added test file output to .gitignore

.gitignore

@@ -143,3 +143,6 @@ dmypy.json
 
 # gitignore exception for docs files (otherwise docs/build does not get saved in repo
 !docs/*
+
+# some files created by certain tests need to not be uploaded to the repo
+saved_model.json

splink/term_frequencies.py

@@ -3,6 +3,7 @@
 
 import logging
 import math
+import warnings
 
 try:
     from pyspark.sql.dataframe import DataFrame
@@ -54,7 +55,7 @@ def sql_gen_generate_adjusted_lambda(column_name, params, table_name="df_e"):
     m = params.params["π"][f"gamma_{column_name}"]["prob_dist_match"][f"level_{max_level}"]["probability"]
     u = params.params["π"][f"gamma_{column_name}"]["prob_dist_non_match"][f"level_{max_level}"]["probability"]
 
-    # ensure average adj calculation doesnt divide by zero
+# ensure average adj calculation doesnt divide by zero (see issue 118)
     if ( math.isclose((m+u), 0.0, rel_tol=1e-9, abs_tol=0.0)):
         average_adjustment = 0.5
         warnings.warn( f" Is most of column {column_name} or all of it comprised of NULL values??? There are levels where no comparisons are found.")

tests/test_adj.py

@@ -3,6 +3,7 @@
 import pandas as pd
 import pyspark.sql.functions as f
 import pyspark
+import warnings
 
 
 @pytest.fixture(scope="module")
@@ -109,6 +110,7 @@ def test_freq_adj_divzero(spark, sparkdf):
     # create column weird in a way that could trigger a div by zero on the average adj calculation before the fix
     sparkdf = sparkdf.withColumn("weird",f.lit(None))
 
+
     try:
         linker = Splink(settings, spark, df=sparkdf)
         notpassing = False