full working app

.gitignore

@@ -0,0 +1,3 @@
+.env
+__pycache__
+.DS*

app.py

@@ -0,0 +1,55 @@
+
+from annotated_text import annotation
+from json import JSONDecodeError
+import logging
+from markdown import markdown
+import requests
+
+import streamlit as st
+
+from utils.haystack import query, start_haystack
+from utils.ui import reset_results, set_initial_state, sidebar
+from utils.config import TWITTER_BEARER_TOKEN, SERPER_KEY, OPENAI_API_KEY
+
+set_initial_state()
+
+sidebar()
+
+st.write("# 👩 What would they have tweeted about this?")
+
+# if st.session_state.get("OPENAI_API_KEY"):
+if "last_k_tweets" not in st.session_state:
+    st.session_state["last_k_tweets"] = st.slider("How many tweets should we retrieve?", value=10, step=10, max_value=100, min_value=10, on_change = reset_results)
+
+agent = start_haystack(openai_key=OPENAI_API_KEY, twitter_bearer=TWITTER_BEARER_TOKEN, serper_key=SERPER_KEY, last_k_tweets=st.session_state["last_k_tweets"])
+# st.session_state["api_keys_configured"] = True
+
+# Search bar
+question = st.text_input("If the twitter account tuanacelik were to write a tweet in their style about climate change, what would it be?", on_change=reset_results) 
+run_pressed = st.button("Generate tweet")
+# else:
+#     st.write("Please provide your OpenAI Key to start using the application")
+#     st.write("If you are using a smaller screen, open the sidebar from the top left to provide your OpenAI Key 🙌")
+
+# if st.session_state.get("api_keys_configured"):
+run_query = (
+    run_pressed or question != st.session_state.question
+)
+if run_query and question:
+    reset_results()
+    st.session_state.question = question
+    with st.spinner("🔎"):
+        try:
+            st.session_state.result = query(agent, question)
+        except JSONDecodeError as je:
+            st.error(
+                "👓    An error occurred reading the results. Is the document store working?"
+            )    
+        except Exception as e:
+            logging.exception(e)
+            st.error("🐞    An error occurred during the request.")            
+            
+if st.session_state.result:
+    result = st.session_state.result
+    st.write(result["answers"][0].answer)
+            

custom_nodes/tweet_retriever.py

@@ -0,0 +1,32 @@
+import requests
+from haystack.nodes import BaseComponent
+
+class TwitterRetriever(BaseComponent):
+    outgoing_edges = 1
+    
+    def __init__(self, bearer_token: str, last_k_tweets: int = 10):
+        self.headers = {"Authorization": "Bearer {}".format(bearer_token)}
+        self.last_k_tweets = last_k_tweets
+        
+    def run(self, query: str, last_k_tweets: int = None):
+        if last_k_tweets is None:
+            last_k_tweets = self.last_k_tweets
+
+        url = f"https://api.twitter.com/1.1/statuses/user_timeline.json?screen_name={query}&count={last_k_tweets}"
+        try:
+            response = requests.request("GET", url, headers=self.headers)
+            twitter_stream = ""
+            for tweet in response.json():
+                print(tweet)
+                twitter_stream += tweet["text"] + '\n'
+        except Exception as e:
+            twitter_stream = ["Please make sure you are providing a correct, public twitter account"]
+        
+        output = {
+            "results":  f"{twitter_stream}",
+            "username": query,
+        }
+        return output
+
+    def run_batch(self):
+        pass

prompts/lfqa.yaml

@@ -0,0 +1,18 @@
+description: A long form question answering prompt
+meta:
+  authors:
+  - Tuana
+name: Tuana/lfqa
+tags:
+- question-answering
+text: |
+  Synthesize a comprehensive answer from the following most relevant paragraphs and the given question.
+  Provide a clear and concise response that summarizes the key points and information presented in the paragraphs.
+  Your answer should be in your own words and be no longer than 50 words.
+  
+  Paragraphs: {documents}
+  
+  Question: {query}
+  
+  Answer:
+version: 0.1.1

