Metadata database tracks all uploads with project, package, tag, and timestamp queryable via API
This commit is contained in:
Mondo Diaz
@@ -4,15 +4,14 @@ Test configuration and fixtures for Orchard backend tests.
|
||||
This module provides:
|
||||
- Database fixtures with test isolation
|
||||
- Mock S3 storage using moto
|
||||
- Test data factories for common scenarios
|
||||
- Shared pytest fixtures
|
||||
"""
|
||||
|
||||
import os
|
||||
import pytest
|
||||
import hashlib
|
||||
from typing import Generator, BinaryIO
|
||||
from unittest.mock import MagicMock, patch
|
||||
import io
|
||||
from typing import Generator
|
||||
from unittest.mock import MagicMock
|
||||
|
||||
# Set test environment defaults before importing app modules
|
||||
# Use setdefault to NOT override existing env vars (from docker-compose)
|
||||
@@ -26,54 +25,27 @@ os.environ.setdefault("ORCHARD_S3_BUCKET", "test-bucket")
|
||||
os.environ.setdefault("ORCHARD_S3_ACCESS_KEY_ID", "test")
|
||||
os.environ.setdefault("ORCHARD_S3_SECRET_ACCESS_KEY", "test")
|
||||
|
||||
|
||||
# =============================================================================
|
||||
# Test Data Factories
|
||||
# =============================================================================
|
||||
|
||||
|
||||
def create_test_file(content: bytes = None, size: int = 1024) -> io.BytesIO:
|
||||
"""
|
||||
Create a test file with known content.
|
||||
|
||||
Args:
|
||||
content: Specific content to use, or None to generate random-ish content
|
||||
size: Size of generated content if content is None
|
||||
|
||||
Returns:
|
||||
BytesIO object with the content
|
||||
"""
|
||||
if content is None:
|
||||
content = os.urandom(size)
|
||||
return io.BytesIO(content)
|
||||
|
||||
|
||||
def compute_sha256(content: bytes) -> str:
|
||||
"""Compute SHA256 hash of content as lowercase hex string."""
|
||||
return hashlib.sha256(content).hexdigest()
|
||||
|
||||
|
||||
def compute_md5(content: bytes) -> str:
|
||||
"""Compute MD5 hash of content as lowercase hex string."""
|
||||
return hashlib.md5(content).hexdigest()
|
||||
|
||||
|
||||
def compute_sha1(content: bytes) -> str:
|
||||
"""Compute SHA1 hash of content as lowercase hex string."""
|
||||
return hashlib.sha1(content).hexdigest()
|
||||
|
||||
|
||||
# Known test data with pre-computed hashes
|
||||
TEST_CONTENT_HELLO = b"Hello, World!"
|
||||
TEST_HASH_HELLO = "dffd6021bb2bd5b0af676290809ec3a53191dd81c7f70a4b28688a362182986f"
|
||||
TEST_MD5_HELLO = "65a8e27d8879283831b664bd8b7f0ad4"
|
||||
TEST_SHA1_HELLO = "0a0a9f2a6772942557ab5355d76af442f8f65e01"
|
||||
|
||||
TEST_CONTENT_EMPTY = b""
|
||||
# Note: Empty content should be rejected by the storage layer
|
||||
|
||||
TEST_CONTENT_BINARY = bytes(range(256))
|
||||
TEST_HASH_BINARY = compute_sha256(TEST_CONTENT_BINARY)
|
||||
# Re-export factory functions for backward compatibility
|
||||
from tests.factories import (
|
||||
create_test_file,
|
||||
compute_sha256,
|
||||
compute_md5,
|
||||
compute_sha1,
|
||||
upload_test_file,
|
||||
TEST_CONTENT_HELLO,
|
||||
TEST_HASH_HELLO,
|
||||
TEST_MD5_HELLO,
|
||||
TEST_SHA1_HELLO,
|
||||
TEST_CONTENT_EMPTY,
|
||||
TEST_CONTENT_BINARY,
|
||||
TEST_HASH_BINARY,
|
||||
get_s3_client,
|
||||
get_s3_bucket,
|
||||
list_s3_objects_by_hash,
|
||||
count_s3_objects_by_prefix,
|
||||
s3_object_exists,
|
||||
delete_s3_object_by_hash,
|
||||
)
|
||||
|
||||
|
||||
# =============================================================================
|
||||
@@ -289,126 +261,3 @@ def test_content():
|
||||
content = f"test-content-{uuid.uuid4().hex}".encode()
|
||||
sha256 = compute_sha256(content)
|
||||
return (content, sha256)
|
||||
|
||||
|
||||
def upload_test_file(
|
||||
client,
|
||||
project: str,
|
||||
package: str,
|
||||
content: bytes,
|
||||
filename: str = "test.bin",
|
||||
tag: str = None,
|
||||
) -> dict:
|
||||
"""
|
||||
Helper function to upload a test file.
