#!/usr/bin/env python2.2

import string
import sys
import os
import getopt
import random

class Dataset:
	fname = ""
	label = ""
	events = None

	def __init__(self,fname,label):
		self.fname = fname
		self.label = label
		self.events = []
		F = open(fname,"r")
		while 1:
			line = F.readline()
			if not line: break

			self.events.append(line)

def GenerateSteps(datasets,name,initial_step_size,step_size_multiplier,min_step_size,num_steps):
	print "Writing event stream to " + name + ".events."
	print "Writing event labels to " + name + ".labels."
	print "Writing summary to " + name + ".smry."

	event_F = open(name + ".events","w")
	label_F = open(name + ".labels","w")
	smry_F = open(name + ".smry","w")

	smry_F.write("\n")
	for dataset in datasets:
		smry_F.write("#" + dataset.fname + " as " + dataset.label + "\n")
	smry_F.write("\n")
	smry_F.write("#initial_step_size = " + str(initial_step_size) + "\n")
	smry_F.write("#step_size_multiplier = " + str(step_size_multiplier) + "\n")
	smry_F.write("#min_step_size = " + str(min_step_size) + "\n")
	smry_F.write("#num_steps = " + str(num_steps) + "\n")
	smry_F.write("\n")
	smry_F.write("#steps taken:\n")

	event_count = 0
	step_count = 0
	step_size = initial_step_size
	dataset_index = len(datasets)
	while step_count < num_steps:
		if dataset_index >= len(datasets):
			random.shuffle(datasets)
			dataset_index = 0
		dataset = datasets[dataset_index]
		dataset_index = dataset_index + 1
		smry_F.write(dataset.label + " " + str(step_size) + "\n")
		for i in range(0,step_size):
			event = random.choice(dataset.events)
			event_F.write(event)
			label_F.write("Event " + str(event_count) + " " + dataset.label + "\n")
			event_count = event_count + 1
		step_count = step_count + 1
		step_size = long(step_size * step_size_multiplier)
		if step_size < min_step_size: step_size = min_step_size

def PrintUsage():
	msg = """
USAGE: mixer.py options input_file1 [as label1] input_file2 [as label2] ...

Mixes together a stream of records produced from two or more samples.
Each "step" taken by the algorithm switches between input samples.

Output files:
  *.events    (output stream of mixed events)
  *.labels    (label of each event)
  *.smry      (summary of actions taken)

OPTIONS:
  --output_name=X
  --initial_step_size=10000
  --step_size_multiplier=1
  --min_step_size=10
  --steps=20
"""
	sys.stderr.write(msg)

long_options = [
	"initial_step_size=",
	"step_size_multiplier=",
	"min_step_size=",
	"steps=",
	"output_name=",
	"help"
]

options,args = getopt.getopt(sys.argv[1:],"h",long_options)

initial_step_size = 10000
step_size_multiplier = 1
min_step_size = 10
num_steps = 20
name = None

for option,value in options:
	if option == "--initial_step_size":
		initial_step_size = long(value)
	elif option == "--step_size_multiplier":
		step_size_multiplier = float(value)
	elif option == "--min_step_size":
		min_step_size = long(value)
	elif option == "--steps":
		num_steps = long(value)
	elif option == "--output_name":
		name = value
	elif option == "--help" or option == "-h":
		PrintUsage()
		sys.exit(0)

if not name:
	sys.stderr.write("You must specify an output name.\n")
	PrintUsage()
	sys.exit(2)

#labels = []
#for c in string.ascii_uppercase:
#	if len(labels) == len(args): break
#	labels.append(c)
#random.shuffle(labels)

datasets = []
i = 0
while i < len(args):
	fname = args[i]
	i = i + 1
	#label = labels[len(datasets)]
	label = os.path.basename(fname)
	if i+1<len(args) and args[i] == "as":
		label = args[i+1]
		i = i + 2
	dataset = Dataset(fname,label)
	datasets.append(dataset)
	if label == fname:
		print fname, len(dataset.events)
	else:
		print label, fname, len(dataset.events), "events."

GenerateSteps(
	datasets=datasets,
	name=name,
	initial_step_size=initial_step_size,
	step_size_multiplier=step_size_multiplier,
	min_step_size=min_step_size,
	num_steps=num_steps)