fix: typos of sql, sqllogictest and substrait packages (apache#11548)

datafusion/sql/src/expr/function.rs

@@ -67,7 +67,7 @@ pub fn suggest_valid_function(
     find_closest_match(valid_funcs, input_function_name)
 }
 
-/// Find the closest matching string to the target string in the candidates list, using edit distance(case insensitve)
+/// Find the closest matching string to the target string in the candidates list, using edit distance(case insensitive)
 /// Input `candidates` must not be empty otherwise it will panic
 fn find_closest_match(candidates: Vec<String>, target: &str) -> String {
     let target = target.to_lowercase();

datafusion/sql/src/parser.rs

@@ -218,7 +218,7 @@ impl fmt::Display for CreateExternalTable {
 ///
 /// This can either be a [`Statement`] from [`sqlparser`] from a
 /// standard SQL dialect, or a DataFusion extension such as `CREATE
-/// EXTERAL TABLE`. See [`DFParser`] for more information.
+/// EXTERNAL TABLE`. See [`DFParser`] for more information.
 ///
 /// [`Statement`]: sqlparser::ast::Statement
 #[derive(Debug, Clone, PartialEq, Eq)]
@@ -1101,7 +1101,7 @@ mod tests {
         });
         expect_parse_ok(sql, expected)?;
 
-        // positive case: column definiton allowed in 'partition by' clause
+        // positive case: column definition allowed in 'partition by' clause
         let sql =
             "CREATE EXTERNAL TABLE t(c1 int) STORED AS CSV PARTITIONED BY (p1 int) LOCATION 'foo.csv'";
         let expected = Statement::CreateExternalTable(CreateExternalTable {

datafusion/sql/src/relation/mod.rs

@@ -105,7 +105,7 @@ impl<'a, S: ContextProvider> SqlToRel<'a, S> {
                 // Unnest table factor has empty input
                 let schema = DFSchema::empty();
                 let input = LogicalPlanBuilder::empty(true).build()?;
-                // Unnest table factor can have multiple arugments.
+                // Unnest table factor can have multiple arguments.
                 // We treat each argument as a separate unnest expression.
                 let unnest_exprs = array_exprs
                     .into_iter()

datafusion/sql/src/select.rs

@@ -306,7 +306,7 @@ impl<'a, S: ContextProvider> SqlToRel<'a, S> {
         let mut intermediate_select_exprs = select_exprs;
         // Each expr in select_exprs can contains multiple unnest stage
         // The transformation happen bottom up, one at a time for each iteration
-        // Ony exaust the loop if no more unnest transformation is found
+        // Only exaust the loop if no more unnest transformation is found
         for i in 0.. {
             let mut unnest_columns = vec![];
             // from which column used for projection, before the unnest happen

datafusion/sql/src/unparser/ast.rs

@@ -497,7 +497,7 @@ impl Default for DerivedRelationBuilder {
 pub(super) struct UninitializedFieldError(&'static str);
 
 impl UninitializedFieldError {
-    /// Create a new `UnitializedFieldError` for the specified field name.
+    /// Create a new `UninitializedFieldError` for the specified field name.
     pub fn new(field_name: &'static str) -> Self {
         UninitializedFieldError(field_name)
     }

datafusion/sql/src/unparser/plan.rs

@@ -214,12 +214,12 @@ impl Unparser<'_> {
         } else {
             let mut derived_builder = DerivedRelationBuilder::default();
             derived_builder.lateral(false).alias(None).subquery({
-                let inner_statment = self.plan_to_sql(plan)?;
-                if let ast::Statement::Query(inner_query) = inner_statment {
+                let inner_statement = self.plan_to_sql(plan)?;
+                if let ast::Statement::Query(inner_query) = inner_statement {
                     inner_query
                 } else {
                     return internal_err!(
-                        "Subquery must be a Query, but found {inner_statment:?}"
+                        "Subquery must be a Query, but found {inner_statement:?}"
                     );
                 }
             });

datafusion/sql/src/unparser/utils.rs

@@ -31,7 +31,7 @@ pub(crate) enum AggVariant<'a> {
 /// Recursively searches children of [LogicalPlan] to find an Aggregate or window node if one exists
 /// prior to encountering a Join, TableScan, or a nested subquery (derived table factor).
 /// If an Aggregate or window node is not found prior to this or at all before reaching the end
-/// of the tree, None is returned. It is assumed that a Window and Aggegate node cannot both
+/// of the tree, None is returned. It is assumed that a Window and Aggregate node cannot both
 /// be found in a single select query.
 pub(crate) fn find_agg_node_within_select<'a>(
     plan: &'a LogicalPlan,
@@ -82,7 +82,7 @@ pub(crate) fn unproject_agg_exprs(expr: &Expr, agg: &Aggregate) -> Result<Expr>
     expr.clone()
         .transform(|sub_expr| {
             if let Expr::Column(c) = sub_expr {
-                // find the column in the agg schmea
+                // find the column in the agg schema
                 if let Ok(n) = agg.schema.index_of_column(&c) {
                     let unprojected_expr = agg
                         .group_expr

