fix: Resolve multiple SQL JOIN issues (pola-rs#16507)

crates/polars-lazy/src/frame/mod.rs

@@ -1085,8 +1085,13 @@ impl LazyFrame {
 
     /// Creates the Cartesian product from both frames, preserving the order of the left keys.
     #[cfg(feature = "cross_join")]
-    pub fn cross_join(self, other: LazyFrame) -> LazyFrame {
-        self.join(other, vec![], vec![], JoinArgs::new(JoinType::Cross))
+    pub fn cross_join(self, other: LazyFrame, suffix: Option<String>) -> LazyFrame {
+        self.join(
+            other,
+            vec![],
+            vec![],
+            JoinArgs::new(JoinType::Cross).with_suffix(suffix),
+        )
     }
 
     /// Left outer join this query with another lazy query.
@@ -1237,9 +1242,7 @@ impl LazyFrame {
         if let Some(suffix) = args.suffix {
             builder = builder.suffix(suffix);
         }
-
         // Note: args.slice is set by the optimizer
-
         builder.finish()
     }
 

crates/polars-lazy/src/tests/cse.rs

@@ -305,7 +305,7 @@ fn test_cse_columns_projections() -> PolarsResult<()> {
     ]?
     .lazy();
 
-    let left = left.cross_join(right.clone().select([col("A")]));
+    let left = left.cross_join(right.clone().select([col("A")]), None);
     let q = left.join(
         right.rename(["B"], ["C"]),
         [col("A"), col("C")],

crates/polars-lazy/src/tests/projection_queries.rs

@@ -47,7 +47,7 @@ fn test_cross_join_pd() -> PolarsResult<()> {
         "price" => [5, 4]
     ]?;
 
-    let q = food.lazy().cross_join(drink.lazy()).select([
+    let q = food.lazy().cross_join(drink.lazy(), None).select([
         col("name").alias("food"),
         col("name_right").alias("beverage"),
         (col("price") + col("price_right")).alias("total"),

crates/polars-lazy/src/tests/queries.rs

@@ -1171,7 +1171,7 @@ fn test_cross_join() -> PolarsResult<()> {
         "b" => [None, Some(12)]
     ]?;
 
-    let out = df1.lazy().cross_join(df2.lazy()).collect()?;
+    let out = df1.lazy().cross_join(df2.lazy(), None).collect()?;
     assert_eq!(out.shape(), (6, 4));
     Ok(())
 }

crates/polars-lazy/src/tests/streaming.rs

@@ -85,7 +85,7 @@ fn test_streaming_union_order() -> PolarsResult<()> {
 fn test_streaming_union_join() -> PolarsResult<()> {
     let q = get_csv_glob();
     let q = q.select([col("sugars_g"), col("calories")]);
-    let q = q.clone().cross_join(q);
+    let q = q.clone().cross_join(q, None);
 
     assert_streaming_with_default(q, true, true);
     Ok(())
@@ -166,18 +166,22 @@ fn test_streaming_cross_join() -> PolarsResult<()> {
         "a" => [1 ,2, 3]
     ]?;
     let q = df.lazy();
-    let out = q.clone().cross_join(q).with_streaming(true).collect()?;
+    let out = q
+        .clone()
+        .cross_join(q, None)
+        .with_streaming(true)
+        .collect()?;
     assert_eq!(out.shape(), (9, 2));
 
     let q = get_parquet_file().with_projection_pushdown(false);
     let q1 = q
         .clone()
         .select([col("calories")])
-        .cross_join(q.clone())
+        .cross_join(q.clone(), None)
         .filter(col("calories").gt(col("calories_right")));
     let q2 = q1
         .select([all().name().suffix("_second")])
-        .cross_join(q)
+        .cross_join(q, None)
         .filter(col("calories_right_second").lt(col("calories")))
         .select([
             col("calories"),
@@ -266,7 +270,7 @@ fn test_streaming_slice() -> PolarsResult<()> {
     ]?
     .lazy();
 
-    let q = lf_a.clone().cross_join(lf_a).slice(10, 20);
+    let q = lf_a.clone().cross_join(lf_a, None).slice(10, 20);
     let a = q.with_streaming(true).collect().unwrap();
     assert_eq!(a.shape(), (20, 2));
 

crates/polars-ops/src/frame/join/args.rs

@@ -88,6 +88,11 @@ impl JoinArgs {
         self
     }
 
+    pub fn with_suffix(mut self, suffix: Option<String>) -> Self {
+        self.suffix = suffix;
+        self
+    }
+
     pub fn suffix(&self) -> &str {
         self.suffix.as_deref().unwrap_or("_right")
     }

crates/polars-plan/src/dsl/expr.rs

@@ -304,7 +304,7 @@ pub enum Excluded {
 impl Expr {
     /// Get Field result of the expression. The schema is the input data.
     pub fn to_field(&self, schema: &Schema, ctxt: Context) -> PolarsResult<Field> {
-        // this is not called much and th expression depth is typically shallow
+        // this is not called much and the expression depth is typically shallow
         let mut arena = Arena::with_capacity(5);
         self.to_field_amortized(schema, ctxt, &mut arena)
     }

crates/polars-plan/src/logical_plan/optimizer/projection_pushdown/joins.rs

@@ -444,7 +444,7 @@ fn resolve_join_suffixes(
         .iter()
         .map(|proj| {
             let name = column_node_to_name(*proj, expr_arena);
-            if name.contains(suffix) && schema_after_join.get(&name).is_none() {
+            if name.ends_with(suffix) && schema_after_join.get(&name).is_none() {
                 let downstream_name = &name.as_ref()[..name.len() - suffix.len()];
                 let col = AExpr::Column(ColumnName::from(downstream_name));
                 let node = expr_arena.add(col);

crates/polars-sql/Cargo.toml

@@ -13,6 +13,7 @@ arrow = { workspace = true }
 polars-core = { workspace = true }
 polars-error = { workspace = true }
 polars-lazy = { workspace = true, features = ["abs", "binary_encoding", "concat_str", "cross_join", "cum_agg", "dtype-date", "dtype-decimal", "is_in", "list_eval", "log", "meta", "regex", "round_series", "sign", "string_reverse", "strings", "timezones", "trigonometry"] }
+polars-ops = { workspace = true }
 polars-plan = { workspace = true }
 
 hex = { workspace = true }