|
||||
|
||||
Returns the upload response as a dict.
|
||||
"""
|
||||
files = {"file": (filename, io.BytesIO(content), "application/octet-stream")}
|
||||
data = {}
|
||||
if tag:
|
||||
data["tag"] = tag
|
||||
|
||||
response = client.post(
|
||||
f"/api/v1/project/{project}/{package}/upload",
|
||||
files=files,
|
||||
data=data if data else None,
|
||||
)
|
||||
assert response.status_code == 200, f"Upload failed: {response.text}"
|
||||
return response.json()
|
||||
|
||||
|
||||
# =============================================================================
|
||||
# S3 Direct Access Helpers (for integration tests)
|
||||
# =============================================================================
|
||||
|
||||
|
||||
def get_s3_client():
|
||||
"""
|
||||
Create a boto3 S3 client for direct S3 access in integration tests.
|
||||
|
||||
Uses environment variables for configuration (same as the app).
|
||||
Note: When running in container, S3 endpoint should be 'minio:9000' not 'localhost:9000'.
|
||||
"""
|
||||
import boto3
|
||||
from botocore.config import Config
|
||||
|
||||
config = Config(s3={"addressing_style": "path"})
|
||||
|
||||
# Use the same endpoint as the app (minio:9000 in container, localhost:9000 locally)
|
||||
endpoint = os.environ.get("ORCHARD_S3_ENDPOINT", "http://minio:9000")
|
||||
|
||||
return boto3.client(
|
||||
"s3",
|
||||
endpoint_url=endpoint,
|
||||
region_name=os.environ.get("ORCHARD_S3_REGION", "us-east-1"),
|
||||
aws_access_key_id=os.environ.get("ORCHARD_S3_ACCESS_KEY_ID", "minioadmin"),
|
||||
aws_secret_access_key=os.environ.get(
|
||||
"ORCHARD_S3_SECRET_ACCESS_KEY", "minioadmin"
|
||||
),
|
||||
config=config,
|
||||
)
|
||||
|
||||
|
||||
def get_s3_bucket():
|
||||
"""Get the S3 bucket name from environment."""
|
||||
return os.environ.get("ORCHARD_S3_BUCKET", "orchard-artifacts")
|
||||
|
||||
|
||||
def list_s3_objects_by_hash(sha256_hash: str) -> list:
|
||||
"""
|
||||
List S3 objects that match a specific SHA256 hash.
|
||||
|
||||
Uses the fruits/{hash[:2]}/{hash[2:4]}/{hash} key pattern.
|
||||
Returns list of matching object keys.
|
||||
"""
|
||||
client = get_s3_client()
|
||||
bucket = get_s3_bucket()
|
||||
prefix = f"fruits/{sha256_hash[:2]}/{sha256_hash[2:4]}/{sha256_hash}"
|
||||
|
||||
response = client.list_objects_v2(Bucket=bucket, Prefix=prefix)
|
||||
|
||||
if "Contents" not in response:
|
||||
return []
|
||||
|
||||
return [obj["Key"] for obj in response["Contents"]]
|
||||
|
||||
|
||||
def count_s3_objects_by_prefix(prefix: str) -> int:
|
||||
"""
|
||||
Count S3 objects with a given prefix.
|
||||
|
||||
Useful for checking if duplicate uploads created multiple objects.
|
||||
"""
|
||||
client = get_s3_client()
|
||||
bucket = get_s3_bucket()
|
||||
|
||||
response = client.list_objects_v2(Bucket=bucket, Prefix=prefix)
|
||||
|
||||
if "Contents" not in response:
|
||||
return 0
|
||||
|
||||
return len(response["Contents"])
|
||||
|
||||
|
||||
def s3_object_exists(sha256_hash: str) -> bool:
|
||||
"""
|
||||
Check if an S3 object exists for a given SHA256 hash.
|
||||
"""
|
||||
objects = list_s3_objects_by_hash(sha256_hash)
|
||||
return len(objects) > 0
|
||||
|
||||
|
||||
def delete_s3_object_by_hash(sha256_hash: str) -> bool:
|
||||
"""
|
||||
Delete an S3 object by its SHA256 hash (for test cleanup).
|
||||
"""
|
||||
client = get_s3_client()
|
||||
bucket = get_s3_bucket()
|
||||
s3_key = f"fruits/{sha256_hash[:2]}/{sha256_hash[2:4]}/{sha256_hash}"
|
||||
|
||||
try:
|
||||
client.delete_object(Bucket=bucket, Key=s3_key)
|
||||
return True
|
||||
except Exception:
|
||||
return False
|
||||
|
||||
Reference in New Issue
Block a user