eodag 3.0.0b3__py3-none-any.whl → 3.1.0__py3-none-any.whl

eodag/plugins/download/base.py

@@ -26,17 +26,7 @@ import tempfile
 import zipfile
 from datetime import datetime, timedelta
 from time import sleep
-from typing import (
-    TYPE_CHECKING,
-    Any,
-    Callable,
-    Dict,
-    List,
-    Optional,
-    Tuple,
-    TypeVar,
-    Union,
-)
+from typing import TYPE_CHECKING, Any, Callable, Optional, TypeVar, Union
 
 from eodag.plugins.base import PluginTopic
 from eodag.utils import (
@@ -60,6 +50,7 @@ if TYPE_CHECKING:
     from eodag.api.product import EOProduct
     from eodag.api.search_result import SearchResult
     from eodag.config import PluginConfig
+    from eodag.types import S3SessionKwargs
     from eodag.types.download_args import DownloadConf
     from eodag.utils import DownloadedCallback, Unpack
 
@@ -81,15 +72,15 @@ class Download(PluginTopic):
 
     - download data in the ``output_dir`` folder defined in the plugin's
       configuration or passed through kwargs
-    - extract products from their archive (if relevant) if ``extract`` is set to True
-      (True by default)
+    - extract products from their archive (if relevant) if ``extract`` is set to ``True``
+      (``True`` by default)
     - save a product in an archive/directory (in ``output_dir``) whose name must be
       the product's ``title`` property
     - update the product's ``location`` attribute once its data is downloaded (and
       eventually after it's extracted) to the product's location given as a file URI
-      (e.g. 'file:///tmp/product_folder' on Linux or
-      'file:///C:/Users/username/AppData/LOcal/Temp' on Windows)
-    - save a *record* file in the directory ``output_dir/.downloaded`` whose name
+      (e.g. ``file:///tmp/product_folder`` on Linux or
+      ``file:///C:/Users/username/AppData/Local/Temp`` on Windows)
+    - save a *record* file in the directory ``{output_dir}/.downloaded`` whose name
       is built on the MD5 hash of the product's ``product_type`` and ``properties['id']``
       attributes (``hashlib.md5((product.product_type+"-"+product.properties['id']).encode("utf-8")).hexdigest()``)
       and whose content is the product's ``remote_location`` attribute itself.
@@ -110,10 +101,10 @@ class Download(PluginTopic):
     def download(
         self,
         product: EOProduct,
-        auth: Optional[Union[AuthBase, Dict[str, str]]] = None,
+        auth: Optional[Union[AuthBase, S3SessionKwargs]] = None,
         progress_callback: Optional[ProgressCallback] = None,
-        wait: int = DEFAULT_DOWNLOAD_WAIT,
-        timeout: int = DEFAULT_DOWNLOAD_TIMEOUT,
+        wait: float = DEFAULT_DOWNLOAD_WAIT,
+        timeout: float = DEFAULT_DOWNLOAD_TIMEOUT,
         **kwargs: Unpack[DownloadConf],
     ) -> Optional[str]:
         r"""
@@ -130,8 +121,8 @@ class Download(PluginTopic):
                         and will override any other values defined in a configuration
                         file or with environment variables.
         :returns: The absolute path to the downloaded product in the local filesystem
-            (e.g. '/tmp/product.zip' on Linux or
-            'C:\\Users\\username\\AppData\\Local\\Temp\\product.zip' on Windows)
+            (e.g. ``/tmp/product.zip`` on Linux or
+            ``C:\\Users\\username\\AppData\\Local\\Temp\\product.zip`` on Windows)
         """
         raise NotImplementedError(
             "A Download plugin must implement a method named download"
@@ -140,10 +131,10 @@ class Download(PluginTopic):
     def _stream_download_dict(
         self,
         product: EOProduct,
-        auth: Optional[Union[AuthBase, Dict[str, str]]] = None,
+        auth: Optional[Union[AuthBase, S3SessionKwargs]] = None,
         progress_callback: Optional[ProgressCallback] = None,
-        wait: int = DEFAULT_DOWNLOAD_WAIT,
-        timeout: int = DEFAULT_DOWNLOAD_TIMEOUT,
+        wait: float = DEFAULT_DOWNLOAD_WAIT,
+        timeout: float = DEFAULT_DOWNLOAD_TIMEOUT,
         **kwargs: Unpack[DownloadConf],
     ) -> StreamResponse:
         r"""
@@ -155,11 +146,11 @@ class Download(PluginTopic):
         :param wait: (optional) If download fails, wait time in minutes between two download tries
         :param timeout: (optional) If download fails, maximum time in minutes before stop retrying
                         to download
-        :param kwargs: `output_dir` (str), `extract` (bool), `delete_archive` (bool)
-                        and `dl_url_params` (dict) can be provided as additional kwargs
+        :param kwargs: ``output_dir`` (str), ``extract`` (bool), ``delete_archive`` (bool)
+                        and ``dl_url_params`` (dict) can be provided as additional kwargs
                         and will override any other values defined in a configuration
                         file or with environment variables.
-        :returns: Dictionnary of :class:`~fastapi.responses.StreamingResponse` keyword-arguments
+        :returns: Dictionary of :class:`~fastapi.responses.StreamingResponse` keyword-arguments
         """
         raise NotImplementedError(
             "Download streaming must be implemented using a method named _stream_download_dict"
@@ -170,7 +161,7 @@ class Download(PluginTopic):
         product: EOProduct,
         progress_callback: Optional[ProgressCallback] = None,
         **kwargs: Unpack[DownloadConf],
-    ) -> Tuple[Optional[str], Optional[str]]:
+    ) -> tuple[Optional[str], Optional[str]]:
         """Check if file has already been downloaded, and prepare product download
 
