nothing@1: from os.path import join
nothing@1: from rdflib import Graph, BNode, Namespace, RDF, RDFS, Literal, URIRef, XSD, OWL
nothing@1: import glob, wave
nothing@1: 
nothing@1: class Adc2004Converter:
nothing@1:     def __init__(self):
nothing@1:         self.data_dir = "./data"
nothing@1:         self.destination = "./rdf"
nothing@1: 
nothing@1:     def run(self):
nothing@1:         for path in glob.glob(self.data_dir + "/*REF.txt"):
nothing@1:             self.createGraph()
nothing@1:             with open(path, "r") as adc_file:
nothing@1:                 data = adc_file.read()
nothing@1:                 adc_file.close()
nothing@1:             audio_data = self.get_audio_data(path.replace("REF.txt", ".wav"))
nothing@1:             self.convert(data, audio_data)
nothing@1:             write_path = self.destination + "/" + path.split("/")[-1].split(".")[0] + ".n3"
nothing@1:             self.graph.serialize(write_path, format="n3")
nothing@1: 
nothing@1:     def bindNamespaces(self):
nothing@1:         self.ns = {
nothing@1:             'afv': Namespace("https://w3id.org/afo/vocab/1.1#"),
nothing@1:             'afo': Namespace("https://w3id.org/afo/onto/1.1#"),
nothing@1:             'tl': Namespace("http://purl.org/NET/c4dm/timeline.owl#"),
nothing@1:             'event': Namespace("http://purl.org/NET/c4dm/event.owl#"),
nothing@1:             'mo': Namespace("http://purl.org/ontology/mo/"),
nothing@1:             'sxsd': Namespace("https://www.w3.org/TR/speech-synthesis11/synthesis-nonamespace.xsd#")
nothing@1:         }
nothing@1:         for key in self.ns:
nothing@1:             self.graph.bind(key, self.ns[key])
nothing@1: 
nothing@1:     def createGraph(self):
nothing@1:         self.graph = Graph()
nothing@1:         self.bindNamespaces()
nothing@1: 
nothing@1:     def convert(self, data, audio_data):
nothing@1:         self.signal = BNode()
nothing@2:         self.file = URIRef(audio_data['path'].split("/")[-1])
nothing@1:         self.timeline = BNode()
nothing@1:         self.interval = BNode()
nothing@1:         duration = audio_data['n_frames'] / audio_data['f_rate']
nothing@1: 
nothing@1:         self.graph.add(( self.signal, RDF.type, self.ns['mo']['Signal'] ))
nothing@1:         self.graph.add(( self.file, RDF.type, self.ns['mo']['AudioFile'] ))
nothing@1:         self.graph.add(( self.timeline, RDF.type, self.ns['mo']['Timeline'] ))
nothing@1:         self.graph.add(( self.interval, RDF.type, self.ns['tl']['Interval'] ))
nothing@1:         self.graph.add(( self.file, self.ns['mo']['encodes'], self.signal ))
nothing@1:         self.graph.add(( self.signal, self.ns['mo']['sample_rate'], Literal(audio_data['f_rate']) ))
nothing@1:         self.graph.add(( self.signal, self.ns['mo']['channels'], Literal(audio_data['n_channels']) ))
nothing@1:         self.graph.add(( self.signal, self.ns['mo']['time'], self.interval ))
nothing@1:         self.graph.add(( self.interval, self.ns['tl']['duration'], Literal(str(duration), datatype=XSD.duration) ))
nothing@1:         self.graph.add(( self.interval, self.ns['tl']['timeline'], self.timeline ))
nothing@1: 
nothing@1:         index = 0
nothing@1:         for row in data.split("\n"):
nothing@1:             if row != "":
nothing@1:                 time, freq = row.split("     ")
nothing@1:                 event_id = BNode("event_" + str(index))
nothing@1:                 interval_id = BNode()
nothing@1:                 self.graph.add(( event_id, RDF.type, self.ns['afv']['FundamentalFrequency'] ))
nothing@1:                 self.graph.add(( event_id, self.ns['afo']['value'], Literal(str(float(freq)), datatype=self.ns['sxsd']['hertz.number']) ))
nothing@1:                 self.graph.add(( event_id, self.ns['event']['time'], interval_id ))
nothing@1:                 self.graph.add(( interval_id, self.ns['tl']['at'], Literal(time, datatype=XSD.float) ))
nothing@1:                 self.graph.add(( interval_id, self.ns['tl']['duration'], Literal((256.0/44100.0), datatype=XSD.duration) ))
nothing@1:                 self.graph.add(( interval_id, self.ns['tl']['timeline'], self.timeline ))
nothing@1:                 index += 1
nothing@1: 
nothing@1:     def get_audio_data(self, path):
nothing@1:         audio_data = {}
nothing@1:         wave_read = wave.open(path, 'rb')
nothing@1:         audio_data['n_channels'] = wave_read.getnchannels()
nothing@1:         audio_data['n_frames'] = wave_read.getnframes()
nothing@1:         audio_data['s_width']= wave_read.getsampwidth()
nothing@1:         audio_data['f_rate'] = wave_read.getframerate()
nothing@1:         audio_data['path'] = path
nothing@1:         wave_read.close()
nothing@1:         return audio_data
nothing@1: 
nothing@1: def main():
nothing@1:     Adc2004Converter().run()
nothing@1: 
nothing@1: if __name__ == "__main__":
nothing@1:     main()