Merge pull request #16 from AutoLLM/no-email-in-config

advanced_usage.md

@@ -61,8 +61,8 @@ An alternative way to get started using this repository is to:
    - `MAIL_CONNECTION` (see above)
    - `MAIL_PASSWORD` (only if you don't have `MAIL_CONNECTION` set)
    - `MAIL_USERNAME` (only if you don't have `MAIL_CONNECTION` set)
-   - `FROM_EMAIL` (only if you don't have it set in `config.yaml`)
-   - `TO_EMAIL` (only if you don't have it set in `config.yaml`)
+   - `FROM_EMAIL`
+   - `TO_EMAIL`
 6. Manually trigger the action or wait until the scheduled action takes place.
 
 ### Running as a github action without emails 
@@ -84,8 +84,8 @@ If you do not wish to fork this repository, and would prefer to clone and run it
 5. Set the following secrets as environment variables: 
    - `OPENAI_API_KEY`
    - `SENDGRID_API_KEY` (only if using SendGrid)
-   - `FROM_EMAIL` (only if using SendGrid and if you don't have it set in `config.yaml`. Note that this value must match the email you used to create the SendGrid Api Key.)
-   - `TO_EMAIL` (only if using SendGrid and if you don't have it set in `config.yaml`)
+   - `FROM_EMAIL` (only if using SendGrid. Note that this value must match the email you used to create the SendGrid Api Key.)
+   - `TO_EMAIL` (only if using SendGrid)
 6. Run `python action.py`.
 7. If you are not using SendGrid, the html of the digest will be written to `digest.html`. You can then use your favorite webbrowser to view it.
 

config.yaml

@@ -5,17 +5,6 @@ topic: "Computer Science"
 # Including more categories will result in more calls to the large language model
 categories: ["Artificial Intelligence", "Computation and Language"]
 
-# The email address that the digest will be sent from. must be the address matching
-# your sendgrid api key.
-# Leaving this empty will cause the script to use the
-# FROM_EMAIL environment variable instead
-from_email: ""
-
-# The email address you are going to send the digest to
-# Leaving this empty will cause the script to use the
-# TO_EMAIL environment variable instead
-to_email: ""
-
 # Relevance score threshold. abstracts that receive a score less than this from the large language model
 # will have their papers filtered out.
 #

src/action.py

@@ -116,27 +116,27 @@ if __name__ == "__main__":
 
     topic = config["topic"]
     categories = config["categories"]
-    from_email = config.get("from_email") or os.environ.get("FROM_EMAIL")
-    to_email = config.get("to_email") or os.environ.get("TO_EMAIL")
+    from_email = os.environ.get("FROM_EMAIL")
+    to_email = os.environ.get("TO_EMAIL")
     threshold = config["threshold"]
     interest = config["interest"]
+    body = generate_body(topic, categories, interest, threshold)
     with open("digest.html", "w") as f:
-        body = generate_body(topic, categories, interest, threshold)
         f.write(body)
-        if os.environ.get('SENDGRID_API_KEY', None):
-            sg = SendGridAPIClient(api_key=os.environ.get('SENDGRID_API_KEY'))
-            from_email = Email(from_email)  # Change to your verified sender
-            to_email = To(to_email)
-            subject = date.today().strftime("Personalized arXiv Digest, %d %b %Y")
-            content = Content("text/html", body)
-            mail = Mail(from_email, to_email, subject, content)
-            mail_json = mail.get()
-
-            # Send an HTTP POST request to /mail/send
-            response = sg.client.mail.send.post(request_body=mail_json)
-            if response.status_code >= 200 and response.status_code <= 300:
-                print("Send test email: Success!")
-            else:
-                print("Send test email: Failure ({response.status_code}, {response.text})")
+    if os.environ.get('SENDGRID_API_KEY', None):
+        sg = SendGridAPIClient(api_key=os.environ.get('SENDGRID_API_KEY'))
+        from_email = Email(from_email)  # Change to your verified sender
+        to_email = To(to_email)
+        subject = date.today().strftime("Personalized arXiv Digest, %d %b %Y")
+        content = Content("text/html", body)
+        mail = Mail(from_email, to_email, subject, content)
+        mail_json = mail.get()
+
+        # Send an HTTP POST request to /mail/send
+        response = sg.client.mail.send.post(request_body=mail_json)
+        if response.status_code >= 200 and response.status_code <= 300:
+            print("Send test email: Success!")
         else:
-            print("No sendgrid api key found. Skipping email")
+            print("Send test email: Failure ({response.status_code}, {response.text})")
+    else:
+        print("No sendgrid api key found. Skipping email")

src/utils.py

@@ -96,6 +96,8 @@ def openai_completion(
     ):
         batch_decoding_args = copy.deepcopy(decoding_args)  # cloning the decoding_args
 
+        backoff = 3
+
         while True:
             try:
                 shared_kwargs = dict(
@@ -115,7 +117,7 @@ def openai_completion(
                     completion_batch = openai.Completion.create(prompt=prompt_batch, **shared_kwargs)
 
                 choices = completion_batch.choices
-    
+
                 for choice in choices:
                     choice["total_tokens"] = completion_batch.usage.total_tokens
                 completions.extend(choices)
@@ -125,7 +127,11 @@ def openai_completion(
                 if "Please reduce your prompt" in str(e):
                     batch_decoding_args.max_tokens = int(batch_decoding_args.max_tokens * 0.8)
                     logging.warning(f"Reducing target length to {batch_decoding_args.max_tokens}, Retrying...")
+                elif not backoff:
+                    logging.error("Hit too many failures, exiting")
+                    raise e
                 else:
+                    backoff -= 1
                     logging.warning("Hit request rate limit; retrying...")
                     time.sleep(sleep_time)  # Annoying rate limit on requests.