Projekt Structure changed

aligner/bitext_align.py

@@ -1,18 +1,13 @@
 # -*- coding: utf-8 -*- 
 
 
-import os,sys
+from itertools import product as cp
-import re
+
-import pandas as pd
 import numpy as np
-from numpy import cumsum
+import pandas as pd
-from pandas import DataFrame
-from nltk import word_tokenize, sent_tokenize
-#import xml.etree.ElementTree as ET
-from jellyfish import levenshtein_distance as lev
-#import six
 from google.cloud import translate_v2 as translate
-from itertools import product as cp
+from jellyfish import levenshtein_distance as lev
+from nltk import sent_tokenize
 
 translate_client = translate.Client()
 

run.py

@@ -17,10 +17,10 @@ def save_validated_files_to_db():
         books_list = books_json[book_code]
         for book in books_list:
             if not book['is_validated']:
-                print(const.WARNING, 'Book : ', book['xml_file'], ' is not validated against XSD', const.END)
+                print(const.WARNING, 'XML File :: ', book['xml_file'], ' is not validated against XSD', const.END)
                 continue
             if not book['is_saved_to_db']:
-                print(const.BLUE, 'Adding Book : ', book['xml_file'], ' to the DB', const.END)
+                print(const.BLUE, 'Adding Book Data : ', book['xml_file'], ' to the DB', const.END)
                 book_dict = read_xml.parse_xml_file(book['xml_file_path'])
                 result = adb.add_book_to_db(book_code, book_dict)
                 book['is_saved_to_db'] = result
@@ -35,4 +35,4 @@ def save_validated_files_to_db():
 
 if env.check_env_variables():
     validate_all_xml_files()
-    save_validated_files_to_db()
+    # save_validated_files_to_db()

utils/constants.py

@@ -1,6 +1,6 @@
 JSON_PATH = 'json/books.json'
 
-XSD_PATH = 'xml_files/book.xsd'
+XSD_PATH = 'xml_parser/book.xsd'
 
 TRANSLATE_ENV_VAR = 'GOOGLE_APPLICATION_CREDENTIALS'