diff --git a/umbra/controller.py b/umbra/controller.py
index 9f696e975e70e767bde4c2e8dea0b18a090d7f3e..dd629561dfae346178fd2d7c72cf7abcd702d1d8 100644
--- a/umbra/controller.py
+++ b/umbra/controller.py
@@ -48,7 +48,7 @@ class Controller:
elif action == 'save_txt':
self._filewriter = TxtWriter()
elif action == 'select_folder':
- self._select_folder()
+ self._select_folder(type[0])
elif action == 'rm_all':
self._delete_files(type[0], True)
@@ -59,11 +59,16 @@ class Controller:
folder_path = self._view.dir_dialog()
file_names = os.listdir(folder_path)
file_paths = [folder_path + "/" + x for x in file_names]
- self._shadow_files = file_paths
+ self._set_path(file_paths, type)
self._view.update_files(file_paths, type)
if file_paths:
self._view.button_status("select {}".format(type), "disabled")
# TODO: Call 'buttons' by functionality, not object (type)
+ def _set_path(self, paths, type):
+ if type == 'source':
+ self._source_files = paths
+ else:
+ self._shadow_files = paths
def _select_files(self, type):
"""Select files and add them to files list corresponding to type.
@@ -76,8 +81,9 @@ class Controller:
# TODO: Check if not already present
files.extend(selection)
self._view.update_files(files, type)
- if type == "shadow" and files:
- self._view.button_status("select_folder", "disabled")
+ if files:
+ self._view.button_status("select_folder {}".format(type),
+ "disabled")
def _delete_files(self, type, all=False): # TODO: Not yet implemented fully
"""Remove selected file from files list corresponding to type.
@@ -97,7 +103,7 @@ class Controller:
else:
selection = [file for file in files if selection.lower() in file.lower()][0]
files.remove(selection)
- self._view.button_status("select_folder".format(type), "normal")
+ self._view.button_status("select_folder {}".format(type), "normal")
self._view.update_files(files, type)
def _compare_files(self):
@@ -115,10 +121,14 @@ class Controller:
self._read_folder("shadow")
else:
self._read_files(self._shadow_files[0], "shadow") # TODO: Path is temp fix
-
- self._read_files(self._source_files[0], "source")
+ if len(self._source_files) > 1:
+ self._read_folder("source")
+ else:
+ self._read_files(self._source_files[0], "source")
self._model.compare()
- self._view.update_message('comparison complete')
+ if self._model.analysis_complete:
+ self._view.update_message('comparison complete')
+ self._view.button_status("compare", "disabled")
def _save_results(self):
"""Save analysis results to file. If there is multiple results,
@@ -167,8 +177,13 @@ class Controller:
if "AO" not in file_path or ".csv" not in file_path:
# Very dirty
pass # TODO: raise error
- else:
- dict = self._model.multi_data_shadow
+ elif type == "shadow":
+ dict = self._model.get_multi_data(type)
key = ut.shadow_regex(file_path)
data = self._filereader.read(file_path, type)
ut.add_to_dict(key, data, dict, alternative=True)
+ else:
+ dict = self._model.get_multi_data(type)
+ key = ut.id_regex(file_path)
+ data = self._filereader.read(file_path, type)
+ ut.add_to_dict(key, data, dict, alternative=True)
diff --git a/umbra/mistake_finder.py b/umbra/mistake_finder.py
index 40fe64a7b0c9e9ddadf0af2c0e345df88b6a66f4..fb18835008b2d022fd28bbe0c43d7a7c055435c7 100644
--- a/umbra/mistake_finder.py
+++ b/umbra/mistake_finder.py
@@ -134,6 +134,7 @@ class MistakeFinder:
while chain and index-i > 1:
diff += 1
i += 1
+ print(i, index)
chain = self._shadow[i].word == self._shadow[index+diff].word
if chain:
@@ -193,4 +194,3 @@ class MistakeFinder:
phonetic mistake.
