 1)
mas01mc@354: 	if debug:
mas01mc@354: 		print 'Client received: ',response
mas01mc@354: 
mas01mc@354: 	sSock.close()
mas01mc@354: 	return response
mas01mc@354: 
mas01mc@354: 	
mas01mc@354: ###############  Main Program - Figure out which query we want ###########
mas01mc@354: 
mas01mc@354: # Argument processing scheme described at: http://docs.python.org/lib/module-getopt.html
mas01mc@354: 
mas01mc@354: import sys
mas01mc@354: if __name__=="__main__":
mas01mc@354: 	cmdname = sys.argv[0]
mas01mc@354: 	if len(sys.argv) == 1:
mas01mc@354: 		print "Syntax: " + sys.argv[0] + " -q feature_file pos len"
mas01mc@354: 		sys.exit(1)
mas01mc@354: 
mas01mc@354: 	queryType = sys.argv[1]
mas01mc@354: 	if queryType == '-s' or queryType == 'status':
mas01mc@354: 		response = RunStatusQuery()
mas01mc@354: 		for k, v in response.iteritems():
mas01mc@354: 			print k, v
mas01mc@354: 	elif queryType == '-q' or queryType == 'query':
mas01mc@354: 		RunSequenceQuery(sys.argv)
mas01mc@354: 	elif queryType == '-l' or queryType == 'list':
mas01mc@354: 		response = RunListQuery()
mas01mc@354: 		# print response
mas01mc@354: 		results = ParseListXML(response)
mas01mc@354: 		for (f,l) in results:
mas01mc@354: 			print "%s\t%s" % (f,l)
mas01mc@354: 	elif queryType == '-v' or queryType == 'vector':
mas01mc@354: 		response = RunShingleQuery()
mas01mc@354: 		
mas01mc@354: