email

app.py

@@ -22,18 +22,15 @@ def set_profile(name):
 
 
 def edit_profile(profile, author_name):
-    # names = author_name.split(" ")
-    # for n in names:
-    #     if len(n) == 0: 
-    #         gr.Info("Please input standard name format.")
-    #         return "", False
-    #     elif n[0].islower(): 
-    #         gr.Info("Please input standard name format.")
-    #         return "", False
 
     msg = agent.edit_profile(profile, author_name)
     gr.Info("Edit profile successfully!")
     return profile
+    
+def sign_email(profile, email):
+    msg = agent.sign_email(profile, email)
+    gr.Info(email + " signs up successfully!")
+    # return profile
 
 def confirm_date(date, profile_input):
     # Simulate fetching data based on the selected date
@@ -97,7 +94,7 @@ with gr.Blocks(css="""#chat_container {height: 820px; width: 1000px; margin-left
                 
                 ✨ **Guidance**: 
                 
-                Step (1) Enter researcher name and generate research profile in "Set your profile!"🧑‍💼
+                Step (1) Enter researcher name and generate research profile in "Set your profile!"; Here you also can sign up with email to get monthly research news 🧑‍💼
                 
                 Step (2) Select time range and get relevant topic trend and ideas in "Get research trend and ideas!"💡
 
@@ -114,33 +111,41 @@ with gr.Blocks(css="""#chat_container {height: 820px; width: 1000px; margin-left
     with gr.Accordion("Set your profile!", open=True):
         gr.Markdown(
             """
-            You can input your name in standard format to get your profile from arxiv here. Standard examples: Yoshua Bengio. Wrong examples: yoshua bengio, Yoshua bengio, yoshua Bengio.
+            Input your name: You can input your name in standard format to get your profile from arxiv here. Standard examples: Yoshua Bengio. Wrong examples: yoshua bengio, Yoshua bengio, yoshua Bengio.
             """
         )
         with gr.Row():
-            with gr.Column(scale=2, min_width=300):
+            with gr.Column(scale=2, min_width=300, variant = 'compact'):
                 name_input = gr.Textbox(label="Input your name:")
                 set_button = gr.Button("Set Profile")
-            profile_text = gr.Textbox(label="Generated profile (can be edited):", interactive=True, scale=7, lines=5, max_lines=5)
-            edit_button = gr.Button("Edit Profile", scale=1)
+            with gr.Column(scale=8, min_width=300):
+                with gr.Row(variant = 'compact'):
+                    profile_text = gr.Textbox(label="Generated profile (can be edited):", interactive=True, scale=7, lines=5, max_lines=5)
+                    edit_button = gr.Button("Edit Profile", scale=1)
         set_button.click(set_profile, inputs=name_input, outputs=[profile_text])
         edit_button.click(edit_profile, inputs=[profile_text, name_input], outputs=[profile_text])
-
+        
     with gr.Accordion("Get research trend and ideas!", open=True):
         gr.Markdown(
             """
-            We will give you personalized research trend and ideas if you have set your profile. Otherwise, general research trend will be provided.
+            (1) Input your email: You can sign up with your email and we will send research trend, ideas, and top papers related to your profile on 1st of every month.
+
+            (2) Select time range: We will give you personalized research trend and ideas under selected time range if you have set your profile. Otherwise, general research trend will be provided.
             """
         )
         with gr.Column():
             with gr.Row():
-                with gr.Column(scale=2, min_width=300):
-                    # gr.Dropdown(
-                    #     ["day", "week", "bird"], label="Select time range", info="Will add more animals later!"
-                    # ),
-                    date_choice = gr.Radio(["day", "week", "all"], label="Select Time Range", value="day")
-                    date_button = gr.Button("Confirm")
-                papers_text = gr.Textbox(label="Trend Papers", interactive=False, scale=8, lines=5, max_lines=5)
+                with gr.Column(scale=2.1, min_width=300):
+                    with gr.Column(scale=2.1, min_width=300, variant = 'compact'):
+                        # gr.Dropdown(
+                        #     ["day", "week", "bird"], label="Select time range", info="Will add more animals later!"
+                        # ),
+                        email_input = gr.Textbox(label="Input your email:")
+                        sign_button = gr.Button("Sign Up")
+                    with gr.Column(scale=2.1, min_width=300, variant = 'compact'):
+                        date_choice = gr.Radio(["day", "week", "all"], label="Select time range: ", value="day")
+                        date_button = gr.Button("Confirm")
+                papers_text = gr.Textbox(label="Trend Papers", interactive=False, scale=8, lines=12, max_lines=15)
 
             with gr.Row():
                 topic_text = gr.Textbox(label="Topic Trend", interactive=False, scale=5, lines=12, max_lines=15)
@@ -148,7 +153,7 @@ with gr.Blocks(css="""#chat_container {height: 820px; width: 1000px; margin-left
                 ideas_text = gr.Textbox(label="Ideas Related to Topic Trend", interactive=False, scale=5, lines=12, max_lines=15)
 
