fix sql format (#385)

timeplus-io · Dec 7, 2023 · 16a408e · 16a408e
1 parent 8066de7
commit 16a408e
Show file tree

Hide file tree

Showing 76 changed files with 1,415 additions and 1,549 deletions.
diff --git a/src/Common/tests/gtest_ident.cpp b/src/Common/tests/gtest_ident.cpp
@@ -21,21 +21,21 @@ String convertQueryToFormat(String query,bool one_line)
 
 TEST(indent, stream_name)
 {
-    //test format 'one_line'=false;
+    /// test format 'one_line'=false;
     String query("with it1 as (with it2 as (select * from numbers(10)) select * from it2) select user.id from user,it1 where it1.number=user.id");
-    String expected_ans("WITH it1 AS\n  (\n    WITH it2 AS\n      (\n        SELECT \n          *\n        FROM \n          numbers(10)\n      )\n    SELECT \n      *\n    FROM \n      it2\n  )\nSELECT \n  user.id\nFROM \n  user,it1\nWHERE \n  it1.number = user.id");
+    String expected_ans("WITH it1 AS\n  (\n    WITH it2 AS\n      (\n        SELECT\n          *\n        FROM\n          numbers(10)\n      )\n    SELECT\n      *\n    FROM\n      it2\n  )\nSELECT\n  user.id\nFROM\n  user, it1\nWHERE\n  it1.number = user.id");
     String ans = convertQueryToFormat(query,false);
     EXPECT_STREQ(expected_ans.c_str(), ans.c_str());
 
-    //test DISTINCT  JOIN  WHERE  PARTITION BY  GROUP BY  HAVING ORDER BY  LIMIT  EMIT  SETTINGS
+    /// test DISTINCT  JOIN  WHERE  PARTITION BY  GROUP BY  HAVING ORDER BY  LIMIT  EMIT  SETTINGS
     query = "with it1 as (select * from numbers(10)) select distinct user.id from user join it1 on user.id=it1.number group by user.id having id>0 order by id limit 10 emit last 1s settings query_mode='stream'";
-    expected_ans = "WITH it1 AS\n  (\n    SELECT \n      *\n    FROM \n      numbers(10)\n  )\nSELECT DISTINCT \n  user.id\nFROM \n  user\nINNER JOIN it1 ON user.id = it1.number\nGROUP BY \n  user.id\nHAVING \n  id > 0\nORDER BY \n  id ASC\nLIMIT 10\nEMIT STREAM LAST 1s\nSETTINGS \n  query_mode = 'stream'";
+    expected_ans = "WITH it1 AS\n  (\n    SELECT\n      *\n    FROM\n      numbers(10)\n  )\nSELECT DISTINCT\n  user.id\nFROM\n  user\nINNER JOIN it1 ON user.id = it1.number\nGROUP BY\n  user.id\nHAVING\n  id > 0\nORDER BY\n  id ASC\nLIMIT 10\nEMIT STREAM LAST 1s\nSETTINGS\n  query_mode = 'stream'";
     ans = convertQueryToFormat(query,false);
     EXPECT_STREQ(expected_ans.c_str(), ans.c_str());
 
-    //test format 'one_line'=true;
+    /// test format 'one_line'=true;
     query = "with it1 as (with it2 as (select * from numbers(10)) select * from it2) select user.id from user,it1 where it1.number=user.id";
-    expected_ans = "WITH it1 AS (WITH it2 AS (SELECT * FROM numbers(10)) SELECT * FROM it2) SELECT user.id FROM user,it1 WHERE it1.number = user.id";
+    expected_ans = "WITH it1 AS (WITH it2 AS (SELECT * FROM numbers(10)) SELECT * FROM it2) SELECT user.id FROM user, it1 WHERE it1.number = user.id";
     ans = convertQueryToFormat(query,true);
     EXPECT_STREQ(expected_ans.c_str(), ans.c_str());
 
