Corrected the issue.

justin13601 · May 13, 2024 · 4837e49 · 4837e49
1 parent 68f7d51
commit 4837e49
Showing 1 changed file with 13 additions and 10 deletions.
diff --git a/src/esgpt_task_querying/aggregate.py b/src/esgpt_task_querying/aggregate.py
@@ -935,8 +935,6 @@ def boolean_expr_bound_sum(
         │ 2          ┆ 1989-12-08 16:22:00 ┆ 1989-12-04 13:14:00 ┆ 1989-12-05 16:22:00 ┆ 0    ┆ 0    ┆ 0    │
         │ 2          ┆ 1989-12-10 03:07:00 ┆ 1989-12-04 13:14:00 ┆ 1989-12-07 03:07:00 ┆ 1    ┆ 1    ┆ 1    │
         └────────────┴─────────────────────┴─────────────────────┴─────────────────────┴──────┴──────┴──────┘
-        >>> print(df)
-        TODO: HERE AND BELOW!!!
         >>> boolean_expr_bound_sum(
         ...     df,
         ...     pl.col("idx").is_in([1, 4, 7]),
@@ -951,14 +949,16 @@ def boolean_expr_bound_sum(
         │ i64        ┆ datetime[μs]        ┆ datetime[μs]        ┆ datetime[μs]        ┆ u16  ┆ u16  ┆ u16  │
         ╞════════════╪═════════════════════╪═════════════════════╪═════════════════════╪══════╪══════╪══════╡
         │ 1          ┆ 1989-12-01 12:03:00 ┆ null                ┆ null                ┆ 0    ┆ 0    ┆ 0    │
-        │ 1          ┆ 1989-12-03 13:14:00 ┆ 1989-12-03 13:14:00 ┆ 1989-12-03 13:14:00 ┆ 0    ┆ 1    ┆ 1    │
-        │ 1          ┆ 1989-12-05 15:17:00 ┆ 1989-12-03 13:14:00 ┆ 1989-12-05 15:17:00 ┆ 1    ┆ 1    ┆ 1    │
-        │ 2          ┆ 1989-12-02 12:03:00 ┆ null                ┆ null                ┆ 0    ┆ 0    ┆ 0    │
-        │ 2          ┆ 1989-12-04 13:14:00 ┆ 1989-12-04 13:14:00 ┆ 1989-12-04 13:14:00 ┆ 1    ┆ 0    ┆ 0    │
-        │ 2          ┆ 1989-12-06 15:17:00 ┆ 1989-12-04 13:14:00 ┆ 1989-12-06 15:17:00 ┆ 2    ┆ 1    ┆ 1    │
-        │ 2          ┆ 1989-12-08 16:22:00 ┆ 1989-12-04 13:14:00 ┆ 1989-12-08 16:22:00 ┆ 2    ┆ 2    ┆ 1    │
-        │ 2          ┆ 1989-12-10 03:07:00 ┆ 1989-12-10 03:07:00 ┆ 1989-12-10 03:07:00 ┆ 0    ┆ 1    ┆ 1    │
+        │ 1          ┆ 1989-12-03 13:14:00 ┆ null                ┆ null                ┆ 0    ┆ 0    ┆ 0    │
+        │ 1          ┆ 1989-12-05 15:17:00 ┆ null                ┆ null                ┆ 0    ┆ 0    ┆ 0    │
+        │ 2          ┆ 1989-12-02 12:03:00 ┆ 1989-12-05 12:03:00 ┆ 1989-12-10 03:07:00 ┆ 1    ┆ 3    ┆ 2    │
+        │ 2          ┆ 1989-12-04 13:14:00 ┆ 1989-12-07 13:14:00 ┆ 1989-12-10 03:07:00 ┆ 0    ┆ 2    ┆ 1    │
+        │ 2          ┆ 1989-12-06 15:17:00 ┆ 1989-12-09 15:17:00 ┆ 1989-12-10 03:07:00 ┆ 0    ┆ 1    ┆ 1    │
+        │ 2          ┆ 1989-12-08 16:22:00 ┆ null                ┆ null                ┆ 0    ┆ 0    ┆ 0    │
+        │ 2          ┆ 1989-12-10 03:07:00 ┆ null                ┆ null                ┆ 0    ┆ 0    ┆ 0    │
         └────────────┴─────────────────────┴─────────────────────┴─────────────────────┴──────┴──────┴──────┘
+        >>> print(df)
+        TODO: HERE AND BELOW!!!
         >>> boolean_expr_bound_sum(
         ...     df,
         ...     pl.col("idx").is_in([1, 4, 7]),
@@ -1122,7 +1122,10 @@ def boolean_expr_bound_sum(
             ).alias(c)
             for c in cols
         }
-        if closed in ("left", "both"):
+        if (closed in ("left", "both") and offset <= timedelta(0)) or offset < timedelta(0):
+            # If we either do include the left endpoint due to the closed value and the lack of a positive
+            # offset or we have a negative offset, we want to add in the left endpoint's (the row's) counts
+            # As they will always be included.
             sum_exprs = {c: expr + pl.col(c) for c, expr in sum_exprs.items()}
 
     at_boundary_df = df.filter(boundary_expr).select(