# Copyright 1999 taliesin
from shortcuts import *
import re, string

#--- LATIN-1
Vuml = ('ä', 'ï', 'ë', 'ö', 'ü', 'ÿ')
Vgrave = ('à', 'ì', 'è', 'ò', 'ù')
Vacute = ('á', 'í', 'é', 'ó', 'ú', 'ý')
Vcirc = ('â', 'î', 'ê', 'ô', 'û')
Vnas = ('ã', 'õ') 
Vdia = Vuml + Vgrave + Vacute + Vcirc + Vnas

"""
max = n
counted = list of 1 to n dicts
analysed = list of 3 dicts
type = 0 to n-1 for counted, -1 for analysed
"""

class freq:

	def __init__(self, data, phonemes=[], ignore=(), max=3):
		assert data
		if ignore: 
			self.ignore = ignore
			ignores = re.compile('[' + string.join(ignore, '') + ']')
			data = ignores.sub('', data)
	
		if not phonemes:
			for i in range(3):
				phonemes.append(map(None, string.lowercase))
		self.phonemes = phonemes
		for type in range(3):
			self.phonemes[type].sort(lambda x, y: cmp(len(y), len(x)))
		self.max = max
		
		self.countedsum = []
		self.countedstats = []
		counted = []
		for i in range(self.max):
			self.countedsum.append(0)
			self.countedstats.append([])
			counted.append({})
		counted = self.counter(counted, data)
		self.counted = counted
		
		self.zerofound = [[], [], []]
		self.analysedsum = [0, 0, 0]
		self.analysedstats = [[], [], []]
		analysed = [{}, {}, {}]
		analysed = self.analyser(analysed, data)
		self.analysed = analysed

		self.makestats()
	
	def counter(self, counted, data, replace=string.replace):
		pos = 0
		while pos < len(data):
			if data[pos] != '\000':
				for type in range(self.max):
					if pos >= type:
						entry = data[pos-type:pos+1]
						if '\000' in entry: break
						entry = replace(entry, '\000', '')
						if entry: counted[type][entry] = counted[type].get(entry, 0) + 1
			pos = pos + 1
		self.counted = counted
		return counted
	
	def analyser(self, analysed, data):
		assert len(self.phonemes) == 3
		type = 0
		zerofound = self.zerofound
		while type < 3:
			phonemes = self.phonemes[type]
			rawdata = data[:]
			for phoneme in phonemes:
				rawletters = string.split(rawdata, phoneme)
				if type == 1: phoneme = phoneme[1:]
				if type == 2: phoneme = phoneme[:-1]
				rawdata = string.join(rawletters, '')
				phonemecount = len(rawletters) - 1
				if phonemecount: analysed[type][phoneme] = \
					analysed[type].get(phoneme, 0) + phonemecount 
				else: zerofound[type].append(phoneme)	
				if not rawdata: break
			type = type + 1
		self.zerofound = zerofound
		self.analysed = analysed
		return analysed
	
	def makestats(self):

		def statit(self, dictlist, sums):
			assert dictlist and sums
			i = 0
			statlist = []
			while i < len(sums):
				dict = dictlist[i]
				assert dict
				keys = dict.keys()
				letters = reduce(inc, dict.values())
				stats = []
				for key in keys:
					stats.append(float(dict[key])/letters*100.0, dict[key], key)
				sums[i] = letters
				statlist.append(stats)
				i = i + 1
			return statlist, sums

		stats, sums = statit(self, self.counted, self.countedstats)
		self.countedstats = stats
		self.countedsum = sums
		stats, sums = statit(self, self.analysed, self.analysedstats)
		self.analysedstats = stats
		self.analysedsum = sums
	
	def printstats(self, list=[], type=0, format='perc', sort='0-9', break_at=65536, ana=0):
	
		def printit(list, format="<%3s> %4.1f%%", order=(2,0), break_at=break_at):
			i = 0
			printorder = []
			for entry in list:
				i = i + 1
				print format % (entry[order[0]], entry[order[1]])
				if i >= break_at: break
		
		"list: [(%f, %u, %s)]"

		assert list
		list = list[type]
		print 'Letter frequencies:'
		if format == 'num':
			printformat = "%7u <%3s>"
			list = map(lambda (x, y, z): (y, x, z), list)
			order = (0, 2)
		elif format == 'perc':
			printformat = "<%3s> %4.1f%%"
			order = (2,0)
		elif format == 'both':
			pass
		if sort == '0-9':
			list.sort()
			list.reverse()
			#print 'Processing:', list
			printit(list, printformat, order)
		elif sort == 'a-z':
			list = map(lambda (x, y, z): (z, x, y), list)
			list.sort()
			printit(list, order=(0,1))
		if ana: 
			print 'Not found:'
			for t in self.zerofound[type]:
				print t,