diff --git a/umbra/dutch_mmetaphone.py b/umbra/dutch_mmetaphone.py
index 6a3d1cbadbebb5d8afb577dd7d1c4cbb600e170b..caa5dc53e3cb0efb03a3305f02881481853460c0 100644
--- a/umbra/dutch_mmetaphone.py
+++ b/umbra/dutch_mmetaphone.py
@@ -213,8 +213,6 @@ class DutchPhonetics:
source = source[:amb_pos] + "+" + source[amb_pos + 1:]
amb_pos = source.find(amb)
- print(source, representations)
-
return source, representations
@staticmethod
diff --git a/umbra/mistake_finder.py b/umbra/mistake_finder.py
index f90ce08e48ec227ad435b9e73edf919fee759ee1..90ac984de2a947b19cd92e25189221cb30062f74 100644
--- a/umbra/mistake_finder.py
+++ b/umbra/mistake_finder.py
@@ -5,7 +5,7 @@ from dutch_mmetaphone import DutchPhonetics
class MistakeFinder:
"""Finds the mistakes in already aligned lists of Words"""
- def __init__(self, seman_checker):
+ def __init__(self, seman_checker, form_checker):
"""
Constructor
@@ -13,6 +13,7 @@ class MistakeFinder:
seman_checker: Instance of the SemanticChecker class
"""
self._seman_checker = seman_checker
+ self._form_checker = form_checker
self._source = None
self._shadow = None
@@ -38,7 +39,9 @@ class MistakeFinder:
# PRINT FOR TEST OF SO FAR IMPLEMENTED MISTAKE CHECKS:
for word in self._shadow:
- if not word.correct and word.mistake != Mistake.RANDOM:
+ if word.mistake == Mistake.REPETITION or word.mistake\
+ == Mistake.SEMANTIC or word.mistake == Mistake.FORM or\
+ word.mistake == Mistake.PHONETIC:
print(str(word) + " " + str(word.mistake))
def _determine_mistake(self, index, word):
@@ -50,12 +53,13 @@ class MistakeFinder:
word: Instance of the Words class
"""
- if not self._check_repetition(word, index): # If not a repetition,
- # if not self._check_form_mistake(word): # form,
- if not self._check_semantic_mistake(word, index): # semantic,
- if not self._check_phonetic_mistake(word, index): # phon;
+ # Check for all the types of mistakes, and if it is none of them ...
+ if not self._check_repetition(word, index):
+ if not self._check_form_mistake(word, index):
+ if not self._check_semantic_mistake(word, index):
+ if not self._check_phonetic_mistake(word, index):
# ... then shadow word is random:
- word.mistake = Mistake.RANDOM
+ word.mistake = Mistake.RANDOM
def _check_phonetic_mistake(self, shd_word, index):
"""Checks if shd_word is a phonetic mistake or not. If yes, it flags
@@ -79,10 +83,12 @@ class MistakeFinder:
not phonetic_mistake and src_index < (
len(self._source) - 1): # Crude fix, but removes error
src_word = self._source[src_index]
- if DutchPhonetics.compare(src_word.word, shd_word.word):
+ phonetic_mistake = self.phonetically_related(src_word, shd_word)
+ if phonetic_mistake:
shd_word.mistake = Mistake.PHONETIC
- src_word.mistake = Mistake.PHONETIC
- phonetic_mistake = True
+ shd_word.source = src_word
+ if not src_word.shadowed and src_word.mistake is None:
+ src_word.mistake = Mistake.PHONETIC
src_index += 1
return phonetic_mistake
@@ -105,16 +111,38 @@ class MistakeFinder:
src_index = self._source.index(
self._shadow[shd_anchor_index].anchor)
while self._source[src_index].get_difference(shd_word) > 0 and\
- not semantic_mistake and src_index < (
- len(self._source)-1): # Crude fix, but removes error
+ not semantic_mistake and src_index < (len(self._source)-1):
src_word = self._source[src_index]
- if self.semantically_related(src_word, shd_word):
+ semantic_mistake = self.semantically_related(src_word, shd_word)
+ if semantic_mistake:
shd_word.mistake = Mistake.SEMANTIC
- src_word.mistake = Mistake.SEMANTIC
- semantic_mistake = True
+ shd_word.source = src_word
+ if not src_word.shadowed and src_word.mistake is None:
+ src_word.mistake = Mistake.SEMANTIC
src_index += 1
return semantic_mistake
+ def _check_form_mistake(self, shd_word, index):
+ if shd_word.word == "te":
+ print("te")
+ form_mistake = False
+ last_shd_index = self._shadow.find_last_matched_shadow(index)
+ if last_shd_index < 0:
+ src_index = 0
+ else:
+ src_index = self._source.index(self._shadow[last_shd_index].source)
+ while self._source[src_index].get_difference(shd_word) > 0 and\
+ not form_mistake and src_index < len(self._source) - 1:
+ src_word = self._source[src_index]
+ form_mistake = self.form_related(src_word, shd_word)
+ if form_mistake:
+ shd_word.mistake = Mistake.FORM
+ shd_word.source = src_word
+ if not src_word.shadowed and src_word.mistake is None:
+ src_word.mistake = Mistake.FORM
+ src_index += 1
+ return form_mistake
+
def _check_repetition(self, word, index):
"""
Check whether a word can be seen as a repetition mistake
@@ -170,8 +198,7 @@ class MistakeFinder:
return chain
def semantically_related(self, src_word, shd_word):
- """Checks if src_word and shd_word are semantically related. If yes, it
- flags both as semantic mistakes.
