Spaces:
Runtime error
Runtime error
Upload folder using huggingface_hub
Browse files- app.py +195 -0
- requirements.txt +3 -0
app.py
ADDED
|
@@ -0,0 +1,195 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import os
|
| 3 |
+
import requests
|
| 4 |
+
import re
|
| 5 |
+
from urllib.parse import urljoin, urlparse
|
| 6 |
+
from pathlib import Path
|
| 7 |
+
import time
|
| 8 |
+
from bs4 import BeautifulSoup
|
| 9 |
+
|
| 10 |
+
def extract_images_from_url(url):
|
| 11 |
+
"""Extrahiert alle Bild-URLs von einer Webseite."""
|
| 12 |
+
try:
|
| 13 |
+
# Hole die Webseite
|
| 14 |
+
response = requests.get(url, timeout=10)
|
| 15 |
+
response.raise_for_status()
|
| 16 |
+
|
| 17 |
+
# Parse HTML mit BeautifulSoup
|
| 18 |
+
soup = BeautifulSoup(response.content, 'html.parser')
|
| 19 |
+
|
| 20 |
+
# Finde alle Bild-Tags
|
| 21 |
+
img_tags = soup.find_all('img')
|
| 22 |
+
|
| 23 |
+
# Extrahiere URLs
|
| 24 |
+
img_urls = []
|
| 25 |
+
for img in img_tags:
|
| 26 |
+
src = img.get('src') or img.get('data-src')
|
| 27 |
+
if src:
|
| 28 |
+
# Erstelle absolute URL
|
| 29 |
+
absolute_url = urljoin(url, src)
|
| 30 |
+
img_urls.append(absolute_url)
|
| 31 |
+
|
| 32 |
+
# Entferne Duplikate und sortiere
|
| 33 |
+
img_urls = sorted(list(set(img_urls)))
|
| 34 |
+
|
| 35 |
+
return img_urls, len(img_urls)
|
| 36 |
+
|
| 37 |
+
except Exception as e:
|
| 38 |
+
raise gr.Error(f"Fehler beim Abrufen der URL: {str(e)}")
|
| 39 |
+
|
| 40 |
+
def download_image(url, folder, index):
|
| 41 |
+
"""Lädt ein einzelnes Bild herunter."""
|
| 42 |
+
try:
|
| 43 |
+
# Hole den Dateinamen aus der URL
|
| 44 |
+
parsed = urlparse(url)
|
| 45 |
+
filename = os.path.basename(parsed.path)
|
| 46 |
+
|
| 47 |
+
# Wenn kein Dateiname vorhanden ist, generiere einen
|
| 48 |
+
if not filename:
|
| 49 |
+
filename = f"image_{index}.jpg"
|
| 50 |
+
|
| 51 |
+
# Erstelle den Dateipfad
|
| 52 |
+
filepath = os.path.join(folder, filename)
|
| 53 |
+
|
| 54 |
+
# Füge Index hinzu, falls Datei bereits existiert
|
| 55 |
+
counter = 1
|
| 56 |
+
while os.path.exists(filepath):
|
| 57 |
+
name, ext = os.path.splitext(filename)
|
| 58 |
+
filepath = os.path.join(folder, f"{name}_{counter}{ext}")
|
| 59 |
+
counter += 1
|
| 60 |
+
|
| 61 |
+
# Lade das Bild herunter
|
| 62 |
+
response = requests.get(url, timeout=30)
|
| 63 |
+
response.raise_for_status()
|
| 64 |
+
|
| 65 |
+
# Speichere das Bild
|
| 66 |
+
with open(filepath, 'wb') as f:
|
| 67 |
+
f.write(response.content)
|
| 68 |
+
|
| 69 |
+
return filepath, True
|
| 70 |
+
|
| 71 |
+
except Exception as e:
|
| 72 |
+
return str(e), False
|
| 73 |
+
|
| 74 |
+
def download_all_images(url, progress=gr.Progress()):
|
| 75 |
+
"""Lädt alle Bilder von einer Webseite herunter."""
|
| 76 |
+
if not url:
|
| 77 |
+
raise gr.Error("Bitte geben Sie eine URL ein.")
|
| 78 |
+
|
| 79 |
+
# Validiere URL
|
| 80 |
+
if not url.startswith(('http://', 'https://')):
|
| 81 |
+
url = 'https://' + url
|
| 82 |
+
|
| 83 |
+
# Erstelle Download-Verzeichnis
|
| 84 |
+
timestamp = time.strftime("%Y%m%d_%H%M%S")
|
| 85 |
+
folder_name = f"downloaded_images_{timestamp}"
|
| 86 |
+
download_folder = Path.cwd() / folder_name
|
| 87 |
+
download_folder.mkdir(exist_ok=True)
|
| 88 |
+
|
| 89 |
+
progress(0, desc="Rufe Bilder ab...")
|
| 90 |
+
|
| 91 |
+
# Extrahiere Bild-URLs
|
| 92 |
+
try:
|
| 93 |
+
img_urls, total_count = extract_images_from_url(url)
|
| 94 |
+
except Exception as e:
|
| 95 |
+
raise gr.Error(str(e))
|
| 96 |
+
|
| 97 |
+
if total_count == 0:
|
| 98 |
+
return None, f"Keine Bilder auf {url} gefunden."
|
| 99 |
+
|
| 100 |
+
progress(0.3, desc=f"Starte Download von {total_count} Bildern...")
