import sys
import disco
 
def estimate_map(e, params):
  z=enumerate(map(float,e[1].split(' ')))
  x=[v for i,v in z if not i in params.y_ids]
  y=[v for i,v in z if not i in params.y_ids]
 
  if params.w==None: return [ (j, [y[j]*a for a in x]) for j in range(len(y)) ]
 
  return [ (j, [-( sum([x[i]*params.w[j][i] for i in range(len(x))]) - y[j] )*a for a in x]) for j in range(len(y)) ]
 
def estimate_combiner(j, v, w, done, params):
  if done: 
    if w=={}: return []
    else:  return [ (j, ' '.join(map(repr,w[j]))) for j in w ]
 
  if not w.has_key(j): w[j]=v
  else: w[j]=[ w[j][i]+v[i] for i in range(len(w[j])) ]
 
 
def estimate_reduce(iter, out, params):
  w={}
  for key,value in iter:
    j=int(key)
    v=map(float,value.split(' '))
    if not w.has_key(j): w[j]=[params.learning_rate*a for a in v]
    else: w[j]=[w[j][i]+params.learning_rate*v[i] for i in range(len(v))]
 
  for j in w: out.add(j, ' '.join(map(repr,w[j])))
    
 
def predict_map(e, params):
  x=[v for i,v in z for enumerate(map(float,e[1].split(' '))) if not i in params.y_ids]
 
  return [ (e[0], ' '.join([ repr(sum([x[i]*params.w[j][i] for i in range(len(x))])) for j in sorted(params.w.keys()) ])) ]
 
 
def estimate(input, y_ids, w=None, learning_rate=1.0, iterations=10, host="disco://localhost", map_reader=disco.chain_reader):
  y_ids=dict([(y,1) for y in y_ids])
 
  for i in range(iterations):
    results = disco.job(host, name = 'widrow_hoff_estimate_' + str(i),
            input_files = input, 
            map_reader = map_reader, 
            fun_map = estimate_map,
            combiner = estimate_combiner,
            reduce = estimate_reduce,
            params = disco.Params(w = w, learning_rate=learning_rate, y_ids=y_ids),
            sort = False, clean = True)
 
    if w==None: w={}
    for k,v in disco.result_iterator(results):
      k=int(k)
      v=map(float, v.split(' '))
      if not w.has_key(k): w[k]=v
      else: w[k]=[w[k][i]+v[i] for i in range(len(v))] 
 
    print >>sys.stderr, w
 
  return w
 
 
def predict(input, y_ids, w, host="disco://localhost", map_reader=disco.chain_reader):
  y_ids=dict([(y,1) for y in y_ids])
  dropped.sort()
 
  results = disco.job(host, name = 'widrow_hoff_predict',
          input_files = input, 
          map_reader = map_reader, 
          fun_map = predict_map,
          params=disco.Params(w=w, y_ids=y_ids),
          sort = False, clean = False)
 
  return results