compatibility mysql insert and select

.github/actions/test_stateless_cluster_linux/action.yml

@@ -28,6 +28,11 @@ runs:
       run: |
         chmod +x ./target/debug/databend-*
 
+    - name: Install python dependences
+      shell: bash
+      run: |
+        pip3 install --user boto3 "moto[all]" yapf shfmt-py mysql-connector pymysql sqlalchemy clickhouse_driver
+
     - name: Run Stateless Tests with Cluster mode
       shell: bash
       run: |

.github/actions/test_stateless_cluster_macos/action.yml

@@ -33,7 +33,7 @@ runs:
     - name: Install python dependences
       shell: bash
       run: |
-        pip3 install --user boto3 "moto[all]" yapf shfmt-py mysql-connector pymysql sqlalchemy
+        pip3 install --user boto3 "moto[all]" yapf shfmt-py mysql-connector pymysql sqlalchemy clickhouse_driver
 
     - name: Run Stateless Tests with Cluster mode
       shell: bash

.github/actions/test_stateless_standalone_linux/action.yml

@@ -28,6 +28,11 @@ runs:
       run: |
         chmod +x ./target/debug/databend-*
 
+    - name: Install python dependences
+      shell: bash
+      run: |
+        pip3 install --user boto3 "moto[all]" yapf shfmt-py mysql-connector pymysql sqlalchemy clickhouse_driver
+
     - name: Run Stateless Tests with Standalone mode, with embedded meta-store
       shell: bash
       run: |

.github/actions/test_stateless_standalone_macos/action.yml

@@ -28,7 +28,7 @@ runs:
     - name: Install python dependences
       shell: bash
       run: |
-        pip3 install --user boto3 "moto[all]" yapf shfmt-py mysql-connector pymysql sqlalchemy
+        pip3 install --user boto3 "moto[all]" yapf shfmt-py mysql-connector pymysql sqlalchemy clickhouse_driver
 
     - name: Set up file as executable
       shell: bash

common/ast/Cargo.toml

@@ -19,7 +19,7 @@ common-functions = { path = "../functions" }
 # TODO (andylokandy): Use the version from crates.io once
 # https://github.com/brendanzab/codespan/pull/331 is released.
 codespan-reporting = { git = "https://github.com/brendanzab/codespan", rev = "c84116f5" }
-sqlparser = { git = "https://github.com/datafuse-extras/sqlparser-rs", rev = "f88fb32" }
+sqlparser = { git = "https://github.com/datafuse-extras/sqlparser-rs", rev = "818c0f9" }
 
 # Crates.io dependencies
 async-trait = "0.1.53"

common/exception/Cargo.toml

@@ -26,4 +26,4 @@ tonic = "=0.6.2"
 
 # Github dependencies
 bincode = { git = "https://github.com/datafuse-extras/bincode", rev = "fd3f9ff" }
-sqlparser = { git = "https://github.com/datafuse-extras/sqlparser-rs", rev = "f88fb32" }
+sqlparser = { git = "https://github.com/datafuse-extras/sqlparser-rs", rev = "818c0f9" }

common/functions/Cargo.toml

@@ -44,7 +44,7 @@ serde_json = "1.0.79"
 sha1 = "0.10.1"
 sha2 = "0.10.2"
 simdutf8 = "0.1.4"
-sqlparser = { git = "https://github.com/datafuse-extras/sqlparser-rs", rev = "f88fb32" }
+sqlparser = { git = "https://github.com/datafuse-extras/sqlparser-rs", rev = "818c0f9" }
 strength_reduce = "0.2.3"
 twox-hash = "1.6.2"
 uuid = { version = "0.8.2", features = ["v4"] }

query/Cargo.toml

@@ -54,7 +54,7 @@ common-tracing = { path = "../common/tracing" }
 bincode = { git = "https://github.com/datafuse-extras/bincode", rev = "fd3f9ff" }
 opensrv-clickhouse = { git = "https://github.com/datafuselabs/opensrv", rev = "9690be9", package = "opensrv-clickhouse" }
 opensrv-mysql = { git = "https://github.com/datafuselabs/opensrv", rev = "967477f1", package = "opensrv-mysql" }
-sqlparser = { git = "https://github.com/datafuse-extras/sqlparser-rs", rev = "f88fb32" }
+sqlparser = { git = "https://github.com/datafuse-extras/sqlparser-rs", rev = "818c0f9" }
 
