nltk/cltk auto-setup provisionally added

dhelp/files/folder.py

@@ -53,8 +53,8 @@ def files(self):
         """
         dir_files = []
         for folder_item in self.contents:
-            if os.path.isfile(folder_item):
-                dir_files.append(folder_item)
+            if os.path.isfile(os.path.join(self.data, folder_item)):
+                dir_files.append(os.path.join(self.data, folder_item))
         return dir_files
 
     @property
@@ -73,8 +73,8 @@ def folders(self):
         """
         dir_subdirs = []
         for folder_item in self.contents:
-            if os.path.isdir(folder_item):
-                dir_subdirs.append(folder_item)
+            if os.path.isdir(os.path.join(self.data, folder_item)):
+                dir_subdirs.append(os.path.join(self.data, folder_item))
         return dir_subdirs
 
     @property

dhelp/files/tests/test_folder.py

@@ -49,6 +49,16 @@ def test_contents(self):
         exempla = self.make_test_obj()
         return self.assertTrue(len(exempla.contents) == 5)
 
+    def test_files(self):
+        # should be 5 items in the folder
+        exempla = self.make_test_obj()
+        return self.assertTrue(len(exempla.files) == 5)
+
+    def test_folders(self):
+        # should be no length since there area no folders
+        exempla = self.make_test_obj()
+        return self.assertEqual(len(exempla.folders), 0)
+
     def test_length(self):
         # should have 5 items in the folder
         exempla = self.make_test_obj()

dhelp/text/__init__.py

@@ -10,3 +10,4 @@
 """
 
 from .basic_text import BasicText
+from .install import NLTKInstall, CLTKInstall

dhelp/text/install.py

@@ -0,0 +1,179 @@
+#!/usr/bin/python
+
+""" dhelp/text/install.py
+
+David J. Thomas
+
+Contains functions to automatically download necessary nltk packages and
+trainer data sets.
+
+"""
+
+import os
+from collections import UserString
+import nltk
+
+
+class BaseInstall(UserString):
+    """
+    Parent class for both NLTK and CLTK installer classes. Not meant to be used
+    directly.
+
+    Parameters
+    ----------
+    language : :obj:`str`, optional
+        Desired language for install, defaults to english
+
+    Raises
+    ------
+    Exception
+        If argument 'language' is non-string
+    """
+
+    def __init__(self, language=None):
+        if not language:
+            language = 'english'
+        elif type(language) is not str:
+            raise Exception('"language" was not a string.')
+        self.data = language
+
+
+class NLTKInstall(BaseInstall):
+    """
+    Provides functions to automatically dicover all possible packages, check
+    it against local data, and download any missing data. If you have not
+    already downloaded the NLTK trainer data, you should must use this.
+
+    Parameters
+    ----------
+    language : :obj:`str`, optional
+        Desired language for install, defaults to english
+
+    Raises
+    ------
+    Exception
+        If argument 'language' is non-string
+
+    Example
+    -------
+    >>> from dhelp.text import NLTKInstall
+    >>> NLTKInstall('english')
+    'english'
+    """
+
+    @property
+    def data_root(self):
+        """
+        Returns the absolute path pointing to the cltk data root on the
+        local machine.
+
+        Returns
+        -------
+        Absolute path to the nltk data directory
+
+        Example
+        -------
+        >>> NLTKInstall().data_root
+        """
+        return os.path.expanduser('~/nltk_data')
+
+    def setup(self):
+        """
+        Launches NLTK download interface, either GUI or CLI, depending on OS
+
+        Returns
+        -------
+        :obj:`bool`
+            True if successful
+        """
+        nltk.download()
+        return True
+
+
+class CLTKInstall(BaseInstall):
+    """
+    Provides functions to automatically dicover all possible packages, check
+    it against local data, and download any missing data. If you have not
+    already downloaded the CLTK trainer data, you should must use this.
+
+    Parameters
+    ----------
+    language : :obj:`str`, optional
+        Desired language for install, defaults to latin
+
+    Raises
+    ------
+    Exception
+        If argument 'language' is non-string
+
+    Example
+    -------
+    >>> from dhelp.text import CLTKInstall
+    >>> CLTKInstall('latin')
+    'latin'
+    """
+
+    def __init__(self, language=None):
+        if not language:
+            language = 'latin'
+        super().__init__(language=language)
+
+    @property
+    def data_root(self):
+        """
+        Returns the absolute path pointing to the cltk data root on the
+        local machine.
+
+        Returns
+        -------
+        Absolute path to the cltk data directory
+
+        Example
+        -------
+        >>> CLTKInstall('latin').data_root
+        """
+        return os.path.expanduser('~/cltk_data')
+
+    @property
+    def corpora_list(self):
+        """
+        Gets a list of all possible corpora, downloaded from CLTK.
+
+        Returns
+        -------
+        :obj:`list`
+            List of cltk corpora for specified language
+
+        Example
+        -------
+        >>> CLTKInstall('latin').corpora_list
+        ['latin_text_perseus', 'latin_treebank_perseus', 'latin_text_latin_library', 'phi5', 'phi7', 'latin_proper_names_cltk', 'latin_models_cltk', 'latin_pos_lemmata_cltk', 'latin_treebank_index_thomisticus', 'latin_lexica_perseus', 'latin_training_set_sentence_cltk', 'latin_word2vec_cltk', 'latin_text_antique_digiliblt', 'latin_text_corpus_grammaticorum_latinorum', 'latin_text_poeti_ditalia'] # noqa
+        """
+        # import cltk inline
+        from cltk.corpus.utils.importer import CorpusImporter
+        return CorpusImporter(self.data).list_corpora
+
+    def setup(self):
+        """
+        Gets list of all corpora and downloads them.
+
+        Returns
+        -------
+        :obj:`bool`
+            True if downloading was successful, or at least no fatal errors
+
+        Example
+        -------
+        >>> CLTKInstall('latin').missing_corpora
+        ['latin_text_corpus_grammaticorum_latinorum', 'latin_text_poeti_ditalia'] # noqa
+        """
+        from cltk.corpus.utils.importer import CorpusImporter
+        corpus_importer = CorpusImporter(self.data)
+        for missing_corpus in self.corpora_list:
+            print('Downloading', missing_corpus)
+            try:
+                corpus_importer.import_corpus(missing_corpus)
+            except:
+                print('Problem downloading', missing_corpus, '(skipping)')
+        print('Finished downloading corpora')
+        return True

dhelp/text/tests/test_install.py

@@ -0,0 +1,19 @@
+#!/usr/bin/python
+
+import unittest
+
+from ..install import NLTKInstall, CLTKInstall
+
+
+class TestNLTKInstall(unittest.TestCase):
+    test_class = NLTKInstall
+
+    def test_some_method(self):
+        pass
+
+
+class TestCLTKInstall(unittest.TestCase):
+    test_class = CLTKInstall
+
+    def test_some_method(self):
+        pass

setup.cfg

@@ -4,9 +4,3 @@ universal = 1
 [metadata]
 license_file = LICENSE
 description-file = README.md
-
-[nosetests]
-verbosity=1
-detailed-errors=1
-with-coverage=1
-cover-package=nose

unittest.cfg

@@ -5,6 +5,5 @@ plugins = nose2.plugins.layers
 
 [coverage]
 always-on = True
-coverage-report =
-    term-missing
-    html
+coverage-report = term-missing
+                  html