WE HIGH TEA

README.md

@@ -1,13 +1,16 @@
 # Electhon - 2023
 
-#### Team Name -
-#### Problem Statement - 
-#### Team Leader Email -
+#### Team Name - We High Tea
+#### Problem Statement - Devise a solution that can boost voter turnout on election day at the polling stations by creation of election anthems, jingles, posters etc.
+#### Team Leader Email - manomayjamble15@gmail.com
 
 ## A Brief of the Prototype:
-  What is your solution? and how it works.
+Our solution is a generative AI model that uses natural language processing (NLP) and deep learning algorithms to create unique and engaging advertisements for each community. Our solution is built on the GPT-3.5 architecture, a large language model trained by OpenAI. The model is pre-trained on a large corpus of text data, and fine-tuned on specific election-related keywords and phrases.
+The input to the model is a set of parameters such as location, demographics, and popular local references. The model generates an advertisement based on these parameters, which is then reviewed by a team of human experts to ensure quality and relevance.
+We have also designed a plan for efficient distrubution of generated content using google analytics and google ads.
 
 ## Code Execution Instruction:
-  *[If your solution is **not** application based, you can ignore this para]
+  `pip install -r requirements.txt`
+
+  `flask run`
 
- *The Repository must contain your **Execution Plan PDF**.

