Use stacked ambiguous date checking

Previously, the ambiguous date checks were 1:1 with the generated date
columns. This should not be the case since month needs to check for
ambiguous year, and week needs to check for anything that is ambiguous.

Separate the ambiguous date checking from the column generation, and
update the conditions for the former to be "stacking" (i.e. year is
always checked, month is checked for month/week, and day is checked for
week only).

augur/filter.py

@@ -1049,8 +1049,10 @@ def get_groups_for_subsampling(strains, metadata, group_by=None):
             # Extend metadata with generated date columns
             # Drop the 'date' column since it should not be used for grouping.
             metadata = pd.concat([metadata.drop('date', axis=1), df_dates], axis=1)
-            if 'year' in generated_columns_requested:
-                # Skip ambiguous years.
+
+            # Skip ambiguous dates.
+            if True:
+                # Skip ambiguous years (always, since generated columns are requested).
                 df_skip = metadata[metadata[f'{temp_prefix}year'].isnull()]
                 metadata.drop(df_skip.index, inplace=True)
                 for strain in df_skip.index:
@@ -1059,11 +1061,7 @@ def get_groups_for_subsampling(strains, metadata, group_by=None):
                         "filter": "skip_group_by_with_ambiguous_year",
                         "kwargs": "",
                     })
-
-                # Make a generated 'year' column available for grouping.
-                metadata['year'] = metadata[f'{temp_prefix}year']
-
-            if 'month' in generated_columns_requested:
+            if 'month' in generated_columns_requested or 'week' in generated_columns_requested:
                 # Skip ambiguous months.
                 df_skip = metadata[metadata[f'{temp_prefix}month'].isnull()]
                 metadata.drop(df_skip.index, inplace=True)
@@ -1073,13 +1071,6 @@ def get_groups_for_subsampling(strains, metadata, group_by=None):
                         "filter": "skip_group_by_with_ambiguous_month",
                         "kwargs": "",
                     })
-
-                # Make a generated 'month' column available for grouping.
-                metadata['month'] = list(zip(
-                    metadata[f'{temp_prefix}year'],
-                    metadata[f'{temp_prefix}month']
-                ))
-
             if 'week' in generated_columns_requested:
                 # Skip ambiguous days.
                 df_skip = metadata[metadata[f'{temp_prefix}day'].isnull()]
@@ -1091,7 +1082,15 @@ def get_groups_for_subsampling(strains, metadata, group_by=None):
                         "kwargs": "",
                     })
 
-                # Make a generated 'week' column available for grouping.
+            # Generate columns.
+            if 'year' in generated_columns_requested:
+                metadata['year'] = metadata[f'{temp_prefix}year']
+            if 'month' in generated_columns_requested:
+                metadata['month'] = list(zip(
+                    metadata[f'{temp_prefix}year'],
+                    metadata[f'{temp_prefix}month']
+                ))
+            if 'week' in generated_columns_requested:
                 # Note that week = (year, week) from the date.isocalendar().
                 # Do not combine the raw year with the ISO week number alone,
                 # since raw year ≠ ISO year.

tests/functional/filter/cram/subsample-skip-ambiguous-dates.t

@@ -38,5 +38,9 @@ Group by 'week'. Check the number of strains that have been dropped due to ambig
   >  --subsample-seed 0 \
   >  --output-strains "$TMP/filtered_strains.txt" \
   >  --output-log "$TMP/filtered_log.tsv" > /dev/null
+  $ grep "skip_group_by_with_ambiguous_year" "$TMP/filtered_log.tsv" | wc -l
+  \s*1 (re)
+  $ grep "skip_group_by_with_ambiguous_month" "$TMP/filtered_log.tsv" | wc -l
+  \s*1 (re)
   $ grep "skip_group_by_with_ambiguous_day" "$TMP/filtered_log.tsv" | wc -l
-  \s*5 (re)
+  \s*3 (re)