@@ -45,3 +45,33 @@ TEST(indent, stream_name)
     EXPECT_STREQ(expected_ans.c_str(), ans.c_str());
 
 }
+
+TEST(indent, complex_query)
+{
+    /// Covering: WITH  SELECT  DISTINCT  FROM  JOIN  WHERE  PARTITION BY  SHUFFLE BY  GROUP BY  HAVING  ORDER BY  LIMIT  EMIT  SETTINGS UNION
+    /// test format 'one_line'=false;
+    String query = "with it1 as (select * from numbers(10)), it2 as (select * from it1) select distinct user.id from user join it1 on user.id=it1.number array join it2.number as number where number > 1 or number < 5 partition by user.id group by user.id having id>0 order by id limit 10 emit last 1s settings query_mode='stream' union select * from it2";
+    String expected_ans = """WITH it1 AS\n  (\n    SELECT\n      *\n    FROM\n      numbers(10)\n  ), it2 AS\n  (\n    SELECT\n      *\n    FROM\n      it1\n  )\nSELECT DISTINCT\n  user.id\nFROM\n  user\nINNER JOIN it1 ON user.id = it1.number\nARRAY JOIN it2.number AS number\nWHERE\n  (number > 1) OR (number < 5)\nPARTITION BY\n  user.id\nGROUP BY\n  user.id\nHAVING\n  id > 0\nORDER BY\n  id ASC\nLIMIT 10\nEMIT STREAM LAST 1s\nSETTINGS\n  query_mode = 'stream'\nUNION\nSELECT\n  *\nFROM\n  it2";
+    String ans = convertQueryToFormat(query,false);
+    EXPECT_STREQ(expected_ans.c_str(), ans.c_str());
+
+    /// test format 'one_line'=true;
+    expected_ans = "WITH it1 AS (SELECT * FROM numbers(10)), it2 AS (SELECT * FROM it1) SELECT DISTINCT user.id FROM user INNER JOIN it1 ON user.id = it1.number ARRAY JOIN it2.number AS number WHERE (number > 1) OR (number < 5) PARTITION BY user.id GROUP BY user.id HAVING id > 0 ORDER BY id ASC LIMIT 10 EMIT STREAM LAST 1s SETTINGS query_mode = 'stream' UNION SELECT * FROM it2";
+    ans = convertQueryToFormat(query,true);
+    EXPECT_STREQ(expected_ans.c_str(), ans.c_str());
+}
+
+TEST(indent, join_query)
+{
+    /// Covering: JOIN, INNER JOIN, LEFT JOIN, ASOF JOIN, LEFT ASOF JOIN, INNER ASOF JOIN, LATEST JOIN, LEFT LATEST JOIN, INNER LATEST JOIN, ARRAY JOIN
+    /// test format 'one_line'=false;
+    String query = "with tmp as (select * from t1 join t2 using(id) inner join t3 on t1.id=t3.id left join t4 on t1.id = t4.id asof join t as t5 on t1.id=t5.id and t1.time < t5.time array join t1.time as time, t2.time as time2 inner asof join (select * from t6) as t6 on t1.id=t2.id and t1.time < t6.time left asof join t7 on t1.id=t7.id and t1.time > t7.time latest join t8 using(id) inner latest join t9 on t1.id=t9.id left latest join cte as t10 on t1.id=t10.id) select * from tmp";
+    String expected_ans = "WITH tmp AS\n  (\n    SELECT\n      *\n    FROM\n      t1\n    INNER JOIN t2 USING (id)\n    INNER JOIN t3 ON t1.id = t3.id\n    LEFT JOIN t4 ON t1.id = t4.id\n    ASOF INNER JOIN t AS t5 ON (t1.id = t5.id) AND (t1.time < t5.time)\n    ARRAY JOIN t1.time AS time, t2.time AS time2\n    ASOF INNER JOIN (\n        SELECT\n          *\n        FROM\n          t6\n      ) AS t6 ON (t1.id = t2.id) AND (t1.time < t6.time)\n    ASOF LEFT JOIN t7 ON (t1.id = t7.id) AND (t1.time > t7.time)\n    LATEST INNER JOIN t8 USING (id)\n    LATEST INNER JOIN t9 ON t1.id = t9.id\n    LATEST LEFT JOIN cte AS t10 ON t1.id = t10.id\n  )\nSELECT\n  *\nFROM\n  tmp";
