upload to new gitea

news_fetch/app/utils_storage/models.py

@@ -45,7 +45,11 @@ class ArticleDownload(DownloadBaseModel):
     # ... are added through foreignkeys
 
     def __str__(self) -> str:
-        return f"ART [{self.title} -- {self.source_name}]"
+        if self.title != '' and self.source_name != '':
+            desc = f"{shorten_name(self.title)} -- {self.source_name}"
+        else:
+            desc = f"{self.article_url}"
+        return f"ART [{desc}]"
 
     ## Useful Properties
     @property
@@ -255,7 +259,7 @@ class Message(ChatBaseModel):
     # reaction
 
     def __str__(self) -> str:
-        return "MSG [{}]".format(self.text[:min(len(self.text), 30)].replace('\n','/') + '...')
+        return "MSG [{}]".format(shorten_name(self.text).replace('\n','/'))
 
     @property
     def slack_ts(self):
@@ -319,4 +323,9 @@ def clear_path_name(path):
     keepcharacters = (' ','.','_', '-')
     converted = "".join([c if (c.isalnum() or c in keepcharacters) else "_" for c in path]).rstrip()
     return converted
-    
+
+def shorten_name(name, offset = 50):
+    if len(name) > offset:
+        return name[:offset] + "..."
+    else:
+        return name

news_fetch/app/utils_worker/download/browser.py

@@ -155,11 +155,11 @@ class PDFDownloader:
             hrefs = [e.get_attribute("href") for e in self.driver.find_elements_by_xpath("//a[@href]")]
         except:
             hrefs = []
-        len_old = len(hrefs)
+        # len_old = len(hrefs)
         hrefs = [h for h in hrefs \
             if not sum([(domain in h) for domain in blacklisted]) # sum([True, False, False, False]) == 1 (esp. not 0)
             ] # filter a tiny bit at least
-        self.logger.info(f"Hrefs filtered (before: {len_old}, after: {len(hrefs)})")
+        # self.logger.info(f"Hrefs filtered (before: {len_old}, after: {len(hrefs)})")
         return hrefs