Snowflake dialect update for MERGE INTO predicates

benchmarks/bench_002/bench_002_pearson_fix.sql

@@ -13,158 +13,158 @@ raw_effect_sizes AS (
         --  numerator this is the slope of the regression line considering STDDEV_POP(open_rate_su) to be
         --  the "y variable" and uses_small_subject_line to be the "x variable" in terms of the regression line.
         SAFE_DIVIDE(
-             SAFE_MULTIPLY(
+            SAFE_MULTIPLY(
                 CORR(open_rate_su, uses_small_subject_line),
                 STDDEV_POP(open_rate_su)
             ),
-             STDDEV_POP(uses_small_subject_line)
+            STDDEV_POP(uses_small_subject_line)
         ) AS open_uses_small_subject_line,
 
         -- The following is the slope of the regression line. Note that CORR (which is the Pearson's correlation
         --  coefficient is symmetric in its arguments, but since STDDEV_POP(open_rate_su) appears in the
         --  numerator this is the slope of the regression line considering STDDEV_POP(open_rate_su) to be
         --  the "y variable" and uses_personal_subject to be the "x variable" in terms of the regression line.
         SAFE_DIVIDE(
-             SAFE_MULTIPLY(
+            SAFE_MULTIPLY(
                 CORR(open_rate_su, uses_personal_subject),
                 STDDEV_POP(open_rate_su)
             ),
-             STDDEV_POP(uses_personal_subject)
+            STDDEV_POP(uses_personal_subject)
         ) AS open_uses_personal_subject,
 
         -- The following is the slope of the regression line. Note that CORR (which is the Pearson's correlation
         --  coefficient is symmetric in its arguments, but since STDDEV_POP(open_rate_su) appears in the
         --  numerator this is the slope of the regression line considering STDDEV_POP(open_rate_su) to be
         --  the "y variable" and uses_timewarp to be the "x variable" in terms of the regression line.
         SAFE_DIVIDE(
-             SAFE_MULTIPLY(
+            SAFE_MULTIPLY(
                 CORR(open_rate_su, uses_timewarp), STDDEV_POP(open_rate_su)
             ),
-             STDDEV_POP(uses_timewarp)
+            STDDEV_POP(uses_timewarp)
         ) AS open_uses_timewarp,
 
         -- The following is the slope of the regression line. Note that CORR (which is the Pearson's correlation
         --  coefficient is symmetric in its arguments, but since STDDEV_POP(open_rate_su) appears in the
         --  numerator this is the slope of the regression line considering STDDEV_POP(open_rate_su) to be
         --  the "y variable" and uses_small_preview to be the "x variable" in terms of the regression line.
         SAFE_DIVIDE(
-             SAFE_MULTIPLY(
+            SAFE_MULTIPLY(
                 CORR(open_rate_su, uses_small_preview), STDDEV_POP(open_rate_su)
             ),
-             STDDEV_POP(uses_small_preview)
+            STDDEV_POP(uses_small_preview)
         ) AS open_uses_small_preview,
 
         -- The following is the slope of the regression line. Note that CORR (which is the Pearson's correlation
         --  coefficient is symmetric in its arguments, but since STDDEV_POP(open_rate_su) appears in the
         --  numerator this is the slope of the regression line considering STDDEV_POP(open_rate_su) to be
         --  the "y variable" and uses_personal_to to be the "x variable" in terms of the regression line.
         SAFE_DIVIDE(
-             SAFE_MULTIPLY(
+            SAFE_MULTIPLY(
                 CORR(open_rate_su, uses_personal_to), STDDEV_POP(open_rate_su)
             ),
-             STDDEV_POP(uses_personal_to)
+            STDDEV_POP(uses_personal_to)
         ) AS open_uses_personal_to,
 
         -- The following is the slope of the regression line. Note that CORR (which is the Pearson's correlation
         --  coefficient is symmetric in its arguments, but since STDDEV_POP(open_rate_su) appears in the
         --  numerator this is the slope of the regression line considering STDDEV_POP(open_rate_su) to be
         --  the "y variable" and uses_ab_test_subject to be the "x variable" in terms of the regression line.
         SAFE_DIVIDE(
-             SAFE_MULTIPLY(
+            SAFE_MULTIPLY(
                 CORR(open_rate_su, uses_ab_test_subject),
                 STDDEV_POP(open_rate_su)
             ),
-             STDDEV_POP(uses_ab_test_subject)
+            STDDEV_POP(uses_ab_test_subject)
         ) AS open_uses_ab_test_subject,
 
