print 'max comp:', maxcomp
print 'input avgsize:', line_avg
print 'input steps:', math.log(len(lines), 2)
print 'target size:', target
print
steps = [0]
cac = {}
def compo(i):
output = zlib.compress(''.join(lines[:i]))
cac[i] = output
print [i, len(output) - target]
steps[0] += 1
return len(output) - target
t0 = time.time()
maxlines, actual = sortedfile.bisect_func_right(0, 1, len(lines), compo)
maxlines -= 1
output = cac[maxlines]
print
print 'time taken:', 1000 * (time.time() - t0)
print 'output steps:', steps
print 'maxlines:', maxlines
print 'maxlines len:', sum(map(len, lines[:maxlines]))
print 'compression ratio:', sum(map(len, lines[:maxlines])) / float(len(output))
print 'outlen:', len(output)
print 'input steps:', math.log(len(lines), 2)
print 'target size:', target
print
steps = [0]
cac = {}
def compo(i):
output = zlib.compress(''.join(lines[:i]))
cac[i] = output
print[i, len(output) - target]
steps[0] += 1
return len(output) - target
t0 = time.time()
maxlines, actual = sortedfile.bisect_func_right(0, 1, len(lines), compo)
maxlines -= 1
output = cac[maxlines]
print
print 'time taken:', 1000 * (time.time() - t0)
print 'output steps:', steps
print 'maxlines:', maxlines
print 'maxlines len:', sum(map(len, lines[:maxlines]))
print 'compression ratio:', sum(map(len, lines[:maxlines])) / float(
len(output))
print 'outlen:', len(output)
steps = [0]
def compo(i):
dat = zlib.compress(''.join(lines[:i]))
actual = len(dat)
ratio = actual/float(target)
out = (ratio, dat)
print[i, actual, ratio]
steps[0]+=1
return out
t0 = time.time()
doink = sortedfile.bisect_func_right((1.0,), 1, len(lines), compo)
maxlines, (ratio, output) = doink
if maxlines and (ratio > 1.0):
maxlines -= 1
ratio, output = compo(maxlines)
print
print 'time taken:', 1000 * (time.time() - t0)
print 'output steps:', steps
print 'maxlines:', maxlines
print 'maxlines len:', sum(map(len, lines[:maxlines]))
print 'target ratio:', ratio
print 'outlen:', len(output)
#print compo(50, (15000,))
steps = [0]
def compo(i):
dat = zlib.compress(''.join(lines[:i]))
actual = len(dat)
ratio = actual / float(target)
out = (ratio, dat)
print[i, actual, ratio]
steps[0] += 1
return out
t0 = time.time()
doink = sortedfile.bisect_func_right((1.0, ), 1, len(lines), compo)
maxlines, (ratio, output) = doink
if maxlines and (ratio > 1.0):
maxlines -= 1
ratio, output = compo(maxlines)
print
print 'time taken:', 1000 * (time.time() - t0)
print 'output steps:', steps
print 'maxlines:', maxlines
print 'maxlines len:', sum(map(len, lines[:maxlines]))
print 'target ratio:', ratio
print 'outlen:', len(output)
#print compo(50, (15000,))
