From 8d44566464ee26ba98c3e5d50c177cd3f2deb0b1 Mon Sep 17 00:00:00 2001
From: Sylvain Brunato <sylvain.brunato@c-s.fr>
Date: Wed, 22 May 2024 11:22:39 +0200
Subject: [PATCH] fix: harmonized regex address retrieval mechanism

---
 eodag_cube/api/product/drivers/generic.py     | 33 +++++++++----
 eodag_cube/api/product/drivers/stac_assets.py | 49 ++++++-------------
 2 files changed, 38 insertions(+), 44 deletions(-)

diff --git a/eodag_cube/api/product/drivers/generic.py b/eodag_cube/api/product/drivers/generic.py
index 59e7156..c029e92 100644
--- a/eodag_cube/api/product/drivers/generic.py
+++ b/eodag_cube/api/product/drivers/generic.py
@@ -17,6 +17,8 @@
 # limitations under the License.
 from __future__ import annotations
 
+import logging
+import re
 from pathlib import Path
 from typing import TYPE_CHECKING
 
@@ -29,6 +31,8 @@
 if TYPE_CHECKING:
     from eodag.api.product._product import EOProduct
 
+logger = logging.getLogger("eodag-cube.driver.generic")
+
 
 class GenericDriver(DatasetDriver):
     """Generic Driver for products that need to be downloaded"""
@@ -48,16 +52,27 @@ def get_data_address(self, eo_product: EOProduct, band: str) -> str:
         product_location_scheme = eo_product.location.split("://")[0]
         if product_location_scheme == "file":
 
-            filenames = Path(uri_to_path(eo_product.location)).glob(f"**/*{band}*")
+            p = re.compile(rf"{band}", re.IGNORECASE)
+            matching_files = []
+            for f_path in Path(uri_to_path(eo_product.location)).glob("**/*"):
+                f_str = str(f_path.resolve())
+                if p.search(f_str):
+                    try:
+                        # files readable by rasterio
+                        rasterio.drivers.driver_from_extension(f_path)
+                        matching_files.append(f_str)
+                        logger.debug(f"Matching band: {f_str}")
+                    except ValueError:
+                        pass
+
+            if len(matching_files) == 1:
+                return matching_files[0]
+
+            raise AddressNotFound(
+                rf"Please adapt given band parameter ('{band}') to match only file: "
+                rf"{len(matching_files)} files found matching {p}"
+            )
 
-            for filename in filenames:
-                try:
-                    # return the first file readable by rasterio
-                    rasterio.drivers.driver_from_extension(filename)
-                    return str(filename.resolve())
-                except ValueError:
-                    pass
-            raise AddressNotFound
         raise UnsupportedDatasetAddressScheme(
             "eo product {} is accessible through a location scheme that is not yet "
             "supported by eodag: {}".format(eo_product, product_location_scheme)
diff --git a/eodag_cube/api/product/drivers/stac_assets.py b/eodag_cube/api/product/drivers/stac_assets.py
index 2e63349..73543b0 100644
--- a/eodag_cube/api/product/drivers/stac_assets.py
+++ b/eodag_cube/api/product/drivers/stac_assets.py
@@ -17,6 +17,7 @@
 # limitations under the License.
 from __future__ import annotations
 
+import logging
 import re
 from typing import TYPE_CHECKING
 
@@ -26,6 +27,8 @@
 if TYPE_CHECKING:
     from eodag.api.product._product import EOProduct
 
+logger = logging.getLogger("eodag-cube.driver.stac_assets")
+
 
 class StacAssets(DatasetDriver):
     """Driver for Stac Assets"""
@@ -42,47 +45,23 @@ def get_data_address(self, eo_product: EOProduct, band: str) -> str:
         :raises: :class:`~eodag.utils.exceptions.AddressNotFound`
         :raises: :class:`~eodag.utils.exceptions.UnsupportedDatasetAddressScheme`
         """
-        error_message = ""
-
-        # try using exact
-        p = re.compile(rf"^{band}$", re.IGNORECASE)
-        matching_keys = [
-            s
-            for s in eo_product.assets.keys()
+        p = re.compile(rf"{band}", re.IGNORECASE)
+        matching_keys = []
+        for s in eo_product.assets.keys():
             if (
                 (
                     "roles" in eo_product.assets[s]
                     and "data" in eo_product.assets[s]["roles"]
                 )
                 or ("roles" not in eo_product.assets[s])
-            )
-            and p.match(s)
-        ]
+            ) and p.search(s):
+                matching_keys.append(s)
+                logger.debug(f"Matching asset key: {s}")
+
         if len(matching_keys) == 1:
             return str(eo_product.assets[matching_keys[0]]["href"])
-        else:
-            error_message += (
-                rf"{len(matching_keys)} assets keys found matching {p} AND "
-            )
 
-            # try to find keys containing given band
-            p = re.compile(rf"^.*{band}.*$", re.IGNORECASE)
-            matching_keys = [
-                s
-                for s in eo_product.assets.keys()
-                if (
-                    (
-                        "roles" in eo_product.assets[s]
-                        and "data" in eo_product.assets[s]["roles"]
-                    )
-                    or ("roles" not in eo_product.assets[s])
-                )
-                and p.match(s)
-            ]
-            if len(matching_keys) == 1:
-                return str(eo_product.assets[matching_keys[0]]["href"])
-            else:
-                raise AddressNotFound(
-                    rf"Please adapt given band parameter ('{band}') to match only one asset: {error_message}"
-                    rf"{len(matching_keys)} assets keys found matching {p}"
-                )
+        raise AddressNotFound(
+            rf"Please adapt given band parameter ('{band}') to match only one asset: "
+            rf"{len(matching_keys)} assets keys found matching {p}"
+        )