testing updates

school escort seems to interact poorly with pipeline recoding

Author: jpn--

test/ext-configs/settings.yaml

@@ -42,4 +42,4 @@ output_tables:
     - vehicles
     - proto_disaggregate_accessibility
 
-recode_pipeline_columns: False
+recode_pipeline_columns: True

test/ext-configs_chunkless/network_los.yaml

@@ -0,0 +1,6 @@
+inherit_settings: True
+
+# read cached skims (using numpy memmap) from output directory (memmap is faster than omx )
+read_skim_cache: False
+# write memmapped cached skims to output directory after reading from omx, for use in subsequent runs
+write_skim_cache: False

test/ext-configs_chunkless/settings.yaml

@@ -0,0 +1,31 @@
+inherit_settings: True
+
+# treat warnings as errors
+strict: True
+
+# number of households to simulate
+households_sample_size:  10
+chunk_size: 0
+chunk_training_mode: disabled
+
+# - shadow pricing global switches
+use_shadow_pricing: False
+
+# turn writing of sample_tables on and off for all models
+# (if True, tables will be written if DEST_CHOICE_SAMPLE_TABLE_NAME is specified in individual model settings)
+want_dest_choice_sample_tables: False
+
+cleanup_pipeline_after_run: True
+
+output_tables:
+  h5_store: False
+  action: include
+  prefix: final_
+  sort: True
+  tables:
+    - tablename: trips
+      decode_columns:
+        origin: land_use.zone_id
+        destination: land_use.zone_id
+
+recode_pipeline_columns: True

test/ext-configs_mp/settings.yaml

@@ -29,7 +29,7 @@ trace_hh_id:
 # global switch to turn on or off presampling of destination alternatives at TAZ level (multizone models only)
 want_dest_choice_presampling: True
 
-recode_pipeline_columns: False
+recode_pipeline_columns: True
 
 output_tables:
   h5_store: False

test/test_mtc.py

@@ -25,12 +25,12 @@ def run_test_mtc(
     multiprocess=False, chunkless=False, recode=False, sharrow=False, extended=False
 ):
 
-    def regress(ext):
+    def regress(ext, out_dir):
         if ext:
             regress_trips_df = pd.read_csv(_test_path("regress/final_trips-ext.csv"))
         else:
             regress_trips_df = pd.read_csv(_test_path("regress/final_trips.csv"))
-        final_trips_df = pd.read_csv(_test_path("output/final_trips.csv"))
+        final_trips_df = pd.read_csv(_test_path(out_dir.joinpath("final_trips.csv")))
 
         # column order may not match, so fix it before checking
         assert sorted(regress_trips_df.columns) == sorted(final_trips_df.columns)
@@ -65,12 +65,20 @@ def regress(ext):
                 ]
             )
     elif chunkless:
-        run_args.extend(
-            [
-                "-c",
-                _test_path("configs_chunkless"),
-            ]
-        )
+        if extended:
+            run_args.extend(
+                [
+                    "-c",
+                    _test_path("ext-configs_chunkless"),
+                ]
+            )
+        else:
+            run_args.extend(
+                [
+                    "-c",
+                    _test_path("configs_chunkless"),
+                ]
+            )
     elif recode:
         run_args.extend(
             [
@@ -136,7 +144,7 @@ def regress(ext):
     else:
         subprocess.run([sys.executable, file_path] + run_args, check=True)
 
-    regress(extended)
+    regress(extended, Path(out_dir))
 
 
 def test_mtc():
@@ -171,10 +179,6 @@ def test_mtc_mp_ext():
     run_test_mtc(multiprocess=True, extended=True)
 
 
-def test_mtc_recode_ext():
-    run_test_mtc(recode=True, extended=True)
-
-
 def test_mtc_sharrow_ext():
     run_test_mtc(sharrow=True, extended=True)
 
@@ -244,6 +248,8 @@ def test_mtc_extended_progressive():
         output_dir=out_dir,
     )
     state.filesystem.persist_sharrow_cache()
+    state.logging.config_logger()
+    state.settings.trace_hh_id = 1196298
 
     assert state.settings.models == EXPECTED_MODELS
     assert state.settings.chunk_size == 0
@@ -263,6 +269,50 @@ def test_mtc_extended_progressive():
             print(f"> prototype_mtc_extended {step_name}: ok")
 
 
+@testing.run_if_exists("reference-pipeline-extended.zip")
+def test_mtc_extended_progressive_chunkless():
+
+    import activitysim.abm  # register components # noqa: F401
+
+    out_dir = _test_path("output-progressive-2")
+    Path(out_dir).mkdir(exist_ok=True)
+    Path(out_dir).joinpath(".gitignore").write_text("**\n")
+
+    state = workflow.State.make_default(
+        configs_dir=(
+            _test_path("configs_chunkless"),
+            _test_path("ext-configs"),
+            _example_path("ext-configs"),
+            _test_path("configs"),
+            _example_path("configs"),
+        ),
+        data_dir=_example_path("data"),
+        data_model_dir=_example_path("data_model"),
+        output_dir=out_dir,
+    )
+    state.filesystem.persist_sharrow_cache()
+    state.logging.config_logger()
+    state.settings.trace_hh_id = 1196298
+
+    assert state.settings.models == EXPECTED_MODELS
+    assert state.settings.chunk_size == 0
+    assert not state.settings.sharrow
+
+    for step_name in EXPECTED_MODELS:
+        state.run.by_name(step_name)
+        try:
+            pass
+            # state.checkpoint.check_against(
+            #     Path(__file__).parent.joinpath("reference-pipeline-extended.zip"),
+            #     checkpoint_name=step_name,
+            # )
+        except Exception:
+            print(f"> prototype_mtc_extended {step_name}: ERROR")
+            raise
+        else:
+            print(f"> prototype_mtc_extended {step_name}: ok")
+
+
 if __name__ == "__main__":
     run_test_mtc(multiprocess=False)
     run_test_mtc(multiprocess=True)