prompts/twitter_agent.yaml

@@ -0,0 +1,27 @@
+description: An experimental Agent prompt for tweet creations based on twitter voice or last tweets
+meta:
+  authors:
+  - Tuana
+name: Tuana/twitter-agent
+tags:
+- agent
+text: | 
+    You are a helpful and knowledgeable agent. To achieve your goal of generating new tweets for twitter users, you have access to the following tools:
+    {tool_names_with_descriptions}
+    To answer questions, you'll need to go through multiple steps involving step-by-step thinking and selecting appropriate tools and their inputs; tools 
+    will respond with observations. When you are ready for a final answer, respond with the `Final Answer:`
+    Use the following format:
+    Question: the question to be answered
+    Thought: Reason if you have the final answer. If yes, answer the question. If not, find out the missing information needed to answer it.
+    Tool: pick one of {tool_names}
+    Tool Input: the input for the tool
+    Observation: the tool will respond with the result
+    ...
+
+    Final Answer: the final answer to the question, make it a catchy tweet with the users voice
+    Thought, Tool, Tool Input, and Observation steps can be repeated multiple times, but sometimes we can find an answer in the first pass
+    ---
+
+    Question: {query}
+    Thought: Let's think step-by-step, I first need to
+version: 0.1.1

prompts/twitter_voice.yaml

@@ -0,0 +1,30 @@
+description: An experimental few-shot prompt to syntehsize the 'twitter voice' of a uesrname
+meta:
+  authors:
+  - Tuana
+name: Tuana/twitter-voice
+tags:
+- summarization
+text: | 
+    You will be given a twitter stream belonging to a specific profile. Answer with a summary of what they've lately been tweeting about and in what languages.
+    You may go into some detail about what topics they tend to like tweeting about. Please also mention their overall tone, for example: positive,
+    negative, political, sarcastic or something else.
+    Examples: 
+    
+    Twitter stream: RT @deepset_ai: Come join our Haystack server for our first Discord event tomorrow, a deepset AMA session with @rusic_milos @malte_pietsch…
+    RT @deepset_ai: Join us for a chat! On Thursday 25th we are hosting a 'deepset - Ask Me Anything' session on our brand new Discord. Come…
+    RT @deepset_ai: Curious about how you can use @OpenAI GPT3 in a Haystack pipeline? This week we released Haystack 1.7 with which we introdu…
+    RT @tuanacelik: So many updates from @deepset_ai today! 
+    
+    Summary: This user has lately been retweeting tweets fomr @deepset_ai. The topics of the tweets have been around the Haystack community, NLP and GPT. They've
+    been posting in English, and have had a positive, informative tone.
+    
+    Twitter stream: I've directed my team to set sharper rules on how we deal with unidentified objects.\n\nWe will inventory, improve ca… 
+    the incursion by China’s high-altitude balloon, we enhanced radar to pick up slower objects.\n \nBy doing so, w…
+    I gave an update on the United States’ response to recent aerial objects. 
+    Summary: This user has lately been tweeting about having sharper rules to deal with unidentified objects and an incursuin by China's high-altitude
+    baloon. Their tweets have mostly been neutral but determined in tone. They mostly post in English.
+    Twitter stream: {tweets} 
+    
+    Summary: 
+version: 0.1.1

requirements.txt

@@ -1,5 +1,5 @@
-farm-haystack==1.18.0
-streamlit==1.17.0
+farm-haystack[inference]==1.18.0
+streamlit==1.21.0
 markdown
 st-annotated-text
 python-dotenv

utils/config.py

@@ -0,0 +1,7 @@
+import os
+from dotenv import load_dotenv
+
+load_dotenv()
+TWITTER_BEARER_TOKEN = os.getenv('TWITTER_BEARER_TOKEN')
+SERPER_KEY = os.getenv('SERPER_KEY')
+OPENAI_API_KEY = os.getenv('OPENAI_API_KEY')