 # Crates.io dependencies
 ahash = "0.7.6"

query/src/servers/http/clickhouse_handler.rs

@@ -127,8 +127,11 @@ pub async fn clickhouse_handler_get(
         .map_err(InternalServerError)
 }
 
-fn try_parse_insert_formatted(sql: &str) -> Result<Option<(Format, Vec<DfStatement>)>> {
-    if let Ok((statements, _)) = DfParser::parse_sql(sql) {
+fn try_parse_insert_formatted(
+    sql: &str,
+    typ: SessionType,
+) -> Result<Option<(Format, Vec<DfStatement>)>> {
+    if let Ok((statements, _)) = DfParser::parse_sql(sql, typ) {
         if statements.is_empty() {
             return Ok(None);
         }
@@ -178,30 +181,33 @@ pub async fn clickhouse_handler_post(
     let mut sql = params.query;
 
     // Insert into format sql
-    let (plan, input_stream) =
-        if let Some((format, statements)) = try_parse_insert_formatted(&sql).map_err(BadRequest)? {
-            let plan = PlanParser::build_plan(statements, ctx.clone())
-                .await
-                .map_err(InternalServerError)?;
-            ctx.attach_query_str(&sql);
-
-            let input_stream = match format {
-                Format::NDJson => build_ndjson_stream(&plan, body).await.map_err(BadRequest)?,
-            };
-            (plan, Some(input_stream))
-        } else {
-            // Other sql
-            let body = body.into_string().await.map_err(BadRequest)?;
-            let sql = format!("{}\n{}", sql, body);
-            let (statements, _) = DfParser::parse_sql(&sql).map_err(BadRequest)?;
-
-            let plan = PlanParser::build_plan(statements, ctx.clone())
-                .await
-                .map_err(InternalServerError)?;
-            ctx.attach_query_str(&sql);
-
-            (plan, None)
+    let (plan, input_stream) = if let Some((format, statements)) =
+        try_parse_insert_formatted(&sql, ctx.get_current_session().get_type())
+            .map_err(BadRequest)?
+    {
+        let plan = PlanParser::build_plan(statements, ctx.clone())
+            .await
+            .map_err(InternalServerError)?;
+        ctx.attach_query_str(&sql);
+
+        let input_stream = match format {
+            Format::NDJson => build_ndjson_stream(&plan, body).await.map_err(BadRequest)?,
         };
+        (plan, Some(input_stream))
+    } else {
+        // Other sql
+        let body = body.into_string().await.map_err(BadRequest)?;
+        let sql = format!("{}\n{}", sql, body);
+        let (statements, _) =
+            DfParser::parse_sql(&sql, ctx.get_current_session().get_type()).map_err(BadRequest)?;
+
+        let plan = PlanParser::build_plan(statements, ctx.clone())
+            .await
+            .map_err(InternalServerError)?;
+        ctx.attach_query_str(&sql);
+
+        (plan, None)
+    };
 
     execute(ctx, plan, input_stream)
         .await

query/src/sessions/session_type.rs

@@ -24,11 +24,15 @@ pub enum SessionType {
     FlightRPC,
     HTTPAPI(String),
     Test,
+    Fuzz,
 }
 
 impl SessionType {
     pub fn is_user_session(&self) -> bool {
-        !matches!(self, SessionType::HTTPAPI(_) | SessionType::Test)
+        !matches!(
+            self,
+            SessionType::HTTPAPI(_) | SessionType::Test | SessionType::Fuzz
+        )
     }
 }
 
