setup for nltk/cltk fixed

thePortus · May 23, 2018 · d786954 · d786954
1 parent 4c163bc
commit d786954
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 24 deletions.
diff --git a/dhelp/text/_bases.py b/dhelp/text/_bases.py
@@ -21,14 +21,13 @@ class BaseText(UserString):
         >>> print(text)
         'Lorem ipsum dolor sit amet...'
     """ # noqa
-    options = {}
+    options = {
+        'encoding': 'utf-8',
+        'language': 'english'
+    }
 
     def __init__(self, text, *args, **kwargs):
         super().__init__(str)
-        self.options = {
-            'encoding': 'utf-8',
-            'language': 'english'
-        }
         # update .options if options keyword arg passed
         if 'options' in kwargs:
             if type(kwargs['options']) == dict:

diff --git a/dhelp/text/cltk.py b/dhelp/text/cltk.py
@@ -256,10 +256,10 @@ class LatinText(CLTKMixin, BaseText):
         >>> print(text.lemmatize())
         gallia edo1 omne divido in pars tres
     """
-
-    def __init__(self, text, options={}):
-        options['language'] = 'latin'
-        super().__init__(text=text, options=options)
+    options = {
+        'encoding': 'utf-8',
+        'language': 'latin'
+    }
 
     def macronize(self, mode='tag_ngram_123_backoff'):
         """Adds macrons (long vowel marks).
@@ -366,10 +366,10 @@ class AncientGreekText(CLTKMixin, BaseText):
         >>> print(text.lemmatize())
         εἰμί δὲ σύμπας οὗτος τὰ σύγγραμμα ἐκεῖνος μάλιστα οὐ ὠφέλιμος , ὅστις ὡς πρὸς οἶδα συγγράφω.
     """ # noqa
-
-    def __init__(self, text, options={}):
-        options['language'] = 'greek'
-        super().__init__(text=text, options=options)
+    options = {
+        'encoding': 'utf-8',
+        'language': 'greek'
+    }
 
     def normalize(self):
         """Fixes problems with differences in greek accent encoding.

diff --git a/dhelp/text/nltk.py b/dhelp/text/nltk.py
@@ -26,7 +26,7 @@ class NLTKMixin:
     """
 
     @classmethod
-    def setup(self, lang_pkgs_info):
+    def setup(cls):
         """Download NLTK packages and trainer corpora.
 
         Launches the NLTK package download interface. Method is invoked by
@@ -42,7 +42,7 @@ def setup(self, lang_pkgs_info):
         # (1) pkg name (2) list of path segs where pkg data is stored locally
         pkgs_and_path_segments = settings.NLTK_PACKAGES['all']
         # join common list with language specific packages
-        for package_info in lang_pkgs_info:
+        for package_info in settings.NLTK_PACKAGES[cls.options['language']]:
             pkgs_and_path_segments.append(package_info)
         # loop through list of tuples, each with pkg name and path info
         for package, package_path_segments in pkgs_and_path_segments:
@@ -269,12 +269,7 @@ class EnglishText(NLTKMixin, BaseText):
         >>> english_text.rm_lines().rm_nonchars().rm_spaces()
         The quick brown fox jumped over the lazy dog
     """ # noqa
-
-    def __init__(self, text, options={}):
-        options['language'] = 'english'
-        super().__init__(text=text, options=options)
-
-    @classmethod
-    def setup(self):
-        # invoke parent setup method, sending it the pkg info for specific lang
-        super(self.__class__).setup(settings.NLTK_PACKAGES['english'])
+    options = {
+        'encoding': 'utf-8',
+        'language': 'english'
+    }