Checked if msds are valid

conversion_utils/jos_msds_and_properties.py

@@ -53,12 +53,16 @@ LEVEL_EXCEPTIONS = {('pronoun', 2, 'čezme'), ('zaimek', 2, 'čezme'),
 class Specifications:
     """JOS specifications with list of all word categories."""
 
-    def __init__(self, *categories):
-        self.categories = list(categories)
+    def __init__(self):
+        self.categories = []
+        self.codes_map = {'en':set(), 'sl':set()}
 
     def add_category(self, category):
         self.categories.append(category)
         
+    def add_code(self, code, language):
+        self.codes_map[language].add(code)
+
     def find_category_by_code(self, char, language):
         return next((category for category in self.categories if category.codes.get(language) == char), None)
         
@@ -145,10 +149,17 @@ class SpecificationsParser:
 
     def parse(self, file_name):
         root = lxml.parse(file_name).getroot()
-        specifications = Specifications()
         div_elements = xpath_find(root, 'tei:div')
+        specifications = Specifications()
         for div_element in div_elements:
-            if (re.match(r'^msd\..-sl', get_xml_id(div_element))):
+            xml_id = get_xml_id(div_element)
+            if (xml_id == 'msd.msds-sl'):
+                msd_elements = xpath_find(div_element, 'tei:table/tei:row[@role="msd"]')
+                for msd_element in msd_elements:
+                    msd_codes = self.get_cell_pair(msd_element, 'msd')
+                    specifications.add_code(msd_codes.get('en').capitalize(), 'en')
+                    specifications.add_code(msd_codes.get('sl').capitalize(), 'sl')
+            elif (re.match(r'^msd\..-sl', xml_id)):
                 category_element = xpath_find(div_element, 'tei:table/tei:row[@role="type"]')[0]
                 category_names = self.get_cell_pair(category_element, 'value')
                 category_codes = self.get_cell_pair(category_element, 'code')
@@ -214,6 +225,9 @@ class Converter:
     def msd_to_properties(self, msd, language, lemma=None):
         """Convert Msd to Properties (possibly in the other language)."""
  
+        if (msd.code not in self.specifications.codes_map[msd.language]):
+            exit('[ERROR] msd {} is unknown'.format(msd.code))
+
         category_char = msd.code[0].lower()
         value_chars = msd.code[1:]
         category = self.specifications.find_category_by_code(category_char, msd.language)