datafusion/sql/src/utils.rs

@@ -325,7 +325,7 @@ pub(crate) fn transform_bottom_unnest(
             let (data_type, _) = arg.data_type_and_nullable(input.schema())?;
 
             if let DataType::Struct(_) = data_type {
-                return internal_err!("unnest on struct can ony be applied at the root level of select expression");
+                return internal_err!("unnest on struct can only be applied at the root level of select expression");
             }
 
             let mut transformed_exprs = transform(&expr, arg)?;

datafusion/sql/tests/common/mod.rs

@@ -56,7 +56,7 @@ pub(crate) struct MockContextProvider {
 }
 
 impl MockContextProvider {
-    // Surpressing dead code warning, as this is used in integration test crates
+    // Suppressing dead code warning, as this is used in integration test crates
     #[allow(dead_code)]
     pub(crate) fn options_mut(&mut self) -> &mut ConfigOptions {
         &mut self.options

datafusion/sql/tests/sql_integration.rs

@@ -1144,7 +1144,7 @@ fn select_aggregate_with_group_by_with_having_that_reuses_aggregate_multiple_tim
 }
 
 #[test]
-fn select_aggregate_with_group_by_with_having_using_aggreagate_not_in_select() {
+fn select_aggregate_with_group_by_with_having_using_aggregate_not_in_select() {
     let sql = "SELECT first_name, MAX(age)
                    FROM person
                    GROUP BY first_name
@@ -1185,7 +1185,7 @@ fn select_aggregate_compound_aliased_with_group_by_with_having_referencing_compo
 }
 
 #[test]
-fn select_aggregate_with_group_by_with_having_using_derived_column_aggreagate_not_in_select(
+fn select_aggregate_with_group_by_with_having_using_derived_column_aggregate_not_in_select(
 ) {
     let sql = "SELECT first_name, MAX(age)
                    FROM person

datafusion/sqllogictest/test_files/aggregate.slt

@@ -3705,7 +3705,7 @@ select
   column3 as tag
 from t_source;
 
-# Demonstate the contents
+# Demonstrate the contents
 query PPPPPPPPTT
 select * from t;
 ----
@@ -3816,7 +3816,7 @@ select
   column3 as tag
 from t_source;
 
-# Demonstate the contents
+# Demonstrate the contents
 query DDTT
 select * from t;
 ----
@@ -3914,7 +3914,7 @@ select
   column3 as tag
 from t_source;
 
-# Demonstate the contents
+# Demonstrate the contents
 query DDDDTT
 select * from t;
 ----
@@ -4108,7 +4108,7 @@ select sum(c1), arrow_typeof(sum(c1)) from d_table;
 ----
 100 Decimal128(20, 3)
 
-# aggregate sum with deciaml
+# aggregate sum with decimal
 statement ok
 create table t (c decimal(35, 3)) as values (10), (null), (20);
 

datafusion/sqllogictest/test_files/array.slt

@@ -5690,7 +5690,7 @@ select
 ----
 [] [] [0] [0]
 
-# Test range for other egde cases
+# Test range for other edge cases
 query ????????
 select 
   range(9223372036854775807, 9223372036854775807, -1) as c1,
@@ -5828,7 +5828,7 @@ select
 [-9223372036854775808] [9223372036854775807] [0, -9223372036854775808] [0, 9223372036854775807]
 
 
-# Test generate_series for other egde cases
+# Test generate_series for other edge cases
 query ????
 select 
   generate_series(9223372036854775807, 9223372036854775807, -1) as c1,

datafusion/sqllogictest/test_files/binary.slt

@@ -25,7 +25,7 @@ SELECT X'FF01', arrow_typeof(X'FF01');
 ----
 ff01 Binary
 
-# Invaid hex values
+# Invalid hex values
 query error DataFusion error: Error during planning: Invalid HexStringLiteral 'Z'
 SELECT X'Z'
 

datafusion/sqllogictest/test_files/copy.slt

@@ -271,7 +271,7 @@ select * from validate_struct_with_array;
 {c0: foo, c1: [1, 2, 3], c2: {c0: bar, c1: [2, 3, 4]}}
 
 
-# Copy parquet with all supported statment overrides
+# Copy parquet with all supported statement overrides
 query IT
 COPY source_table
 TO 'test_files/scratch/copy/table_with_options/'

datafusion/sqllogictest/test_files/explain.slt

@@ -394,7 +394,7 @@ physical_plan_with_schema
 statement ok
 set datafusion.execution.collect_statistics = false;
 
