Replaced JOS XML specifications with preprocessed pickle

MANIFEST.in

@@ -0,0 +1 @@
+include conversion_utils/resources/jos_specifications.pickle

conversion_utils/jos_msds_and_properties.py

@@ -1,8 +1,12 @@
 import lxml.etree as lxml
 import re
+import pickle
+import importlib.resources as pkg_resources
 
 from conversion_utils.utils import xpath_find, get_xml_id
 
+JOS_SPECIFICATIONS_PICKLE_RESOURCE = 'jos_specifications.pickle'
+
 ## Positions of lexeme-level features for each category
 LEXEME_FEATURE_MAP = {'noun':{1,2},
                       'verb':{1,2},
@@ -219,8 +223,22 @@ class Msd:
 class Converter:
     """Converter between Msd and Properties objects."""
 
-    def __init__(self, specifications):
-        self.specifications = specifications
+    def __init__(self, xml_file_name=None):
+        if (xml_file_name is None):
+            if (pkg_resources.is_resource('conversion_utils.resources', JOS_SPECIFICATIONS_PICKLE_RESOURCE)):
+                try:
+                    with pkg_resources.open_binary('conversion_utils.resources', JOS_SPECIFICATIONS_PICKLE_RESOURCE) as pickle_file:
+                        self.specifications = pickle.load(pickle_file)
+                except:
+                    exit('Could not parse specifications pickle file installed.')
+            else:
+                exit('No pickle installed or xml provided.')
+        else:
+            parser = SpecificationsParser()
+            try:
+                self.specifications = parser.parse(xml_file_name)
+            except:
+                exit('Could not parse specifications xml file provided.')
 
     def msd_to_properties(self, msd, language, lemma=None):
         """Convert Msd to Properties (possibly in the other language)."""

scripts/install_jos_specifications.py

@@ -0,0 +1,13 @@
+import pickle
+import argparse
+from conversion_utils.jos_msds_and_properties import SpecificationsParser
+
+arg_parser = argparse.ArgumentParser(description='Parse source TEI specifications file and save as pickle.')
+arg_parser.add_argument('-xml', type=str, help='input XML file', required=True)
+arg_parser.add_argument('-pickle', type=str, help='output pickle file', required=True)
+arguments = arg_parser.parse_args()
+
+parser = SpecificationsParser()
+specifications = parser.parse(arguments.xml)
+with open(arguments.pickle, 'wb') as pickle_file:
+    pickle.dump(specifications, pickle_file)

setup.py

@@ -6,5 +6,6 @@ setup(name='conversion_utils',
       url='https://gitea.cjvt.si/generic/conversion_utils',
       author='Cyprian Laskowski',
       author_email='cyp@cjvt.si',
-      packages=['conversion_utils'],
+      packages=['conversion_utils', 'conversion_utils.resources'],
+      include_package_data=True,
       zip_safe=True)

tests/test_jos_msd_to_properties.py

@@ -6,10 +6,7 @@ from conversion_utils.jos_msds_and_properties import SpecificationsParser, Conve
 class JosMsdToPropertiesTestCase(unittest.TestCase):
 
     def setUp(self):
-        specifications_file_name = os.path.join(os.path.dirname(__file__), '../resources/msd-sl.spc.xml')
-        parser = SpecificationsParser()
-        specifications = parser.parse(specifications_file_name)
-        self.converter = Converter(specifications)
+        self.converter = Converter()
 
     def test_en_en(self):
         properties = self.converter.msd_to_properties(Msd('Ncfpd', 'en'), 'en')

tests/test_jos_properties_to_msd.py

@@ -6,10 +6,7 @@ from conversion_utils.jos_msds_and_properties import SpecificationsParser, Conve
 class JosPropertiesToMsdTestCase(unittest.TestCase):
 
     def setUp(self):
-        specifications_file_name = os.path.join(os.path.dirname(__file__), '../resources/msd-sl.spc.xml')
-        parser = SpecificationsParser()
-        specifications = parser.parse(specifications_file_name)
-        self.converter = Converter(specifications)
+        self.converter = Converter()
 
     def test_en_en(self):
         msd = self.converter.properties_to_msd(Properties('noun', {'type':'common', 'gender':'feminine'}, {'number':'dual', 'case':'nominative'}, 'en'), 'en')