Merge pull request #268 from Qrlew/bug_joins

Preserve columns names when SELECT * and JOIN

CHANGELOG.md

@@ -6,8 +6,10 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 ## [Unreleased]
+## [0.9.15] - 2024-02-20
 ### Changed
-- removing the coalesce of the privacy unit with since it looses the UNIQUE property and changing the JOIN to left instead of INNER after the clipping.
+- SELECT * with JOINs preserve the column names when there is no ambiguity [#268](https://github.com/Qrlew/qrlew/pull/268)
+- removing the coalesce of the privacy unit with since it looses the UNIQUE property and changing the JOIN to left instead of INNER after the clipping. [#272](https://github.com/Qrlew/qrlew/pull/272)
 
 ## [0.9.14] - 2024-01-30
 ### Added

Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 authors = ["Nicolas Grislain <[email protected]>"]
 name = "qrlew"
-version = "0.9.14"
+version = "0.9.15"
 edition = "2021"
 description = "Sarus Qrlew Engine"
 documentation = "https://docs.rs/qrlew"

src/dialect_translation/postgresql.rs

@@ -99,13 +99,7 @@ mod tests {
 
     use super::*;
     use crate::{
-        builder::{Ready, With},
-        data_type::{DataType, Value as _},
-        display::Dot,
-        expr::Expr,
-        namer,
-        relation::{schema::Schema, Relation, TableBuilder},
-        sql::{parse, relation::QueryWithRelations},
+        builder::{Ready, With}, data_type::{DataType, Value as _}, display::Dot, expr::Expr, io::{postgresql, Database as _}, namer, relation::{schema::Schema, Relation, TableBuilder}, sql::{parse, relation::QueryWithRelations}
     };
     use std::sync::Arc;
 
@@ -154,36 +148,22 @@ mod tests {
 
     #[test]
     fn test_table_special() -> Result<()> {
-        let table: Relation = TableBuilder::new()
-            .path(["MY SPECIAL TABLE"])
-            .name("my_table")
-            .size(100)
-            .schema(
-                Schema::empty()
-                    .with(("Id", DataType::integer_interval(0, 1000)))
-                    .with(("Na.Me", DataType::text()))
-                    .with(("inc&ome", DataType::float_interval(100.0, 200000.0)))
-                    .with(("normal_col",  DataType::text())),
-            )
-            .build();
-        let relations = Hierarchy::from([(["schema", "MY SPECIAL TABLE"], Arc::new(table))]);
+        let mut database = postgresql::test_database();
+        let relations = database.relations();
         let query_str = r#"SELECT "Id", NORMAL_COL, "Na.Me" FROM "MY SPECIAL TABLE" ORDER BY "Id" "#;
         let translator = PostgreSqlTranslator;
         let query = parse_with_dialect(query_str, translator.dialect())?;
         let query_with_relation = QueryWithRelations::new(&query, &relations);
         let relation = Relation::try_from((query_with_relation, translator))?;
         println!("\n {} \n", relation);
         let rel_with_traslator = RelationWithTranslator(&relation, translator);
-        let retranslated = ast::Query::from(rel_with_traslator);
-        print!("{}", retranslated);
-        let translated = r#"
-        WITH "map_mou5" ("Id","normal_col","Na.Me") AS (
-            SELECT "Id" AS "Id", "normal_col" AS "normal_col", "Na.Me" AS "Na.Me" FROM "MY SPECIAL TABLE"
-        ), "map_0swv"("Id","normal_col","Na.Me") AS (
-            SELECT "Id" AS "Id", "normal_col" AS "normal_col", "Na.Me" AS "Na.Me" FROM "map_mou5" ORDER BY "Id" ASC
-        ) SELECT * FROM "map_0swv"
-        "#;
-        assert_same_query_str(&retranslated.to_string(), translated);
+        let translated = ast::Query::from(rel_with_traslator);
+        print!("{}", translated);
+        _ = database
+            .query(translated.to_string().as_str())
+            .unwrap()
+            .iter()
+            .map(ToString::to_string);
         Ok(())
     }
 }

