Spaces:
Sleeping
Sleeping
Commit
·
5838c7c
1
Parent(s):
8c333c0
updated app.py
Browse files
app.py
CHANGED
|
@@ -16,7 +16,8 @@ from huggingface_hub import HfApi, hf_hub_download
|
|
| 16 |
DATASET_REPO_ID = os.environ.get("DATASET_REPO_ID", "fast-stager/property-labels")
|
| 17 |
HF_TOKEN = os.environ.get("HF_TOKEN")
|
| 18 |
|
| 19 |
-
CACHE_DIR = "
|
|
|
|
| 20 |
URL_FILE = "new_urls.json"
|
| 21 |
LABEL_FILE = os.path.join(CACHE_DIR, "annotations.csv")
|
| 22 |
VERIFY_FILE = os.path.join(CACHE_DIR, "verifications.csv")
|
|
@@ -34,8 +35,7 @@ def sync_pull():
|
|
| 34 |
filename=filename,
|
| 35 |
repo_type="dataset",
|
| 36 |
local_dir=CACHE_DIR,
|
| 37 |
-
token=token
|
| 38 |
-
local_dir_use_symlinks=False
|
| 39 |
)
|
| 40 |
print(f"✅ Loaded {filename}")
|
| 41 |
except Exception as e:
|
|
@@ -102,17 +102,15 @@ def load_all_urls():
|
|
| 102 |
data = json.load(f)
|
| 103 |
if "groups" in data:
|
| 104 |
for group in data["groups"]:
|
| 105 |
-
|
| 106 |
-
urls.extend(group["images"])
|
| 107 |
elif isinstance(data, dict):
|
| 108 |
-
for
|
| 109 |
if isinstance(rows, list):
|
| 110 |
for row in rows:
|
| 111 |
if "unstaged_images" in row:
|
| 112 |
urls.append(row["unstaged_images"])
|
| 113 |
except Exception as e:
|
| 114 |
print(f"Error loading URLs: {e}")
|
| 115 |
-
|
| 116 |
return urls
|
| 117 |
|
| 118 |
def get_ordered_groups():
|
|
|
|
| 16 |
DATASET_REPO_ID = os.environ.get("DATASET_REPO_ID", "fast-stager/property-labels")
|
| 17 |
HF_TOKEN = os.environ.get("HF_TOKEN")
|
| 18 |
|
| 19 |
+
CACHE_DIR = os.path.join(os.getcwd(), "data")
|
| 20 |
+
os.makedirs(CACHE_DIR, exist_ok=True)
|
| 21 |
URL_FILE = "new_urls.json"
|
| 22 |
LABEL_FILE = os.path.join(CACHE_DIR, "annotations.csv")
|
| 23 |
VERIFY_FILE = os.path.join(CACHE_DIR, "verifications.csv")
|
|
|
|
| 35 |
filename=filename,
|
| 36 |
repo_type="dataset",
|
| 37 |
local_dir=CACHE_DIR,
|
| 38 |
+
token=token
|
|
|
|
| 39 |
)
|
| 40 |
print(f"✅ Loaded {filename}")
|
| 41 |
except Exception as e:
|
|
|
|
| 102 |
data = json.load(f)
|
| 103 |
if "groups" in data:
|
| 104 |
for group in data["groups"]:
|
| 105 |
+
urls.extend(group.get("images", []))
|
|
|
|
| 106 |
elif isinstance(data, dict):
|
| 107 |
+
for rows in data.values():
|
| 108 |
if isinstance(rows, list):
|
| 109 |
for row in rows:
|
| 110 |
if "unstaged_images" in row:
|
| 111 |
urls.append(row["unstaged_images"])
|
| 112 |
except Exception as e:
|
| 113 |
print(f"Error loading URLs: {e}")
|
|
|
|
| 114 |
return urls
|
| 115 |
|
| 116 |
def get_ordered_groups():
|