Initial commit of document-service

app/s3.py

@@ -0,0 +1,101 @@
+import boto3
+import tempfile
+import os
+from botocore.client import Config
+from fastapi import UploadFile
+from app.config import settings
+from app.logger import get_logger
+
+logger = get_logger(__name__)
+
+def get_client():
+    return boto3.client(
+        "s3",
+        endpoint_url=settings.s3_endpoint,
+        aws_access_key_id=settings.s3_access_key,
+        aws_secret_access_key=settings.s3_secret_key,
+        config=Config(signature_version="s3v4"),
+        region_name=settings.s3_region
+    )
+
+def ensure_bucket_exists() -> None:
+    """Ensure the S3 bucket exists, create it if it doesn't exist.
+    
+    Raises:
+        Exception: If bucket creation fails (service will fail to start)
+    """
+    client = get_client()
+    try:
+        client.head_bucket(Bucket=settings.s3_bucket)
+        logger.info(f"Bucket '{settings.s3_bucket}' already exists")
+    except client.exceptions.ClientError as e:
+        error_code = e.response['Error']['Code']
+        if error_code == '404':
+            try:
+                client.create_bucket(
+                    Bucket=settings.s3_bucket,
+                    CreateBucketConfiguration={
+                        'LocationConstraint': settings.s3_region
+                    }
+                )
+                logger.info(f"Created bucket '{settings.s3_bucket}'")
+            except Exception as create_error:
+                logger.error(f"Failed to create bucket '{settings.s3_bucket}': {create_error}")
+                raise
+        else:
+            logger.error(f"Error checking bucket: {e}")
+            raise
+
+def upload_file(file: UploadFile, s3_key: str, content_type: str, metadata: dict = None) -> str:
+    """Upload file to S3 with metadata"""
+    client = get_client()
+
+    extra_args = {"ContentType": content_type}
+    if metadata:
+        extra_args["Metadata"] = metadata
+
+    client.upload_fileobj(
+        file.file,
+        settings.s3_bucket,
+        s3_key,
+        ExtraArgs=extra_args
+    )
+    return s3_key
+
+def delete_file(s3_key: str) -> None:
+    """Delete file from S3"""
+    client = get_client()
+    client.delete_object(Bucket=settings.s3_bucket, Key=s3_key)
+
+def file_exists(s3_key: str) -> bool:
+    """Check if file exists in S3"""
+    client = get_client()
+    try:
+        client.head_object(Bucket=settings.s3_bucket, Key=s3_key)
+        return True
+    except client.exceptions.ClientError:
+        return False
+
+def get_file_metadata(s3_key: str) -> dict:
+    """Get file metadata from S3"""
+    client = get_client()
+    response = client.head_object(Bucket=settings.s3_bucket, Key=s3_key)
+    return response.get("Metadata", {})
+
+def download_to_temp(s3_key: str) -> str:
+    """Download file from S3 to temp file"""
+    client = get_client()
+    suffix = os.path.splitext(s3_key)[-1] or ".tmp"
+    tmp = tempfile.NamedTemporaryFile(delete=False, suffix=suffix)
+    client.download_fileobj(settings.s3_bucket, s3_key, tmp)
+    tmp.close()
+    return tmp.name
+
+def presigned_download_url(s3_key: str, expires_in: int = 3600) -> str:
+    """Generate presigned download URL"""
+    client = get_client()
+    return client.generate_presigned_url(
+        "get_object",
+        Params={"Bucket": settings.s3_bucket, "Key": s3_key},
+        ExpiresIn=expires_in
+    )