Fixing S2 metadata retrieval

Getting masks from granule metadata

core/registrar/scheme.py

@@ -34,7 +34,7 @@ def parse_footprint(value):
 
 
 class Sentinel2RegistrationScheme(RegistrationScheme):
-    MTD_TL_SCHEMA = {
+    MTD_MSIL2A_SCHEMA = {
         'begin_time': Parameter('/n1:Level-2A_User_Product/n1:General_Info/Product_Info/PRODUCT_START_TIME/text()', False, parse_datetime),
         'end_time': Parameter('/n1:Level-2A_User_Product/n1:General_Info/Product_Info/PRODUCT_STOP_TIME/text()', False, parse_datetime),
         'identifier': Parameter('/n1:Level-2A_User_Product/n1:General_Info/Product_Info/PRODUCT_URI/text()'),
@@ -44,35 +44,46 @@ class Sentinel2RegistrationScheme(RegistrationScheme):
         'generation_time': Parameter('/n1:Level-2A_User_Product/n1:General_Info/Product_Info/GENERATION_TIME/text()', False, parse_datetime),
         'cloud_cover': Parameter('/n1:Level-2A_User_Product/n1:Quality_Indicators_Info/Cloud_Coverage_Assessment/text()'),
         'image_file_paths': Parameter('/n1:Level-2A_User_Product/n1:General_Info/Product_Info/Product_Organisation/Granule_List/Granule/IMAGE_FILE/text()', True),
-        'mask_file_paths': Parameter('/n1:Level-2A_Tile_ID/n1:Quality_Indicators_Info/Pixel_Level_QI/MASK_FILENAME', True),
     }
 
-    S2_NAMESPACES = {
+    MTD_TL_SCHEMA = {
+        'mask_file_paths': Parameter('/n1:Level-2A_Tile_ID/n1:Quality_Indicators_Info/Pixel_Level_QI/MASK_FILENAME/text()', True),
+    }
+
+    MTD_MSIL2A_NAMESPACES = {
         'n1': "https://psd-14.sentinel2.eo.esa.int/PSD/User_Product_Level-2A.xsd"
     }
 
+    MTD_TL_NAMESPACES = {
+        'n1': 'https://psd-14.sentinel2.eo.esa.int/PSD/S2_PDI_Level-2A_Tile_Metadata.xsd'
+    }
+
     def get_context(self, source: Source, path: str):
         metadata_file = join(path, 'MTD_MSIL2A.xml')
-        mtd_tree = read_xml(source, metadata_file)
-
-        # get MTD metadata
+        tree = read_xml(source, metadata_file)
 
-        metadata = parse_metadata_schema(mtd_tree, self.MTD_TL_SCHEMA, self.S2_NAMESPACES)
+        # get product metadata
+        metadata = parse_metadata_schema(tree, self.MTD_MSIL2A_SCHEMA, self.MTD_MSIL2A_NAMESPACES)
 
-        band_re = re.compile(r'.*([A-Z0-9]{3})_([0-9]{2}m)$')
+        band_re = re.compile(r'.*([A-Z0-9]{3}_[0-9]{2}m)$')
         raster_files = {
             band_re.match(image_file_path).groups()[0]: f'{join(path, image_file_path)}.jp2'
             for image_file_path in metadata['image_file_paths']
         }
 
+        # get granule metadata
+        mtd_files = source.list_files(join(path, 'GRANULE'), '*/MTD_TL.xml')
+        logger.info(f'{mtd_files}')
+        tl_tree = read_xml(source, mtd_files[0])
+        tile_metadata = parse_metadata_schema(tl_tree, self.MTD_TL_SCHEMA, self.MTD_TL_NAMESPACES)
+
         mask_type_re = re.compile(r'.*/MSK_([A-Z]*)_([A-Z0-9]{3}).[a-z0-9]+$')
         mask_files = {
-            mask_type_re.match(mask_file_path).groups[0]: join(path, mask_file_path)
-            for mask_file_path in metadata['mask_file_paths']
+            mask_type_re.match(mask_file_path).groups()[0]: join(path, mask_file_path)
+            for mask_file_path in tile_metadata['mask_file_paths']
+            if mask_type_re.match(mask_file_path) is not None
         }
 
-        logger.info(f'{mask_files} {metadata["mask_file_paths"]}')
-
         return Context(
             identifier=metadata['identifier'],
             path=path,