+    String ans = convertQueryToFormat(query,false);
+    EXPECT_STREQ(expected_ans.c_str(), ans.c_str());
+
+    /// test format 'one_line'=true;
+    expected_ans = "WITH tmp AS (SELECT * FROM t1 INNER JOIN t2 USING (id) INNER JOIN t3 ON t1.id = t3.id LEFT JOIN t4 ON t1.id = t4.id ASOF INNER JOIN t AS t5 ON (t1.id = t5.id) AND (t1.time < t5.time) ARRAY JOIN t1.time AS time, t2.time AS time2 ASOF INNER JOIN (SELECT * FROM t6) AS t6 ON (t1.id = t2.id) AND (t1.time < t6.time) ASOF LEFT JOIN t7 ON (t1.id = t7.id) AND (t1.time > t7.time) LATEST INNER JOIN t8 USING (id) LATEST INNER JOIN t9 ON t1.id = t9.id LATEST LEFT JOIN cte AS t10 ON t1.id = t10.id) SELECT * FROM tmp";
+    ans = convertQueryToFormat(query,true);
+    EXPECT_STREQ(expected_ans.c_str(), ans.c_str());
+}
diff --git a/src/Interpreters/Streaming/tests/gtest_eliminate_subquery.cpp b/src/Interpreters/Streaming/tests/gtest_eliminate_subquery.cpp
@@ -65,7 +65,7 @@ TEST(EliminateSubquery, OptimizedQuery)
     EXPECT_EQ(
         optimizeSubquery("SELECT count(a.productid) FROM (SELECT a.*, b.* FROM access1 as a, db2.product_info as b WHERE (a.productid = "
                          "b.productid))"),
-        "SELECT count(a.productid) FROM access1 AS a,db2.product_info AS b WHERE a.productid = b.productid");
+        "SELECT count(a.productid) FROM access1 AS a, db2.product_info AS b WHERE a.productid = b.productid");
     EXPECT_EQ(
         optimizeSubquery("SELECT count(*) FROM (SELECT SIGN FROM (SELECT SIGN, query FROM tablea WHERE query_id='mock-1') WHERE "
                          "query = "
@@ -76,7 +76,7 @@ TEST(EliminateSubquery, OptimizedQuery)
         optimizeSubquery("SELECT sum(a.price), sum(b.sale) FROM (SELECT a.*, b.* FROM access1 as a, product_info as b WHERE (a.productid "
                          "= "
                          "b.productid)) GROUP BY a.name, b.sign"),
-        "SELECT sum(a.price), sum(b.sale) FROM access1 AS a,product_info AS b WHERE a.productid = b.productid GROUP BY a.name, b.sign");
+        "SELECT sum(a.price), sum(b.sale) FROM access1 AS a, product_info AS b WHERE a.productid = b.productid GROUP BY a.name, b.sign");
     EXPECT_EQ(
         optimizeSubquery("SELECT sum(a.price), sum(b.sale) FROM (SELECT a.*, b.* FROM access1 as a JOIN product_info as b ON a.productid "
                          "= "
@@ -97,11 +97,11 @@ TEST(EliminateSubquery, OptimizedQuery)
     EXPECT_EQ(
         optimizeSubquery("SELECT count(b.code) FROM (SELECT a.code, b.code FROM product_info as a, product_info as b WHERE (a.productid = "
                          "b.productid))"),
-        "SELECT count(b.code) FROM product_info AS a,product_info AS b WHERE a.productid = b.productid");
+        "SELECT count(b.code) FROM product_info AS a, product_info AS b WHERE a.productid = b.productid");
     EXPECT_EQ(