+ """Checks if src_word and shd_word are semantically related.
Args:
src_word: SourceWord instance
@@ -183,10 +210,40 @@ class MistakeFinder:
shd_string = shd_word.word
src_string = src_word.word
related = self._seman_checker.semantically_related(src_string,
- shd_string)
- if related:
- shd_word.mistake = Mistake.SEMANTIC
- src_word.mistake = Mistake.SEMANTIC
+ shd_string)\
+ and shd_string != src_string
+ return related
+
+ def phonetically_related(self, src_word, shd_word):
+ """Checks if src_word and shd_word are phonetically related.
+
+ Args:
+ src_word: SourceWord instance
+ shd_word: ShadowWord instance
+ Returns:
+ related: True if src_word and shd_word are phonetically related,
+ False if not.
+ """
+ shd_string = shd_word.word
+ src_string = src_word.word
+ related = DutchPhonetics.compare(src_string, shd_string)\
+ and shd_string != src_string
+ return related
+
+ def form_related(self, src_word, shd_word):
+ """Checks if src_word and shd_word are related in form.
+
+ Args:
+ src_word: SourceWord instance
+ shd_word: ShadowWord instance
+ Returns:
+ related: True if src_word and shd_word are related in form,
+ False if not.
+ """
+ shd_string = shd_word.word
+ src_string = src_word.word
+ related = self._form_checker.form_related(src_string, shd_string) and\
+ shd_string != src_string
return related
def print_for_nw(self, source, shadow):
diff --git a/umbra/statistics.py b/umbra/statistics.py
index df0f04bfb20af725e2b8fd4d48f71c7ebfe7b43d..ef135a88089c90a9a9760404ae00ca7bd0e6c92e 100644
--- a/umbra/statistics.py
+++ b/umbra/statistics.py
@@ -20,11 +20,12 @@ class Statistics:
# currently the case, so the two operate seperately in this class.
# _parser and _seman_checker should be moved to MistakeFinder when
# possible
- self.path = ut.get_path("OpenDutchWordnet/resources/odwn/odwn_orbn_gwg-LMF_1.3.xml.gz")
-
+ self.path = ut.get_path(
+ "OpenDutchWordnet/resources/odwn/odwn_orbn_gwg-LMF_1.3.xml.gz")
self._seman_checker = SemanticChecker()
self._form_checker = FormChecker()
- self._mistake_finder = MistakeFinder(self._seman_checker)
+ self._mistake_finder = MistakeFinder(self._seman_checker,
+ self._form_checker)
self._mistake_counter = MistakeCounter()
@property
diff --git a/umbra/words.py b/umbra/words.py
index 9db5f7cd0118d79447ac7fd37c3494575068ddef..f2dc49ee80c514b3a3bbb0dec94a4d04a9b934f0 100644
--- a/umbra/words.py
+++ b/umbra/words.py
@@ -231,6 +231,25 @@ class Sentence(list):
return index
index -= 1
return -1
+
+ def find_last_matched_shadow(self, index):
+ """Find the index of the last shadowed word.
+
+ Args:
+ index: the index before which should be sought for a shadow word
+ that is matched with a source word.
+
+ Returns:
+ last_index: the index of the previous matched shadow. This is -1
+ if there is no matches shadow before or on the specified index, or
+ if the index was out of bounds.
+ """
+ if 0 <= index < len(self):
+ while index >= 0:
+ if self[index].has_source():
+ return index
+ index -= 1
+ return index
def __str__(self):
return ' '.join([word.word for word in self])