         date_button.click(confirm_date, inputs=[date_choice, profile_text], outputs=[topic_text, papers_text, ideas_text])
-
+        sign_button.click(sign_email, inputs=[profile_text, email_input])
     with gr.Accordion("Chat with Arxiv Copilot!", open=True):
         gr.Markdown(
             """

arxiv_agent.py

@@ -10,11 +10,13 @@ from pathlib import Path
 import requests
 from datasets import load_dataset_builder
 import warnings
+from mail_simple import *
+
 warnings.filterwarnings("ignore")
 os.environ['KMP_DUPLICATE_LIB_OK']='True'
 from utils import *
 import thread6
-MAX_DAILY_PAPER = 200
+MAX_DAILY_PAPER = int(os.environ['MAX_DAILY_PAPER'])
 DAY_TIME = 60 * 60 * 24
 DAY_TIME_MIN = 60 * 24
 DATA_REPO_ID = "cmulgy/ArxivCopilot_data"
@@ -150,6 +152,10 @@ def dailySave(agent_ls):
                 
             with open(agent.profile_path,"w") as f:
                 json.dump(agent.profile,f)
+
+            with open(agent.email_pool_path,"w") as f:
+                json.dump(agent.email_pool,f)
+
             with open(agent.comment_path,"w") as f:
                 json.dump(agent.comment,f)
 
@@ -160,6 +166,8 @@ class ArxivAgent:
         self.thought_path = DATASET_DIR / "dataset/thought.json"
         self.trend_idea_path = DATASET_DIR / "dataset/trend_idea.json"
         self.profile_path = DATASET_DIR / "dataset/profile.json"
+        self.email_pool_path = DATASET_DIR / "dataset/email.json"
+
         self.comment_path = DATASET_DIR / "dataset/comment.json"
 
         self.embedding_path = DATASET_DIR / "dataset/paper_embedding.pkl"
@@ -179,7 +187,9 @@ class ArxivAgent:
         self.download()
         try:
             thread6.run_threaded(dailyDownload, [self])
+            thread6.run_threaded(monthlyEmail, [self])
             thread6.run_threaded(dailySave, [self])
+            
         except:
             print("Error: unable to start thread")
 
@@ -188,7 +198,13 @@ class ArxivAgent:
         self.profile[author_name]=profile
 
         return "Successfully edit profile!"
-    
+
+    def sign_email(self, profile, email):
+
+        self.email_pool[email]=profile
+        sign_up_email(email, profile)
+        return "Successfully sign up!"
+
     def get_profile(self, author_name):
         if author_name == "": return None
 
@@ -213,6 +229,15 @@ class ArxivAgent:
                 offset_day = today - datetime.timedelta(days=i)
                 str_day = offset_day.strftime("%m/%d/%Y")
                 
+                if str_day in self.paper:
+                    # print(str_day)
+                    paper_by_date[str_day] = self.paper[str_day]
+                    chunk_embedding_date[str_day] = self.paper_embedding[str_day]
+        elif method == "month":
+            for i in range(30):
+                offset_day = today - datetime.timedelta(days=i)
+                str_day = offset_day.strftime("%m/%d/%Y")
+                
                 if str_day in self.paper:
                     # print(str_day)
                     paper_by_date[str_day] = self.paper[str_day]
@@ -237,8 +262,7 @@ class ArxivAgent:
                     idea = self.trend_idea[profile][key_update][method]["idea"]
                     isQuery = True 
 
-        # import pdb
-        # pdb.set_trace()
+
         if not(isQuery):
             trend, paper_link = summarize_research_field(profile, "Machine Learning", dataset,data_chunk_embedding) # trend
             reference = papertitleAndLink(paper_link)
@@ -424,6 +448,20 @@ class ArxivAgent:
                 m = {}
         self.profile = m.copy()
 
+        filename = self.email_pool_path
+        # if os.path.exists(filename):
+        try:
+            hf_hub_download(repo_id=DATA_REPO_ID, filename="dataset/email.json", local_dir = ".", repo_type="dataset")    
+            with open(filename,"rb") as f:
+                content = f.read()
+                if not content:
+                    m = {}
+                else:
+                    m = json.loads(content)
+        except:
+            with open(filename, mode='w', encoding='utf-8') as ff:
+                m = {}
+        self.email_pool = m.copy()
 