         optimizeSubquery("SELECT count(a.code) FROM (SELECT a.code, b.code FROM product_info as a, product_info as b WHERE (a.productid = "
                          "b.productid))"),
-        "SELECT count(a.code) FROM product_info AS a,product_info AS b WHERE a.productid = b.productid");
+        "SELECT count(a.code) FROM product_info AS a, product_info AS b WHERE a.productid = b.productid");
     EXPECT_EQ(
         optimizeSubquery("SELECT * FROM (SELECT _raw AS c, t._raw FROM default.frontier_integration_test AS t)"),
         "SELECT _raw AS c, t._raw FROM default.frontier_integration_test AS t");
@@ -187,19 +187,19 @@ TEST(EliminateSubquery, FailedOptimizedQuery)
     EXPECT_EQ(
         optimizeSubquery("SELECT count(t.code) FROM (SELECT a.code, b.code FROM product_info as a, product_info as b WHERE a.productid = "
                          "b.productid) AS t"),
-        "SELECT count(t.code) FROM (SELECT a.code, b.code FROM product_info AS a,product_info AS b WHERE a.productid = b.productid) AS t");
+        "SELECT count(t.code) FROM (SELECT a.code, b.code FROM product_info AS a, product_info AS b WHERE a.productid = b.productid) AS t");
     EXPECT_EQ(
         optimizeSubquery("SELECT count(t.b.code) FROM (SELECT a.code, b.code FROM product_info as a, product_info as b WHERE a.productid = "
                          "b.productid) AS t"),
-        "SELECT count(t.b.code) FROM (SELECT a.code, b.code FROM product_info AS a,product_info AS b WHERE a.productid = b.productid) AS t");
+        "SELECT count(t.b.code) FROM (SELECT a.code, b.code FROM product_info AS a, product_info AS b WHERE a.productid = b.productid) AS t");
     EXPECT_EQ(
         optimizeSubquery(
             "SELECT count(t.code) FROM (SELECT code, code FROM product_info as a, product_info as b WHERE a.productid = b.productid) AS t"),
-        "SELECT count(t.code) FROM (SELECT code, code FROM product_info AS a,product_info AS b WHERE a.productid = b.productid) AS t");
+        "SELECT count(t.code) FROM (SELECT code, code FROM product_info AS a, product_info AS b WHERE a.productid = b.productid) AS t");
     EXPECT_EQ(optimizeSubquery("SELECT _time FROM (SELECT name FROM price)"), "SELECT _time FROM (SELECT name FROM price)");
     EXPECT_EQ(
         optimizeSubquery(
             "SELECT count(code) FROM (SELECT a.code, b.code FROM product_info as a, product_info as b WHERE a.productid = b.productid)"),
-        "SELECT count(code) FROM (SELECT a.code, b.code FROM product_info AS a,product_info AS b WHERE a.productid = b.productid)");
+        "SELECT count(code) FROM (SELECT a.code, b.code FROM product_info AS a, product_info AS b WHERE a.productid = b.productid)");
     EXPECT_EQ(optimizeSubquery("SELECT b FROM (SELECT b AS a FROM product)"), "SELECT b FROM (SELECT b AS a FROM product)");
 }
diff --git a/src/Parsers/ASTAlterQuery.cpp b/src/Parsers/ASTAlterQuery.cpp
@@ -550,7 +550,9 @@ void ASTAlterQuery::formatQueryImpl(const FormatSettings & settings, FormatState
 {
     frame.need_parens = false;
 
