hestia-earth-aggregation 0.21.7__tar.gz → 0.21.9__tar.gz

{hestia_earth_aggregation-0.21.7/hestia_earth_aggregation.egg-info → hestia_earth_aggregation-0.21.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: hestia_earth_aggregation
-Version: 0.21.7
+Version: 0.21.9
 Summary: HESTIA's aggregation engine.
 Home-page: https://gitlab.com/hestia-earth/hestia-aggregation-engine
 Author: HESTIA Team

{hestia_earth_aggregation-0.21.7 → hestia_earth_aggregation-0.21.9}/hestia_earth/aggregation/__init__.py

@@ -16,20 +16,10 @@ def _mock_nb_distribution(include_distribution: bool):
     not include_distribution and logger.warning("Not generating distribution.")
 
 
-def _get_aggregate_function(country: dict, is_global: bool | None):
-    if is_global is True and not is_global_country(country):
-        raise ValueError(
-            "country must be a region (e.g., World, Continent) when using global aggregation"
-        )
-
+def _get_aggregate_function(country: dict, filter_by_country: bool = True):
     return (
         aggregate_global
-        if is_global
-        or (
-            # automatic detection
-            is_global is None
-            and is_global_country(country)
-        )
+        if is_global_country(country) and filter_by_country
         else aggregate_country
     )
 
@@ -42,7 +32,7 @@ def aggregate(
     source: dict = None,
     include_distribution: bool = True,
     filter_by_country: bool = True,
-    is_global: bool | None = None,
+    include_covariance: bool = True,
 ):
     """
     Aggregates data from HESTIA.
@@ -64,10 +54,9 @@ def aggregate(
         Include a `distribution` for each aggregated data point. Included by default.
     filter_by_country : bool
         If set to `False`, Cycles from all countries will be used.
-        Only used for country-level aggregations.
-    is_global : bool | None
-        Run global aggregation, which will find aggregated Country-level data, and weight the result by country.
-        If the parameter is not set, it will be inferred from the `country`.
+        When used with `Worl` or `Continent` aggregation, setting `False` will also use non-aggregated Cycles.
+    include_covariance : bool
+        Include the `covarianceMatrix`. Included by default.
 
     Returns
     -------
@@ -76,7 +65,9 @@ def aggregate(
     """
     # mock nb distributions depending on the parameter
     _mock_nb_distribution(include_distribution)
-    aggregate_function = _get_aggregate_function(country=country, is_global=is_global)
+    aggregate_function = _get_aggregate_function(
+        country=country, filter_by_country=filter_by_country
+    )
 
     now = current_time_ms()
     logger.info(
@@ -88,7 +79,14 @@ def aggregate(
         end_year,
     )
     aggregations, countries = aggregate_function(
-        country, product, source, start_year, end_year, filter_by_country
+        country,
+        product,
+        source,
+        start_year,
+        end_year,
+        include_distribution=include_distribution,
+        filter_by_country=filter_by_country,
+        include_covariance=include_covariance,
     )
     logger.info("time=%s, unit=ms", current_time_ms() - now)
     aggregations = (

{hestia_earth_aggregation-0.21.7 → hestia_earth_aggregation-0.21.9}/hestia_earth/aggregation/aggregate_cycles.py

@@ -91,6 +91,9 @@ def aggregate_country(
     start_year: int,
     end_year: int,
     filter_by_country: bool = True,
+    include_covariance: bool = True,
+    *args,
+    **kwargs,
 ) -> Tuple[list, list]:
     """
     Create 1 to many country-level aggregations.
@@ -130,7 +133,11 @@ def aggregate_country(
 
     # combine cycles into a "master" cycle with multiple values
     cycles_aggregated = aggregate_cycles(
-        cycles=cycles, product=product, start_year=start_year, end_year=end_year
+        cycles=cycles,
+        product=product,
+        start_year=start_year,
+        end_year=end_year,
+        include_covariance=include_covariance,
     )
     if not cycles_aggregated:
         logger.info("2 - No aggregated cycles.")
@@ -181,7 +188,14 @@ def aggregate_country(
     log_memory_usage()
 
     country_cycle = (
-        (country_cycle | generate_covariance_country(weights=weights))
+        (
+            country_cycle
+            | (
+                generate_covariance_country(weights=weights)
+                if include_covariance
+                else {}
+            )
+        )
         if country_cycle
         else None
     )

{hestia_earth_aggregation-0.21.7 → hestia_earth_aggregation-0.21.9}/hestia_earth/aggregation/utils/aggregate_country_nodes.py

@@ -720,7 +720,13 @@ def _aggregate_sites(site_ids: List[str], start_year: int, end_year: int) -> Sit
     return site
 
 
-def aggregate_cycles(cycles: list, product: dict, start_year: int, end_year: int):
+def aggregate_cycles(
+    cycles: list,
+    product: dict,
+    start_year: int,
+    end_year: int,
+    include_covariance: bool = True,
+):
     batches = range(0, len(cycles), _BATCH_SIZE)
     logger.info(f"Processing {len(cycles)} cycles in {len(batches)} batches.")
 
