hf folder path change
Browse files- github_storage.py +77 -77
github_storage.py
CHANGED
|
@@ -1,77 +1,77 @@
|
|
| 1 |
-
import os
|
| 2 |
-
from github import Github
|
| 3 |
-
import base64
|
| 4 |
-
import shutil
|
| 5 |
-
import zipfile
|
| 6 |
-
from io import BytesIO
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
# Global Variables
|
| 10 |
-
|
| 11 |
-
# HF ------------
|
| 12 |
-
hf_folder_path = '
|
| 13 |
-
zip_name = 'embeddings'
|
| 14 |
-
|
| 15 |
-
# Github -------
|
| 16 |
-
github_token = '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' # Replace with your GitHub token
|
| 17 |
-
repo_name = 'AdityaMetkar/Patseer-Database' # Replace with your repository, e.g., 'octocat/Hello-World'
|
| 18 |
-
folder_path = 'Manual Database/embeddings.zip' # Replace with the path to the folder in the repository
|
| 19 |
-
|
| 20 |
-
# Authenticate to GitHub
|
| 21 |
-
g = Github(github_token)
|
| 22 |
-
repo = g.get_repo(repo_name)
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
# Functions -------------------------------
|
| 27 |
-
def zip_folder():
|
| 28 |
-
shutil.make_archive(zip_name, 'zip', hf_folder_path)
|
| 29 |
-
return zip_name + '.zip'
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
def update_db():
|
| 33 |
-
|
| 34 |
-
try:
|
| 35 |
-
# Check if the file already exists in the repository
|
| 36 |
-
existing_file = repo.get_contents(folder_path)
|
| 37 |
-
|
| 38 |
-
compressed_zip = zip_folder()
|
| 39 |
-
with open(compressed_zip, 'rb') as file:
|
| 40 |
-
file_content = file.read()
|
| 41 |
-
|
| 42 |
-
# Update the existing file
|
| 43 |
-
repo.update_file(existing_file.path, "New DB Update", file_content, existing_file.sha)
|
| 44 |
-
print(f"Updated {folder_path} in GitHub repository.")
|
| 45 |
-
|
| 46 |
-
except Exception as e:
|
| 47 |
-
print(f"Error: {e}")
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
def download_db():
|
| 51 |
-
if not os.path.exists(hf_folder_path):
|
| 52 |
-
os.makedirs(hf_folder_path)
|
| 53 |
-
|
| 54 |
-
file_content = repo.get_contents(folder_path)
|
| 55 |
-
|
| 56 |
-
try:
|
| 57 |
-
# Download the zip file content from GitHub
|
| 58 |
-
file_content = repo.get_contents(folder_path)
|
| 59 |
-
zip_data = base64.b64decode(file_content.content)
|
| 60 |
-
|
| 61 |
-
# Extract the downloaded zip file directly to hf_folder_path using shutil
|
| 62 |
-
with zipfile.ZipFile(BytesIO(zip_data)) as zip_ref:
|
| 63 |
-
for file in zip_ref.namelist():
|
| 64 |
-
zip_ref.extract(file, hf_folder_path)
|
| 65 |
-
|
| 66 |
-
print(f"Successfully unzipped files to {hf_folder_path}")
|
| 67 |
-
|
| 68 |
-
except Exception as e:
|
| 69 |
-
print(f"Error: {e}")
|
| 70 |
-
|
| 71 |
-
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
# Download the folder
|
| 75 |
-
# download_folder()
|
| 76 |
-
# update_db()
|
| 77 |
-
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
from github import Github
|
| 3 |
+
import base64
|
| 4 |
+
import shutil
|
| 5 |
+
import zipfile
|
| 6 |
+
from io import BytesIO
|
| 7 |
+
|
| 8 |
+
|
| 9 |
+
# Global Variables
|
| 10 |
+
|
| 11 |
+
# HF ------------
|
| 12 |
+
hf_folder_path = '/home/user/app/embeddings'
|
| 13 |
+
zip_name = 'embeddings'
|
| 14 |
+
|
| 15 |
+
# Github -------
|
| 16 |
+
github_token = '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' # Replace with your GitHub token
|
| 17 |
+
repo_name = 'AdityaMetkar/Patseer-Database' # Replace with your repository, e.g., 'octocat/Hello-World'
|
| 18 |
+
folder_path = 'Manual Database/embeddings.zip' # Replace with the path to the folder in the repository
|
| 19 |
+
|
| 20 |
+
# Authenticate to GitHub
|
| 21 |
+
g = Github(github_token)
|
| 22 |
+
repo = g.get_repo(repo_name)
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
# Functions -------------------------------
|
| 27 |
+
def zip_folder():
|
| 28 |
+
shutil.make_archive(zip_name, 'zip', hf_folder_path)
|
| 29 |
+
return zip_name + '.zip'
|
| 30 |
+
|
| 31 |
+
|
| 32 |
+
def update_db():
|
| 33 |
+
|
| 34 |
+
try:
|
| 35 |
+
# Check if the file already exists in the repository
|
| 36 |
+
existing_file = repo.get_contents(folder_path)
|
| 37 |
+
|
| 38 |
+
compressed_zip = zip_folder()
|
| 39 |
+
with open(compressed_zip, 'rb') as file:
|
| 40 |
+
file_content = file.read()
|
| 41 |
+
|
| 42 |
+
# Update the existing file
|
| 43 |
+
repo.update_file(existing_file.path, "New DB Update", file_content, existing_file.sha)
|
| 44 |
+
print(f"Updated {folder_path} in GitHub repository.")
|
| 45 |
+
|
| 46 |
+
except Exception as e:
|
| 47 |
+
print(f"Error: {e}")
|
| 48 |
+
|
| 49 |
+
|
| 50 |
+
def download_db():
|
| 51 |
+
if not os.path.exists(hf_folder_path):
|
| 52 |
+
os.makedirs(hf_folder_path)
|
| 53 |
+
|
| 54 |
+
file_content = repo.get_contents(folder_path)
|
| 55 |
+
|
| 56 |
+
try:
|
| 57 |
+
# Download the zip file content from GitHub
|
| 58 |
+
file_content = repo.get_contents(folder_path)
|
| 59 |
+
zip_data = base64.b64decode(file_content.content)
|
| 60 |
+
|
| 61 |
+
# Extract the downloaded zip file directly to hf_folder_path using shutil
|
| 62 |
+
with zipfile.ZipFile(BytesIO(zip_data)) as zip_ref:
|
| 63 |
+
for file in zip_ref.namelist():
|
| 64 |
+
zip_ref.extract(file, hf_folder_path)
|
| 65 |
+
|
| 66 |
+
print(f"Successfully unzipped files to {hf_folder_path}")
|
| 67 |
+
|
| 68 |
+
except Exception as e:
|
| 69 |
+
print(f"Error: {e}")
|
| 70 |
+
|
| 71 |
+
|
| 72 |
+
|
| 73 |
+
|
| 74 |
+
# Download the folder
|
| 75 |
+
# download_folder()
|
| 76 |
+
# update_db()
|
| 77 |
+
|