Fixing extraction method bugs.

setup.py

@@ -65,6 +65,8 @@
             "header = stac_generator.plugins.extraction_methods.header.header:HeaderExtract",
             "iso19115 = stac_generator.plugins.extraction_methods.iso19115:ISO19115Extract",
             "xml = stac_generator.plugins.extraction_methods.xml:XMLExtract",
+            "open_zip = stac_generator.plugins.extraction_methods.open_zip:ZipExtract",
+            "datetime_bound_to_centroid = stac_generator.plugins.extraction_methods.datetime_bound_to_centroid:DatetimeBoundToCentroidExtract",
             "elasticsearch_aggregation = stac_generator.plugins.extraction_methods.elasticsearch_aggregation:ElasticsearchAggregationExtract",
             "json_file = stac_generator.plugins.extraction_methods.json_file:JsonFileExtract",
             "path_parts = stac_generator.plugins.extraction_methods.path_parts:PathPartsExtract",
@@ -84,6 +86,7 @@
             "geometry_line = stac_generator.plugins.extraction_methods.geometry_line:GeometryLineExtract",
             "geometry_point = stac_generator.plugins.extraction_methods.geometry_point:GeometryPointExtract",
             "geometry_polygon = stac_generator.plugins.extraction_methods.geometry_polygon:GeometryPolygonExtract",
+            "geometry_to_bbox = stac_generator.plugins.extraction_methods.geometry_to_bbox:GeometryToBboxExtract",
             "string_join = stac_generator.plugins.extraction_methods.string_join:StringJoinExtract",
             "string_template = stac_generator.plugins.extraction_methods.string_template:StringTemplateExtract",
             "facet_prefix = stac_generator.plugins.extraction_methods.facet_prefix:FacetPrefixExtract",

stac_generator/core/baker.py

@@ -105,9 +105,6 @@ def __init__(self, root_path: str):
         for file_path in Path(root_path).rglob("*.y*ml"):
             _ = self._load_data(file_path)
 
-        # only used during the
-        del self.location_map
-
     def _load_data(self, file: Path) -> Recipe:
         """
         Loads the yaml files from the root path and builds the recipe dictionary and map of
@@ -159,7 +156,7 @@ def get(self, path: str, stac_type: str) -> Recipe:
         :param path: Path for which to retrieve the recipe
         """
         if path in self.recipes[stac_type]:
-            return self.load_recipe(path)
+            return self.load_recipe(path, stac_type)
 
         for parent in chain([path], Path(path).parents):
             if parent in self.paths_map[stac_type]:
@@ -168,3 +165,6 @@ def get(self, path: str, stac_type: str) -> Recipe:
                 return self.load_recipe(key, stac_type)
 
         raise ValueError(f"No Recipe found for path: {path}")
+
+    def get_maps(self):
+        return self.paths_map, self.location_map

stac_generator/core/extraction_method.py

@@ -29,6 +29,9 @@ def __init__(self, **kwargs):
         # Override with specific processor settings
         self._set_attrs(kwargs)
 
+        if not hasattr(self, "exists_key"):
+            self.exists_key = "$"
+
     def _set_attrs(self, conf: dict) -> None:
         """
         Set instance attributes

stac_generator/plugins/extraction_methods/asset_aggregator.py

@@ -124,20 +124,22 @@ def run(self, body: dict, **kwargs) -> dict:
 
         for asset in body["assets"].values():
             for list_term in self.list_terms:
-                body[list_term["name"]].append(asset[list_term["key"]])
+                if list_term["key"] in asset:
+                    body[list_term["name"]].append(asset[list_term["key"]])
 
             for sum_term in self.sum_terms:
-                body[sum_term["name"]] += asset[sum_term["key"]]
+                if sum_term["key"] in asset:
+                    body[sum_term["name"]] += asset[sum_term["key"]]
 
             for avg_term in self.avg_terms:
                 body[avg_term["name"]] /= len(body["assets"])
 
             for min_term in self.min_terms:
-                if asset[min_term["key"]] < body[min_term["name"]]:
+                if min_term["key"] in asset and asset[min_term["key"]] < body[min_term["name"]]:
                     body[min_term["name"]] = asset[min_term["key"]]
 
             for max_term in self.max_terms:
-                if asset[max_term["key"]] < body[max_term["name"]]:
+                if max_term["key"] in asset and asset[max_term["key"]] < body[max_term["name"]]:
                     body[max_term["name"]] = asset[max_term["key"]]
 
