#!/usr/bin/python
import os,sys,glob,re
from fpformat import fix
pass_dir = '/home/jbueno/pass8'
nevts_fn = '/home/jbueno/QOD/Part1/nevts.txt'

## __________
def steal_numbers(my_string): return re.findall(r'(\d+)', my_string)

## __________
def get_n_fid():

    n_fid = {}
    michel_logs = glob.glob(pass_dir+'/*.michel.log')
    michel_logs.sort()
    for l in michel_logs:
        
        l_clean = l.split('/')[-1]
        set = int(l_clean[3:5])
        if set not in [68,70,71,72,74,75,76,83,84,86,87,91,92,93]:
            print 'ERROR: something wrong with this filename:'
            print l
            sys.exit(1)

        n_fiducial = None
        fl = file(l,'r').readlines()
        for i,l in enumerate(fl):
            if 'Data: HistoDesc' in l:
                n_fiducial = int(float(fl[i+1].split('=')[2]))/1E6
                break

        if n_fiducial == None:
            print 'ERROR: could not find line in michel log file:'
            print l
            sys.exit(1)

        n_fid[set] = n_fiducial

    return n_fid

## __________
def get_good_set_ana():
    sa = []
    michel_logs = glob.glob(pass_dir+'/*.michel.log')
    for l in michel_logs:
        try:
            nums = steal_numbers(l.split('/')[-1])
            sa.append((int(nums[0]),int(nums[1])))
        except:
            print 'ERROR: weird format for michel log file:'
            print l
            sys.exit(1)
    return sa

## __________
def get_nevts():
    nevts = {}
    f = file(nevts_fn,'r')
    for l in f:
        if not l.startswith('#'):
            ls = l.split()
            nevts[int(ls[0])] = int(float(ls[1]))
    return nevts

## __________
def count_nevts(sa,nevts,pathbit):
    counted = {}
    for (s,a) in sa:
        path = '/twist/tw00y/systematics/data/set'+str(s)+'/anal'+str(a)+'/'+pathbit+'/'
        print 'reading',path
        l = []
        for r in glob.glob(path+'run*'):
            try:
                l.append(nevts[int(r.split('/')[-1][3:8])])
            except KeyError:
                continue
            counted[s] = sum(l)
    return counted

## __________
if __name__=='__main__':

    n_fid = get_n_fid()

    sa    = get_good_set_ana()
    #following are for the sets not used for MPs:
#    sa.extend([(73,8),(80,1),(82,2),(89,1)])
    sa.extend([(73,8),(80,1),(89,1)])

    nevts = get_nevts()
    n_pre_checks  = count_nevts(sa,nevts,'root')
    n_post_checks = count_nevts(sa,nevts,'goodlinks')

    sets = [s for (s,a) in sa]
    sets.sort()
    for s in sets:
        print s,
        try:
            print fix(n_post_checks[s]/1E6,0),
        except KeyError:
            print '-','-',
        try:
            print fix(n_fid[s],0),
        except KeyError:
            print '-',
        print

# /twist/data36/systematics/data/set68/anal2
# print set,fix(n_fiducial,0)