gamora commited on
Commit
5daa268
·
1 Parent(s): 8f4a7ee

concurnecy

Browse files
Files changed (2) hide show
  1. app.py +1 -1
  2. upload_to_s3.py +78 -0
app.py CHANGED
@@ -17,7 +17,7 @@ from concurrent.futures import ThreadPoolExecutor
17
  import random
18
  import string
19
 
20
- from experiments.upload_to_s3 import upload_multiple_files
21
  load_dotenv()
22
 
23
  # Access the variables
 
17
  import random
18
  import string
19
 
20
+ from upload_to_s3 import upload_multiple_files
21
  load_dotenv()
22
 
23
  # Access the variables
upload_to_s3.py ADDED
@@ -0,0 +1,78 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ from dotenv import load_dotenv
2
+ import boto3
3
+ import os
4
+ from botocore.exceptions import ClientError
5
+ import time
6
+ from boto3.s3.transfer import TransferConfig# Configuration
7
+ REGION = "us-east-1"
8
+ BUCKET = "shortsai-us" # Replace with your bucket name
9
+ S3_KEY = "test/IMG_8280.MOV" # S3 path for the file
10
+ LOCAL_FILE_PATH = "/Users/georgia.bucea/products/ShortsAI/21_aug/mcp_video-3028_singular_display.MOV" # Replace with your local file path
11
+
12
+ load_dotenv()
13
+
14
+ # Access the variables
15
+
16
+
17
+ from concurrent.futures import ThreadPoolExecutor
18
+
19
+ def upload_multiple_files(file_list, bucket):
20
+
21
+ uploaded_files = []
22
+
23
+ def upload_to_s3(file_path, bucket, s3_key):
24
+ """Upload a local file to an S3 bucket."""
25
+
26
+ start_time = time.time()
27
+ try:
28
+
29
+ config = TransferConfig(
30
+ multipart_threshold=16*1024*1024, # 8MB threshold
31
+ max_concurrency=30, # Parallel upload threads
32
+ multipart_chunksize=8*1024*1024, # Chunk size
33
+ use_threads=True,
34
+ )
35
+ # Initialize S3 client
36
+ s3 = boto3.client("s3",endpoint_url='https://s3-accelerate.amazonaws.com', region_name=REGION )
37
+
38
+ # Verify file exists
39
+ if not os.path.exists(file_path):
40
+ raise FileNotFoundError(f"File not found: {file_path}")
41
+
42
+ # # Upload file
43
+ s3.upload_file(file_path, bucket, s3_key,ExtraArgs={'ACL': 'bucket-owner-full-control'}, Config=config)
44
+ print(f"Successfully uploaded {file_path} to s3://{bucket}/{s3_key}")
45
+
46
+ uploaded_files.append(s3_key)
47
+ end_time = time.time()
48
+ # Calculate elapsed time
49
+ elapsed_time = end_time - start_time
50
+
51
+ print(f"In {elapsed_time} time ")
52
+ return True
53
+
54
+ except FileNotFoundError as e:
55
+ print(f"Error: {e}")
56
+ return False
57
+ except ClientError as e:
58
+ print(f"AWS Client Error: {e}")
59
+ return False
60
+ except Exception as e:
61
+ print(f"Unexpected error: {e}")
62
+ return False
63
+
64
+ with ThreadPoolExecutor(max_workers=10) as executor:
65
+ executor.map(lambda f: upload_to_s3(f[0], bucket, f[1]), file_list)
66
+ return uploaded_files
67
+
68
+
69
+ if __name__ == "__main__":
70
+ # upload_to_s3(LOCAL_FILE_PATH,BUCKET,S3_KEY)
71
+ bucket = os.getenv("BUCKET")
72
+
73
+
74
+ file_list = [
75
+ ("/Users/georgia.bucea/products/ShortsAI/21_aug/mcp_video-3028_singular_display.MOV", "test/21_aug/mcp_video-3028_singular_display.MOV"),
76
+ ]
77
+
78
+ uploaded = upload_multiple_files(file_list, bucket)