from cachecow.Cache.LRU import LRU
from cachecow.CacheReader.basicCacheReader import basicCacheReader
from cachecow.CacheReader.csvCacheReader import csvCacheReader
from cachecow.CacheReader.vscsiReader import vscsiReader
from cachecow.Profiler.getMRCBasicLRU import getMRCBasicLRU
from cachecow.Profiler.parda import parda

# first step: construct a reader for reading any kind of trace

# this one is the most basic one, each line is a label/tag/record
reader1 = basicCacheReader("../Data/parda.trace")

# this one reads csv file and choose one column as label/tag
reader2 = csvCacheReader("../Data/trace_CloudPhysics_txt", column=4)

# this one reads binary cloudphysics trace file
reader3 = vscsiReader("../Data/trace_CloudPhysics_bin")

# reader is also a generator, for readers you can do the following thing:
# read one trace element at one time:
reader1.read_one_element()
# for loop:
for element in reader1:
     do something
# reset, after read some elements, you want to go back
reader1.reset()

# second step: construct a profiler for analyze

# basic mattson profiler (toooooo slow)
basic_profiler = getMRCBasicLRU(LRU, cache_size=20000, bin_size=10, reader=reader1)
# or you can use a short way to construct, see below
basic_profiler = getMRCBasicLRU(LRU, 20000, 10, reader1)

# now let's run it!
basic_profiler.run()
# after run, you can either plot it, print it, save the list, save the plot
basic_profiler.plotHRC()  # the plot is also saved as figure_temp in the data folder in case you forget to save it
basic_profiler.plotMRC()  # Wow, after run, you can also obtain MRC
basic_profiler.printHRC()
basic_profiler.printMRC()
basic_profiler.outputHRC("I can be a folder")
basic_profiler.outputMRC("I can also be a file name")



# the second profiler now it supports now is parda
p = parda(LRU, 30000, reader1)      # construction is same, but you don't need to specify bin_size
p.run(parda_mode.seq)               # let's run, it supports two mode, sequential mode and openmp mode
# p.run(parda_mode.openmp, threads=4) # parallel analysis, not stable
p.plotHRC()                         # the rest is the same as all other profilers, plot, print, output
p.plotMRC()                         # in case the specified cache-size is not large enough, I also saved
                                     # parda histogram list in the same folder with data