         :param product: The EO product to download
@@ -202,8 +193,8 @@ class Download(PluginTopic):
             or getattr(self.config, "output_dir", tempfile.gettempdir())
             or tempfile.gettempdir()
         )
-        output_extension = kwargs.get("output_extension", None) or getattr(
-            self.config, "output_extension", ".zip"
+        output_extension = kwargs.get("output_extension") or getattr(
+            self.config, "output_extension", ""
         )
 
         # Strong asumption made here: all products downloaded will be zip files
@@ -233,9 +224,13 @@ class Download(PluginTopic):
                 logger.warning(
                     f"Unable to create records directory. Got:\n{tb.format_exc()}",
                 )
+        url_hash = hashlib.md5(url.encode("utf-8")).hexdigest()
+        old_record_filename = os.path.join(download_records_dir, url_hash)
         record_filename = os.path.join(
             download_records_dir, self.generate_record_hash(product)
         )
+        if os.path.isfile(old_record_filename):
+            os.rename(old_record_filename, record_filename)
         if os.path.isfile(record_filename) and os.path.isfile(fs_path):
             logger.info(
                 f"Product already downloaded: {fs_path}",
@@ -339,13 +334,7 @@ class Download(PluginTopic):
             if delete_archive is not None
             else getattr(self.config, "delete_archive", True)
         )
-        output_extension = kwargs.pop("output_extension", ".zip")
-
-        product_path = (
-            fs_path[: fs_path.index(output_extension)]
-            if output_extension in fs_path
-            else fs_path
-        )
+        product_path, _ = os.path.splitext(fs_path)
         product_path_exists = os.path.exists(product_path)
         if product_path_exists and os.path.isfile(product_path):
             logger.info(
@@ -422,10 +411,10 @@ class Download(PluginTopic):
 
             tmp_dir.cleanup()
 
