Feature/multiple coruptures (#37)

* added list support for corupture queries;
* update changelog;
* detox;

CHANGELOG.md

@@ -1,5 +1,10 @@
 # Changelog
 
+## [0.8.2] - 2023-07-04
+
+### Changed
+ - added list support for corupture queries
+
 ## [0.8.1] - 2023-07-03
 
 ### Changed

solvis_graphql_api/composite_solution/cached.py

@@ -5,7 +5,7 @@
 import time
 from functools import lru_cache
 from pathlib import Path
-from typing import Callable, Iterable, Iterator, List, Tuple, Union
+from typing import Callable, Iterable, Iterator, List, Set, Tuple, Union
 
 import geopandas as gpd
 import nzshm_model
@@ -93,6 +93,11 @@ def get_fss(slt, fault_system):
     return get_rupture_ids(rupture_set_id=rupture_set_id, locations=location_ids, radius=radius_km, union=union)
 
 
+@lru_cache
+def get_rupture_ids_for_parent_fault(fault_system_solution: InversionSolutionProtocol, fault_name: str) -> Set[int]:
+    return set(fault_system_solution.get_ruptures_for_parent_fault(fault_name))
+
+
 @lru_cache
 def matched_rupture_sections_gdf(
     model_id: str,
@@ -104,17 +109,20 @@ def matched_rupture_sections_gdf(
     min_mag: float,
     max_mag: float,
     union: bool = False,
-    corupture_parent_fault_name: str = "",
+    corupture_fault_names: Union[None, Tuple[str]] = None,
 ) -> gpd.GeoDataFrame:
     """
     Query the solvis.CompositeSolution instance identified by model ID.
 
     return a dataframe of the matched ruptures.
     """
-
+    tic0 = time.perf_counter()
     composite_solution = get_composite_solution(model_id)
 
     fss = composite_solution._solutions[fault_system]
+    tic1 = time.perf_counter()
+    log.debug('matched_rupture_sections_gdf(): time to load fault system solution: %2.3f seconds' % (tic1 - tic0))
+
     df0 = fss.ruptures_with_rates
 
     # attribute filters
@@ -123,21 +131,43 @@ def matched_rupture_sections_gdf(
     df0 = df0 if not max_rate else df0[df0.rate_weighted_mean <= max_rate]
     df0 = df0 if not min_rate else df0[df0.rate_weighted_mean > min_rate]
 
+    tic2 = time.perf_counter()
+    log.debug('matched_rupture_sections_gdf(): time apply attribute filters: %2.3f seconds' % (tic2 - tic1))
+
     # co-rupture filter
-    if corupture_parent_fault_name and len(corupture_parent_fault_name):
-        rupture_ids = list(fss.get_ruptures_for_parent_fault(corupture_parent_fault_name))
-        df0 = df0[df0["Rupture Index"].isin(rupture_ids)]
+    if corupture_fault_names and len(corupture_fault_names):
+        first = True
+        rupture_ids: Set[int]
+        for fault_name in corupture_fault_names:
+            if fault_name not in parent_fault_names(fss):
+                raise ValueError("Invalid fault name: %s" % fault_name)
+            tic22 = time.perf_counter()
+            fault_rupture_ids = get_rupture_ids_for_parent_fault(fss, fault_name)
+            tic23 = time.perf_counter()
+            log.debug('fss.get_ruptures_for_parent_fault %s: %2.3f seconds' % (fault_name, (tic23 - tic22)))
+
+            if first:
+                rupture_ids = fault_rupture_ids
+                first = False
+            else:
+                rupture_ids = rupture_ids.intersection(fault_rupture_ids)
+
+        df0 = df0[df0["Rupture Index"].isin(list(rupture_ids))]
+
+    tic3 = time.perf_counter()
+    log.debug('matched_rupture_sections_gdf(): time apply co-rupture filter: %2.3f seconds' % (tic3 - tic2))
 
