Add Google Cloud Storage backend

cogeo_mosaic/backends/__init__.py

@@ -6,6 +6,7 @@
 from cogeo_mosaic.backends.base import BaseBackend
 from cogeo_mosaic.backends.dynamodb import DynamoDBBackend
 from cogeo_mosaic.backends.file import FileBackend
+from cogeo_mosaic.backends.gs import GCSBackend
 from cogeo_mosaic.backends.memory import MemoryBackend
 from cogeo_mosaic.backends.s3 import S3Backend
 from cogeo_mosaic.backends.sqlite import SQLiteBackend
@@ -29,6 +30,10 @@ def MosaicBackend(url: str, *args: Any, **kwargs: Any) -> BaseBackend:
     elif parsed.scheme == "s3":
         return S3Backend(url, *args, **kwargs)
 
+    # `gs://{bucket}/{key}`
+    elif parsed.scheme == "gs":
+        return GCSBackend(url, *args, **kwargs)
+
     # `dynamodb://{region}/{table}:{mosaic}`
     elif parsed.scheme == "dynamodb":
         return DynamoDBBackend(url, *args, **kwargs)

cogeo_mosaic/backends/gs.py

@@ -0,0 +1,103 @@
+"""cogeo-mosaic Google Cloud Storage backend."""
+
+import json
+from typing import Any
+from urllib.parse import urlparse
+
+import attr
+from cachetools import TTLCache, cached
+from cachetools.keys import hashkey
+
+from cogeo_mosaic.backends.base import BaseBackend
+from cogeo_mosaic.backends.utils import _compress_gz_json, _decompress_gz
+from cogeo_mosaic.cache import cache_config
+from cogeo_mosaic.errors import _HTTP_EXCEPTIONS, MosaicError, MosaicExistsError
+from cogeo_mosaic.mosaic import MosaicJSON
+
+try:
+    from google.auth.exceptions import GoogleAuthError
+    from google.cloud.storage import Client as gcp_session
+except ImportError:  # pragma: nocover
+    gcp_session = None  # type: ignore
+    GoogleAuthError = None  # type: ignore
+
+
+@attr.s
+class GCSBackend(BaseBackend):
+    """GCS Backend Adapter"""
+
+    client: Any = attr.ib(default=None)
+    bucket: str = attr.ib(init=False)
+    key: str = attr.ib(init=False)
+
+    _backend_name = "Google Cloud Storage"
+
+    def __attrs_post_init__(self):
+        """Post Init: parse path and create client."""
+        assert (
+            gcp_session is not None
+        ), "'google-cloud-storage' must be installed to use GCSBackend"
+
+        parsed = urlparse(self.path)
+        self.bucket = parsed.netloc
+        self.key = parsed.path.strip("/")
+        self.client = self.client or gcp_session()
+        super().__attrs_post_init__()
+
+    def write(self, overwrite: bool = False, **kwargs: Any):
+        """Write mosaicjson document to Google Cloud Storage."""
+        if not overwrite and self._head_object(self.key, self.bucket):
+            raise MosaicExistsError("Mosaic file already exist, use `overwrite=True`.")
+
+        mosaic_doc = self.mosaic_def.dict(exclude_none=True)
+        if self.key.endswith(".gz"):
+            body = _compress_gz_json(mosaic_doc)
+        else:
+            body = json.dumps(mosaic_doc).encode("utf-8")
+
+        self._put_object(self.key, self.bucket, body, **kwargs)
+
+    @cached(
+        TTLCache(maxsize=cache_config.maxsize, ttl=cache_config.ttl),
+        key=lambda self: hashkey(self.path),
+    )
+    def _read(self) -> MosaicJSON:  # type: ignore
+        """Get mosaicjson document."""
+        body = self._get_object(self.key, self.bucket)
+        self._file_byte_size = len(body)
+
+        if self.key.endswith(".gz"):
+            body = _decompress_gz(body)
+
+        return MosaicJSON(**json.loads(body))
+
+    def _get_object(self, key: str, bucket: str) -> bytes:
+        try:
+            gcs_bucket = self.client.bucket(bucket)
+            response = gcs_bucket.blob(key).download_as_bytes()
+        except GoogleAuthError as e:
+            status_code = e.response["ResponseMetadata"]["HTTPStatusCode"]
+            exc = _HTTP_EXCEPTIONS.get(status_code, MosaicError)
+            raise exc(e.response["Error"]["Message"]) from e
+
+        return response
+
+    def _put_object(self, key: str, bucket: str, body: bytes, **kwargs) -> str:
+        try:
+            gcs_bucket = self.client.bucket(bucket)
+            blob = gcs_bucket.blob(key)
+            blob.upload_from_string(body)
+        except GoogleAuthError as e:
+            status_code = e.response["ResponseMetadata"]["HTTPStatusCode"]
+            exc = _HTTP_EXCEPTIONS.get(status_code, MosaicError)
+            raise exc(e.response["Error"]["Message"]) from e
+
+        return key
+
+    def _head_object(self, key: str, bucket: str) -> bool:
+        try:
+            gcs_bucket = self.client.bucket(bucket)
+            blob = gcs_bucket.blob(key)
+            return blob.exists()
+        except GoogleAuthError:
+            return False