-            if delete_archive:
+            if delete_archive and os.path.isfile(fs_path):
                 logger.info(f"Deleting archive {os.path.basename(fs_path)}")
                 os.unlink(fs_path)
-            else:
+            elif os.path.isfile(fs_path):
                 logger.info(
                     f"Archive deletion is deactivated, keeping {os.path.basename(fs_path)}"
                 )
@@ -441,13 +430,13 @@ class Download(PluginTopic):
     def download_all(
         self,
         products: SearchResult,
-        auth: Optional[Union[AuthBase, Dict[str, str]]] = None,
+        auth: Optional[Union[AuthBase, S3SessionKwargs]] = None,
         downloaded_callback: Optional[DownloadedCallback] = None,
         progress_callback: Optional[ProgressCallback] = None,
-        wait: int = DEFAULT_DOWNLOAD_WAIT,
-        timeout: int = DEFAULT_DOWNLOAD_TIMEOUT,
+        wait: float = DEFAULT_DOWNLOAD_WAIT,
+        timeout: float = DEFAULT_DOWNLOAD_TIMEOUT,
         **kwargs: Unpack[DownloadConf],
-    ) -> List[str]:
+    ) -> list[str]:
         """
         Base download_all method.
 
@@ -458,7 +447,7 @@ class Download(PluginTopic):
         :param auth: (optional) authenticated object
         :param downloaded_callback: (optional) A method or a callable object which takes
                                     as parameter the ``product``. You can use the base class
-                                    :class:`~eodag.api.product.DownloadedCallback` and override
+                                    :class:`~eodag.utils.DownloadedCallback` and override
                                     its ``__call__`` method. Will be called each time a product
                                     finishes downloading
         :param progress_callback: (optional) A progress callback
@@ -476,7 +465,7 @@ class Download(PluginTopic):
         # Products are going to be removed one by one from this sequence once
         # downloaded.
         products = products[:]
-        paths: List[str] = []
+        paths: list[str] = []
         # initiate retry loop
         start_time = datetime.now()
         stop_time = start_time + timedelta(minutes=timeout)
@@ -541,7 +530,7 @@ class Download(PluginTopic):
                             )
                             raise
 
-                        except RuntimeError:
+                        except (RuntimeError, Exception):
                             import traceback as tb
 
                             logger.error(
@@ -549,16 +538,9 @@ class Download(PluginTopic):
                                 "Skipping it"
                             )
                             logger.debug(f"\n{tb.format_exc()}")
-                            stop_time = datetime.now()
 
-                        except Exception:
-                            import traceback as tb
-
-                            logger.warning(
-                                f"A problem occurred during download of product: {product}. "
-                                "Skipping it",
-                            )
-                            logger.debug(f"\n{tb.format_exc()}")
+                            # product skipped, to not retry it
+                            products.remove(product)
 
                 if (
                     len(products) > 0
@@ -585,14 +567,14 @@ class Download(PluginTopic):
 
         return paths
 
-    def _download_retry(
-        self, product: EOProduct, wait: int, timeout: int
+    def _order_download_retry(
+        self, product: EOProduct, wait: float, timeout: float
     ) -> Callable[[Callable[..., T]], Callable[..., T]]:
         """
-        Download retry decorator.
+        Order download retry decorator.
 
-        Retries the wrapped  download method after `wait` minutes if a NotAvailableError
-        exception is thrown until `timeout` minutes.
+        Retries the wrapped order_download method after ``wait`` minutes if a
+        ``NotAvailableError`` exception is thrown until ``timeout`` minutes.
 
         :param product: The EO product to download
         :param wait: If download fails, wait time in minutes between two download tries
@@ -601,7 +583,7 @@ class Download(PluginTopic):
         :returns: decorator
         """
 
-        def decorator(download: Callable[..., T]) -> Callable[..., T]:
+        def decorator(order_download: Callable[..., T]) -> Callable[..., T]:
             def download_and_retry(*args: Any, **kwargs: Unpack[DownloadConf]) -> T:
                 # initiate retry loop
                 start_time = datetime.now()
@@ -618,7 +600,7 @@ class Download(PluginTopic):
                     if datetime_now >= product.next_try:
                         product.next_try += timedelta(minutes=wait)
                         try:
-                            return download(*args, **kwargs)
+                            return order_download(*args, **kwargs)
 
