feat(spark, bigquery): Add support for UNIX_SECONDS(...)

tobymao · Nov 5, 2024 · e290d11 · e290d11
1 parent 71f4a47
commit e290d11
Show file tree

Hide file tree

Showing 5 changed files with 40 additions and 0 deletions.
diff --git a/sqlglot/dialects/bigquery.py b/sqlglot/dialects/bigquery.py
@@ -704,6 +704,7 @@ class Generator(generator.Generator):
         WITH_PROPERTIES_PREFIX = "OPTIONS"
         SUPPORTS_EXPLODING_PROJECTIONS = False
         EXCEPT_INTERSECT_SUPPORT_ALL_CLAUSE = False
+        SUPPORTS_UNIX_SECONDS = True
 
         TRANSFORMS = {
             **generator.Generator.TRANSFORMS,

diff --git a/sqlglot/dialects/spark.py b/sqlglot/dialects/spark.py
@@ -137,6 +137,7 @@ class Generator(Spark2.Generator):
         PAD_FILL_PATTERN_IS_REQUIRED = False
         SUPPORTS_CONVERT_TIMEZONE = True
         SUPPORTS_MEDIAN = True
+        SUPPORTS_UNIX_SECONDS = True
 
         TYPE_MAPPING = {
             **Spark2.Generator.TYPE_MAPPING,

diff --git a/sqlglot/expressions.py b/sqlglot/expressions.py
@@ -6516,6 +6516,10 @@ class UnixToTimeStr(Func):
     pass
 
 
+class UnixSeconds(Func):
+    pass
+
+
 class Uuid(Func):
     _sql_names = ["UUID", "GEN_RANDOM_UUID", "GENERATE_UUID", "UUID_STRING"]
 

diff --git a/sqlglot/generator.py b/sqlglot/generator.py
@@ -436,6 +436,9 @@ class Generator(metaclass=_Generator):
     # Whether MEDIAN(expr) is supported; if not, it will be generated as PERCENTILE_CONT(expr, 0.5)
     SUPPORTS_MEDIAN = True
 
+    # Whether UNIX_SECONDS(timestamp) is supported
+    SUPPORTS_UNIX_SECONDS = False
+
     # The name to generate for the JSONPath expression. If `None`, only `this` will be generated
     PARSE_JSON_NAME: t.Optional[str] = "PARSE_JSON"
 
@@ -4472,3 +4475,15 @@ def overflowtruncatebehavior_sql(self, expression: exp.OverflowTruncateBehavior)
         filler = f" {filler}" if filler else ""
         with_count = "WITH COUNT" if expression.args.get("with_count") else "WITHOUT COUNT"
         return f"TRUNCATE{filler} {with_count}"
+
+    def unixseconds_sql(self, expression: exp.UnixSeconds) -> str:
+        if self.SUPPORTS_UNIX_SECONDS:
+            return self.function_fallback_sql(expression)
+
+        start_ts = exp.cast(
+            exp.Literal.string("1970-01-01 00:00:00+00"), to=exp.DataType.Type.TIMESTAMPTZ
+        )
+
+        return self.sql(
+            exp.TimestampDiff(this=expression.this, expression=start_ts, unit=exp.var("SECONDS"))
+        )
diff --git a/tests/dialects/test_bigquery.py b/tests/dialects/test_bigquery.py
@@ -2095,3 +2095,22 @@ def test_json_extract_scalar(self):
                     "snowflake": """SELECT JSON_EXTRACT_PATH_TEXT('{"name": "Jakob", "age": "6"}', 'age')""",
                 },
             )
+
+    def test_unix_seconds(self):
+        self.validate_all(
+            "SELECT UNIX_SECONDS('2008-12-25 15:30:00+00')",
+            read={
+                "bigquery": "SELECT UNIX_SECONDS('2008-12-25 15:30:00+00')",
+                "spark": "SELECT UNIX_SECONDS('2008-12-25 15:30:00+00')",
+                "databricks": "SELECT UNIX_SECONDS('2008-12-25 15:30:00+00')",
+            },
+            write={
+                "spark": "SELECT UNIX_SECONDS('2008-12-25 15:30:00+00')",
+                "databricks": "SELECT UNIX_SECONDS('2008-12-25 15:30:00+00')",
+                "duckdb": "SELECT DATE_DIFF('SECONDS', CAST('1970-01-01 00:00:00+00' AS TIMESTAMPTZ), '2008-12-25 15:30:00+00')",
+                "snowflake": "SELECT TIMESTAMPDIFF(SECONDS, CAST('1970-01-01 00:00:00+00' AS TIMESTAMPTZ), '2008-12-25 15:30:00+00')",
+            },
+        )
+
+        for dialect in ("bigquery", "spark", "databricks"):
+            parse_one("UNIX_SECONDS(col)", dialect=dialect).assert_is(exp.UnixSeconds)