utils/haystack.py

@@ -0,0 +1,46 @@
+import streamlit as st
+import requests
+from custom_nodes.tweet_retriever import TwitterRetriever
+from haystack.agents import Agent, Tool
+from haystack.nodes import PromptNode, PromptTemplate, WebRetriever
+from haystack.pipelines import WebQAPipeline
+
+
+def start_haystack(openai_key, twitter_bearer, serper_key, last_k_tweets):
+    prompt_node = PromptNode(
+        "text-davinci-003",
+        default_prompt_template=PromptTemplate(prompt="./prompts/lfqa.yaml"),
+        api_key=openai_key,
+        max_length=256,
+    )
+
+    web_retriever = WebRetriever(api_key=serper_key, top_search_results=2, mode="preprocessed_documents")
+    web_pipeline = WebQAPipeline(retriever=web_retriever, prompt_node=prompt_node)
+
+    twitter_retriver = TwitterRetriever(bearer_token=twitter_bearer, last_k_tweets=last_k_tweets)
+
+    pn = PromptNode(model_name_or_path="gpt-4", api_key=openai_key, stop_words=["Observation:"])
+    agent = Agent(prompt_node=pn, prompt_template="./prompts/twitter_agent.yaml")
+
+    tweet_retriver_tool = Tool(name="TwitterRetriever", pipeline_or_node=twitter_retriver, 
+                               description="Useful for when you need to retrive the latest tweets from a username to get an understanding of their style", 
+                               output_variable="results")
+    web_tool = Tool(name="WebSearch", pipeline_or_node=web_pipeline, description="Uesful for when you need to research the latest about a new topic")
+
+    agent.add_tool(tweet_retriver_tool)
+    agent.add_tool(web_tool)
+    
+
+    st.session_state["haystack_started"] = True     
+    return agent                                                
+    
+
+@st.cache_data(show_spinner=True)
+def query(_agent, question):
+    print("USER Q: {question}")
+    try:
+        result = _agent.run(question)
+    except Exception as e:
+        print(e)
+        result = ["Life isn't ideal sometimes and this Agent fails at doing a good job.. Maybe try another query..."]
+    return result

utils/ui.py

@@ -0,0 +1,60 @@
+import streamlit as st
+from PIL import Image
+
+def set_state_if_absent(key, value):
+    if key not in st.session_state:
+        st.session_state[key] = value
+
+def set_initial_state():
+    set_state_if_absent("question", "Provide a Twitter username")
+    set_state_if_absent("result", None)
+    set_state_if_absent("haystack_started", False)
+
+def reset_results(*args):
+    st.session_state.result = None
+
+# def set_openai_api_key(api_key: str):
+#     st.session_state["OPENAI_API_KEY"] = api_key
+
+def sidebar():
+    with st.sidebar:
+        image = Image.open('logo/haystack-logo-colored.png')
+        st.markdown("Thanks for coming to this 🤗 Space.\n\n"
+        "This is a project for fun and is a sister to the [should-i-follow](https://huggingface.co/spaces/deepset/should-i-follow) Space."
+        " There's a lot that can be improved to make this app better.\n\n"
+        "**Take results with a grain of** 🧂\n\n"
+        "")
+
+        st.markdown(
+            "## How to use\n"
+            # "1. Enter your [OpenAI API key](https://platform.openai.com/account/api-keys) below\n"
+            "1. Enter a query that includes a twitter username and be descriptive about wanting a tweet as a result.\n"
+            "2. Enjoy 🤗\n"
+        )
+
+        # api_key_input = st.text_input(
+        #     "OpenAI API Key",
+        #     type="password",
+        #     placeholder="Paste your OpenAI API key here (sk-...)",
+        #     help="You can get your API key from https://platform.openai.com/account/api-keys.",
+        #     value=st.session_state.get("OPENAI_API_KEY", ""),
+        # )
+
+        # if api_key_input:
+        #     set_openai_api_key(api_key_input)
+
+        st.markdown("---")
+        st.markdown(
+            "## How this works\n"
+            "This app was built with [Haystack](https://haystack.deepset.ai) using the"
+            " [`Agent`](https://docs.haystack.deepset.ai/docs/agent) custom [`PromptTemplates`](https://docs.haystack.deepset.ai/docs/prompt_node#templates)\n\n"
+            "as well as a custom `TweetRetriever` node\n"
+            " The source code is also on [GitHub](https://github.com/TuanaCelik/what-would-mother-say)"
+            " with instructions to run locally.\n"
+            "You can see how the `PromptNode` was set up [here](https://github.com/TuanaCelik/should-i-follow/blob/main/utils/haystack.py)")
+        st.markdown("---")
+        st.markdown("Made by [tuanacelik](https://twitter.com/tuanacelik)")
+        st.markdown("---")
+        st.markdown("""Thanks to [mmz_001](https://twitter.com/mm_sasmitha) 
+                        for open sourcing [KnowledgeGPT](https://knowledgegpt.streamlit.app/) which helped me with this sidebar 🙏🏽""")
+        st.image(image, width=250)