[SPARK-50055][SQL] Add TryMakeInterval alternative

### What changes were proposed in this pull request?
Addition of two new expression try_make_interval.

### Why are the changes needed?
This is a split PR from apache#48499 so that we divide the reasonings for PRs.

### Does this PR introduce _any_ user-facing change?
Yes, new expressions added.

### How was this patch tested?
Tests added.

### Was this patch authored or co-authored using generative AI tooling?
No.

Closes apache#48580 from mihailom-db/addTryConv-TryMakeInterval.

Authored-by: Mihailo Milosevic <mihailo.milosevic@databricks.com>
Signed-off-by: Max Gekk <max.gekk@gmail.com>

connector/connect/client/jvm/src/test/scala/org/apache/spark/sql/PlanGenerationTestSuite.scala

@@ -1925,6 +1925,47 @@ class PlanGenerationTestSuite
     fn.make_interval()
   }
 
+  functionTest("try_make_interval years months weeks days hours mins secs") {
+    fn.try_make_interval(
+      fn.col("a"),
+      fn.col("a"),
+      fn.col("a"),
+      fn.col("a"),
+      fn.col("a"),
+      fn.col("a"),
+      fn.col("b"))
+  }
+
+  functionTest("try_make_interval years months weeks days hours mins") {
+    fn.try_make_interval(
+      fn.col("a"),
+      fn.col("a"),
+      fn.col("a"),
+      fn.col("a"),
+      fn.col("a"),
+      fn.col("a"))
+  }
+
+  functionTest("try_make_interval years months weeks days hours") {
+    fn.try_make_interval(fn.col("a"), fn.col("a"), fn.col("a"), fn.col("a"), fn.col("a"))
+  }
+
+  functionTest("try_make_interval years months weeks days") {
+    fn.try_make_interval(fn.col("a"), fn.col("a"), fn.col("a"), fn.col("a"))
+  }
+
+  functionTest("try_make_interval years months weeks") {
+    fn.try_make_interval(fn.col("a"), fn.col("a"), fn.col("a"))
+  }
+
+  functionTest("try_make_interval years months") {
+    fn.try_make_interval(fn.col("a"), fn.col("a"))
+  }
+
+  functionTest("try_make_interval years") {
+    fn.try_make_interval(fn.col("a"))
+  }
+
   functionTest("make_timestamp with timezone") {
     fn.make_timestamp(
       fn.col("a"),

docs/sql-ref-ansi-compliance.md

@@ -383,7 +383,7 @@ When ANSI mode is on, it throws exceptions for invalid operations. You can use t
   - `try_make_timestamp`: identical to the function `make_timestamp`, except that it returns `NULL` result instead of throwing an exception on error.
   - `try_make_timestamp_ltz`: identical to the function `make_timestamp_ltz`, except that it returns `NULL` result instead of throwing an exception on error.
   - `try_make_timestamp_ntz`: identical to the function `make_timestamp_ntz`, except that it returns `NULL` result instead of throwing an exception on error.
-
+  - `try_make_interval`: identical to the function `make_interval`, except that it returns `NULL` result instead of throwing an exception on invalid interval.
 
 ### SQL Keywords (optional, disabled by default)
 

python/docs/source/reference/pyspark.sql/functions.rst

@@ -301,6 +301,7 @@ Date and Timestamp Functions
     to_unix_timestamp
     to_utc_timestamp
     trunc
+    try_make_interval
     try_make_timestamp
     try_make_timestamp_ltz
     try_make_timestamp_ntz

python/pyspark/sql/connect/functions/builtin.py

@@ -3714,6 +3714,31 @@ def make_dt_interval(
 make_dt_interval.__doc__ = pysparkfuncs.make_dt_interval.__doc__
 
 
+def try_make_interval(
+    years: Optional["ColumnOrName"] = None,
+    months: Optional["ColumnOrName"] = None,
+    weeks: Optional["ColumnOrName"] = None,
+    days: Optional["ColumnOrName"] = None,
+    hours: Optional["ColumnOrName"] = None,
+    mins: Optional["ColumnOrName"] = None,
+    secs: Optional["ColumnOrName"] = None,
+) -> Column:
+    _years = lit(0) if years is None else _to_col(years)
+    _months = lit(0) if months is None else _to_col(months)
+    _weeks = lit(0) if weeks is None else _to_col(weeks)
+    _days = lit(0) if days is None else _to_col(days)
+    _hours = lit(0) if hours is None else _to_col(hours)
+    _mins = lit(0) if mins is None else _to_col(mins)
+    _secs = lit(decimal.Decimal(0)) if secs is None else _to_col(secs)
+
+    return _invoke_function_over_columns(
+        "try_make_interval", _years, _months, _weeks, _days, _hours, _mins, _secs
+    )
+
+
+try_make_interval.__doc__ = pysparkfuncs.try_make_interval.__doc__
+
+
 def make_interval(
     years: Optional["ColumnOrName"] = None,
     months: Optional["ColumnOrName"] = None,