                         except NotAvailableError as e:
                             if not getattr(self.config, "order_enabled", False):
@@ -634,7 +616,7 @@ class Download(PluginTopic):
                         ).seconds
                         retry_count += 1
                         retry_info = (
-                            f"[Retry #{retry_count}] Waited {wait_seconds}s, trying again to download ordered product"
+                            f"[Retry #{retry_count}] Waited {wait_seconds}s, checking order status again"
                             f" (retry every {wait}' for {timeout}')"
                         )
                         logger.info(not_available_info)
@@ -656,8 +638,8 @@ class Download(PluginTopic):
                         ).microseconds / 1e6
                         retry_count += 1
                         retry_info = (
-                            f"[Retry #{retry_count}] Waiting {wait_seconds}s until next download try"
-                            f" for ordered product (retry every {wait}' for {timeout}')"
+                            f"[Retry #{retry_count}] Waiting {wait_seconds}s until next order status check"
+                            f" (retry every {wait}' for {timeout}')"
                         )
                         logger.info(not_available_info)
                         # Retry-After info from Response header
@@ -678,12 +660,12 @@ class Download(PluginTopic):
                         logger.info(not_available_info)
                         raise NotAvailableError(
                             f"{product.properties['title']} is not available ({product.properties['storageStatus']})"
-                            f" and could not be downloaded, timeout reached"
+                            f" and order was not successfull, timeout reached"
                         )
                     elif datetime_now >= stop_time:
                         raise NotAvailableError(not_available_info)
 
-                return download(*args, **kwargs)
+                return order_download(*args, **kwargs)
 
             return download_and_retry
 

eodag/plugins/download/creodias_s3.py

@@ -15,17 +15,28 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+from typing import Optional
 
 import boto3
 from botocore.exceptions import ClientError
 
+from eodag import EOProduct
 from eodag.plugins.download.aws import AwsDownload
 from eodag.utils.exceptions import MisconfiguredError
 
 
 class CreodiasS3Download(AwsDownload):
     """
-    Download on creodias s3 from their VMs
+    Download on creodias s3 from their VMs (extension of :class:`~eodag.plugins.download.aws.AwsDownload`)
+
+    :param provider: provider name
+    :param config: Download plugin configuration:
+
+        * :attr:`~eodag.config.PluginConfig.type` (``str``) (**mandatory**): CreodiasS3Download
+        * :attr:`~eodag.config.PluginConfig.base_uri` (``str``) (**mandatory**): s3 endpoint url
+        * :attr:`~eodag.config.PluginConfig.s3_bucket` (``str``) (**mandatory**): bucket where the products can be found
+        * :attr:`~eodag.config.PluginConfig.ssl_verify` (``bool``): if the ssl certificates should be
+          verified in requests; default: ``True``
     """
 
     def _get_authenticated_objects_unsigned(self, bucket_name, prefix, auth_dict):
@@ -50,9 +61,36 @@ class CreodiasS3Download(AwsDownload):
 
         s3_session = boto3.session.Session(**auth_dict)
         s3_resource = s3_session.resource(
-            "s3", endpoint_url=getattr(self.config, "base_uri", None)
+            "s3", endpoint_url=getattr(self.config, "s3_endpoint", None)
         )
         objects = s3_resource.Bucket(bucket_name).objects.filter()
         list(objects.filter(Prefix=prefix).limit(1))
         self.s3_session = s3_session
         return objects
+
+    def _get_bucket_names_and_prefixes(
+        self,
+        product: EOProduct,
+        asset_filter: Optional[str] = None,
+        ignore_assets: Optional[bool] = False,
+    ) -> list[tuple[str, Optional[str]]]:
+        """
+        Retrieves the bucket names and path prefixes for the assets
+
+        :param product: product for which the assets shall be downloaded
+        :param asset_filter: text for which the assets should be filtered
+        :param ignore_assets: if product instead of individual assets should be used
+        :return: tuples of bucket names and prefixes
+        """
+        # if assets are defined, use them instead of scanning product.location
+        if len(product.assets) > 0 and not ignore_assets:
+            bucket_names_and_prefixes = super()._get_bucket_names_and_prefixes(
+                product, asset_filter, ignore_assets
+            )
+        else:
+            # if no assets are given, use productIdentifier to get S3 path for download
+            s3_url = "s3:/" + product.properties["productIdentifier"]
+            bucket_names_and_prefixes = [
+                self.get_product_bucket_name_and_prefix(product, s3_url)
+            ]
+        return bucket_names_and_prefixes