Spaces:
Sleeping
Sleeping
fix dockerfile
Browse files- Dockerfile +5 -2
- app.py +7 -7
- download_images.py +22 -3
Dockerfile
CHANGED
|
@@ -37,8 +37,11 @@ ENV ENV=production
|
|
| 37 |
|
| 38 |
# Add script to download images from Hugging Face during build
|
| 39 |
COPY download_images.py .
|
| 40 |
-
|
| 41 |
-
|
|
|
|
|
|
|
|
|
|
| 42 |
python download_images.py; \
|
| 43 |
else \
|
| 44 |
echo "Hugging Face credentials not set, skipping image download."; \
|
|
|
|
| 37 |
|
| 38 |
# Add script to download images from Hugging Face during build
|
| 39 |
COPY download_images.py .
|
| 40 |
+
# Use build secrets to access HF_TOKEN during build time
|
| 41 |
+
RUN --mount=type=secret,id=HF_TOKEN \
|
| 42 |
+
if [ -n "$HF_USERNAME" ] && [ -f /run/secrets/HF_TOKEN ] && [ -n "$HF_DATASET_REPO" ]; then \
|
| 43 |
+
export HF_TOKEN=$(cat /run/secrets/HF_TOKEN) && \
|
| 44 |
+
echo "Downloading images from Hugging Face dataset for $HF_USERNAME/$HF_DATASET_REPO..." && \
|
| 45 |
python download_images.py; \
|
| 46 |
else \
|
| 47 |
echo "Hugging Face credentials not set, skipping image download."; \
|
app.py
CHANGED
|
@@ -221,7 +221,7 @@ def get_hf_image_url(filename):
|
|
| 221 |
"""Get the URL for an image in the Hugging Face repo."""
|
| 222 |
# Return the direct Space URL if available, otherwise use dataset URL
|
| 223 |
if SPACE_NAME and HF_USERNAME:
|
| 224 |
-
return f"https://{HF_USERNAME}-{SPACE_NAME}.hf.space/static/uploads/{filename}"
|
| 225 |
elif HF_USERNAME:
|
| 226 |
return f"https://huggingface.co/datasets/{HF_USERNAME}/{DATASET_REPO}/resolve/main/{IMAGES_PATH}/{filename}"
|
| 227 |
return None
|
|
@@ -533,7 +533,7 @@ async def upload_image(
|
|
| 533 |
and HF_USERNAME
|
| 534 |
and os.environ.get("ENV", "development") == "production"
|
| 535 |
):
|
| 536 |
-
full_url = f"https://{HF_USERNAME}-{SPACE_NAME}.hf.space/static/uploads/{unique_filename}"
|
| 537 |
else:
|
| 538 |
full_url = f"{request.base_url}static/uploads/{unique_filename}"
|
| 539 |
|
|
@@ -648,7 +648,7 @@ async def upload_with_replace(
|
|
| 648 |
and HF_USERNAME
|
| 649 |
and os.environ.get("ENV", "development") == "production"
|
| 650 |
):
|
| 651 |
-
full_url = f"https://{HF_USERNAME}-{SPACE_NAME}.hf.space/static/uploads/{unique_filename}"
|
| 652 |
else:
|
| 653 |
full_url = f"{request.base_url}static/uploads/{unique_filename}"
|
| 654 |
|
|
@@ -693,9 +693,7 @@ async def view_image(request: Request, file_name: str):
|
|
| 693 |
and HF_USERNAME
|
| 694 |
and os.environ.get("ENV", "development") == "production"
|
| 695 |
):
|
| 696 |
-
embed_url = (
|
| 697 |
-
f"https://{HF_USERNAME}-{SPACE_NAME}.hf.space/static/uploads/{file_name}"
|
| 698 |
-
)
|
| 699 |
else:
|
| 700 |
embed_url = f"{request.base_url}static/uploads/{file_name}"
|
| 701 |
|
|
@@ -803,7 +801,9 @@ async def startup_event():
|
|
| 803 |
print(f" - Dataset: {DATASET_REPO}")
|
| 804 |
if SPACE_NAME:
|
| 805 |
print(f" - Space: {SPACE_NAME}")
|
| 806 |
-
print(
|
|
|
|
|
|
|
| 807 |
else:
|
| 808 |
print(" - Space name not set. Direct URLs will use dataset links.")
|
| 809 |
else:
|
|
|
|
| 221 |
"""Get the URL for an image in the Hugging Face repo."""