python/pyspark/sql/functions/builtin.py

@@ -21850,6 +21850,162 @@ def make_dt_interval(
     return _invoke_function_over_columns("make_dt_interval", _days, _hours, _mins, _secs)
 
 
+@_try_remote_functions
+def try_make_interval(
+    years: Optional["ColumnOrName"] = None,
+    months: Optional["ColumnOrName"] = None,
+    weeks: Optional["ColumnOrName"] = None,
+    days: Optional["ColumnOrName"] = None,
+    hours: Optional["ColumnOrName"] = None,
+    mins: Optional["ColumnOrName"] = None,
+    secs: Optional["ColumnOrName"] = None,
+) -> Column:
+    """
+    This is a special version of `make_interval` that performs the same operation, but returns a
+    NULL value instead of raising an error if interval cannot be created.
+
+    .. versionadded:: 4.0.0
+
+    Parameters
+    ----------
+    years : :class:`~pyspark.sql.Column` or str, optional
+        The number of years, positive or negative.
+    months : :class:`~pyspark.sql.Column` or str, optional
+        The number of months, positive or negative.
+    weeks : :class:`~pyspark.sql.Column` or str, optional
+        The number of weeks, positive or negative.
+    days : :class:`~pyspark.sql.Column` or str, optional
+        The number of days, positive or negative.
+    hours : :class:`~pyspark.sql.Column` or str, optional
+        The number of hours, positive or negative.
+    mins : :class:`~pyspark.sql.Column` or str, optional
+        The number of minutes, positive or negative.
+    secs : :class:`~pyspark.sql.Column` or str, optional
+        The number of seconds with the fractional part in microsecond precision.
+
+    Returns
+    -------
+    :class:`~pyspark.sql.Column`
+        A new column that contains an interval.
+
+    Examples
+    --------
+
+    Example 1: Try make interval from years, months, weeks, days, hours, mins and secs.
+
+    >>> import pyspark.sql.functions as sf
+    >>> df = spark.createDataFrame([[100, 11, 1, 1, 12, 30, 01.001001]],
+    ...     ["year", "month", "week", "day", "hour", "min", "sec"])
+    >>> df.select(sf.try_make_interval(
+    ...     df.year, df.month, df.week, df.day, df.hour, df.min, df.sec)
+    ... ).show(truncate=False)
+    +---------------------------------------------------------------+
+    |try_make_interval(year, month, week, day, hour, min, sec)      |
+    +---------------------------------------------------------------+
+    |100 years 11 months 8 days 12 hours 30 minutes 1.001001 seconds|
+    +---------------------------------------------------------------+
+
+    Example 2: Try make interval from years, months, weeks, days, hours and mins.
+
+    >>> import pyspark.sql.functions as sf
+    >>> df = spark.createDataFrame([[100, 11, 1, 1, 12, 30, 01.001001]],
+    ...     ["year", "month", "week", "day", "hour", "min", "sec"])
+    >>> df.select(sf.try_make_interval(
+    ...     df.year, df.month, df.week, df.day, df.hour, df.min)
+    ... ).show(truncate=False)
+    +-------------------------------------------------------+
+    |try_make_interval(year, month, week, day, hour, min, 0)|
+    +-------------------------------------------------------+
+    |100 years 11 months 8 days 12 hours 30 minutes         |
+    +-------------------------------------------------------+
+
+    Example 3: Try make interval from years, months, weeks, days and hours.
+
+    >>> import pyspark.sql.functions as sf
+    >>> df = spark.createDataFrame([[100, 11, 1, 1, 12, 30, 01.001001]],
+    ...     ["year", "month", "week", "day", "hour", "min", "sec"])
+    >>> df.select(sf.try_make_interval(
+    ...     df.year, df.month, df.week, df.day, df.hour)
+    ... ).show(truncate=False)
+    +-----------------------------------------------------+
+    |try_make_interval(year, month, week, day, hour, 0, 0)|
+    +-----------------------------------------------------+
+    |100 years 11 months 8 days 12 hours                  |
+    +-----------------------------------------------------+
+
+    Example 4: Try make interval from years, months, weeks and days.
+
+    >>> import pyspark.sql.functions as sf