docs/advanced/backends.md

@@ -8,6 +8,8 @@ Starting in version `3.0.0`, we introduced specific `backend` to abstract mosaic
 
 - **S3Backend** (`s3://`)
 
+- **GCSBackend** (`gs://`)
+
 - **DynamoDBBackend** (`dynamodb://{region}/{table_name}`). If `region` is not passed, it reads the value of the `AWS_REGION` environment variable. If that environment variable does not exist, it falls back to `us-east-1`. If you choose not to pass a `region`, you still need three `/` before the table name, like so `dynamodb:///{table_name}`.
 
 - **SQLiteBackend** (`sqlite:///{file.db}:{mosaic_name}`)
@@ -28,7 +30,7 @@ with MemoryBackend(mosaic_def=mosaicjson) as mosaic:
     img = mosaic.tile(1, 1, 1)
 ```
 
-### Absctract Class
+### Abstract Class
 
 All backends are built from a `BaseBackend` which is a sub-class or `rio-tiler.io.BaseReader`.
 
@@ -42,6 +44,9 @@ from cogeo_mosaic.backends import MosaicBackend
 with MosaicBackend("s3://mybucket/amosaic.json") as mosaic:
     assert isinstance(mosaic, cogeo_mosaic.backends.s3.S3Backend)
 
+with MosaicBackend("gs://mybucket/amosaic.json") as mosaic:
+    assert isinstance(mosaic, cogeo_mosaic.backends.gs.GCSBackend)
+
 with MosaicBackend("dynamodb://us-east-1/amosaic") as mosaic:
     assert isinstance(mosaic, cogeo_mosaic.backends.dynamodb.DynamoDBBackend)
 
@@ -70,6 +75,18 @@ with MosaicBackend(None, mosaic_def=mosaic) as mosaic:
     assert isinstance(mosaic, cogeo_mosaic.backends.memory.MemoryBackend)
 ```
 
+## GCS Backend
+
+The GCS Backend allows read and write operations from Google Cloud Storage.
+
+When using this backend is necessary to set the appropriate roles and IAM
+permissions to let cogeo-mosaic access the files. For example:
+
+* Read-only bucket - IAM Role `roles/storage.objectViewer`. It is possible
+  to restrict the read-only operation to a single bucket by using the
+  following condition: `resource.type == "storage.googleapis.com/Object"
+  && resource.name.startsWith("projects/_/buckets/mybucket")`
+
 ## STAC Backend
 
 The STACBackend is a **read-only** backend, meaning it can't be used to write a file. This backend will POST to the input url looking for STAC items which will then be used to create the mosaicJSON in memory.

setup.py

@@ -21,6 +21,7 @@
 
 extra_reqs = {
     "aws": ["boto3"],
+    "gcp": ["google-cloud-storage"],
     "test": ["pytest", "pytest-cov"],
     "dev": ["pytest", "pytest-cov", "pre-commit"],
     "docs": ["mkdocs", "mkdocs-material", "pygments", "mkapi", "mkdocs-jupyter"],

tests/test_backends.py

@@ -16,7 +16,7 @@
 from requests.exceptions import HTTPError, RequestException
 from rio_tiler.errors import PointOutsideBounds
 
-from cogeo_mosaic.backends import MosaicBackend
+from cogeo_mosaic.backends import GCSBackend, MosaicBackend
 from cogeo_mosaic.backends.dynamodb import DynamoDBBackend
 from cogeo_mosaic.backends.file import FileBackend
 from cogeo_mosaic.backends.memory import MemoryBackend
@@ -307,6 +307,103 @@ def test_s3_backend(session):
     session.reset_mock()
 