         -- The following is the slope of the regression line. Note that CORR (which is the Pearson's correlation
         --  coefficient is symmetric in its arguments, but since STDDEV_POP(open_rate_su) appears in the
         --  numerator this is the slope of the regression line considering STDDEV_POP(open_rate_su) to be
         --  the "y variable" and uses_ab_test_content to be the "x variable" in terms of the regression line.
         SAFE_DIVIDE(
-             SAFE_MULTIPLY(
+            SAFE_MULTIPLY(
                 CORR(open_rate_su, uses_ab_test_content),
                 STDDEV_POP(open_rate_su)
             ),
-             STDDEV_POP(uses_ab_test_content)
+            STDDEV_POP(uses_ab_test_content)
         ) AS open_uses_ab_test_content,
 
         -- The following is the slope of the regression line. Note that CORR (which is the Pearson's correlation
         --  coefficient is symmetric in its arguments, but since STDDEV_POP(open_rate_su) appears in the
         --  numerator this is the slope of the regression line considering STDDEV_POP(open_rate_su) to be
         --  the "y variable" and uses_preview_text to be the "x variable" in terms of the regression line.
         SAFE_DIVIDE(
-             SAFE_MULTIPLY(
+            SAFE_MULTIPLY(
                 CORR(open_rate_su, uses_preview_text), STDDEV_POP(open_rate_su)
             ),
-             STDDEV_POP(uses_preview_text)
+            STDDEV_POP(uses_preview_text)
         ) AS open_uses_preview_text,
 
         -- The following is the slope of the regression line. Note that CORR (which is the Pearson's correlation
         --  coefficient is symmetric in its arguments, but since STDDEV_POP(open_rate_su) appears in the
         --  numerator this is the slope of the regression line considering STDDEV_POP(open_rate_su) to be
         --  the "y variable" and uses_sto to be the "x variable" in terms of the regression line.
         SAFE_DIVIDE(
-             SAFE_MULTIPLY(
+            SAFE_MULTIPLY(
                 CORR(open_rate_su, uses_sto), STDDEV_POP(open_rate_su)
             ),
-             STDDEV_POP(uses_sto)
+            STDDEV_POP(uses_sto)
         ) AS open_uses_sto,
 
         -- The following is the slope of the regression line. Note that CORR (which is the Pearson's correlation
         --  coefficient is symmetric in its arguments, but since STDDEV_POP(open_rate_su) appears in the
         --  numerator this is the slope of the regression line considering STDDEV_POP(open_rate_su) to be
         --  the "y variable" and uses_freemail_from to be the "x variable" in terms of the regression line.
         SAFE_DIVIDE(
-             SAFE_MULTIPLY(
+            SAFE_MULTIPLY(
                 CORR(open_rate_su, uses_freemail_from), STDDEV_POP(open_rate_su)
             ),
-             STDDEV_POP(uses_freemail_from)
+            STDDEV_POP(uses_freemail_from)
         ) AS open_uses_freemail_from,
 
         -- The following is the slope of the regression line. Note that CORR (which is the Pearson's correlation
         --  coefficient is symmetric in its arguments, but since STDDEV_POP(open_rate_su) appears in the
         --  numerator this is the slope of the regression line considering STDDEV_POP(open_rate_su) to be
         --  the "y variable" and uses_resend_non_openers to be the "x variable" in terms of the regression line.
         SAFE_DIVIDE(
-             SAFE_MULTIPLY(
+            SAFE_MULTIPLY(
                 CORR(open_rate_su, uses_resend_non_openers),
                 STDDEV_POP(open_rate_su)
             ),
-             STDDEV_POP(uses_resend_non_openers)
+            STDDEV_POP(uses_resend_non_openers)
         ) AS open_uses_resend_non_openers,
 
         -- The following is the slope of the regression line. Note that CORR (which is the Pearson's correlation
         --  coefficient is symmetric in its arguments, but since STDDEV_POP(open_rate_su) appears in the
         --  numerator this is the slope of the regression line considering STDDEV_POP(open_rate_su) to be
         --  the "y variable" and uses_promo_code to be the "x variable" in terms of the regression line.
         SAFE_DIVIDE(
-             SAFE_MULTIPLY(
+            SAFE_MULTIPLY(
                 CORR(open_rate_su, uses_promo_code), STDDEV_POP(open_rate_su)
             ),
-             STDDEV_POP(uses_promo_code)
+            STDDEV_POP(uses_promo_code)
         ) AS open_uses_promo_code,
 