@@ -43,6 +47,7 @@ impl fmt::Display for SessionType {
             SessionType::Test => "Test".to_string(),
             SessionType::FlightRPC => "FlightRPC".to_string(),
             SessionType::HTTPAPI(usage) => format!("HTTPAPI({})", usage),
+            SessionType::Fuzz => "Fuzz".to_string(),
         };
         write!(f, "{}", name)
     }

query/src/sql/plan_parser.rs

@@ -35,15 +35,15 @@ pub struct PlanParser;
 
 impl PlanParser {
     pub async fn parse(ctx: Arc<QueryContext>, query: &str) -> Result<PlanNode> {
-        let (statements, _) = DfParser::parse_sql(query)?;
+        let (statements, _) = DfParser::parse_sql(query, ctx.get_current_session().get_type())?;
         PlanParser::build_plan(statements, ctx).await
     }
 
     pub async fn parse_with_hint(
         query: &str,
         ctx: Arc<QueryContext>,
     ) -> (Result<PlanNode>, Vec<DfHint>) {
-        match DfParser::parse_sql(query) {
+        match DfParser::parse_sql(query, ctx.get_current_session().get_type()) {
             Err(cause) => (Err(cause), vec![]),
             Ok((statements, hints)) => (PlanParser::build_plan(statements, ctx).await, hints),
         }

query/src/sql/sql_parser.rs

@@ -24,6 +24,7 @@ use sqlparser::ast::Value;
 use sqlparser::dialect::keywords::Keyword;
 use sqlparser::dialect::Dialect;
 use sqlparser::dialect::GenericDialect;
+use sqlparser::dialect::MySqlDialect;
 use sqlparser::dialect::SnowflakeDialect;
 use sqlparser::parser::Parser;
 use sqlparser::parser::ParserError;
@@ -32,6 +33,7 @@ use sqlparser::tokenizer::Tokenizer;
 use sqlparser::tokenizer::Whitespace;
 
 use super::statements::DfShowRoles;
+use crate::sessions::SessionType;
 use crate::sql::statements::DfShowEngines;
 use crate::sql::statements::DfShowMetrics;
 use crate::sql::statements::DfShowProcessList;
@@ -55,12 +57,6 @@ pub struct DfParser<'a> {
 }
 
 impl<'a> DfParser<'a> {
-    /// Parse the specified tokens
-    pub fn new(sql: &'a str) -> Result<Self, ParserError> {
-        let dialect = &GenericDialect {};
-        DfParser::new_with_dialect(sql, dialect)
-    }
-
     /// Parse the specified tokens with dialect
     pub fn new_with_dialect(sql: &'a str, dialect: &'a dyn Dialect) -> Result<Self, ParserError> {
         let mut tokenizer = Tokenizer::new(dialect, sql);
@@ -73,12 +69,26 @@ impl<'a> DfParser<'a> {
     }
 
     /// Parse a SQL statement and produce a set of statements with dialect
-    pub fn parse_sql(sql: &'a str) -> Result<(Vec<DfStatement<'a>>, Vec<DfHint>), ErrorCode> {
-        let dialect = &GenericDialect {};
-        let start = Instant::now();
-        let result = DfParser::parse_sql_with_dialect(sql, dialect)?;
-        histogram!(super::metrics::METRIC_PARSER_USEDTIME, start.elapsed());
-        Ok(result)
+    pub fn parse_sql(
+        sql: &'a str,
+        typ: SessionType,
+    ) -> Result<(Vec<DfStatement<'a>>, Vec<DfHint>), ErrorCode> {
+        match typ {
+            SessionType::MySQL => {
+                let dialect = &MySqlDialect {};
+                let start = Instant::now();
+                let result = DfParser::parse_sql_with_dialect(sql, dialect)?;
+                histogram!(super::metrics::METRIC_PARSER_USEDTIME, start.elapsed());
+                Ok(result)
+            }
+            _ => {
+                let dialect = &GenericDialect {};
+                let start = Instant::now();
+                let result = DfParser::parse_sql_with_dialect(sql, dialect)?;
+                histogram!(super::metrics::METRIC_PARSER_USEDTIME, start.elapsed());
+                Ok(result)
+            }
+        }
     }
 