"""
pass
-
diff --git a/umbra/model.py b/umbra/model.py
index 045f16701779caf4ae2d6d1205f6d2b368877c4b..7f67dd85c393eb1811e76c9a87b3dc58be3a68a7 100644
--- a/umbra/model.py
+++ b/umbra/model.py
@@ -7,8 +7,9 @@ class Model:
self._stats = Statistics(None)
self._data_source = None
self._multi_data_shadow = {} # dict because fast and indexing
+ self._multi_data_source = {}
self._data_shadow = None
- self._analysis_results = None
+ self._analysis_results = {}
self._multi_results = {}
self._save_pref = None
self._id = "" # this is a string
@@ -77,6 +78,14 @@ class Model:
def multi_data_shadow(self, data):
self._multi_data_shadow = data
+ @property
+ def multi_data_source(self):
+ return self._multi_data_source
+
+ @multi_data_source.setter
+ def multi_data_source(self, data):
+ self._multi_data_source = data
+
@property
def data_shadow(self):
return self._data_shadow
@@ -91,7 +100,8 @@ class Model:
def has_source(self):
"""Check whether self._data_source has a value."""
- return self._data_source is not None
+ return self._data_source is not None \
+ or self._multi_data_source is not None
def has_shadow(self):
"""Check whether self._data_shadow has a value."""
@@ -100,23 +110,44 @@ class Model:
def analysis_complete(self):
"""Check whether self._analysis_results has a value."""
- return self._analysis_results is not None \
- or self._multi_results is not None
+ print(self.analysis_results, self.multi_results)
+ return self._analysis_results or self._multi_results
+
+ def get_multi_data(self, type):
+ if type == "source":
+ return self._multi_data_source
+ else:
+ return self._multi_data_shadow
def compare(self):
""""Run the analyses and saves the results."""
if self.has_source() and self.has_shadow():
- if self._multi_data_shadow:
+ if self._multi_data_shadow and not self._multi_data_source:
for key, data in self._multi_data_shadow.items():
-
if self._id is ut.filter_key(key):
result = self._stats.analyze(self._data_source, data)
ut.add_to_dict(key+"_res", result, self._multi_results)
else:
pass
- else:
-
+ elif self._multi_data_shadow and self._multi_data_source:
+ for key, source in self._multi_data_source.items():
+ for key2, shadow in self._multi_data_shadow.items():
+ #TODO: Problem: what if filename does not comply
+ if key in key2:
+ print("result of", key, key2)
+ print("*"*100)
+ result = self._stats.analyze(source, shadow)
+
+ ut.add_to_dict(key+"_res",
+ result, self._multi_results)
+ else:
+ pass
+
+ elif self._data_source and self._data_shadow:
self._analysis_results = \
self._stats.analyze(self._data_source, self._data_shadow)
+
+ else:
+ print("cannot do this yet")
else:
pass
diff --git a/umbra/saa_Romeo.py b/umbra/saa_Romeo.py
index 96162e6adcfc78c642ee32871519d34eeac8d004..4d462c7ff8bbf9df6bf94fa72af0427375e4df27 100644
--- a/umbra/saa_Romeo.py
+++ b/umbra/saa_Romeo.py
@@ -115,7 +115,7 @@ class SaaRomeo(AlignmentStrategy):
if matrix[i][j]['pointer'] == 'diag':
alignment_source.append(self._source[j - 1])
alignment_shadow.append(self._shadow[i - 1])
- new_source,new_shadow = self._equals_checker(
+ new_source, new_shadow = self._equals_checker(
self._source[j-1],self._shadow[i-1])
self._source[j-1] = new_source
self._shadow[i-1] = new_shadow
@@ -153,5 +153,4 @@ class SaaRomeo(AlignmentStrategy):
source.shadowed = True
source.mistake = Mistake.SEMANTIC
shadow.mistake = Mistake.SEMANTIC
- return source,shadow
-
+ return source, shadow
diff --git a/umbra/semantic_checker.py b/umbra/semantic_checker.py
index d5c11376386d924434a23ed7d58612779717a86e..b7df79222a2dc8b7868d9ed43808bd14f70337f0 100644