-# Explain ArrayFuncions
+# Explain ArrayFunctions
 
 statement ok
 set datafusion.explain.physical_plan_only = false

datafusion/sqllogictest/test_files/interval.slt

@@ -325,7 +325,7 @@ select
 ----
 Interval(MonthDayNano) Interval(MonthDayNano)
 
-# cast with explicit cast sytax
+# cast with explicit cast syntax
 query TT
 select
   arrow_typeof(cast ('5 months' as interval)),

datafusion/sqllogictest/test_files/math.slt

@@ -112,7 +112,7 @@ SELECT iszero(1.0), iszero(0.0), iszero(-0.0), iszero(NULL)
 ----
 false true true NULL
 
-# abs: empty argumnet
+# abs: empty argument
 statement error
 SELECT abs();
 

datafusion/sqllogictest/test_files/options.slt

@@ -42,7 +42,7 @@ physical_plan
 statement ok
 set datafusion.execution.coalesce_batches = false
 
-# expect no coalsece
+# expect no coalescence
 query TT
 explain SELECT * FROM a WHERE c0 < 1;
 ----

datafusion/sqllogictest/test_files/scalar.slt

@@ -1203,7 +1203,7 @@ FROM t1
 999
 999
 
-# case_when_else_with_null_contant()
+# case_when_else_with_null_constant()
 query I
  SELECT
 CASE WHEN c1 = 'a' THEN 1

datafusion/sqllogictest/test_files/timestamps.slt

@@ -1161,7 +1161,7 @@ ts_data_secs 2020-09-08T00:00:00
 ts_data_secs 2020-09-08T00:00:00
 ts_data_secs 2020-09-08T00:00:00
 
-# Test date trun on different granularity
+# Test date turn on different granularity
 query TP rowsort
 SELECT 'millisecond', DATE_TRUNC('millisecond', ts) FROM ts_data_nanos
   UNION ALL

datafusion/sqllogictest/test_files/unnest.slt

@@ -484,7 +484,7 @@ query error DataFusion error: type_coercion\ncaused by\nThis feature is not impl
 select sum(unnest(generate_series(1,10)));
 
 ## TODO: support unnest as a child expr
-query error DataFusion error: Internal error: unnest on struct can ony be applied at the root level of select expression
+query error DataFusion error: Internal error: unnest on struct can only be applied at the root level of select expression
 select arrow_typeof(unnest(column5)) from unnest_table;
 
 
@@ -517,7 +517,7 @@ select unnest(unnest(unnest(column3)['c1'])), column3 from recursive_unnest_tabl
 3 [{c0: [2], c1: [[3], [4]]}]
 4 [{c0: [2], c1: [[3], [4]]}]
 
-## tripple list unnest
+## triple list unnest
 query I?
 select unnest(unnest(unnest(column2))), column2 from recursive_unnest_table;
 ----

datafusion/sqllogictest/test_files/update.slt

@@ -74,7 +74,7 @@ logical_plan
 statement ok
 create table t3(a int, b varchar, c double, d int);
 
-# set from mutiple tables, sqlparser only supports from one table
+# set from multiple tables, sqlparser only supports from one table
 query error DataFusion error: SQL error: ParserError\("Expected end of statement, found: ,"\)
 explain update t1 set b = t2.b, c = t3.a, d = 1 from t2, t3 where t1.a = t2.a and t1.a = t3.a;
 

datafusion/sqllogictest/test_files/window.slt

@@ -3968,7 +3968,7 @@ CREATE TABLE table_with_pk (
 # However, if we know that <expr> contains a unique column (e.g. a PRIMARY KEY),
 # it can be treated as `OVER (ORDER BY <expr> ROWS BETWEEN UNBOUNDED PRECEDING
 # AND CURRENT ROW)` where window frame units change from `RANGE` to `ROWS`. This
-# conversion makes the window frame manifestly causal by eliminating the possiblity
+# conversion makes the window frame manifestly causal by eliminating the possibility
 # of ties explicitly (see window frame documentation for a discussion of causality
 # in this context). The Query below should have `ROWS` in its window frame.
 query TT

datafusion/substrait/src/logical_plan/producer.rs

@@ -994,7 +994,7 @@ pub fn make_binary_op_scalar_func(
 ///
 /// * `expr` - DataFusion expression to be parse into a Substrait expression
 /// * `schema` - DataFusion input schema for looking up field qualifiers
-/// * `col_ref_offset` - Offset for caculating Substrait field reference indices.
+/// * `col_ref_offset` - Offset for calculating Substrait field reference indices.
 ///                     This should only be set by caller with more than one input relations i.e. Join.
 ///                     Substrait expects one set of indices when joining two relations.
 ///                     Let's say `left` and `right` have `m` and `n` columns, respectively. The `right`