 
+@patch("cogeo_mosaic.backends.gs.gcp_session")
+def test_gs_backend(session):
+    """Test GS backend."""
+    with open(mosaic_gz, "rb") as f:
+        session.return_value.bucket.return_value.blob.return_value.download_as_bytes.return_value = (
+            f.read()
+        )
+
+    session.return_value.bucket.return_value.blob.return_value.upload_from_string.return_value = (
+        True
+    )
+    session.return_value.bucket.return_value.blob.return_value.exists.return_value = (
+        False
+    )
+
+    with MosaicBackend("gs://mybucket/mymosaic.json.gz") as mosaic:
+        assert mosaic._backend_name == "Google Cloud Storage"
+        assert isinstance(mosaic, GCSBackend)
+        assert (
+            mosaic.mosaicid
+            == "24d43802c19ef67cc498c327b62514ecf70c2bbb1bbc243dda1ee075"
+        )
+        assert mosaic.quadkey_zoom == 7
+        assert list(mosaic.metadata.dict(exclude_none=True).keys()) == [
+            "mosaicjson",
+            "version",
+            "minzoom",
+            "maxzoom",
+            "quadkey_zoom",
+            "bounds",
+            "center",
+        ]
+        assert mosaic.assets_for_tile(150, 182, 9) == ["cog1.tif", "cog2.tif"]
+        assert mosaic.assets_for_point(-73, 45) == ["cog1.tif", "cog2.tif"]
+        session.return_value.bucket.assert_called_once_with("mybucket")
+        session.return_value.bucket.return_value.blob.assert_called_once_with(
+            "mymosaic.json.gz"
+        )
+
+        session.return_value.bucket.return_value.blob.return_value.upload_from_string.assert_not_called()
+        session.return_value.bucket.return_value.blob.return_value.exists.assert_not_called()
+        session.reset_mock()
+
+    with open(mosaic_gz, "rb") as f:
+        session.return_value.bucket.return_value.blob.return_value.download_as_bytes.return_value = (
+            f.read()
+        )
+
+    session.return_value.bucket.return_value.blob.return_value.upload_from_string.return_value = (
+        True
+    )
+    session.return_value.bucket.return_value.blob.return_value.exists.return_value = (
+        False
+    )
+
+    with MosaicBackend(
+        "gs://mybucket/mymosaic.json.gz", mosaic_def=mosaic_content
+    ) as mosaic:
+        assert isinstance(mosaic, GCSBackend)
+        mosaic.write()
+    session.return_value.bucket.return_value.blob.return_value.download_as_bytes.assert_not_called()
+
+    session.return_value.bucket.return_value.blob.return_value.exists.assert_called_once()
+    session.reset_mock()
+
+    session.return_value.bucket.return_value.blob.return_value.exists.return_value = (
+        False
+    )
+    with MosaicBackend("gs://mybucket/00000.json", mosaic_def=mosaic_content) as mosaic:
+        assert isinstance(mosaic, GCSBackend)
+        mosaic.write()
+    session.return_value.bucket.return_value.blob.return_value.download_as_bytes.assert_not_called()
+    session.return_value.bucket.return_value.blob.return_value.exists.assert_called_once()
+    session.reset_mock()
+
+    session.return_value.bucket.return_value.blob.return_value.exists.return_value = (
+        True
+    )
+    with MosaicBackend("gs://mybucket/00000.json", mosaic_def=mosaic_content) as mosaic:
+        assert isinstance(mosaic, GCSBackend)
+        with pytest.raises(MosaicExistsError):
+            mosaic.write()
+    session.return_value.client.return_value.get_object.assert_not_called()
+    session.return_value.bucket.return_value.blob.return_value.exists.assert_called_once()
+    session.reset_mock()
+
+    session.return_value.bucket.return_value.blob.return_value.exists.return_value = (
+        True
+    )
+    with MosaicBackend("gs://mybucket/00000.json", mosaic_def=mosaic_content) as mosaic:
+        assert isinstance(mosaic, GCSBackend)
+        mosaic.write(overwrite=True)
+    session.return_value.client.return_value.get_object.assert_not_called()
+    session.return_value.client.return_value.head_object.assert_not_called()
+    session.reset_mock()
+
+
 class MockMeta(object):
     """Mock Meta."""