Redmine #1487: updated pipeline for classla/obeliks changes

package/structure_assignment/constants.py

@@ -24,6 +24,6 @@ FILE_MAP = {'strings-list': 'strings.txt',
 NLP_CONFIG_MAP = {
     'type': 'standard_jos',
     'processors': 'tokenize,pos,lemma,depparse',
-    'tokenize_pretokenized': True,
+    'tokenize_pretokenized': 'conllu',
     'pos_use_lexicon': True,
 }

package/structure_assignment/pipeline.py

@@ -5,10 +5,7 @@ from types import SimpleNamespace
 import lxml.etree as lxml
 
 import obeliks
-
 import classla
-from classla import Document
-from classla.utils.conll import CoNLL
 
 from structure_assignment.constants import *
 from structure_assignment.tweak_conllu import tweak as tweak_conllu
@@ -134,11 +131,11 @@ class Pipeline:
         print('Parsing with classla ...')
         input_file_name = self.file_map['obeliks-tweaked']
         output_file_name = self.file_map['classla-parsed']
-        doc = Document(text=None)
-        conll_file = CoNLLFile(filename=input_file_name)
-        doc.conll_file = conll_file
-        result = nlp(doc)
-        result.conll_file.write_conll(output_file_name)
+        with open(input_file_name, 'r') as input_file:
+            input_conllu = input_file.read()
+        doc = self.nlp(input_conllu)
+        with open(output_file_name, 'w') as output_file:
+            output_file.write(doc.to_conll())
 
     def do_translate_jos(self):
         print('Translating JOS ...')

scripts/setup.sh

@@ -17,7 +17,7 @@ pip install lxml
 pip install psycopg2cffi
 pip install sqlalchemy
 pip install classla
-python -c "import classla; classla.download('sl_ssj_jos')" <<< $'Y\nresources/classla'
+python -c "import classla; classla.download('sl', type='standard_jos')" <<< $'Y\nresources/classla'
 pip install obeliks
 pip install nova_slovnica/python/package/
 pip install luscenje_struktur/