Test and fix debug mode

tests/test_debug_mode.py

@@ -0,0 +1,274 @@
+import os
+from functools import wraps
+
+import pandas as pd
+
+import splink.comparison_library as cl
+from splink import Linker, SettingsCreator, block_on
+from splink.exploratory import profile_columns
+
+from .decorator import mark_with_dialects_excluding
+
+
+class DebugModeError(Exception): ...
+
+
+df = pd.read_csv("./tests/datasets/fake_1000_from_splink_demos.csv")
+settings = SettingsCreator(
+    link_type="dedupe_only",
+    comparisons=[
+        cl.ExactMatch("first_name"),
+        cl.ExactMatch("surname"),
+        cl.ExactMatch("dob"),
+        cl.ExactMatch("city").configure(term_frequency_adjustments=True),
+    ],
+    blocking_rules_to_generate_predictions=[
+        block_on("dob"),
+        block_on("first_name", "surname"),
+    ],
+    retain_intermediate_calculation_columns=True,
+)
+
+
+def debug_mode_test_wrapper(test_function):
+    @wraps(test_function)
+    def wrapper(*args, **kwargs):
+        errors = {}
+        for debug_mode in (False, True):
+            kwargs["debug_mode"] = debug_mode
+            try:
+                test_function(*args, **kwargs)
+            except Exception as e:
+                errors[debug_mode] = e
+        if errors:
+            # in this case we get an error in normal test execution
+            # raise, but this is not a debug_mode issue
+            if errors.get(False, False):
+                raise errors[False]
+            # this is a debug mode error only!
+            raise DebugModeError(
+                f"Error occurs in debug mode only: {errors[True]}"
+            ) from errors[True]
+
+    return wrapper
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_u_training(test_helpers, dialect, debug_mode=None):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    linker = Linker(df, settings, db_api)
+    db_api.debug_mode = debug_mode
+
+    linker.training.estimate_u_using_random_sampling(max_pairs=6e5)
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_ptrrm_train(test_helpers, dialect, debug_mode=None):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    linker = Linker(df, settings, db_api)
+    db_api.debug_mode = debug_mode
+
+    linker.training.estimate_u_using_random_sampling(max_pairs=6e5)
+    linker.training.estimate_probability_two_random_records_match(
+        [block_on("first_name", "surname")],
+        recall=0.7,
+    )
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_em_training(test_helpers, dialect, debug_mode=None):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    linker = Linker(df, settings, db_api)
+    db_api.debug_mode = debug_mode
+
+    linker.training.estimate_parameters_using_expectation_maximisation(block_on("dob"))
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_combined_training(test_helpers, dialect, debug_mode=None):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    linker = Linker(df, settings, db_api)
+    db_api.debug_mode = debug_mode
+
+    linker.training.estimate_probability_two_random_records_match(
+        [block_on("first_name", "surname")],
+        recall=0.7,
+    )
+    linker.training.estimate_u_using_random_sampling(max_pairs=6e5)
+    linker.training.estimate_parameters_using_expectation_maximisation(block_on("dob"))
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_predict(test_helpers, dialect, debug_mode=None):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    linker = Linker(df, settings, db_api)
+    db_api.debug_mode = debug_mode
+
+    linker.inference.predict(0.8)
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_clustering(test_helpers, dialect, debug_mode=None):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    linker = Linker(df, settings, db_api)
+    db_api.debug_mode = debug_mode
+
+    df_e = linker.inference.predict(0.8)
+    linker.clustering.cluster_pairwise_predictions_at_threshold(df_e, 0.9)
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_match_weights_chart(test_helpers, dialect, debug_mode=None):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    linker = Linker(df, settings, db_api)
+    db_api.debug_mode = debug_mode
+
+    linker.visualisations.match_weights_chart()
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_m_u_chart(test_helpers, dialect, debug_mode=None):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    linker = Linker(df, settings, db_api)
+    db_api.debug_mode = debug_mode
+
+    linker.visualisations.m_u_parameters_chart()
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_tf_adjustments_chart(test_helpers, dialect, debug_mode=None):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    linker = Linker(df, settings, db_api)
+    db_api.debug_mode = debug_mode
+
+    linker.visualisations.tf_adjustment_chart("city")
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_interactive_match_weights_chart(
+    test_helpers, dialect, debug_mode=None
+):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    linker = Linker(df, settings, db_api)
+    db_api.debug_mode = debug_mode
+
+    em_session = linker.training.estimate_parameters_using_expectation_maximisation(
+        block_on("dob")
+    )
+    em_session.match_weights_interactive_history_chart()
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_interactive_m_u_chart(test_helpers, dialect, debug_mode=None):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    linker = Linker(df, settings, db_api)
+    db_api.debug_mode = debug_mode
+
+    em_session = linker.training.estimate_parameters_using_expectation_maximisation(
+        block_on("dob")
+    )
+    em_session.m_u_values_interactive_history_chart()
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_unlinkables_chart(test_helpers, dialect, debug_mode=None):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    linker = Linker(df, settings, db_api)
+    db_api.debug_mode = debug_mode
+
+    linker.evaluation.unlinkables_chart()
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_profile_columns(test_helpers, dialect, debug_mode=None):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    db_api.debug_mode = debug_mode
+
+    profile_columns(df, db_api)
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_comparison_viewer(test_helpers, dialect, tmp_path, debug_mode=None):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    linker = Linker(df, settings, db_api)
+    db_api.debug_mode = debug_mode
+
+    df_e = linker.inference.predict(0.8)
+
+    linker.visualisations.comparison_viewer_dashboard(
+        df_e, os.path.join(tmp_path, "test_cvd.html"), overwrite=True
+    )
+
+
+@mark_with_dialects_excluding()
+@debug_mode_test_wrapper
+def test_debug_mode_cluster_studio(test_helpers, dialect, tmp_path, debug_mode=None):
+    helper = test_helpers[dialect]
+
+    db_api = helper.DatabaseAPI(**helper.db_api_args())
+
+    linker = Linker(df, settings, db_api)
+    db_api.debug_mode = debug_mode
+
+    df_e = linker.inference.predict(0.8)
+    df_c = linker.clustering.cluster_pairwise_predictions_at_threshold(df_e, 0.9)
+
+    linker.visualisations.cluster_studio_dashboard(
+        df_e, df_c, os.path.join(tmp_path, "test_csd.html"), overwrite=True
+    )