def calculateFrequenciesFromXML(self):
input_file = open(sys.argv[1], "r")
lines = input_file.readlines()
num_lines_parsed = 0
# Set up the first context
current_context = Context(self.extractTimestamp(lines[0]))
for line in lines:
line_timestamp = self.extractTimestamp(line)
# Handles malformed timestamps
if line_timestamp is None:
continue
# Extract other components of message
line_message = line[line.find("<msg>")+5:line.find("</msg>")]
# If we hit a new second, that is the two timestamps differ by more than the context duration
if line_timestamp - current_context.timestamp > timedelta(seconds=Globals.CONTEXT_DURATION-1):
self.writeContextToFile(current_context)
self.context_history.append(current_context)
current_context = Context(line_timestamp)
# Else, we're in the same second, so update the current context
else:
current_context.addMessageToContext(line_message)
num_lines_parsed += 1
if num_lines_parsed % 1000000 == 0:
print("Finished: " + str(num_lines_parsed))
# Add the latest context
self.context_history.append(current_context)
self.writeContextToFile(current_context)
self.output_file.close()