+    >>> df = spark.createDataFrame([[100, 11, 1, 1, 12, 30, 01.001001]],
+    ...     ["year", "month", "week", "day", "hour", "min", "sec"])
+    >>> df.select(sf.try_make_interval(df.year, df.month, df.week, df.day)).show(truncate=False)
+    +--------------------------------------------------+
+    |try_make_interval(year, month, week, day, 0, 0, 0)|
+    +--------------------------------------------------+
+    |100 years 11 months 8 days                        |
+    +--------------------------------------------------+
+
+    Example 5: Try make interval from years, months and weeks.
+
+    >>> import pyspark.sql.functions as sf
+    >>> df = spark.createDataFrame([[100, 11, 1, 1, 12, 30, 01.001001]],
+    ...     ["year", "month", "week", "day", "hour", "min", "sec"])
+    >>> df.select(sf.try_make_interval(df.year, df.month, df.week)).show(truncate=False)
+    +------------------------------------------------+
+    |try_make_interval(year, month, week, 0, 0, 0, 0)|
+    +------------------------------------------------+
+    |100 years 11 months 7 days                      |
+    +------------------------------------------------+
+
+    Example 6: Try make interval from years and months.
+
+    >>> import pyspark.sql.functions as sf
+    >>> df = spark.createDataFrame([[100, 11, 1, 1, 12, 30, 01.001001]],
+    ...     ["year", "month", "week", "day", "hour", "min", "sec"])
+    >>> df.select(sf.try_make_interval(df.year, df.month)).show(truncate=False)
+    +---------------------------------------------+
+    |try_make_interval(year, month, 0, 0, 0, 0, 0)|
+    +---------------------------------------------+
+    |100 years 11 months                          |
+    +---------------------------------------------+
+
+    Example 7: Try make interval from years.
+
+    >>> import pyspark.sql.functions as sf
+    >>> df = spark.createDataFrame([[100, 11, 1, 1, 12, 30, 01.001001]],
+    ...     ["year", "month", "week", "day", "hour", "min", "sec"])
+    >>> df.select(sf.try_make_interval(df.year)).show(truncate=False)
+    +-----------------------------------------+
+    |try_make_interval(year, 0, 0, 0, 0, 0, 0)|
+    +-----------------------------------------+
+    |100 years                                |
+    +-----------------------------------------+
+
+    Example 8: Try make interval from years with overflow.
+
+    >>> import pyspark.sql.functions as sf
+    >>> df = spark.createDataFrame([[2147483647, 11, 1, 1, 12, 30, 01.001001]],
+    ...     ["year", "month", "week", "day", "hour", "min", "sec"])
+    >>> df.select(sf.try_make_interval(df.year)).show(truncate=False)
+    +-----------------------------------------+
+    |try_make_interval(year, 0, 0, 0, 0, 0, 0)|
+    +-----------------------------------------+
+    |NULL                                     |
+    +-----------------------------------------+
+
+    """
+    _years = lit(0) if years is None else years
+    _months = lit(0) if months is None else months
+    _weeks = lit(0) if weeks is None else weeks
+    _days = lit(0) if days is None else days
+    _hours = lit(0) if hours is None else hours
+    _mins = lit(0) if mins is None else mins
+    _secs = lit(decimal.Decimal(0)) if secs is None else secs
+    return _invoke_function_over_columns(
+        "try_make_interval", _years, _months, _weeks, _days, _hours, _mins, _secs
+    )
+
+
 @_try_remote_functions
 def make_interval(
     years: Optional["ColumnOrName"] = None,

python/pyspark/sql/tests/test_functions.py

@@ -452,6 +452,11 @@ def test_collation(self):
         actual = df.select(F.collation(F.collate("name", "UNICODE"))).distinct().collect()
         self.assertEqual([Row("UNICODE")], actual)
 
+    def test_try_make_interval(self):
+        df = self.spark.createDataFrame([(2147483647,)], ["num"])
+        actual = df.select(F.isnull(F.try_make_interval("num"))).collect()
+        self.assertEqual([Row(True)], actual)
+
     def test_octet_length_function(self):
         # SPARK-36751: add octet length api for python
         df = self.spark.createDataFrame([("cat",), ("\U0001F408",)], ["cat"])