feat: handle download and extraction for tarfiles with '.tar' extension

CS-SI · Nov 20, 2023 · 759a87e · 759a87e
1 parent 4cb0f70
commit 759a87e
Show file tree

Hide file tree

Showing 3 changed files with 35 additions and 7 deletions.
diff --git a/eodag/plugins/download/base.py b/eodag/plugins/download/base.py
@@ -377,7 +377,7 @@ def _finalize(self, fs_path, progress_callback=None, **kwargs):
                     os.makedirs(outputs_dir)
                 shutil.move(product_extraction_path, outputs_dir)
 
-            elif fs_path.endswith(".tar.gz"):
+            elif fs_path.endswith(".tar") or fs_path.endswith(".tar.gz"):
                 with tarfile.open(fs_path, "r") as zfile:
                     progress_callback.reset(total=1)
                     zfile.extractall(path=extraction_dir)

diff --git a/eodag/plugins/download/http.py b/eodag/plugins/download/http.py
@@ -373,8 +373,18 @@ def download(
             )
             progress_callback = ProgressCallback(disable=True)
 
+        kwargs.pop("outputs_extension", "parameter not available in kwargs")
+        outputs_extension = getattr(self.config, "products", {}).get(
+            product.product_type, {}
+        ).get("outputs_extension", None) or getattr(
+            self.config, "outputs_extension", ".zip"
+        )
+
         fs_path, record_filename = self._prepare_download(
-            product, progress_callback=progress_callback, **kwargs
+            product,
+            progress_callback=progress_callback,
+            outputs_extension=outputs_extension,
+            **kwargs,
         )
         if not fs_path or not record_filename:
             if fs_path:
@@ -417,14 +427,14 @@ def download_request(product, auth, progress_callback, wait, timeout, **kwargs):
         logger.debug("Download recorded in %s", record_filename)
 
         # Check that the downloaded file is really a zip file
-        outputs_extension = kwargs.get("outputs_extension", None) or getattr(
-            self.config, "outputs_extension", ".zip"
-        )
         if not zipfile.is_zipfile(fs_path) and outputs_extension == ".zip":
             logger.warning(
                 "Downloaded product is not a Zip File. Please check its file type before using it"
             )
-            new_fs_path = fs_path[: fs_path.index(".zip")]
+            new_fs_path = os.path.join(
+                os.path.dirname(fs_path),
+                sanitize(product.properties["title"]),
+            )
             if os.path.isfile(fs_path) and not tarfile.is_tarfile(fs_path):
                 if not os.path.isdir(new_fs_path):
                     os.makedirs(new_fs_path)
@@ -433,6 +443,18 @@ def download_request(product, auth, progress_callback, wait, timeout, **kwargs):
                 file_path = os.path.join(new_fs_path, os.listdir(new_fs_path)[0])
                 new_file_path = file_path[: file_path.index(".zip")]
                 shutil.move(file_path, new_file_path)
+            elif tarfile.is_tarfile(fs_path):
+                if not new_fs_path.endswith(".tar"):
+                    new_fs_path += ".tar"
+                shutil.move(fs_path, new_fs_path)
+                product_path = self._finalize(
+                    new_fs_path,
+                    progress_callback=progress_callback,
+                    outputs_extension=".tar",
+                    **kwargs,
+                )
+                product.location = path_to_uri(product_path)
+                return product_path
             else:
                 shutil.move(fs_path, new_fs_path)
             product.location = path_to_uri(new_fs_path)
@@ -451,7 +473,10 @@ def download_request(product, auth, progress_callback, wait, timeout, **kwargs):
             product.location = path_to_uri(new_fs_path)
             return new_fs_path
         product_path = self._finalize(
-            fs_path, progress_callback=progress_callback, **kwargs
+            fs_path,
+            progress_callback=progress_callback,
+            outputs_extension=outputs_extension,
+            **kwargs,
         )
         product.location = path_to_uri(product_path)
         return product_path

diff --git a/eodag/resources/providers.yml b/eodag/resources/providers.yml
@@ -6278,3 +6278,6 @@
     order_status_success:
       status: completed
       message: "Done!"
+    products:
+      COP_DEM_GLO30_DGED:
+        outputs_extension: .tar