(#29)

* Improve `Article` __repr__ method
* Improve logs
* Modify attributes names in `Cluster`
* Add function create_new_cluster into `ClusterUtils`
* Improve `get_similar_articles` route

api/server_api/api_logic.py

@@ -48,7 +48,7 @@ def get_similar_articles_data(self, article_url: str) -> List[ArticleApiData]:
             # collect needed articles data
             self.server_logger.info(f"Got {len(articles)} similar articles")
             for index, article in enumerate(articles):
-                self.server_logger.debug(f"({index + 1}) article data: `{article.convert_to_dict()}`")
+                self.server_logger.debug(f"({index + 1}) Article data: `{str(article)}`")
                 article_api_object = self.__get_convert_article_api_data(article=article)
                 similar_articles.append(article_api_object)
         except Exception as e:

api/server_api/routes.py

@@ -3,6 +3,7 @@
 from api.server_api import app
 from api.server_api.api_logic import APILogic
 from api.server_api.exceptions import ArticleNotFoundException, NoSimilarArticlesException, GetSimilarArticlesException
+from logger import get_current_logger
 
 
 @app.route('/')
@@ -13,20 +14,26 @@ def index():
 
 @app.route('/get_similar_articles', methods=['POST'])
 def get_similar_articles():
+    logger = get_current_logger()
+    logger.debug(f"Try getting similar articles")
     return_data = {"articles_data": list(), "error_msg": "", "succeeded": False}
     if 'url' not in request.args:
         return_data["error_msg"] = "url required"
-        return return_data
-    try:
-        api_logic = APILogic()
-        article_url: str = request.args['url']
-        similar_articles_data = api_logic.get_similar_articles_data(article_url=article_url)
-        return_data["articles_data"] = similar_articles_data
-        return_data["succeeded"] = True
-    except ArticleNotFoundException:
-        return_data["error_msg"] = "article not found in db"
-    except NoSimilarArticlesException:
-        return_data["error_msg"] = "no similar articles found"
-    except GetSimilarArticlesException:
-        return_data["error_msg"] = "error getting similar articles"
+        logger.logger.warning(f"Didn't get the current url of the article")
+    else:
+        try:
+            api_logic = APILogic()
+            article_url: str = request.args['url']
+            similar_articles_data = api_logic.get_similar_articles_data(article_url=article_url)
+            return_data["articles_data"] = similar_articles_data
+            return_data["succeeded"] = True
+            logger.info(f"Got {len(similar_articles_data)} similar articles")
+        except ArticleNotFoundException:
+            return_data["error_msg"] = "article not found in db"
+        except NoSimilarArticlesException:
+            return_data["error_msg"] = "no similar articles found"
+        except GetSimilarArticlesException:
+            return_data["error_msg"] = "error getting similar articles"
+
+    logger.info(f"(get_similar_articles) return data: `{return_data}`")
     return return_data

db_driver/db_objects/article.py

@@ -17,9 +17,7 @@ class Article:
     images: Optional[List[str]] = None
 
     def __repr__(self) -> str:
-        string = ''
-        for prop, value in vars(self).items():
-            string += f"{str(prop)}: {str(value)}\n"
+        string = f'(domain: `{self.domain}`, url: `{self.url}`, title: `{self.title}`)'
         return string
 
     def convert_to_dict(self) -> dict:

db_driver/db_objects/cluster.py

@@ -10,8 +10,8 @@ class Cluster:
     main_article_id: str
     creation_time: datetime
     last_updated: datetime
-    websites: list[str]
-    category: Optional[str] = None
+    domains: list[str]
+    categories: Optional[str] = None
 
     def convert_to_dict(self) -> dict:
         return asdict(self)

server_utils/db_utils/article_utils.py

@@ -66,6 +66,7 @@ def get_article(self, article_url: str) -> Union[Article, None]:
             article = article_object
         except DataNotFoundDBException as e:
             self.logger.warning(f"Error get article by article url: `{article_url}` - {str(e)}")
+        self.logger.info(f"Got article from db, article_id: `{article.article_id}`, url: `{article.url}`")
         return article
 
     def get_articles(self, articles_id: List[str]) -> List[Article]:

server_utils/db_utils/cluster_utils.py

@@ -1,17 +1,53 @@
+import uuid
+from datetime import datetime
+from typing import List
+
 from db_driver import get_current_db_driver
+from db_driver.db_objects.article import Article
 from db_driver.db_objects.cluster import Cluster
 from db_driver.db_objects.db_objects_utils import get_db_object_from_dict
 from db_driver.utils.consts import DBConsts
-# from logger import get_current_logger
+from logger import get_current_logger, log_function
+from server_utils.db_utils.article_utils import ArticleUtils
 
 
 class ClusterUtils:
     def __init__(self):
-        # self.logger = get_current_logger()
+        self.logger = get_current_logger()
         self._db = get_current_db_driver()
 
+    @log_function
+    def create_new_cluster(self, article: Article, classified_categories: List[str] = None) -> str:
+        try:
+            current_time = datetime.now()
+            categories = classified_categories if classified_categories else list()
+            cluster_data = {
+                "cluster_id": str(uuid.uuid4()),
+                "articles_id": [article.article_id],
+                "main_article_id": article.article_id,
+                "creation_time": current_time,
+                "last_updated": current_time,
+                "domains": [article.domain],
+                "categories": categories
+            }
+            cluster: Cluster = Cluster(**cluster_data)
+            _id = self._db.insert_one(table_name=DBConsts.CLUSTERS_TABLE_NAME, data=cluster.convert_to_dict())
+            self.logger.info(f"Inserted cluster inserted_id: `{_id}`, cluster_id: `{cluster.cluster_id}`")
+            return _id
+        except Exception as e:
+            desc = f"Error insert cluster - {str(e)}"
+            self.logger.exception(desc)
+
     def get_cluster(self, cluster_id: str) -> Cluster:
         data_filter = {"cluster_id": cluster_id}
         cluster_data = self._db.get_one(table_name=DBConsts.CLUSTERS_TABLE_NAME, data_filter=data_filter)
         cluster_object: Cluster = get_db_object_from_dict(object_dict=cluster_data, class_instance=Cluster)
         return cluster_object
+
+
+# For debug
+if __name__ == '__main__':
+    article_utils = ArticleUtils()
+    new_article: Article = article_utils.get_article(article_url="https://www.bbc.com/news/world-europe-65471904")
+    cluster_utils = ClusterUtils()
+    cluster_utils.create_new_cluster(article=new_article, classified_categories=["finance", "bitcoin"])