--- a/umbra/semantic_checker.py
+++ b/umbra/semantic_checker.py
@@ -2,7 +2,7 @@ from OpenDutchWordnet.le import Le
from OpenDutchWordnet.synset import Synset
from bisect import bisect_left
from OpenDutchWordnet.wn_grid_parser import Wn_grid_parser
-
+from utils import Utils as ut
class SemanticChecker():
@@ -13,8 +13,8 @@ class SemanticChecker():
Args:
parser: object of the Wn_grid_parser class
"""
- self._parser = Wn_grid_parser(
- "OpenDutchWordnet/resources/odwn/odwn_orbn_gwg-LMF_1.3.xml.gz")
+ self._path = "OpenDutchWordnet/resources/odwn/odwn_orbn_gwg-LMF_1.3.xml.gz"
+ self._parser = Wn_grid_parser(ut.get_path(self._path))
# "./umbra/OpenDutchWordnet/resources/odwn/odwn_orbn_gwg-LMF_1.3.xml.gz") # Ubuntu path, temporary 'fix'
self._parser.load_synonyms_dicts()
self._lemma2synsets = getattr(self._parser,'lemma2synsets')
@@ -212,4 +212,4 @@ class SemanticChecker():
target_synsets = self._find_synsets(target_ids)
all_relations.update(target_synsets)
return self._hypernym_helper(hypernyms,all_relations,depth,
- current_depth+1)
\ No newline at end of file
+ current_depth+1)
diff --git a/umbra/statistics.py b/umbra/statistics.py
index 68cb5a8574af050301dd3697fab7c95d83656020..ab5116e04f3e018f5c5460f9b19247c5c4083b0a 100644
--- a/umbra/statistics.py
+++ b/umbra/statistics.py
@@ -49,7 +49,7 @@ class Statistics:
print('Romeo')
self._strategy = SaaRomeo(self._seman_checker)
source_align, shadow_align = self._strategy.align(source, shadow)
- for s_word in shadow_align:
+ for s_word in shadow_align:
if s_word.has_source():
print(f'source: {s_word.source} shadow: {s_word}')
correctness = self._strategy.correctly_shadowed(source)
diff --git a/umbra/utils.py b/umbra/utils.py
index 14506edbaa1d184277bfc773ec6d49f552ab81b5..59e6f1d619f7d525817d8a7f002bde9e2da9b2d0 100644
--- a/umbra/utils.py
+++ b/umbra/utils.py
@@ -1,6 +1,7 @@
import re
import sys
+
class Utils:
def get_path(path):
@@ -57,10 +58,13 @@ class Utils:
return match.group()[2:]
def id_regex(path):
- match = re.search("\d\.TextGrid", path)
+ match = re.search("\d+\.T", path)
# Make sure that not different number accidentally
- if match:
- return match.group()[0]
+ id = match.group()
+ if len(id) > 3 and ".csv" in path:
+ return id[:2]
+ elif len(id) <= 3 and ".csv" in path:
+ return id[0]
def add_to_dict(key, value, dict, alternative=False):
@@ -77,11 +81,9 @@ class Utils:
dict[key] = value
else:
if alternative:
- cnt = 0
+ cnt = 1
for k, v in dict.items():
- if key is k:
+ if key is k or k in key:
cnt += 1
key += "({})".format(str(cnt))
dict[key] = value
- raise ValueError("Key '{}' already exists in {}."
- .format(key, dict))
diff --git a/umbra/view.py b/umbra/view.py
index 1f178d1a8baf0eb62bb00d64f56d4efafeff4058..d3df7b7d631348b60bfe1e175b9745f3accf877a 100644
--- a/umbra/view.py
+++ b/umbra/view.py
@@ -38,7 +38,7 @@ class View:
# Compare and save buttons
self._create_button('compare', self._frame, 'Compare', 4, 1)
self._create_button('save', self._frame, 'Save result', 4, 2)
- self.button_status("save","disabled")
+ self.button_status("save", "disabled")
# Message
self._create_label('message', self._frame, "", 5, 1)
@@ -143,7 +143,8 @@ class View:
'delete shadow',
'compare',
'save',
- 'select_folder',
+ 'select_folder source',
+ 'select_folder shadow',
'rm_all source',
'rm_all shadow',
]
@@ -200,8 +201,8 @@ class View:
'Delete all {} files'.format(type),
8, column)
- if type == 'shadow':
- self._create_button('select_folder', frame,
+
+ self._create_button('select_folder {}'.format(type), frame,
'Select {} folder'.format(type),
6, column)