updated inverse stemmer

- updated inverse stemmer for better results
- added example for inverse stemmer in example_stemmer_3
- added example_stemmer_3 to test_examples
- fixed minor bugs
- increased version number to 1.1.3

Author: sedthh

examples/example_stemmer_2.py

@@ -14,4 +14,5 @@
 
 	search	= ' '.join(valid)
 
-	print("Keresés erre:",search)
+	print("Keresés erre:",search)
+	

examples/example_stemmer_3.py

@@ -0,0 +1,19 @@
+# -*- coding: UTF-8 -*-
+
+import os.path, sys
+sys.path.append(os.path.join(os.path.dirname(os.path.realpath(__file__)), os.pardir))
+from lara import stemmer, nlp
+
+''' Stemmer and n-gram example '''
+
+if __name__ == "__main__":
+	query	= "Toto - Afrika"
+	
+	parts	= query.split('-')
+	artist	= stemmer.inverse(parts[0],'től')	# "tól" and "től" are both valid
+	title	= stemmer.inverse(parts[1],'t')
+	the		= ('az' if nlp.vowel_beginning(title) else 'a')
+	
+	print('A zenelejátszó program az alábbi számot játssza:')
+	print(artist,the,title)
+	

lara/__init__.py

@@ -3,7 +3,7 @@
 # Lara - Lingusitic Aim Recognizer API
 
 __all__				= 'nlp','parser','stemmer','entities'
-__version__ 		= '1.1.2'
+__version__ 		= '1.1.3'
 __version_info__	= tuple(int(num) for num in __version__.split('.'))
 
 import sys

lara/stemmer.py

@@ -323,59 +323,63 @@ def just_asking(text):
 
 # add affixes to words based on their vowel harmony	
 def inverse(word,affix):
+	word	= lara.nlp.trim(word)
 	if not word:
 		return ''
 	vh		= lara.nlp.vowel_harmony(word)
 	result	= word
 	if affix in ('ra','re'):
+		if word[-1].lower() in ('a','e'):
+			result	= result[:-1]+result[-1].replace('a','á').replace('e','é')
 		if vh == 'magas':
-			return result+'ra'
-		else:
 			return result+'re'
-	if affix == 't':
-		if lara.nlp.is_vowel(word[-1]):
-			if word[-1].lower() in ('a','e'):
-				result	= result[:-1]+result[-1].replace('a','á').replace('e','é')
-			return result+'t'
-		elif vh == 'magas':
-			return result+'et'
 		else:
-			return result+'at'
-	if affix == 'k':
+			return result+'ra'
+	if affix in ('k','s','t'):
 		if lara.nlp.is_vowel(word[-1]):
 			if word[-1].lower() in ('a','e'):
 				result	= result[:-1]+result[-1].replace('a','á').replace('e','é')
-			return result+'k'
-		elif vh == 'magas':
-			return result+'ek'
+			if len(result)==2:
+				if word.lower()=="fű":
+					return result[0]+"üve"+affix
+				elif word.lower()=="tó":
+					return result[0]+"ava"+affix
+				elif word.lower()=="ló":
+					return result[0]+"ova"+affix
+			return result+affix
+		test	= _inverse_only_o(result)	# exceptions
+		if test:
+			test2	= sum([lara.nlp.is_vowel(char) for char in result]) # more exceptions
+			if test2<2:
+				return result+test+affix
+			return result[:-2]+result[-1]+test+affix
+		if vh == 'magas':
+			return result+'e'+affix
+		elif vh == 'vegyes':
+			return result+'o'+affix
 		else:
-			return result+'ak'
+			return result+'a'+affix
 	if affix == 'i':
+		if len(result)==2:
+			if word.lower()=="fű":
+				return result[0]+"üvi"
+			elif word.lower()=="tó":
+				return result[0]+"avi"
+			elif word.lower()=="ló":
+				return result[0]+"ovi"
 		if word[-1]=='i':
 			return result
 		return result+'i'
-	if affix in ('bol','ból','böl','ből'):
-		if word[-1].lower() in ('a','e'):
-			result	= result[:-1]+result[-1].replace('a','á').replace('e','é')
-		if vh in ('magas','vegye'):
-			return result+'ből'
-		return result+'ból'
-	if affix in ('rol','ról','röl','ről'):
-		if word[-1].lower() in ('a','e'):
-			result	= result[:-1]+result[-1].replace('a','á').replace('e','é')
-		if vh in ('magas','vegyes'):
-			return result+'ről'
-		return result+'ról'
-	if affix in ('tol','tól','töl','től'):
+	if affix in ('bol','ból','böl','ből','rol','ról','röl','ről','tol','tól','töl','től'):
 		if word[-1].lower() in ('a','e'):
 			result	= result[:-1]+result[-1].replace('a','á').replace('e','é')
-		if vh in ('magas','vegyes'):
-			return result+'től'
-		return result+'tól'
+		if vh == 'magas':
+			return result+affix[0]+'ől'
+		return result+affix[0]+'ól'
 	if affix in ('nak','nek'):
 		if word[-1].lower() in ('a','e'):
 			result	= result[:-1]+result[-1].replace('a','á').replace('e','é')
-		if vh in ('magas','vegyes'):
+		if vh == 'magas':
 			return result+'nek'
 		return result+'nak'
 	if affix in ('val','vel'):
@@ -394,15 +398,33 @@ def inverse(word,affix):
 		else:
 			if len(word)>1:
 				if word[-2:].lower() in ('cs','gy','ly','ny','sz','ty','zs'):
-					result	= result[:-2]+result[-2]+result[-2]+result[-1]
+					if word[-3].lower()!=word[-2].lower():
+						result	= result[:-2]+result[-2]+result[-2]+result[-1]
+					else:
+						result	= result[:-2]+result[-2]+result[-1]
 					if vh == 'magas':
 						return result+'el'
 					else:
 						return result+'al'				
-			if vh == 'magas':
-				return result+result[-1]+'el'
+			if word[-2].lower()!=word[-1].lower():
+				if vh == 'magas':
+					return result+result[-1]+'el'
+				else:
+					return result+result[-1]+'al'
 			else:
-				return result+result[-1]+'al'		
+				if vh == 'magas':
+					return result+'el'
+				else:
+					return result+'al'
 	raise ValueError('Unsupported affix',affix)
 
-		
+def _inverse_only_o(word):
+	vowel	= ''
+	for char in word:
+		if lara.nlp.is_vowel(char):
+			if char not in ('o','O','ó','Ó','ö','Ö','ő','Ő'):
+				return False
+			else:
+				vowel	= char
+	return vowel.lower()
+				

tests/test_examples.py

@@ -4,7 +4,7 @@
 import os, errno, subprocess
 
 @pytest.mark.parametrize("entity", [
-    "example_chatbot_1","example_chatbot_2","example_chatbot_3","example_chatbot_4","example_chatbot_5","example_chatbot_6","example_chatbot_7","example_huszt","example_news","example_wiki_intents","example_stemmer_1","example_Stemmer_2","example_tweet","example_entities"
+    "example_chatbot_1","example_chatbot_2","example_chatbot_3","example_chatbot_4","example_chatbot_5","example_chatbot_6","example_chatbot_7","example_huszt","example_news","example_wiki_intents","example_stemmer_1","example_stemmer_2","example_stemmer_3","example_tweet","example_entities"
 ])
 def test_entities(entity):
 	file	= os.path.join(os.path.dirname(os.path.realpath(__file__)), os.pardir, 'examples/', entity+'.py')