app.py

@@ -0,0 +1,300 @@
+
+from flask import Flask, render_template, request, redirect, url_for, send_file
+import numpy as np
+import requests
+import json
+import openai
+import os
+import cloudinary
+import cloudinary.uploader
+import textwrap
+import io
+from io import BytesIO
+from PIL import Image, ImageDraw, ImageFont
+import requests
+
+app = Flask(__name__)
+
+
+data = []
+
+@app.route("/")
+def index():
+    return render_template("index.html")
+
+
+def push_cloudinary(data=None, url=None):
+    cloudinary.config(
+    cloud_name="dkjv1ictm",
+    api_key="153874947239145",
+    api_secret="ZkSte7mI4e1s5XNltpJw-NmMQ5Q"
+    )
+    if url!=None:
+        data = url
+    elif data!=None:
+        data = data
+    cloudinary_response = cloudinary.uploader.upload_image(
+    file = data,
+    use_filename=True,
+    folder="/electhon",
+    )
+
+    response = cloudinary_response.metadata.get("url")
+    if response[-3:] != "jpg":
+        response = response.replace(response[len(response) - 3:], "jpg")
+    return response
+
+
+def generate_image():
+    url = "https://v1.genr.ai/api/circuit-element/generate-image"
+
+    payload = {
+        "prompt": "Citizens casting their vote in India",
+        "height": 512,
+        "width": 512,
+        "model": "stable-diffusion-2",
+        "n_images": 1
+    }
+    headers = {"Content-Type": "application/json"}
+
+    response = requests.post(url, json=payload, headers=headers)
+
+    print(response.json())
+    return response.json()['output']
+
+
+
+def get_text(prompt):
+
+    # Note: you need to be using OpenAI Python v0.27.0 for the code below to work
+    openai.api_key = "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"
+    response = openai.ChatCompletion.create(
+    model="gpt-3.5-turbo",
+    messages=[
+            {"role": "system", "content": "You are a creative advertisment generator."},
+            {"role": "user", "content": prompt},
+        ]
+    )
+    # print(response)
+    return response["choices"][0]["message"]["content"]
+
+
+    # url = "https://v1.genr.ai/api/circuit-element/prompt-to-text"
+    # payload = {
+    # "prompt": prompt,
+    # }
+    # headers = {"Content-Type": "application/json"}
+    # response = requests.post(url, json=payload, headers=headers)
+    # print(response)
+    # return json.loads(response.text)["output"]
+
+
+def get_text_ads(location, language, religion, age, keywords, ad_type):
+
+    if ad_type == "slogans":
+        slogan_prompt = f"Generate a tagline as an advertisment to attract voters. The tagline should be relatable for the people of {location} and should be in {language} language. The age distribution of is area is {age} and it is a {religion} majority. Sole local references are - {keywords}. The tagline should be limited to 2-4 lines only. Note: You don't have to use all the references in the tagline."
+        return {"slogans": get_text(slogan_prompt)}
+
+    if ad_type == "jingles":
+        jingle_prompt = f"Generate a jingle as an advertisment to attract voters. The jingle should be relatable for the people of {location} and should be in {language} language. The age distribution of is area is {age} and it is a {religion} majority. Sole local references are - {keywords}. Note: You don't have to use all the references in the jingle, the jingle should be in rhyme scheme AABB and length should not exceed 8-10 lines."
+        return {"jingles": get_text(jingle_prompt)}
+
+    if ad_type == "songs":
+        song_prompt = f"Generate a song as an advertisment to attract voters. The song should be relatable for the people of {location} and should be in {language} language. The age distribution of is area is {age} and it is a {religion} majority. Sole local references are - {keywords}. Note: You don't have to use all the references in the song. the song should have a proper tag of chorus and verse."
+        return {"songs": get_text(song_prompt)}
+
+    if ad_type == "posters":
+        poster_slogan_prompt = f"Generate a tagline as an advertisment to attract voters. The tagline should be relatable for the people of {location} and should be in {language} language. The age distribution of is area is {age} and it is a {religion} majority. Sole local references are - {keywords}. The tagline should be limited to 2-4 lines only. Note: You don't have to use all the references in the tagline."
+        poster_slogan = get_text(poster_slogan_prompt)
+        image_link = generate_image()
+        print(poster_slogan)
+
+        # response = requests.get(image_link)
+        # img = Image.open(io.BytesIO(response.content))
+        #
+        # # Create a draw object for the image
+        # draw = ImageDraw.Draw(img)
+        #
+        # # Set the font and font size
+        # font = ImageFont.truetype("arial.ttf", 20)
+        #
+        # # Get the size of the text
+        # text_size = draw.textsize(poster_slogan, font=font)
+        #
+        # # Calculate the position of the text
+        # text_x = img.width // 2 - text_size[0] // 2
+        # text_y = img.height - text_size[1] - 10
+        #
+        # # Draw the text on the image
+        # draw.text((text_x, text_y), poster_slogan, font=font, fill="white")
+        #
+        # # Save the image to a bytes buffer
+        # img_bytes = io.BytesIO()
+        # img.save(img_bytes, format='PNG')
