From 2e84c19e106cbf855b1ed153080cb6638fd2a666 Mon Sep 17 00:00:00 2001
From: Erickson Silva <erickson.silva@lavid.ufpb.br>
Date: Wed, 21 Jan 2015 16:51:34 -0300
Subject: [PATCH] Adiciona conversor de numeros por extenso

---
 src/new/ConversorExtenso.py | 144 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 144 insertions(+), 0 deletions(-)
 create mode 100644 src/new/ConversorExtenso.py

diff --git a/src/new/ConversorExtenso.py b/src/new/ConversorExtenso.py
new file mode 100644
index 0000000..863f844
--- /dev/null
+++ b/src/new/ConversorExtenso.py
@@ -0,0 +1,144 @@
+# -*- coding: utf-8 -*-
+# Autor: Erickson Silva
+
+import sys
+from unicodedata import normalize
+from Iterator import *
+
+num = {"zero":0, "um":1, "dois":2, "tres":3, "quatro":4, "cinco":5, "seis":6,
+"sete":7, "oito":8, "nove":9}
+
+und = {"mil":1000, "milhao":1000000, "bilhao":1000000000, "trilhao":1000000000000}
+ 
+ext = [{"um":"1", "dois":"2", "tres":"3", "quatro":"4", "cinco":"5", "seis":"6",
+"sete":"7", "oito":"8", "nove":"9", "dez":"10", "onze":"11", "doze":"12",
+"treze":"13", "quatorze":"14", "quinze":"15", "dezesseis":"16", 
+"dezessete":"17", "dezoito":"18", "dezenove":"19"}, {"vinte":"2", "trinta":"3",
+"quarenta":"4", "cinquenta":"5", "sessenta":"6", "setenta":"7", "oitenta":"8",
+"noventa":"9"}, {"cento":"1", "cem":"1", "duzentos":"2", "trezentos":"3",
+"quatrocentos":"4", "quinhentos":"5", "seissentos":"6", "setessentos":"7",
+"oitocentos":"8", "novecentos":"9"}]
+
+unds = {"mil":"000", "milhao":"000000","milhoes":"000000", "bilhao":"000000000","bilhoes":"000000000", "trilhao":"000000000000", "trilhoes":"000000000000"}
+
+
+
+def oneDigit(x):
+	return ext[0][x]
+
+def twoDigit(x):
+	try:
+		return ext[1][x[0]]+ext[0][x[1]]
+	except:
+		return ext[1][x[0]]+"0"
+
+def threeDigit(x):
+	return ext[2][x[0]]+ext[1][x[1]]+ext[0][x[2]]     
+
+def extenso2(n):
+    sn = n.split(",")
+    size = len(sn)
+    firstWord = sn[0]
+    endWord = ""
+    numExt = ""
+
+    if(unds.has_key(sn[size-1])):
+    	size -= 1
+    	endWord = sn[size]
+    	del sn[size]
+
+    if(ext[0].has_key(firstWord)):
+    	numExt = oneDigit(firstWord)
+        
+    elif (ext[1].has_key(firstWord)):
+    	numExt = twoDigit(sn)
+
+    elif (ext[2].has_key(firstWord)):
+    	if(size == 1):
+    		numExt = ext[2][firstWord]+"00"
+    	elif (size == 2):
+    		if(sn[1] == "dez"):
+    			numExt = ext[2][firstWord]+oneDigit(sn[1])
+    		try:
+    			numExt = ext[2][firstWord]+"0"+oneDigit(sn[1])
+    		except:
+    			numExt = ext[2][firstWord]+twoDigit([sn[1]])
+    	else:
+	    	numExt = threeDigit(sn)
+
+    if(endWord != ""):
+    	numExt = numExt+unds[endWord]
+
+    return numExt 
+
+def extenso(extenso):
+	global newToken, auxToken
+	extensoQuebrado = extenso.split(" ")
+	nums = []
+	it = Iterator()
+	it.load(extensoQuebrado)
+	while(it.hasNext()):
+		token = simplifica(it.getToken())
+		tokenAnterior = simplifica(it.getToken('-'))
+		if (und.has_key(token)):
+			#print "cond1"
+			if(it.getCount() == 0):
+				#print "cond2"
+				nums.append(und[token])
+			else:
+				#print "cond3"
+				newToken = und[token] * int(nums[-1])
+				nums[-1] = newToken
+		else:
+			#print "cond4"
+			if (num.has_key(token)):
+				#print "cond5"
+				auxToken = num[token]
+			elif (not und.has_key(token)):
+				#print "cond6"
+				auxToken = extenso2(token)
+				
+			if((not und.has_key(tokenAnterior)) and it.getCount() > 0):
+				#print "cond7"
+				newToken = int(auxToken) + int(nums[-1])
+				nums[-1] = newToken
+			else:
+				#print "cond8"
+				nums.append(auxToken)
+
+	return soma(nums)
+
+def soma(lista):
+	soma = 0
+	for i in lista:
+		soma += int(i)
+	return soma
+
+def simplifica(txt):
+
+	newToken = ""
+	try:
+		newToken = normalize('NFKD', txt.decode('utf-8')).encode('ASCII','ignore')
+	except:
+		newToken = normalize('NFKD', txt.decode('iso-8859-1')).encode('ASCII','ignore')
+
+	if(newToken[-3:] == "oes"): return newToken[:-3] + "ao"
+	return newToken
+
+
+if __name__ == '__main__':
+    n = sys.argv[1]
+    print extenso(n)
+    '''arquivoExts = open('exts', 'r')
+    listaExts = arquivoExts.readlines()
+    arquivoNums = open('nums', 'r')
+    listaNums = arquivoNums.readlines()
+    for i in range(0,500):
+    	n = listaNums[i].replace("\n","")
+    	e = listaExts[i].replace("\n","")
+        numNew = extenso(e)
+        if (str(numNew) != n):
+        	print n + " != " + str(numNew)
+        #else:
+        #	print "OK: " + n + " == " + str(numNew)'''
+
--
libgit2 0.21.2