         -- The following is the slope of the regression line. Note that CORR (which is the Pearson's correlation
         --  coefficient is symmetric in its arguments, but since STDDEV_POP(open_rate_su) appears in the
         --  numerator this is the slope of the regression line considering STDDEV_POP(open_rate_su) to be
         --  the "y variable" and uses_prex to be the "x variable" in terms of the regression line.
         SAFE_DIVIDE(
-             SAFE_MULTIPLY(
+            SAFE_MULTIPLY(
                 CORR(open_rate_su, uses_prex), STDDEV_POP(open_rate_su)
             ),
-             STDDEV_POP(uses_prex)
+            STDDEV_POP(uses_prex)
         ) AS open_uses_prex,
 
         -- The following is the slope of the regression line. Note that CORR (which is the Pearson's correlation
         --  coefficient is symmetric in its arguments, but since STDDEV_POP(open_rate_su) appears in the
         --  numerator this is the slope of the regression line considering STDDEV_POP(open_rate_su) to be
         --  the "y variable" and uses_ab_test_from to be the "x variable" in terms of the regression line.
         SAFE_DIVIDE(
-             SAFE_MULTIPLY(
+            SAFE_MULTIPLY(
                 CORR(open_rate_su, uses_ab_test_from), STDDEV_POP(open_rate_su)
             ),
-             STDDEV_POP(uses_ab_test_from)
+            STDDEV_POP(uses_ab_test_from)
         ) AS open_uses_ab_test_from
 
     FROM
@@ -184,20 +184,20 @@ imputed_effect_sizes AS (
         --  action. In these cases, we assume that campaign outcome is uncorrelated with 
         --  the action because we do not have evidence otherwise.
         COALESCE(
-             IF(
+            IF(
                 IS_NAN(open_uses_small_subject_line),
                 0,
                 open_uses_small_subject_line
             ),
-             0
+            0
         ) AS open_uses_small_subject_line,
 
         -- We now impute the value of the effect size to 0 if it was NaN or NULL. This is to
         --  take into account states where all campaigns either did or did not perform an 
         --  action. In these cases, we assume that campaign outcome is uncorrelated with 
         --  the action because we do not have evidence otherwise.
         COALESCE(
-             IF(
+            IF(
                 IS_NAN(open_uses_personal_subject),
                 0,
                 open_uses_personal_subject
@@ -210,33 +210,31 @@ imputed_effect_sizes AS (
         --  action. In these cases, we assume that campaign outcome is uncorrelated with 
         --  the action because we do not have evidence otherwise.
         COALESCE(
-             IF(IS_NAN(open_uses_timewarp), 0, open_uses_timewarp), 0
+            IF(IS_NAN(open_uses_timewarp), 0, open_uses_timewarp), 0
         ) AS open_uses_timewarp,
 
         -- We now impute the value of the effect size to 0 if it was NaN or NULL. This is to
         --  take into account states where all campaigns either did or did not perform an 
         --  action. In these cases, we assume that campaign outcome is uncorrelated with 
         --  the action because we do not have evidence otherwise.
         COALESCE(
-             IF(IS_NAN(open_uses_small_preview), 0, open_uses_small_preview), 0
+            IF(IS_NAN(open_uses_small_preview), 0, open_uses_small_preview), 0
         ) AS open_uses_small_preview,
 
         -- We now impute the value of the effect size to 0 if it was NaN or NULL. This is to
         --  take into account states where all campaigns either did or did not perform an 
         --  action. In these cases, we assume that campaign outcome is uncorrelated with 
         --  the action because we do not have evidence otherwise.
         COALESCE(
-             IF(IS_NAN(open_uses_personal_to), 0, open_uses_personal_to), 0
+            IF(IS_NAN(open_uses_personal_to), 0, open_uses_personal_to), 0
         ) AS open_uses_personal_to,
 
         -- We now impute the value of the effect size to 0 if it was NaN or NULL. This is to
         --  take into account states where all campaigns either did or did not perform an 
         --  action. In these cases, we assume that campaign outcome is uncorrelated with 
         --  the action because we do not have evidence otherwise.
         COALESCE(
-             IF(
-                IS_NAN(open_uses_ab_test_subject), 0, open_uses_ab_test_subject
-            ),
+            IF(IS_NAN(open_uses_ab_test_subject), 0, open_uses_ab_test_subject),
             0
         ) AS open_uses_ab_test_subject,
 