         filename = self.thought_path
         filename_emb = self.thought_embedding_path

mail_simple.py

@@ -0,0 +1,172 @@
+import smtplib
+from email.mime.multipart import MIMEMultipart
+from email.mime.text import MIMEText
+import time
+import schedule
+from functools import partial
+import markdown
+from datetime import date
+
+DAY_TIME = 60 * 60 * 24
+import os
+APP_PASSWORD = os.environ['APP_PASSWORD']
+def send_email(sender_email, app_password, recipient_email, subject, body):
+    # Create the email message
+    msg = MIMEMultipart()
+    msg['From'] = sender_email
+    msg['Subject'] = subject
+    msg['To'] = recipient_email
+    msg.attach(MIMEText(body, 'html'))
+
+    try:
+        # Set up the SMTP server
+        server = smtplib.SMTP('smtp.gmail.com', 587)
+        server.starttls()  # Secure the connection
+        server.login(sender_email, app_password)  # Log in to your Gmail account
+        
+        # Send the email
+        # for recipient_email in recipient_emails:
+            
+            
+        server.sendmail(sender_email, recipient_email, msg.as_string())
+
+        # print("Email sent successfully!" + recipient_email)
+    except Exception as e:
+        print(f"Failed to send email: {e}")
+    finally:
+        server.quit()  # Terminate the SMTP session
+
+def sign_up_email(email, profile):
+    sender_email = "[email protected]"    
+    app_password = "mqxq jzhq tjhz dwda"
+    recipient_email = email
+    # for recipient_email in email_pool.keys():
+    email_input = {}
+    email_input['profile'] = profile
+    body = \
+('''
+# Welcome to Arxiv Copilot Monthly Updates!
+
+## Hello,
+
+Thank you for signing up for our monthly update service! We are thrilled to have you on board and look forward to providing you with the latest research topics, ideas, and papers tailored to your interests.
+
+The updates will be sent on 1st of every month at 20:30 (EDT).
+
+---
+
+### 📝 **Your Profile Summary**
+
+{profile}
+
+
+---
+
+### 🔔 **What to Expect**
+
+- Each month, you will receive an email with:
+    - **Trending Research Topics**: Stay updated on what's popular in your field.
+    - **Research Ideas**: Get inspired with new ideas tailored to your interests.
+    - **Recommended Papers**: Discover papers that match your research focus.
+
+
+Thank you for choosing Arxiv Copilot. We are excited to support your research journey and keep you informed with the latest developments in your field.
+
+Best regards,  
+**The Arxiv Copilot Team**
+
+---
+
+*If you are unsatisfied with your profile, please edit it at Arxiv Copilot (https://huggingface.co/spaces/ulab-ai/ArxivCopilot) and sign up again!*
+''').format_map(email_input)
+    body = markdown.markdown(body)
+    # body = MIMEText(body, "html")
+
+    subject = "ArxivCopilot Monthly"
+    send_email(sender_email, app_password, recipient_email, subject, body)
+
+
+# Example usage
+def send_to_users(agent):
+    if date.today().day != 1:
+        return
+    sender_email = "[email protected]"    
+    app_password = APP_PASSWORD
+    for recipient_email in agent.email_pool.keys():
+        profile_input = agent.email_pool[recipient_email]
+        topics, papers, ideas = agent.select_date("month", profile_input)
+        email_input = {}
+        email_input['profile'] = profile_input
+        email_input['topics'] = topics[0]
+        email_input['ideas'] = ideas[0]
+        paper_input = ""
+        for paper in papers.split(';\n'):
+            paper_input = paper_input + "- " + paper + ";\n"
+
+        email_input['papers'] = paper_input
+        body =  \
+('''
+# Monthly Update from Arxiv Copilot
+
+## Hello,
+
+We are excited to share your personalized monthly update from Arxiv Copilot! Here’s what we have for you this month:
+
+---
+
+### 📝 **Your Profile Summary**
+
+{profile}
+
+
+---
+
+### 🔥 **Trending Research Topics**
+
+{topics}
+
+---
+
+### 💡 **Research Ideas for You**
+
+{ideas}
+
+---
+
+### 📄 **Recommended Papers**
+
+We have curated a list of papers you might find interesting:
+
+{papers}
+
+
+
+Thank you for being a part of Arxiv Copilot. We look forward to supporting your research journey.
+
+Best regards,  
+**The Arxiv Copilot Team**
+
+---
+
+*If you are unsatisfied with your profile, please edit it at Arxiv Copilot (https://huggingface.co/spaces/ulab-ai/ArxivCopilot) and sign up again!*
+''').format_map(email_input)
+        
+        body = markdown.markdown(body)
+        subject = "ArxivCopilot Monthly"
+        send_email(sender_email, app_password, recipient_email, subject, body)
+        print("sent to"+recipient_email)
+
+
+
+def monthlyEmail(agent_ls):
+
+    agent = agent_ls[0]
+    job_with_para = partial(send_to_users, agent)
+    # schedule.every().minute.do(job_with_para)
+    # schedule.every().month.do(job_with_para).day.at("00:00")
+    schedule.every().day.at("20:30", "America/New_York").do(job_with_para)
+
+    while True:
+        schedule.run_pending()
+        time.sleep(59)
+       

requirements.txt

@@ -5,3 +5,4 @@ arxiv
 requests
 openai==0.28
 thread6
+schedule