Spaces:
Runtime error
Runtime error
File size: 7,377 Bytes
6b366b2 095174f 6b366b2 7a729e8 6b366b2 54c3682 6b366b2 7a729e8 6b366b2 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 |
import streamlit as st
import requests
import cloudinary
import cloudinary.uploader
from PIL import Image
import io
from google_auth_oauthlib.flow import InstalledAppFlow
from googleapiclient.discovery import build
import os
# Configure Cloudinary with your credentials
cloudinary.config(
cloud_name="dvuowbmrz",
api_key="177664162661619",
api_secret="qVMYel17N_C5QUUUuBIuatB5tq0"
)
#
# # Set up OAuth2 client details
# CLIENT_SECRET_FILE = 'client_secret.json'
# SCOPES = ['https://www.googleapis.com/auth/drive.metadata.readonly'] # Adjust scopes as needed
#
# # Set up Streamlit app
# #st.title("Google Authentication Demo")
#
# # Check if the user is authenticated
# if 'credentials' not in st.session_state:
# #st.write("WELCOME")
# flow = InstalledAppFlow.from_client_secrets_file(CLIENT_SECRET_FILE, SCOPES)
# credentials = flow.run_local_server(port=8501, authorization_prompt_message='')
#
# # Save credentials to a file for future use (optional)
# with open('token.json', 'w') as token_file:
# token_file.write(credentials.to_json())
#
# st.session_state.credentials = credentials
# st.success("Authentication successful. You can now use the app.")
#
# # Use authenticated credentials to interact with Google API
# credentials = st.session_state.credentials
# service = build('drive', 'v3', credentials=credentials)
#
# # Fetch user's name from Google API
# try:
# user_info = service.about().get(fields="user").execute()
# user_name = user_info["user"]["displayName"]
# #st.header("Google Profile Information")
# st.markdown(f"<p style='font-size: 24px;'><strong>Userame: {user_name.upper()}</strong></p>", unsafe_allow_html=True)
# except Exception as e:
# st.error(f"Error fetching user profile: {str(e)}")
#
# # Your app's functionality goes here
# # # Display Google Drive contents
# # st.header("Google Drive Contents")
# # results = service.files().list(pageSize=10).execute()
# # files = results.get('files', [])
# # if not files:
# # st.write('No files found in Google Drive.')
# # else:
# # st.write('Files in Google Drive:')
# # for file in files:
# # st.write(f"- {file['name']} ({file['mimeType']})")
#
# # Logout button
# if st.button("Logout"):
# del st.session_state.credentials
# os.remove("token_dir/token.json") # Remove the token file
#
# Set up Hugging Face API endpoint
API_URL = "https://api-inference.huggingface.co/models/CompVis/stable-diffusion-v1-4"
headers = {"Authorization": "Bearer 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"}
def query_model_with_image(image_description):
payload = {
"inputs": image_description
}
response = requests.post(API_URL, headers=headers, json=payload)
image_bytes = response.content
image = Image.open(io.BytesIO(image_bytes))
return image
def upload_to_cloudinary(image, prompt_text):
image_data = io.BytesIO()
image.save(image_data, format="JPEG")
image_data.seek(0)
upload_result = cloudinary.uploader.upload(
image_data,
folder="compvis_app",
public_id=prompt_text
)
return upload_result["secure_url"]
def fetch_latest_images_from_cloudinary(num_images=9):
# Use the Cloudinary Admin API to list resources
url = f"https://api.cloudinary.com/v1_1/{cloudinary.config().cloud_name}/resources/image"
params = {
"max_results": num_images,
"type": "upload"
}
response = requests.get(url, params=params, auth=(cloudinary.config().api_key, cloudinary.config().api_secret))
if response.status_code == 200:
images = response.json()["resources"]
else:
images = []
return images
# Streamlit app
st.markdown("""<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/4.7.0/css/font-awesome.min.css">""", unsafe_allow_html=True)
st.title("Text to Image Generator")
image_description = st.text_input("Enter the image description")
if st.button("Generate Image"):
processed_image = query_model_with_image(image_description)
st.image(processed_image, use_column_width=True, output_format="JPEG") # Use use_column_width=True
st.session_state.processed_image = processed_image
st.session_state.image_description = image_description
st.write("Image generated.")
if st.button("Upload"):
if 'processed_image' in st.session_state:
uploaded_url = upload_to_cloudinary(st.session_state.processed_image, st.session_state.image_description)
st.write("Image uploaded to Cloudinary. Prompt Text:", st.session_state.image_description)
st.write("Image URL on Cloudinary:", uploaded_url)
else:
st.write("Generate an image first before uploading.")
# Fetch and display the latest images from Cloudinary
st.header("Latest Images created")
# Use the 'fetch_latest_images_from_cloudinary' function to get the latest images
latest_images = fetch_latest_images_from_cloudinary()
# Define the number of columns in the grid
num_columns = 3 # You can adjust this number as needed
# Calculate the width for each column
column_width = f"calc(33.33% - {10}px)" # Adjust the width and margin as needed
# Add CSS styling for the grid and rounded images
st.markdown(
f"""
<style>
.responsive-grid {{
display: flex;
flex-wrap: wrap;
justify-content: space-between;
}}
.responsive-grid-item {{
width: {column_width};
margin: 10px;
box-sizing: border-box;
text-align: center;
position: relative;
}}
.image-caption {{
font-weight: bold;
}}
.rounded-image {{
border-radius: 15px; # Adjust the radius as needed for more or less roundness
overflow: hidden;
}}
.download-button {{
background-color: black; # Set button color to black
color: white;
padding: 5px 10px;
border-radius: 5px;
text-decoration: none;
display: inline-block;
position: absolute;
top: 10px; # Adjust top value for vertical positioning
right: 10px; # Adjust right value for horizontal positioning
}}
</style>
""",
unsafe_allow_html=True,
)
# Create the responsive grid layout
st.markdown('<div class="responsive-grid">', unsafe_allow_html=True)
for i, image in enumerate(latest_images):
image_url = image.get('secure_url', '') # Get the image URL
public_id = image.get('public_id', '') # Get the full public_id
# Extract just the filename (without the folder)
filename = public_id.split('/')[-1]
# Add some spacing around the image and its name
st.markdown(f'<div class="responsive-grid-item">', unsafe_allow_html=True)
st.markdown(f'<p class="image-caption">{filename}</p>', unsafe_allow_html=True)
# Add rounded corners to the image using HTML
st.markdown(f'<img src="{image_url}" class="rounded-image" width="{int(1.25 * 300)}">', unsafe_allow_html=True)
# Add an arrow icon instead of "Download" button with black color
download_link = f'<a href="{image_url}" class="download-button" download="{filename}">↓</a>'
st.markdown(download_link, unsafe_allow_html=True)
st.write("") # Add empty spaces for separation
st.markdown('</div>', unsafe_allow_html=True)
# Close the responsive grid layout
st.markdown('</div>', unsafe_allow_html=True) |