-    std::string indent_str = settings.one_line ? "" : std::string(4u * frame.indent, ' ');
+    /// proton: starts
+    std::string indent_str = settings.one_line ? "" : std::string(settings.indent_size * frame.indent, ' ');
+    /// proton: ends
     settings.ostr << (settings.hilite ? hilite_keyword : "") << indent_str;
 
     switch (alter_object)
@@ -576,7 +578,7 @@ void ASTAlterQuery::formatQueryImpl(const FormatSettings & settings, FormatState
             settings.ostr << indent_str << backQuoteIfNeed(getDatabase());
             settings.ostr << ".";
         }
-        settings.ostr << indent_str << backQuoteIfNeed(getTable()) << " ";
+        settings.ostr << indent_str << backQuoteIfNeed(getTable());
     }
     else if (alter_object == AlterObjectType::DATABASE && database)
     {

diff --git a/src/Parsers/ASTCheckQuery.h b/src/Parsers/ASTCheckQuery.h
@@ -28,7 +28,9 @@ struct ASTCheckQuery : public ASTQueryWithTableAndOutput
     {
         std::string nl_or_nothing = settings.one_line ? "" : "\n";
 
-        std::string indent_str = settings.one_line ? "" : std::string(4 * frame.indent, ' ');
+        /// proton: starts
+        std::string indent_str = settings.one_line ? "" : std::string(settings.indent_size * frame.indent, ' ');
+        /// proton: ends
         std::string nl_or_ws = settings.one_line ? " " : "\n";
 
         settings.ostr << (settings.hilite ? hilite_keyword : "") << indent_str << "CHECK TABLE " << (settings.hilite ? hilite_none : "");

diff --git a/src/Parsers/ASTExpressionList.cpp b/src/Parsers/ASTExpressionList.cpp
@@ -14,7 +14,7 @@ ASTPtr ASTExpressionList::clone() const
 
 void ASTExpressionList::formatImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const
 {
-    if (frame.expression_list_prepend_whitespace && !settings.one_line )
+    if (frame.expression_list_prepend_whitespace)
         settings.ostr << ' ';
 
     for (ASTs::const_iterator it = children.begin(); it != children.end(); ++it)
@@ -40,18 +40,19 @@ void ASTExpressionList::formatImpl(const FormatSettings & settings, FormatState
 
 void ASTExpressionList::formatImplMultiline(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const
 {
-    /// proton: starts. expression list in one line
-    std::string indent_str = " ";
+    /// proton: starts.
+    std::string indent_str = "\n" + std::string(settings.indent_size * (frame.indent + 1), ' ');
+    if (frame.expression_list_always_start_on_new_line)
+        settings.ostr << indent_str;
+    else if (frame.expression_list_prepend_whitespace)
+        settings.ostr << " ";
+
+    std::string element_indent_str
+        = (children.size() > 1 || frame.expression_list_always_start_on_new_line) && !frame.expression_list_elements_are_always_on_one_line
+        ? indent_str
+        : " ";
     /// proton: ends.
 
-    if (frame.expression_list_prepend_whitespace)
-    {
-        if (!(children.size() > 1 || frame.expression_list_always_start_on_new_line))
-            /// proton: starts
-            settings.ostr << "";
-            /// proton: ends
-    }
-
     ++frame.indent;
 
     for (ASTs::const_iterator it = children.begin(); it != children.end(); ++it)
@@ -60,12 +61,9 @@ void ASTExpressionList::formatImplMultiline(const FormatSettings & settings, For
         {
             if (separator)
                 settings.ostr << separator;
-        }
 
-        /// proton: starts
-        if (it != children.begin() && (children.size() > 1 || frame.expression_list_always_start_on_new_line))
-        /// proton: ends
-            settings.ostr << indent_str;
+            settings.ostr << element_indent_str;
+        }
 
