from_csv_to_lmf.py

#!/usr/bin/env python

import sys
from lxml import etree


def map_pos(old_pos):
		new_pos = old_pos
		if old_pos == 'n':
				new_pos = 'noun'
		elif old_pos == 'a':
				new_pos = 'adj'
		elif old_pos == 'r':
				new_pos = 'adv'
		elif old_pos == 'v':
				new_pos = 'verb'
		return new_pos

if __name__ == '__main__':
	if sys.stdin.isatty():
		print>>sys.stderr,'Input stream required.'
		print>>sys.stderr,'Example usage: cat my_lexicon.utf8.csv |',sys.argv[0],'-verbose'
		print>>sys.stderr,'\t-verbose option to generate the lexicon with all the lemmas for one synset'
		print>>sys.stderr,'\tInput file format: synset;pos;polarity;value;lemma1,lemma2,lemma3;freq	OR'
		print>>sys.stderr,'\tInput file format: synset;pos;polarity;value;lemma1,lemma2,lemma3;freq;sentiment_modifier (intensifier,shifter...)'
		print>>sys.stderr,'''\nIn case you ran the propagate_wn.py with the output lemmas option the CSV file
can contain 2 types of lines:'''
		print>>sys.stderr,'\tInput file format: unknown;pos;polarity;value;lemma;freq OR'
		print>>sys.stderr,'\tInput file format: unknown;pos;polarity;value;lemma;freq;modifier'
		print>>sys.stderr
		print>>sys.stderr
		sys.exit(-1)
		
		
	lang = sys.argv[1]

	verbose = False
	if False and len(sys.argv)==2:
		if sys.argv[1]=='-verbose':
			verbose = True
		else:
			print>>sys.stderr,'Option ',sys.argv[1],'not recognized. Verbosity=',verbose
	

		
	my_root = etree.Element('LexicalResource')
	my_global = etree.Element('GlobalInformation')
	my_global.set('label','Created with the standard propagation algorithm')
	my_root.append(my_global)
	my_lexicon = etree.Element('Lexicon')
	my_lexicon.set('languageCoding','UTF-8')
	my_lexicon.set('label','sentiment')
	my_lexicon.set('language',lang)

	my_root.append(my_lexicon)
	n=0
	for line in sys.stdin:
		tokens = line.decode('utf-8').strip().split(';')
		if len(tokens)!=6 and len(tokens)!=7:
			print>>sys.stderr,'Skipped line because has not the correct format'
		else:
			synset = tokens[0]
			pos = tokens[1]
			polarity = tokens[2]
			try:
				value = float(tokens[3])
			except:
				value=-1
			lemmas = tokens[4].split(',')
			freq = tokens[5]
			
			my_lemmas = []
			if verbose:
				my_lemmas = lemmas
			else:
				my_lemmas = [lemmas[0]]
				
			if len(tokens) == 7:
				modifier = tokens[6]
			else:
				modifier = None
				
			for my_lemma in my_lemmas:
				if modifier is None:
					lex_ent = etree.Element('LexicalEntry',attrib={'id':'id_'+str(n),'partOfSpeech':map_pos(pos)})
				else:
					lex_ent = etree.Element('LexicalEntry',attrib={'id':'id_'+str(n),'partOfSpeech':map_pos(pos),'type':modifier})
				n+=1
				my_lexicon.append(lex_ent)
				## LEMMA
				l_obj = etree.Element('Lemma',attrib={'writtenForm':my_lemma})
				lex_ent.append(l_obj)
				
				#### SENSE
				sense = etree.Element('Sense')
				if modifier is not None:
					method = 'manual'
					value ='1'
				elif freq=='-1':
					method = 'automatic'
				else:
					method = 'manual'
					value = '1'
				sense.append(etree.Element('Confidence',attrib={'score':str(value),'method':method}))
				refs = etree.Element('MonolingualExternalRefs')
				if synset != 'unknown':
					refs.append(etree.Element('MonolingualExternalRef',attrib={'externalReference':str(synset)}))
				sense.append(refs)
				if modifier is None:
					sense.append(etree.Element('Sentiment',attrib={'polarity':polarity}))
				else:
					sense.append(etree.Element('Sentiment'))
				sense.append(etree.Element('Domain'))
				lex_ent.append(sense)

	my_tree = etree.ElementTree(my_root)
	my_tree.write(sys.stdout,encoding='UTF-8',pretty_print=1, xml_declaration=1)