def sort_data(self, filename, chunk_filename_format):
total_chunks = mcm.get_total_chunk(filename, self.chunk_memory)
file_splitter = FileSplitter(filename, chunk_filename_format)
file_splitter.split(self.chunk_memory)
file_merger = FileMerger(MergeSortFile(total_chunks))
buffer_merge_size = self.chunk_memory / (total_chunks)
output_filename = filename.replace(".txt", "") + '_sorted.txt'
file_merger.merge(file_splitter.get_chunk_filenames(), output_filename, buffer_merge_size)
if self.cleanup:
mcm.cleanup_chunk_file(FileSplitter.chunk_filenames)
for i in range(total_chunk):
file = open(blacklist_temp_format.format(i), "r")
lines = [line.split() for line in file.readlines()]
item_point = bisect.bisect(
lines, check_data) # binary search for each chunk file
if check_data in lines[item_point - 1:item_point]:
is_blacklisted = True
file.close()
break
return is_blacklisted
# Global variables
chunk_memory = mcm.get_parsed_memory("1k")
chunk_filenames = []
blacklist = "blacklist.txt"
blacklist_temp_format = 'blacklist_{0}.txt'
total_chunk = mcm.get_total_chunk(blacklist, chunk_memory)
if __name__ == "__main__":
initialize(blacklist)
# Demo
print(check_blacklist("Siska", "712454"))
print(check_blacklist("Melisa", "8565467"))
print(check_blacklist("Jokho", "081235"))
print()
mcm.cleanup_chunk_file(chunk_filenames)
for i in range(total_chunk):
file = open(blacklist_temp_format.format(i), "r")
lines = [line.split() for line in file.readlines()]
item_point = bisect.bisect(lines, check_data) # binary search for each chunk file
if check_data in lines[item_point-1 : item_point]:
is_blacklisted = True
file.close()
break
return is_blacklisted
# Global variables
chunk_memory = mcm.get_parsed_memory("1k")
chunk_filenames = []
blacklist = "blacklist.txt"
blacklist_temp_format = 'blacklist_{0}.txt'
total_chunk = mcm.get_total_chunk(blacklist, chunk_memory)
if __name__ == "__main__":
initialize(blacklist)
# Demo
print(check_blacklist("Siska", "712454"))
print(check_blacklist("Melisa", "8565467"))
print(check_blacklist("Jokho", "081235"))
print()
mcm.cleanup_chunk_file(chunk_filenames)