     # location filters
     if location_ids is not None and len(location_ids):
         if RESOLVE_LOCATIONS_INTERNALLY:
             df0 = filter_dataframe_by_radius(fss, df0, location_ids, radius_km)
         else:
-            rupture_ids = list(
-                filter_dataframe_by_radius_stored(model_id, fault_system, df0, location_ids, radius_km, union)
-            )
+            rupture_ids = set(filter_dataframe_by_radius_stored(model_id, fault_system, df0, location_ids, radius_km, union))
             df0 = df0[df0["Rupture Index"].isin(rupture_ids)]
 
+    tic4 = time.perf_counter()
+    log.debug('matched_rupture_sections_gdf(): time apply location filters: %2.3f seconds' % (tic4 - tic3))
+
     return df0
 
 
@@ -153,7 +183,7 @@ def fault_section_aggregates_gdf(
     max_mag: float,
     union: bool = False,
     trace_only: bool = False,
-    corupture_parent_fault_name: str = "",
+    corupture_fault_names: Union[None, Tuple[str]] = None,
 ) -> gpd.GeoDataFrame:
 
     tic0 = time.perf_counter()
@@ -173,11 +203,11 @@ def fault_section_aggregates_gdf(
         min_mag,
         max_mag,
         union,
-        corupture_parent_fault_name,
+        corupture_fault_names,
     )
 
     tic2 = time.perf_counter()
-    log.debug('fault_section_aggregates_gdf(): time to filter rutpure sections: %2.3f seconds' % (tic2 - tic1))
+    log.debug('fault_section_aggregates_gdf(): time to filter rupture sections: %2.3f seconds' % (tic2 - tic1))
 
     fsr = fss.fault_sections_with_rates
     fsr = fsr[fsr['Rupture Index'].isin(df0['Rupture Index'].unique())]

solvis_graphql_api/composite_solution/composite_rupture_sections.py

@@ -33,7 +33,7 @@ def get_fault_section_aggregates(filter_args, trace_only=False):
         max_mag=filter_args.maximum_mag,
         union=False,
         trace_only=trace_only,
-        corupture_parent_fault_name=filter_args.corupture_parent_fault_name,
+        corupture_fault_names=tuple(filter_args.corupture_fault_names),
     )
 
 
@@ -132,7 +132,8 @@ def resolve_mfd_histogram(root, info, *args, **kwargs):
             min_mag=filter_args.minimum_mag,
             max_mag=filter_args.maximum_mag,
             union=False,
-            corupture_parent_fault_name=filter_args.corupture_parent_fault_name,
+            corupture_fault_names=tuple(filter_args.corupture_fault_names),
+            # corupture_parent_fault_name=filter_args.corupture_parent_fault_name,
         )
 
         # TODO - move this function into solvis (see solvis.mfd_hist)

solvis_graphql_api/composite_solution/composite_solution.py

@@ -22,9 +22,11 @@ class FilterRupturesArgsBase:
         description="The fault systems [`HIK`, `PUY`, `CRU`]",
     )
 
-    corupture_parent_fault_name = graphene.String(
+    corupture_fault_names = graphene.List(
+        graphene.String,
         required=False,
-        description="The name of parent fault may be supplied to restrict to ruptures that include parent "
+        default_value=[],
+        description="Optional list of parent fault names. Result will only include ruptures that include parent "
         "fault sections",
     )
 

solvis_graphql_api/composite_solution/schema.py

@@ -146,7 +146,7 @@ def paginated_filtered_ruptures(filter_args, sortby_args, **kwargs) -> RuptureDe
         min_mag=filter_args.get('minimum_mag'),
         max_mag=filter_args.get('maximum_mag'),
         union=False,
-        corupture_parent_fault_name=filter_args.corupture_parent_fault_name,
+        corupture_fault_names=tuple(filter_args.corupture_fault_names),
     )
 
     if sortby_args:

tests/test_composite_corupture_sections.py

@@ -11,7 +11,7 @@
   filter_rupture_sections(
       filter:{
       model_id: "NSHM_v1.0.4",
-      corupture_parent_fault_name: "Masterton",
+      corupture_fault_names: ["Masterton"],
       location_ids: [],
       fault_system: "CRU",
       radius_km: 100