src/relation/rewriting.rs

@@ -202,38 +202,52 @@ impl Join {
         self
     }
 
-    /// Replace the duplicates fields specified in `columns` by their coalesce expression
-    /// Its mimicks teh behaviour of USING in SQL
+    /// To mimic the behavior of USING(col) and NATURAL JOIN in SQL we create
+    /// a map where join columns identified by `vec` are coalesced.
+    ///     vec: vector of string identifying input columns present in both _LEFT_
+    ///         and _RIGHT_ relation of the join. 
+    ///     columns: is the Hierarchy mapping input names in the JOIN to name field
+    /// 
+    /// It returns a:
+    ///     - Map build on top the Join with coalesced column along with
+    ///         the other fields of the join and 
+    ///     - coalesced columns mapping (name in join -> name in map)
     pub fn remove_duplicates_and_coalesce(
         self,
         vec: Vec<String>,
         columns: &Hierarchy<Identifier>,
-    ) -> Relation {
-        let fields = self
+    ) -> (Relation, Hierarchy<Identifier>) {
+        let mut coalesced_cols: Vec<(Identifier, Identifier)> = vec![];
+        let coalesced = self
             .field_inputs()
-            .filter_map(|(name, id)| {
-                let col = id.as_ref().last().unwrap();
-                if id.as_ref().first().unwrap().as_str() == LEFT_INPUT_NAME && vec.contains(col) {
+            .filter_map(|(_, input_id)| {
+                let col = input_id.as_ref().last().unwrap();
+                if input_id.as_ref().first().unwrap().as_str() == LEFT_INPUT_NAME && vec.contains(col) {
+                    let left_col = columns[[LEFT_INPUT_NAME, col]].as_ref().last().unwrap();
+                    let right_col = columns[[RIGHT_INPUT_NAME, col]].as_ref().last().unwrap();
+                    coalesced_cols.push((left_col.as_str().into(), col[..].into()));
+                    coalesced_cols.push((right_col.as_str().into(), col[..].into()));
                     Some((
-                        name,
+                        col.clone(),
                         Expr::coalesce(
-                            Expr::col(columns[[LEFT_INPUT_NAME, col]].as_ref().last().unwrap()),
-                            Expr::col(columns[[RIGHT_INPUT_NAME, col]].as_ref().last().unwrap()),
+                            Expr::col(left_col),
+                            Expr::col(right_col),
                         ),
                     ))
                 } else {
                     None
                 }
-            })
+            });
+        let coalesced_with_others = coalesced
             .chain(self.field_inputs().filter_map(|(name, id)| {
                 let col = id.as_ref().last().unwrap();
                 (!vec.contains(col)).then_some((name.clone(), Expr::col(name)))
             }))
             .collect::<Vec<_>>();
-        Relation::map()
+        (Relation::map()
             .input(Relation::from(self))
-            .with_iter(fields)
-            .build()
+            .with_iter(coalesced_with_others)
+            .build(), coalesced_cols.into_iter().collect())
     }
 }
 

src/sql/mod.rs

@@ -118,6 +118,7 @@ mod tests {
     use itertools::Itertools;
     use sqlparser::dialect::BigQueryDialect;
 
+
     #[test]
     fn test_display() {
         let database = postgresql::test_database();
@@ -132,12 +133,12 @@ mod tests {
             t2 AS (SELECT * FROM table_2)
             SELECT max(a), sum(d) FROM t1 INNER JOIN t2 ON t1.d = t2.x CROSS JOIN table_2 GROUP BY t2.y, t1.a",
             "
-            WITH t1 AS (SELECT a,d FROM table_1),
+            WITH t1 AS (SELECT a, d FROM table_1),
             t2 AS (SELECT * FROM table_2)
             SELECT * FROM t1 INNER JOIN t2 ON t1.d = t2.x INNER JOIN table_2 ON t1.d=table_2.x ORDER BY t1.a LIMIT 10",
         ] {
             let relation = Relation::try_from(parse(query).unwrap().with(&database.relations())).unwrap();
-            relation.display_dot();
+            relation.display_dot().unwrap();
         }
     }