|
| 101 |
+
|
| 102 |
+
# Download alle Bilder
|
| 103 |
+
downloaded_count = 0
|
| 104 |
+
failed_count = 0
|
| 105 |
+
results = []
|
| 106 |
+
|
| 107 |
+
for i, img_url in enumerate(progress.tqdm(img_urls, desc="Bilder herunterladen")):
|
| 108 |
+
try:
|
| 109 |
+
filepath, success = download_image(img_url, str(download_folder), i)
|
| 110 |
+
|
| 111 |
+
if success:
|
| 112 |
+
downloaded_count += 1
|
| 113 |
+
results.append(filepath)
|
| 114 |
+
else:
|
| 115 |
+
failed_count += 1
|
| 116 |
+
|
| 117 |
+
# Aktualisiere Fortschritt
|
| 118 |
+
progress((0.3 + 0.7 * (i + 1) / total_count),
|
| 119 |
+
desc=f"Bild {i+1}/{total_count}: {'✓' if success else '✗'}")
|
| 120 |
+
|
| 121 |
+
except Exception as e:
|
| 122 |
+
failed_count += 1
|
| 123 |
+
progress((0.3 + 0.7 * (i + 1) / total_count),
|
| 124 |
+
desc=f"Fehler bei Bild {i+1}: {str(e)}")
|
| 125 |
+
|
| 126 |
+
# Erstelle Ergebnis
|
| 127 |
+
message = f"✅ **Erfolgreich** {downloaded_count} Bilder heruntergeladen.\n"
|
| 128 |
+
if failed_count > 0:
|
| 129 |
+
message += f"⚠️ **Fehler** bei {failed_count} Bildern."
|
| 130 |
+
|
| 131 |
+
# Erstelle eine ZIP-Datei für den Download
|
| 132 |
+
import shutil
|
| 133 |
+
zip_path = download_folder.parent / f"{folder_name}.zip"
|
| 134 |
+
shutil.make_archive(str(zip_path.parent / zip_path.stem), 'zip', download_folder)
|
| 135 |
+
|
| 136 |
+
return str(zip_path), message
|
| 137 |
+
|
| 138 |
+
# Erstelle die Gradio-Anwendung
|
| 139 |
+
with gr.Blocks(theme=gr.themes.Soft(primary_hue="blue", secondary_hue="indigo")) as demo:
|
| 140 |
+
gr.Markdown("# 📷 Bild-Downloader")
|
| 141 |
+
gr.Markdown("Geben Sie eine URL ein, um **alle Bilder** von dieser Seite herunterzuladen.")
|
| 142 |
+
|
| 143 |
+
with gr.Row():
|
| 144 |
+
with gr.Column(scale=1):
|
| 145 |
+
url_input = gr.Textbox(
|
| 146 |
+
label="Web-URL",
|
| 147 |
+
placeholder="https://beispiel.de/seite",
|
| 148 |
+
info="Geben Sie die URL der Seite ein, von der Bilder heruntergeladen werden sollen."
|
| 149 |
+
)
|
| 150 |
+
|
| 151 |
+
download_btn = gr.Button(" Bilder herunterladen", variant="primary", icon="download")
|
| 152 |
+
|
| 153 |
+
gr.Markdown("""
|
| 154 |
+
### ℹ️ Anleitung
|
| 155 |
+
1. Geben Sie eine URL ein (z.B. eine Bildergalerie)
|
| 156 |
+
2. Klicken Sie auf "Bilder herunterladen"
|
| 157 |
+
3. Warten Sie, bis alle Bilder heruntergeladen wurden
|
| 158 |
+
4. Laden Sie die ZIP-Datei mit allen Bildern herunter
|
| 159 |
+
|
| 160 |
+
### 📌 Hinweise
|
| 161 |
+
- Die App extrahiert alle Bilder von `<img>`-Tags
|
| 162 |
+
- Bilder werden als ZIP-Archiv komprimiert
|
| 163 |
+
- Der Download kann einige Sekunden dauern
|
| 164 |
+
""")
|
| 165 |
+
|
| 166 |
+
with gr.Column(scale=1):
|
| 167 |
+
output_zip = gr.File(label="ZIP-Datei mit Bildern")
|
| 168 |
+
status_output = gr.Markdown(label="Status")
|
| 169 |
+
|
| 170 |
+
# Event Handler
|
| 171 |
+
download_btn.click(
|
| 172 |
+
fn=download_all_images,
|
| 173 |
+
inputs=url_input,
|
| 174 |
+
outputs=[output_zip, status_output],
|
| 175 |
+
api_visibility="public"
|
| 176 |
+
)
|
| 177 |
+
|
| 178 |
+
# Beispiele
|
| 179 |
+
gr.Examples(
|
| 180 |
+
examples=[
|
| 181 |
+
["https://picsum.photos/"],
|
| 182 |
+
["https://unsplash.com/s/photos/nature"],
|
| 183 |
+
["https://www.flickr.com/"],
|
| 184 |
+
],
|
| 185 |
+
inputs=url_input
|
| 186 |
+
)
|
| 187 |
+
|
| 188 |
+
# Starte die Anwendung
|
| 189 |
+
if __name__ == "__main__":
|
| 190 |
+
demo.launch(
|
| 191 |
+
theme=gr.themes.Soft(),
|
| 192 |
+
footer_links=[
|
| 193 |
+
{"label": "Built with anycoder", "url": "https://huggingface.co/spaces/akhaliq/anycoder"}
|
| 194 |
+
]
|
| 195 |
+
)
|
requirements.txt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
beautifulsoup4
|
| 2 |
+
gradio>=6.0.2
|
| 3 |
+
requests
|