|
| 222 |
# Return the direct Space URL if available, otherwise use dataset URL
|
| 223 |
if SPACE_NAME and HF_USERNAME:
|
| 224 |
+
return f"https://{HF_USERNAME.lower()}-{SPACE_NAME}.hf.space/static/uploads/{filename}"
|
| 225 |
elif HF_USERNAME:
|
| 226 |
return f"https://huggingface.co/datasets/{HF_USERNAME}/{DATASET_REPO}/resolve/main/{IMAGES_PATH}/{filename}"
|
| 227 |
return None
|
|
|
|
| 533 |
and HF_USERNAME
|
| 534 |
and os.environ.get("ENV", "development") == "production"
|
| 535 |
):
|
| 536 |
+
full_url = f"https://{HF_USERNAME.lower()}-{SPACE_NAME}.hf.space/static/uploads/{unique_filename}"
|
| 537 |
else:
|
| 538 |
full_url = f"{request.base_url}static/uploads/{unique_filename}"
|
| 539 |
|
|
|
|
| 648 |
and HF_USERNAME
|
| 649 |
and os.environ.get("ENV", "development") == "production"
|
| 650 |
):
|
| 651 |
+
full_url = f"https://{HF_USERNAME.lower()}-{SPACE_NAME}.hf.space/static/uploads/{unique_filename}"
|
| 652 |
else:
|
| 653 |
full_url = f"{request.base_url}static/uploads/{unique_filename}"
|
| 654 |
|
|
|
|
| 693 |
and HF_USERNAME
|
| 694 |
and os.environ.get("ENV", "development") == "production"
|
| 695 |
):
|
| 696 |
+
embed_url = f"https://{HF_USERNAME.lower()}-{SPACE_NAME}.hf.space/static/uploads/{file_name}"
|
|
|
|
|
|
|
| 697 |
else:
|
| 698 |
embed_url = f"{request.base_url}static/uploads/{file_name}"
|
| 699 |
|
|
|
|
| 801 |
print(f" - Dataset: {DATASET_REPO}")
|
| 802 |
if SPACE_NAME:
|
| 803 |
print(f" - Space: {SPACE_NAME}")
|
| 804 |
+
print(
|
| 805 |
+
f" - Space URL: https://{HF_USERNAME.lower()}-{SPACE_NAME}.hf.space"
|
| 806 |
+
)
|
| 807 |
else:
|
| 808 |
print(" - Space name not set. Direct URLs will use dataset links.")
|
| 809 |
else:
|
download_images.py
CHANGED
|
@@ -13,6 +13,18 @@ HF_USERNAME = os.environ.get("HF_USERNAME", "")
|
|
| 13 |
HF_TOKEN = os.environ.get("HF_TOKEN", "")
|
| 14 |
DATASET_REPO = os.environ.get("HF_DATASET_REPO", "image-uploader-data")
|
| 15 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
# Set HF cache directory to a writable location
|
| 17 |
os.environ["HF_HOME"] = os.path.join(tempfile.gettempdir(), "huggingface")
|
| 18 |
os.environ["HUGGINGFACE_HUB_CACHE"] = os.path.join(
|
|
@@ -39,6 +51,7 @@ try:
|
|
| 39 |
# Check if repo exists
|
| 40 |
print(f"Checking if repository {HF_USERNAME}/{DATASET_REPO} exists")
|
| 41 |
hf_api.repo_info(repo_id=f"{HF_USERNAME}/{DATASET_REPO}", repo_type="dataset")
|
|
|
|
| 42 |
|
| 43 |
# Download metadata first
|
| 44 |
print(f"Downloading metadata from {HF_USERNAME}/{DATASET_REPO}")
|
|
@@ -58,8 +71,10 @@ try:
|
|
| 58 |
# Save metadata locally
|
| 59 |
with open(METADATA_FILE, "w") as f:
|
| 60 |
json.dump(metadata, f)
|
|
|
|
| 61 |
except Exception as e:
|
| 62 |
print(f"Error downloading metadata: {e}")
|
|
|
|
| 63 |
metadata = {}
|
| 64 |
# Initialize metadata file if it doesn't exist
|
| 65 |
with open(METADATA_FILE, "w") as f:
|
|
@@ -76,10 +91,11 @@ try:
|
|
| 76 |
print(f"Found {len(image_files)} images")
|
| 77 |
|
| 78 |
# Download each image
|