@@ -245,9 +243,7 @@ imputed_effect_sizes AS (
         --  action. In these cases, we assume that campaign outcome is uncorrelated with 
         --  the action because we do not have evidence otherwise.
         COALESCE(
-             IF(
-                IS_NAN(open_uses_ab_test_content), 0, open_uses_ab_test_content
-            ),
+            IF(IS_NAN(open_uses_ab_test_content), 0, open_uses_ab_test_content),
             0
         ) AS open_uses_ab_test_content,
 
@@ -256,7 +252,7 @@ imputed_effect_sizes AS (
         --  action. In these cases, we assume that campaign outcome is uncorrelated with 
         --  the action because we do not have evidence otherwise.
         COALESCE(
-             IF(IS_NAN(open_uses_preview_text), 0, open_uses_preview_text), 0
+            IF(IS_NAN(open_uses_preview_text), 0, open_uses_preview_text), 0
         ) AS open_uses_preview_text,
 
         -- We now impute the value of the effect size to 0 if it was NaN or NULL. This is to
@@ -272,44 +268,44 @@ imputed_effect_sizes AS (
         --  action. In these cases, we assume that campaign outcome is uncorrelated with 
         --  the action because we do not have evidence otherwise.
         COALESCE(
-             IF(IS_NAN(open_uses_freemail_from), 0, open_uses_freemail_from), 0
+            IF(IS_NAN(open_uses_freemail_from), 0, open_uses_freemail_from), 0
         ) AS open_uses_freemail_from,
 
         -- We now impute the value of the effect size to 0 if it was NaN or NULL. This is to
         --  take into account states where all campaigns either did or did not perform an 
         --  action. In these cases, we assume that campaign outcome is uncorrelated with 
         --  the action because we do not have evidence otherwise.
         COALESCE(
-             IF(
+            IF(
                 IS_NAN(open_uses_resend_non_openers),
                 0,
                 open_uses_resend_non_openers
             ),
-             0
+            0
         ) AS open_uses_resend_non_openers,
 
         -- We now impute the value of the effect size to 0 if it was NaN or NULL. This is to
         --  take into account states where all campaigns either did or did not perform an 
         --  action. In these cases, we assume that campaign outcome is uncorrelated with 
         --  the action because we do not have evidence otherwise.
         COALESCE(
-             IF(IS_NAN(open_uses_promo_code), 0, open_uses_promo_code), 0
+            IF(IS_NAN(open_uses_promo_code), 0, open_uses_promo_code), 0
         ) AS open_uses_promo_code,
 
         -- We now impute the value of the effect size to 0 if it was NaN or NULL. This is to
         --  take into account states where all campaigns either did or did not perform an 
         --  action. In these cases, we assume that campaign outcome is uncorrelated with 
         --  the action because we do not have evidence otherwise.
         COALESCE(
-             IF(IS_NAN(open_uses_prex), 0, open_uses_prex), 0
+            IF(IS_NAN(open_uses_prex), 0, open_uses_prex), 0
         ) AS open_uses_prex,
 
         -- We now impute the value of the effect size to 0 if it was NaN or NULL. This is to
         --  take into account states where all campaigns either did or did not perform an 
         --  action. In these cases, we assume that campaign outcome is uncorrelated with 
         --  the action because we do not have evidence otherwise.
         COALESCE(
-             IF(IS_NAN(open_uses_ab_test_from), 0, open_uses_ab_test_from), 0
+            IF(IS_NAN(open_uses_ab_test_from), 0, open_uses_ab_test_from), 0
         ) AS open_uses_ab_test_from
 
     FROM

src/sqlfluff/dialects/dialect_snowflake.py

@@ -3251,13 +3251,15 @@ class MergeNotMatchedClauseSegment(BaseSegment):
             "WHEN",
             "NOT",
             "MATCHED",
+            Sequence("AND", Ref("ExpressionSegment"), optional=True),
             "THEN",
         ),
     )
     parse_grammar = Sequence(
         "WHEN",
         "NOT",
         "MATCHED",
+        Sequence("AND", Ref("ExpressionSegment"), optional=True),
         "THEN",
         Ref("MergeInsertClauseSegment"),
     )

test/fixtures/dialects/snowflake/snowflake_merge_into.sql

@@ -6,4 +6,7 @@ merge into target_table using source_table
         update set target_table.description = source_table.description;
 
 merge into t1 using t2 on t1.t1key = t2.t2key
-    when matched and t2.marked = 1 then delete  
+    when matched and t2.marked = 1 then delete;
+
+merge into t1 using t2 on t1.t1key = t2.t2key
+    when not matched and t2.marked = 1 then insert (marked) values (1)