#!/usr/bin/env python
# -*- coding: utf-8 -*-

PATH = { "DISEASES": "diseases" }

import sys, os

class Disease:
	name     = None
	symptoms = None
	
	def __init__( self, name, symptoms=None ):
		self.name     = name
		self.symptoms = symptoms or []
	# __init__()



	def has_symptom( self, symptom ):
		"""
		Check if the symptom happens in this disease
		"""
		return symptom in self.symptoms
	# has_symptom()



	def load( self, filename ):
		"""
		Load disease's symptoms from filename, which must have
		one symptom per line.
		"""
		try:
			f = open( filename, "r" )
			c = f.read()
			f.close()
		except Exception, e:
			print >> sys.stderr, "ERROR: %s" % e

		c = c.split( '\n' )
		for l in c:
			l = l.strip().lower()
			if l:
				self.symptoms += [ l ]
	# load()
# Disease










class Diseases:
	diseases     = None

	def __init__( self, diseases=None ):
		self.diseases = diseases or []
	# __init__()



	def load_diseases( self, dir ):
		"""
		Load diseases from directory.
		"""
		self.diseases = []
		try:
			df = os.listdir( dir )
			for d in df:
				f = os.path.join( dir, d )
				if os.path.isfile( f ):
					disease = Disease( d )
					disease.load( f )
					self.diseases += [ disease ]
		except Exception, e:
			print >> sys.stderr, "ERROR: %s" % e
	# load_diseases()



	def list_symptoms( self ):
		"""
		Returns a list of every possible symptom
		"""
		symptoms = []
		for d in self.diseases:
			for s in d.symptoms:
				if s not in symptoms:
					symptoms += [ s ]
		return symptoms
	# list_symptoms()



	def check_symptoms_exists( self, symptoms ):
		"""
		Check if symptoms exists in the know diseases.
		Return a list of list of exiting (valid) symptoms and a list
		of non-existant symptoms
		"""
		valid   = []
		invalid = []

		existing_symptoms = self.list_symptoms()

		for s in symptoms:
			if s in existing_symptoms:
				valid += [ s ]
			else:
				invalid += [ s ]

		return [ valid, invalid ]
	# check_symptoms_exists()



	def analyse_count( self, symptoms ):
		"""
		Analyse symptoms against diseases based on how many symptoms
	        match a certain disease.

		Returns a list of pairs ('disease',count) descendent on count.
		"""
		disease_count = {}

		# zero disease counters:
		for d in self.diseases:
			disease_count[ d.name ] = 0

		# count symptoms:
		for s in symptoms:
			for d in self.diseases:
				if d.has_symptom( s ):
					disease_count[ d.name ] += 1

		# Translate dict to list of (key,value):
		analyse = disease_count.items()

		# Sort based on count (or value, the second element in pair)
		# in the inverse order:
		analyse.sort( lambda a, b: cmp( b[ 1 ], a[ 1 ] ) )	

		return analyse
	# analyse_count()



	def analyse_probability( self, symptoms ):
		"""
		Analyse symptoms against diseases based on the percentual
		of symptoms given match.
	                              Symptoms Matching
		   Probability = --------------------------
		                   Total Disease Symptoms

	        Returns a list of tuples ('disease', count, probability)
		descendant on probability.
		"""
		count_analyse = self.analyse_count( symptoms )
		map_disease = {} # Make access faster and easier
		for d in self.diseases:
			map_disease[ d.name ] = d
		
		analyse = []
		for ec in count_analyse:
			# ec is ('disease', count)
			# ep is ('disease', count, probability )

			disease_total = len( map_disease[ ec[ 0 ] ].symptoms )
			
			ep = ec + ( float( ec[ 1 ] ) / disease_total, )
			analyse += [ ep ]
			
		analyse.sort( lambda a,b: cmp( b[ 2 ], a[ 2 ] ) )
		return analyse
	# analyse_probability()
# Diseases
	




def ui_get_symptoms():
	"""
	Get user input with symptoms to be analysed, parse it
	and return a list of symptoms.
	"""
	# Prompt user for symptoms
	symptoms = raw_input( "Sintomas do Paciente (lista " \
			      "separada por virgula): " )
	symptoms = symptoms.split( "," )
	sl = []
	for s in symptoms:
		s = s.strip().lower()
		if s:
			sl += [ s ]
	return sl
# ui_get_symptoms()





if __name__ == "__main__":

	diseases = Diseases()
	diseases.load_diseases( PATH[ "DISEASES" ] )

	# Get user input:
	user_symptoms = ui_get_symptoms()
	# Check for [in]valid symptoms
	user_symptoms, invalid_symptoms = \
		       diseases.check_symptoms_exists( user_symptoms )

	# print out invalid symptoms:
	for s in invalid_symptoms:
		print >> sys.stderr, ( "AVISO: Sintoma '%s' descartado " \
				       "por nao existir em nenhuma doenca " \
				       "coberta por este software." ) % ( s )
	
	print "Sintomas a serem analisados: `%s'" %(", ".join( user_symptoms ))

	print "\nResultados:"
	
 	print "\n* Contagem:"
	analyse = diseases.analyse_count( user_symptoms )

	# Summary:
	disease = analyse[ 0 ][ 0 ].title() # format
	symtom_plural = "" # format between singular and plural
	if analyse[ 0 ][ 1 ] > 1:
		symptom_plural = "sintomas"
	else:
		symptom_plural = "sintoma"
	
	print "   Sumario: `%s' com %d %s." % ( disease,
						analyse[ 0 ][ 1 ],
						symptom_plural)
	print "   Resultado completo:"
	print "\t.--------------------------------.-----."
	print "\t| Doenca                         | No. |"
	print "\t}--------------------------------+-----{"
	for e in analyse:
		print "\t| %s | % 3d |" % ( e[ 0 ][ : 30 ].ljust( 30 ),
					    e[ 1 ] )
	print "\t`--------------------------------'-----'"
		
	

 	print "\n* Porcentagem:"
	analyse = diseases.analyse_probability( user_symptoms )

	# Summary:
	disease = analyse[ 0 ][ 0 ].title() # format
	symtom_plural = "" # format between singular and plural
	if analyse[ 0 ][ 1 ] > 1:
		symptom_plural = "sintomas"
	else:
		symptom_plural = "sintoma"

	di = None
	for d in diseases.diseases:
		if d.name == analyse[ 0 ][ 0 ]:
			di = d
	
	
	print "   Sumario: `%s' com %d %s de %d desta doenca, " \
	      "correspondente a %0.2f%% dos sintomas." % \
	      ( disease,
		analyse[ 0 ][ 1 ],
		symptom_plural,
		len( di.symptoms ),
		analyse[ 0 ][ 2 ] )
	print "   Resultado completo:"
	print "\t.--------------------------------.-----.-------."
	print "\t| Doenca                         | No. |     % |"
	print "\t}--------------------------------+-----+-------{"
	for e in analyse:
		print "\t| %s | % 3d | % 02.2f |" \
		      % ( e[ 0 ][ : 30 ].ljust( 30 ),
			  e[ 1 ],
			  e[ 2 ])
	print "\t`--------------------------------'-----'-------'"