+        # img_bytes.seek(0)
+
+        # # Convert the bytes buffer to a NumPy array
+        # img_array = np.frombuffer(img_bytes.getvalue(), dtype=np.uint8)
+
+        # image_url = push_cloudinary(img_bytes)
+        # return render_template("images.html", img_src=image_url)
+
+
+
+
+        # Download the image from the cloudinary link
+        response = requests.get(image_link)
+
+        # Load the image into a PIL Image object
+        image = Image.open(BytesIO(response.content))
+
+        # Create a draw object on the image
+        draw = ImageDraw.Draw(image)
+
+        # Define the text and font
+        text = poster_slogan
+        font = ImageFont.truetype("arial.ttf", 16)
+
+        # Calculate the size of the text
+        text_size = draw.textsize(text, font)
+
+        # Calculate the position of the text
+        x = (image.width - text_size[0]) // 2  # use floor division to ensure an integer value
+        y = image.height - text_size[1] - 10  # adjust the margin to your liking
+
+        # Draw the text on the image
+        draw.text((x, y), text, font=font, fill=(255, 255, 255))
+
+        img_bytes = io.BytesIO()
+        image.save(img_bytes, format='PNG')
+        img_bytes.seek(0)
+
+        image_url = push_cloudinary(img_bytes)
+        return render_template("images.html", img_src=image_url)
+
+
+
+        # Save the modified image
+        # image.save('path/to/new/image.png')
+
+
+
+
+
+
+
+@app.route("/generate", methods=['GET', 'POST'])
+def generate():
+    if request.method == "POST":
+        state = request.form.get("state")
+        city = request.form.get("city")
+        # location = request.form.get("location")
+        language = request.form.get("language")
+        religion = request.form.get("religion")
+        age = request.form.get("age")
+        keywords = request.form.get("local-references")
+        ad_type = request.form.get("submit-button")
+        print(text_on_image(get_text_ads(city, language, religion, age, keywords, ad_type), ad_type))
+        data.append([state, city, language, religion, age, keywords, ad_type])
+
+        print(city, language, religion,ad_type)
+        urls = [text_on_image(get_text_ads(city, language, religion, age, keywords, ad_type), ad_type), text_on_image(get_text_ads(city, language, religion, age, keywords, ad_type), ad_type), text_on_image(get_text_ads(city, language, religion, age, keywords, ad_type), ad_type)]
+        return redirect(url_for('images', img_src=urls))
+    else:
+        return render_template("generate.html")
+
+
+@app.route("/regenerate", methods=['POST'])
+def regenerate():
+    global data
+    temp = data[0]
+    state = temp[0]
+    city = temp[1]
+    language = temp[2]
+    religion = temp[3]
+    age = temp[4]
+    keywords = temp[5]
+    ad_type = temp[6]
+    urls = [text_on_image(get_text_ads(city, language, religion, age, keywords, ad_type), ad_type), text_on_image(get_text_ads(city, language, religion, age, keywords, ad_type), ad_type), text_on_image(get_text_ads(city, language, religion, age, keywords, ad_type), ad_type)]
+
+    return redirect(url_for('images', img_src=urls))
+
+
+
+
+def text_image(height,width,text):
+    print(text)
+    image = Image.open("bg.png")
+
+    image = image.resize((height, width))
+
+
+    draw = ImageDraw.Draw(image)
+
+    font_size = 50
+    font = ImageFont.truetype("SakalBharati.ttf", font_size)
+
+    max_width = 1500
+
+    wrapped_text = textwrap.wrap(text, width=max_width/font_size)
+
+    y_text = image.height // 2 - font_size * len(wrapped_text) // 2
+
+    for line in wrapped_text:
+        text_width, text_height = font.getsize(line)
+
+        x_text = (image.width - text_width) // 2
+
+        draw.text((x_text, y_text), line, font=font, fill=(0, 0, 0))
+
+        y_text += text_height
+
+    img_byte_arr = io.BytesIO()
+    image.save(img_byte_arr, format='PNG')
+    img_byte_arr = img_byte_arr.getvalue()
+    print(push_cloudinary(img_byte_arr))
+    return push_cloudinary(img_byte_arr)
+
+
+def text_on_image(text,addType):
+    if(addType == "songs"):
+        return text_image(1080,1920,text[addType])
+
+    elif(addType == "slogans"):
+        return text_image(1920,1080,text[addType])
+
+    elif(addType == "jingles"):
+        return text_image(1080,1920,text[addType])
+
+
+
+
+
+
+@app.route("/generate.html")
+def gen_page():
+    return render_template("generate.html")
+
+
+@app.route("/index.html")
+def home_page():
+    return render_template("index.html")
+
+
+
+@app.route("/images")
+def images():
+    # get the output from the query parameter
+    img_src = request.args.getlist('img_src')
+    # render the images.html template with the output
+    return render_template('images.html', img_src=img_src)
+
+
+
+
+if __name__ == '__main__':
+    # print(get_text_ads("Mumbai", "English", "Hindu", "25-35", "Gateway of India, Marine drive", "song")["song"])
+    app.run()
+
+
+

requirements.txt

@@ -0,0 +1,25 @@
+aiohttp==3.8.4
+aiosignal==1.3.1
+async-timeout==4.0.2
+attrs==22.2.0
+certifi==2022.12.7
+charset-normalizer==3.1.0
+click==8.1.3
+cloudinary==1.32.0
+Flask==2.2.3
+frozenlist==1.3.3
+idna==3.4
+itsdangerous==2.1.2
+Jinja2==3.1.2
+MarkupSafe==2.1.2
+multidict==6.0.4
+numpy
+openai==0.27.4
+Pillow==9.5.0
+requests==2.28.2
+six==1.16.0
+tqdm==4.65.0
+urllib3==1.26.15
+Werkzeug==2.2.3
+yarl==1.8.2
+gunicorn