         return body

stac_generator/plugins/extraction_methods/bbox.py

@@ -41,14 +41,19 @@ class BboxExtract(BaseExtractionMethod):
 
     def run(self, body: dict, **kwargs):
         try:
+            west = body[self.coordinate_keys[0]]
+            south = body[self.coordinate_keys[1]]
+            east = body[self.coordinate_keys[2]]
+            north = body[self.coordinate_keys[3]]
+
             coordinates = [
                 [
-                    float(body[self.coordinate_keys[0]]),
-                    float(body[self.coordinate_keys[1]]),
+                    float(west) if west is not None else west,
+                    float(south) if south is not None else south,
                 ],
                 [
-                    float(body[self.coordinate_keys[2]]),
-                    float(body[self.coordinate_keys[3]]),
+                    float(east) if east is not None else east,
+                    float(north) if north is not None else north,
                 ],
             ]
 
@@ -57,6 +62,9 @@ def run(self, body: dict, **kwargs):
                 "coordinates": coordinates,
             }
 
+        except TypeError:
+            LOGGER.warning("Unable to convert bbox.", exc_info=True)
+
         except KeyError:
             LOGGER.warning("Unable to convert bbox.", exc_info=True)
 

stac_generator/plugins/extraction_methods/datetime_bound_to_centroid.py

@@ -0,0 +1,78 @@
+__author__ = "Richard Smith"
+__date__ = "28 May 2021"
+__copyright__ = "Copyright 2018 United Kingdom Research and Innovation"
+__license__ = "BSD - see LICENSE file in top-level package directory"
+__contact__ = "richard.d.smith@stfc.ac.uk"
+
+
+import logging
+
+from datetime import datetime
+
+# Package imports
+from stac_generator.core.extraction_method import BaseExtractionMethod
+
+LOGGER = logging.getLogger(__name__)
+
+
+class DatetimeBoundToCentroidExtract(BaseExtractionMethod):
+    """
+
+    Processor Name: ``datetime_bound_to_centroid``
+
+    Description:
+        Accepts a dictionary of coordinate values and converts to `RFC 7946, section 5 <https://tools.ietf.org/html/rfc7946#section-5>`_
+        formatted bbox.
+
+    Configuration Options:
+        - ``coordinate_keys``: ``REQUIRED`` list of keys to convert to bbox array. Ordering is respected.
+
+    Example Configuration:
+
+    .. code-block:: yaml
+
+        - method: stac_bbox
+            inputs:
+              output_term: polygon
+
+    """
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+
+        if not hasattr(self, "start_term"):
+            self.start_term = {"name": "start_datetime", "format": "%Y-%m-%dT%H:%M:%S"}
+
+        if "name" not in self.start_term:
+            self.start_term["name"] = "start_datetime"
+
+        if "format" not in self.start_term:
+            self.start_term["format"] = "%Y-%m-%dT%H:%M:%S"
+
+        if not hasattr(self, "end_term"):
+            self.end_term = {"name": "end_datetime", "format": "%Y-%m-%dT%H:%M:%S"}
+
+        if "name" not in self.end_term:
+            self.end_term["name"] = "start_datetime"
+
+        if "format" not in self.end_term:
+            self.end_term["format"] = "%Y-%m-%dT%H:%M:%S"
+
+        if not hasattr(self, "output_term"):
+            self.output_term = {"name": "datetime", "format": "%Y-%m-%dT%H:%M:%S"}
+
+        if "name" not in self.output_term:
+            self.output_term["name"] = "datetime"
+
+        if "format" not in self.output_term:
+            self.output_term["format"] = "%Y-%m-%dT%H:%M:%S"
+
+    def run(self, body: dict, **kwargs):
+
+        start_datetime = datetime.strptime(body[self.start_term["name"]], self.start_term["format"])
+        end_datetime = datetime.strptime(body[self.end_term["name"]], self.end_term["format"])
+
+        centroid_datetime = start_datetime + (end_datetime - start_datetime) / 2
+
+        body[self.output_term["name"]] = centroid_datetime.strftime(self.output_term["format"])
+
+        return body

stac_generator/plugins/extraction_methods/default.py

@@ -45,12 +45,6 @@ class DefaultExtract(BaseExtractionMethod):
 
     """
 
-    def __init__(self, **kwargs):
-        super().__init__(**kwargs)
-
-        if not hasattr(self, "exists_key"):
-            self.exists_key = "$"
-
     def run(self, body: dict, **kwargs) -> dict:
         defaults = {}
         for default_key, default_value in self.defaults.items():