         FormatStateStacked frame_nested = frame;
         frame_nested.expression_list_always_start_on_new_line = false;

diff --git a/src/Parsers/ASTFunction.cpp b/src/Parsers/ASTFunction.cpp
@@ -239,7 +239,7 @@ void ASTFunction::formatImplWithoutAlias(const FormatSettings & settings, Format
     {
         std::string nl_or_nothing = settings.one_line ? "" : "\n";
         /// proton: starts
-        std::string indent_str = settings.one_line ? "" : std::string(2u * frame.indent, ' ');
+        std::string indent_str = settings.one_line ? "" : std::string(settings.indent_size * frame.indent, ' ');
         /// proton: ends
         settings.ostr << (settings.hilite ? hilite_function : "") << name << "(" << nl_or_nothing;
         FormatStateStacked frame_nested = frame;

diff --git a/src/Parsers/ASTNameTypePair.cpp b/src/Parsers/ASTNameTypePair.cpp
@@ -24,9 +24,8 @@ ASTPtr ASTNameTypePair::clone() const
 void ASTNameTypePair::formatImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const
 {
     /// proton: starts
-    std::string indent_str = settings.one_line ? "" : std::string(2 * frame.indent, ' ');
+    std::string indent_str = settings.one_line ? "" : std::string(settings.indent_size * frame.indent, ' ');
     /// proton: ends
-
     settings.ostr << indent_str << backQuoteIfNeed(name) << ' ';
     type->formatImpl(settings, state, frame);
 }

diff --git a/src/Parsers/ASTProjectionDeclaration.cpp b/src/Parsers/ASTProjectionDeclaration.cpp
@@ -18,7 +18,9 @@ ASTPtr ASTProjectionDeclaration::clone() const
 void ASTProjectionDeclaration::formatImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const
 {
     settings.ostr << backQuoteIfNeed(name);
-    std::string indent_str = settings.one_line ? "" : std::string(4u * frame.indent, ' ');
+    /// proton: starts
+    std::string indent_str = settings.one_line ? "" : std::string(settings.indent_size * frame.indent, ' ');
+    /// proton: ends
     std::string nl_or_nothing = settings.one_line ? "" : "\n";
     settings.ostr << nl_or_nothing << indent_str << "(" << nl_or_nothing;
     FormatStateStacked frame_nested = frame;

diff --git a/src/Parsers/ASTProjectionSelectQuery.cpp b/src/Parsers/ASTProjectionSelectQuery.cpp
@@ -51,7 +51,9 @@ void ASTProjectionSelectQuery::formatImpl(const FormatSettings & s, FormatState
 {
     frame.current_select = this;
     frame.need_parens = false;
-    std::string indent_str = s.one_line ? "" : std::string(4 * frame.indent, ' ');
+    /// proton: starts
+    std::string indent_str = s.one_line ? "" : std::string(s.indent_size * frame.indent, ' ');
+    /// proton: ends
 
     if (with())
     {

diff --git a/src/Parsers/ASTQueryWithOutput.cpp b/src/Parsers/ASTQueryWithOutput.cpp
@@ -25,9 +25,8 @@ void ASTQueryWithOutput::cloneOutputOptions(ASTQueryWithOutput & cloned) const
 void ASTQueryWithOutput::formatImpl(const FormatSettings & s, FormatState & state, FormatStateStacked frame) const
 {
     formatQueryImpl(s, state, frame);
-
     /// proton: starts
-    std::string indent_str = s.one_line ? "" : std::string(2u * frame.indent, ' ');
+    std::string indent_str = s.one_line ? "" : std::string(s.indent_size * frame.indent, ' ');
     /// proton: ends
 
     if (out_file)

diff --git a/src/Parsers/ASTSelectIntersectExceptQuery.cpp b/src/Parsers/ASTSelectIntersectExceptQuery.cpp
@@ -20,7 +20,9 @@ ASTPtr ASTSelectIntersectExceptQuery::clone() const
 
 void ASTSelectIntersectExceptQuery::formatImpl(const FormatSettings & settings, FormatState & state, FormatStateStacked frame) const
 {
-    std::string indent_str = settings.one_line ? "" : std::string(4 * frame.indent, ' ');
+    /// proton: starts
+    std::string indent_str = settings.one_line ? "" : std::string(settings.indent_size * frame.indent, ' ');
+    /// proton: ends
 
     for (ASTs::const_iterator it = children.begin(); it != children.end(); ++it)
     {