@@ -742,13 +748,17 @@ def aggregate_cycles(cycles: list, product: dict, start_year: int, end_year: int
     logger.info("Finished processing batch, combining all cycles...")
     log_memory_usage()
     # generate covariance on sub cycles
-    covariances = {key: generate_covariance_cycles(key) for key in all_cycles.keys()}
+    covariances = (
+        {key: generate_covariance_cycles(key) for key in all_cycles.keys()}
+        if include_covariance
+        else {}
+    )
     all_cycles = {
         key: _combine_formatted_cycles(cycles=values)
         for key, values in all_cycles.items()
     }
     # add covariances to cycles once they have been combined
-    cycles = [cycle | covariances[key] for key, cycle in all_cycles.items()]
+    cycles = [cycle | covariances.get(key, {}) for key, cycle in all_cycles.items()]
     # discard memory
     all_cycles = {}
 

{hestia_earth_aggregation-0.21.7 → hestia_earth_aggregation-0.21.9}/hestia_earth/aggregation/utils/covariance.py

@@ -136,7 +136,7 @@ def add_covariance_cycles(cycles: list, suffix: str):
 
 
 def _format_covariance_value(value):
-    return None if pd.isna(value) else value
+    return None if any([pd.isna(value), value == -np.inf, value == np.inf]) else value
 
 
 def _format_covariance_matrix(values):

hestia_earth_aggregation-0.21.9/hestia_earth/aggregation/version.py

@@ -0,0 +1 @@
+VERSION = "0.21.9"

{hestia_earth_aggregation-0.21.7 → hestia_earth_aggregation-0.21.9/hestia_earth_aggregation.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: hestia_earth_aggregation
-Version: 0.21.7
+Version: 0.21.9
 Summary: HESTIA's aggregation engine.
 Home-page: https://gitlab.com/hestia-earth/hestia-aggregation-engine
 Author: HESTIA Team

hestia_earth_aggregation-0.21.9/tests/test_aggregation.py

@@ -0,0 +1,60 @@
+import pytest
+from unittest.mock import Mock, patch
+
+from hestia_earth.aggregation.utils.term import DEFAULT_COUNTRY_ID
+from hestia_earth.aggregation import aggregate
+
+class_path = "hestia_earth.aggregation"
+
+country_id = "GADM-FRA"
+product = {}
+start_year = 2000
+end_year = 2009
+
+
+@pytest.mark.parametrize(
+    "country,filter_by_country,called",
+    [
+        ({"@id": country_id}, True, True),
+        ({"@id": country_id}, False, True),
+        ({"@id": DEFAULT_COUNTRY_ID}, True, False),
+        ({"@id": DEFAULT_COUNTRY_ID}, False, True),
+    ],
+)
+@patch(f"{class_path}.aggregate_global", return_value=([], []))
+@patch(f"{class_path}.aggregate_country", return_value=([], []))
+def test_aggregate_country(
+    mock_aggregate: Mock,
+    mock_1: Mock,
+    country: dict,
+    filter_by_country: bool,
+    called: bool,
+):
+    aggregate(
+        country, product, start_year, end_year, filter_by_country=filter_by_country
+    )
+    assert mock_aggregate.called == called
+
+
+@pytest.mark.parametrize(
+    "country,filter_by_country,called",
+    [
+        ({"@id": country_id}, True, False),
+        ({"@id": country_id}, False, False),
+        ({"@id": DEFAULT_COUNTRY_ID}, True, True),
+        ({"@id": DEFAULT_COUNTRY_ID}, False, False),
+    ],
+)
+@patch(f"{class_path}.aggregate_country", return_value=([], []))
+@patch(f"{class_path}.aggregate_global", return_value=([], []))
+def test_aggregate_global(
+    mock_aggregate: Mock,
+    mock_1: Mock,
+    country: dict,
+    filter_by_country: bool,
+    called: bool,
+):
+    aggregate(
+        country, product, start_year, end_year, filter_by_country=filter_by_country
+    )
+    assert mock_aggregate.called == called

hestia_earth_aggregation-0.21.7/hestia_earth/aggregation/version.py

@@ -1 +0,0 @@
-VERSION = "0.21.7"

hestia_earth_aggregation-0.21.7/tests/test_aggregation.py

@@ -1,34 +0,0 @@
-from unittest.mock import Mock, patch
-
-from hestia_earth.aggregation import aggregate
-
-class_path = "hestia_earth.aggregation"
-
-
-@patch(f"{class_path}.aggregate_global", return_value=([], []))
-@patch(f"{class_path}.aggregate_country", return_value=([], []))
-def test_aggregate(mock_aggregate_country: Mock, mock_aggregate_global: Mock):
-    product = {}
-    start_year = 2000
-    end_year = 2009
-
-    country = {"@id": "GADM-FRA"}
-    aggregate(country, product, start_year, end_year)
-    mock_aggregate_country.assert_called_once()
-    mock_aggregate_global.assert_not_called()
-
-    mock_aggregate_global.reset_mock()
-    mock_aggregate_country.reset_mock()
-
-    country = {"@id": "region-world"}
-    aggregate(country, product, start_year, end_year, is_global=False)
-    mock_aggregate_country.assert_called_once()
-    mock_aggregate_global.assert_not_called()
-
-    mock_aggregate_global.reset_mock()
-    mock_aggregate_country.reset_mock()
-
-    country = {"@id": "region-world"}
-    aggregate(country, product, start_year, end_year)
-    mock_aggregate_country.assert_not_called()
-    mock_aggregate_global.assert_called_once()