Replaced JOS XML specifications with preprocessed 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)."""