| 79 |
-
|
|
|
|
| 80 |
try:
|
| 81 |
filename = os.path.basename(image_file)
|
| 82 |
-
print(f"Downloading {filename}")
|
| 83 |
|
| 84 |
# Download file
|
| 85 |
download_path = hf_api.hf_hub_download(
|
|
@@ -96,10 +112,13 @@ try:
|
|
| 96 |
dst.write(src.read())
|
| 97 |
|
| 98 |
print(f"Saved {filename} to {destination}")
|
|
|
|
| 99 |
except Exception as e:
|
| 100 |
print(f"Error downloading {image_file}: {e}")
|
| 101 |
|
| 102 |
-
print(
|
|
|
|
|
|
|
| 103 |
|
| 104 |
except Exception as e:
|
| 105 |
print(f"Error: {e}")
|
|
|
|
| 13 |
HF_TOKEN = os.environ.get("HF_TOKEN", "")
|
| 14 |
DATASET_REPO = os.environ.get("HF_DATASET_REPO", "image-uploader-data")
|
| 15 |
|
| 16 |
+
# Validate required environment variables
|
| 17 |
+
if not HF_USERNAME:
|
| 18 |
+
print("ERROR: HF_USERNAME environment variable is not set")
|
| 19 |
+
exit(1)
|
| 20 |
+
|
| 21 |
+
if not HF_TOKEN:
|
| 22 |
+
print("ERROR: HF_TOKEN environment variable is not set")
|
| 23 |
+
exit(1)
|
| 24 |
+
|
| 25 |
+
print(f"Using Hugging Face credentials for user: {HF_USERNAME}")
|
| 26 |
+
print(f"Dataset repository: {DATASET_REPO}")
|
| 27 |
+
|
| 28 |
# Set HF cache directory to a writable location
|
| 29 |
os.environ["HF_HOME"] = os.path.join(tempfile.gettempdir(), "huggingface")
|
| 30 |
os.environ["HUGGINGFACE_HUB_CACHE"] = os.path.join(
|
|
|
|
| 51 |
# Check if repo exists
|
| 52 |
print(f"Checking if repository {HF_USERNAME}/{DATASET_REPO} exists")
|
| 53 |
hf_api.repo_info(repo_id=f"{HF_USERNAME}/{DATASET_REPO}", repo_type="dataset")
|
| 54 |
+
print(f"Repository {HF_USERNAME}/{DATASET_REPO} exists")
|
| 55 |
|
| 56 |
# Download metadata first
|
| 57 |
print(f"Downloading metadata from {HF_USERNAME}/{DATASET_REPO}")
|
|
|
|
| 71 |
# Save metadata locally
|
| 72 |
with open(METADATA_FILE, "w") as f:
|
| 73 |
json.dump(metadata, f)
|
| 74 |
+
print(f"Metadata saved to {METADATA_FILE}")
|
| 75 |
except Exception as e:
|
| 76 |
print(f"Error downloading metadata: {e}")
|
| 77 |
+
print("Creating empty metadata file")
|
| 78 |
metadata = {}
|
| 79 |
# Initialize metadata file if it doesn't exist
|
| 80 |
with open(METADATA_FILE, "w") as f:
|
|
|
|
| 91 |
print(f"Found {len(image_files)} images")
|
| 92 |
|
| 93 |
# Download each image
|
| 94 |
+
success_count = 0
|
| 95 |
+
for i, image_file in enumerate(image_files):
|
| 96 |
try:
|
| 97 |
filename = os.path.basename(image_file)
|
| 98 |
+
print(f"[{i+1}/{len(image_files)}] Downloading {filename}")
|
| 99 |
|
| 100 |
# Download file
|
| 101 |
download_path = hf_api.hf_hub_download(
|
|
|
|
| 112 |
dst.write(src.read())
|
| 113 |
|
| 114 |
print(f"Saved {filename} to {destination}")
|
| 115 |
+
success_count += 1
|
| 116 |
except Exception as e:
|
| 117 |
print(f"Error downloading {image_file}: {e}")
|
| 118 |
|
| 119 |
+
print(
|
| 120 |
+
f"Image download completed. Successfully downloaded {success_count}/{len(image_files)} images."
|
| 121 |
+
)
|
| 122 |
|
| 123 |
except Exception as e:
|
| 124 |
print(f"Error: {e}")
|