     /// Parse a SQL statement and produce a set of statements

query/src/sql/statements/analyzer_expr.rs

@@ -38,6 +38,7 @@ use sqlparser::ast::Value;
 
 use crate::procedures::ContextFunction;
 use crate::sessions::QueryContext;
+use crate::sessions::SessionType;
 use crate::sql::statements::analyzer_value_expr::ValueExprAnalyzer;
 use crate::sql::statements::AnalyzableStatement;
 use crate::sql::statements::AnalyzedResult;
@@ -61,7 +62,11 @@ impl ExpressionAnalyzer {
         // Build RPN for expr. Because async function unsupported recursion
         for rpn_item in &ExprRPNBuilder::build(self.context.clone(), expr).await? {
             match rpn_item {
-                ExprRPNItem::Value(v) => Self::analyze_value(v, &mut stack)?,
+                ExprRPNItem::Value(v) => Self::analyze_value(
+                    v,
+                    &mut stack,
+                    self.context.get_current_session().get_type(),
+                )?,
                 ExprRPNItem::Identifier(v) => self.analyze_identifier(v, &mut stack)?,
                 ExprRPNItem::QualifiedIdentifier(v) => self.analyze_identifiers(v, &mut stack)?,
                 ExprRPNItem::Function(v) => self.analyze_function(v, &mut stack)?,
@@ -94,8 +99,8 @@ impl ExpressionAnalyzer {
         }
     }
 
-    fn analyze_value(value: &Value, args: &mut Vec<Expression>) -> Result<()> {
-        args.push(ValueExprAnalyzer::analyze(value)?);
+    fn analyze_value(value: &Value, args: &mut Vec<Expression>, typ: SessionType) -> Result<()> {
+        args.push(ValueExprAnalyzer::analyze(value, typ)?);
         Ok(())
     }
 
@@ -205,7 +210,10 @@ impl ExpressionAnalyzer {
         let mut parameters = Vec::with_capacity(info.parameters.len());
 
         for parameter in &info.parameters {
-            match ValueExprAnalyzer::analyze(parameter)? {
+            match ValueExprAnalyzer::analyze(
+                parameter,
+                self.context.get_current_session().get_type(),
+            )? {
                 Expression::Literal { value, .. } => {
                     parameters.push(value);
                 }

query/src/sql/statements/analyzer_value_expr.rs

@@ -19,16 +19,32 @@ use common_planners::Expression;
 use sqlparser::ast::DateTimeField;
 use sqlparser::ast::Value;
 
+use crate::sessions::SessionType;
+
 pub struct ValueExprAnalyzer;
 
 impl ValueExprAnalyzer {
-    pub fn analyze(value: &Value) -> Result<Expression> {
+    pub fn analyze(value: &Value, typ: SessionType) -> Result<Expression> {
         match value {
             Value::Null => Self::analyze_null_value(),
             Value::Boolean(value) => Self::analyze_bool_value(value),
             Value::Number(value, _) => Self::analyze_number_value(value, None),
             Value::HexStringLiteral(value) => Self::analyze_number_value(value, Some(16)),
             Value::SingleQuotedString(value) => Self::analyze_string_value(value),
+            Value::DoubleQuotedString(value) => {
+                // Only MySQL dialect Support insert SQL like this:
+                // INSERT INTO t VALUES("val");
+                // https://github.com/datafuselabs/databend/issues/4861
+                if let SessionType::MySQL = typ {
+                    Self::analyze_string_value(value)
+                } else {
+                    Result::Err(ErrorCode::SyntaxException(format!(
+                        "Unsupported value expression: {}, type: {:?}",
+                        value,
+                        Value::DoubleQuotedString(value.to_string())
+                    )))
+                }
+            }
             Value::Interval {
                 leading_precision: Some(_),
                 ..

query/src/sql/statements/query/query_schema_joined_analyzer.rs

@@ -105,7 +105,8 @@ impl JoinedSchemaAnalyzer {
 
         if tbl_info.engine() == VIEW_ENGINE {
             if let Some(query) = tbl_info.options().get(QUERY) {
-                let (statements, _) = DfParser::parse_sql(query.as_str())?;
+                let (statements, _) =
+                    DfParser::parse_sql(query.as_str(), self.ctx.get_current_session().get_type())?;
                 if statements.len() == 1 {
                     if let DfStatement::Query(subquery) = &statements[0] {
                         if let AnalyzedResult::SelectQuery(state) =