for 4

lab_5_scrapper/scrapper.py

@@ -326,7 +326,7 @@ def _fill_article_with_meta_information(self, article_soup: BeautifulSoup) -> No
         topics = article_soup.find(class_="argcat")
         self.article.topics = topics.text.replace('\n', '')
         time = article_soup.find('time', itemprop="datePublished")
-        self.article.time = time.text.replace('\n', '')
+        self.article.time = self.unify_date_format(time.attrs.get('datetime'))
 
     def unify_date_format(self, date_str: str) -> datetime.datetime:
         """
@@ -339,6 +339,7 @@ def unify_date_format(self, date_str: str) -> datetime.datetime:
             datetime.datetime: Datetime object
         """
 
+
     def parse(self) -> Union[Article, bool, list]:
         """
         Parse each article.

lab_6_pipeline/pipeline.py

@@ -2,7 +2,6 @@
 Pipeline for CONLL-U formatting.
 """
 # pylint: disable=too-few-public-methods, unused-import, undefined-variable, too-many-nested-blocks
-import pathlib
 import stanza
 
 try:
@@ -11,9 +10,13 @@
     DiGraph = None  # type: ignore
     print('No libraries installed. Failed to import.')
 
-from core_utils.article.article import Article, get_article_id_from_filepath, split_by_sentence
+import pathlib
+import spacy_udpipe
+
+from core_utils.article.article import (Article, ArtifactType, get_article_id_from_filepath,
+                                        split_by_sentence)
 from core_utils.article.io import from_raw, to_cleaned
-from core_utils.constants import ASSETS_PATH
+from core_utils.constants import ASSETS_PATH, UDPIPE_MODEL_PATH
 from core_utils.pipeline import (AbstractCoNLLUAnalyzer, CoNLLUDocument, LibraryWrapper,
                                  PipelineProtocol, StanzaDocument, TreeNode)
 

requirements.txt

@@ -1,10 +1,10 @@
 beautifulsoup4==4.12.3
 bs4==0.0.2
-requests~=2.31.0
-pydantic~=2.7.2
-DateTime~=5.5
-stanza~=1.8.2
-spacy~=3.7.4
-networkx~=3.2.1
-pillow~=10.3.0
-matplotlib~=3.8.4
+datetime==5.5
+matplotlib==3.8.4
+networkx==3.2.1
+pillow==10.3.0
+pydantic==2.7.2
+requests==2.31.0
+spacy==3.7.4
+stanza==1.8.2