|
| 1 | +import asyncio |
| 2 | +from pathlib import Path |
| 3 | +from typing import AsyncGenerator |
| 4 | + |
| 5 | +from boto3.session import Session |
| 6 | +from botocore.config import Config |
| 7 | +from botocore.exceptions import ClientError |
| 8 | + |
| 9 | +from . import Storage |
| 10 | + |
| 11 | +class S3Storage(Storage): |
| 12 | + def __init__( |
| 13 | + self, |
| 14 | + bucket: str, |
| 15 | + region: str = "us-east-1", |
| 16 | + endpoint_url: str | None = None, |
| 17 | + access_key_id: str | None = None, |
| 18 | + secret_access_key: str | None = None |
| 19 | + ): |
| 20 | + self.bucket = bucket |
| 21 | + |
| 22 | + config = Config( |
| 23 | + region_name = region, |
| 24 | + connect_timeout = 10, |
| 25 | + read_timeout = 60, |
| 26 | + retries = {"max_attempts": 5, "mode": "adaptive"}, |
| 27 | + signature_version = "s3v4", |
| 28 | + ) |
| 29 | + |
| 30 | + session_config = {} |
| 31 | + if access_key_id is not None: |
| 32 | + session_config["aws_access_key_id"] = access_key_id |
| 33 | + if secret_access_key is not None: |
| 34 | + session_config["aws_secret_access_key"] = secret_access_key |
| 35 | + |
| 36 | + session = Session(**session_config) |
| 37 | + |
| 38 | + self.s3_client = session.client("s3", config=config, endpoint_url=endpoint_url) |
| 39 | + |
| 40 | + async def store_file(self, source_path: Path, key: str) -> str: |
| 41 | + try: |
| 42 | + loop = asyncio.get_event_loop() |
| 43 | + await loop.run_in_executor( |
| 44 | + None, |
| 45 | + self.s3_client.upload_file, |
| 46 | + str(source_path), |
| 47 | + self.bucket, |
| 48 | + key |
| 49 | + ) |
| 50 | + return f"s3://{self.bucket}/{key}" |
| 51 | + except ClientError as e: |
| 52 | + raise ValueError(f"S3 upload failed: {e}") |
| 53 | + |
| 54 | + async def get_file_stream(self, uri: str) -> AsyncGenerator[bytes, None]: |
| 55 | + key = self._extract_key_from_uri(uri) |
| 56 | + |
| 57 | + try: |
| 58 | + loop = asyncio.get_event_loop() |
| 59 | + response = await loop.run_in_executor( |
| 60 | + None, |
| 61 | + self.s3_client.get_object, |
| 62 | + self.bucket, |
| 63 | + key |
| 64 | + ) |
| 65 | + |
| 66 | + body = response["Body"] |
| 67 | + try: |
| 68 | + while True: |
| 69 | + chunk = await loop.run_in_executor(None, body.read, 8192) |
| 70 | + if not chunk: |
| 71 | + break |
| 72 | + yield chunk |
| 73 | + finally: |
| 74 | + await loop.run_in_executor(None, body.close) |
| 75 | + |
| 76 | + except ClientError as e: |
| 77 | + raise ValueError(f"S3 download failed: {e}") |
| 78 | + |
| 79 | + async def get_download_url(self, uri: str) -> str: |
| 80 | + key = self._extract_key_from_uri(uri) |
| 81 | + |
| 82 | + try: |
| 83 | + loop = asyncio.get_event_loop() |
| 84 | + return await loop.run_in_executor( |
| 85 | + None, |
| 86 | + self.s3_client.generate_presigned_url, |
| 87 | + "get_object", |
| 88 | + {"Bucket": self.bucket, "Key": key}, |
| 89 | + 3600 # 1 hour expiration |
| 90 | + ) |
| 91 | + except ClientError as e: |
| 92 | + raise ValueError(f"Failed to generate presigned URL: {e}") |
| 93 | + |
| 94 | + def _extract_key_from_uri(self, uri: str) -> str: |
| 95 | + if not uri.startswith(f"s3://{self.bucket}/"): |
| 96 | + raise ValueError(f"Invalid S3 URI for bucket {self.bucket}: {uri}") |
| 97 | + |
| 98 | + return uri.replace(f"s3://{self.bucket}/", "") |
0 commit comments