bug fixes

news_fetch/utils_storage/models.py

@@ -10,6 +10,8 @@ import datetime
 from . import helpers
 config = configuration.main_config["DOWNLOADS"]
 slack_config = configuration.main_config["SLACK"]
+FILE_SIZE_THRESHOLD = 15 * 1024 * 1024 # 15MB
+
 
 # set the nature of the db at runtime
 download_db = DatabaseProxy()
@@ -94,50 +96,34 @@ class ArticleDownload(DownloadBaseModel):
             desc = f"{self.article_url}"
         return f"ART [{desc}]"
 
-    @property
-    def slack_info(self):
-        status = [":x: No better version available", ":gear: Verification pending", ":white_check_mark: Verified by human"][self.verified + 1]
-        content = "\n>" + "\n>".join(self.summary.split("\n"))
-        file_status, msg = self.file_status()
-        if not file_status:
-            return [msg]
-        
-        # everything alright: generate real content
-        # first the base file
-        if self.file_name[-4:] == ".pdf":
-            answer = [{ # main reply with the base pdf
-                "reply_text" : f"*{self.title}*\n{status}\n{content}",
-                "file_path" : self.save_path + self.file_name 
-            }]
-        else: # don't upload if the file is too big!
-            location = f"Not uploaded to slack, but the file will be on the NAS:\n`{self.fname_nas}`"
-            answer = [{ # main reply with the base pdf
-                "reply_text" : f"*{self.title}*\n{status}\n{content}\n{location}",
-                "file_path" : None 
-            }]
+    def mail_info(self):
+        summary = "\n> " + "\n> ".join(self.summary.split("\n"))
+        answer_text = f"[{self.article_url}]({self.article_url})\n\n" # first the url
+        answer_files = []
+        # displays the summary in a blockquote
+
+        status = self.file_status
+        if status == 1: # file_name was empty
+            return None # there has been an error do not send any message
+        elif status == 2: # no file found at specified location
+            answer_text += f"*{self.title}*\n{summary}\nFilename: {self.file_name}"
+        elif status == 3: # file found but deemed too big
+            location = f"File not sent directly. Location on NAS:\n`{self.fname_nas}`"
+            answer_text += f"*{self.title}*\n{summary}\n{location}"
+        else: # everything nominal
+            answer_text += f"*{self.title}*\n{summary}"
+            answer_files.append(self.save_path + self.file_name)
 
         # then the related files
-        rel_text = ""
-        for r in self.related:
-            fname = r.related_file_name
-            lentry = "\n• `{}` ".format(self.fname_nas(fname))
-            if fname[-4:] == ".pdf": # this is a manageable file, directly upload
-                f_ret = self.save_path + fname
-                answer.append({"reply_text":"", "file_path" : f_ret})
-            else: # not pdf <=> too large. Don't upload but mention its existence
-                lentry += "(not uploaded to slack, but the file will be on the NAS)"
-                
-            rel_text += lentry
-
-        if rel_text:
-            rel_text = answer[0]["reply_text"] = answer[0]["reply_text"] + "\nRelated files:\n" + rel_text
+        if self.related:
+            rel_text = "Related files on NAS:"
+            for r in self.related:
+                fname = r.related_file_name
+                rel_text += f"\n• `{self.fname_nas(fname)}` "
+            
+            answer_text += "\n\n" + rel_text
         
-        return answer
-
-    @property
-    def mail_info(self):
-        base = [{"reply_text": f"[{self.article_url}]({self.article_url})\n", "file_path":None}] + self.slack_info
-        return [{"reply_text": markdown.markdown(m["reply_text"]), "file_path": m["file_path"]} for m in base]
+        return markdown.markdown(answer_text), answer_files
 
 
     def set_authors(self, authors):
@@ -157,18 +143,21 @@ class ArticleDownload(DownloadBaseModel):
                 article = self,
                 related_file_name = r
             )
-
+    
+    @property
     def file_status(self):
+        """0 = file exists, 1 = no file name!, 2 = file does not exit,3 = file exists but is too large"""
         if not self.file_name:
             logger.error(f"Article {self} has no filename!")
-            return False, {"reply_text": "Download failed, no file was saved.", "file_path": None}
-        
+            return 2
         file_path_abs = self.save_path + self.file_name
         if not os.path.exists(file_path_abs):
             logger.error(f"Article {self} has a filename, but the file does not exist at that location!")
-            return False, {"reply_text": "Can't find file. Either the download failed or the file was moved.", "file_path": None}
+            return 2
+        if (os.path.splitext(file_path_abs)[1] != ".pdf") or (os.path.getsize(file_path_abs) > FILE_SIZE_THRESHOLD):
+            logger.warning(f"Article {self} has a file that exceeds the file size limit.")
+            return 3
 
-        return True, {}
 
 
 class ArticleAuthor(DownloadBaseModel):