support geometry type for create table

src/common/io/src/geometry.rs

@@ -19,22 +19,25 @@ use geozero::CoordDimensions;
 use geozero::ToWkb;
 use wkt::TryFromWkt;
 
-pub fn parse_to_ewkb(buf: &[u8]) -> Result<Vec<u8>> {
+pub fn parse_to_ewkb(buf: &[u8], srid: Option<i32>) -> Result<Vec<u8>> {
     let wkt = std::str::from_utf8(buf).map_err(|e| ErrorCode::GeometryError(e.to_string()))?;
-    let mut srid: Option<i32> = None;
     let input_wkt = wkt.trim().to_ascii_uppercase();
 
     let parts: Vec<&str> = input_wkt.split(';').collect();
 
-    if input_wkt.starts_with("SRID=") && parts.len() == 2 {
-        srid = Some(parts[0].replace("SRID=", "").parse()?);
-    }
+    let parsed_srid: Option<i32> = srid.or_else(|| {
+        if input_wkt.starts_with("SRID=") && parts.len() == 2 {
+            parts[0].replace("SRID=", "").parse().ok()
+        } else {
+            None
+        }
+    });
 
     let geo_part = if parts.len() == 2 { parts[1] } else { parts[0] };
 
     let geom: Geometry<f64> = Geometry::try_from_wkt_str(geo_part)
         .map_err(|e| ErrorCode::GeometryError(e.to_string()))?;
 
-    geom.to_ewkb(CoordDimensions::xy(), srid)
+    geom.to_ewkb(CoordDimensions::xy(), parsed_srid)
         .map_err(ErrorCode::from)
 }

src/query/ast/src/ast/expr.rs

@@ -322,6 +322,7 @@ pub enum TypeName {
         fields_type: Vec<TypeName>,
     },
     Variant,
+    Geometry,
     Nullable(Box<TypeName>),
     NotNull(Box<TypeName>),
 }
@@ -899,6 +900,9 @@ impl Display for TypeName {
             TypeName::Variant => {
                 write!(f, "VARIANT")?;
             }
+            TypeName::Geometry => {
+                write!(f, "GEOMETRY")?;
+            }
             TypeName::Nullable(ty) => {
                 write!(f, "{} NULL", ty)?;
             }

src/query/ast/src/parser/expr.rs

@@ -1586,6 +1586,7 @@ pub fn type_name(i: Input) -> IResult<TypeName> {
         rule! { ( STRING | VARCHAR | CHAR | CHARACTER | TEXT ) ~ ( "(" ~ ^#literal_u64 ~ ^")" )? },
     );
     let ty_variant = value(TypeName::Variant, rule! { VARIANT | JSON });
+    let ty_geometry = value(TypeName::Geometry, rule! { GEOMETRY });
     map_res(
         alt((
             rule! {
@@ -1614,6 +1615,7 @@ pub fn type_name(i: Input) -> IResult<TypeName> {
             | #ty_binary
             | #ty_string
             | #ty_variant
+            | #ty_geometry
             | #ty_nullable
             ) ~ #nullable? : "type name" },
         )),

src/query/ast/src/parser/token.rs

@@ -597,6 +597,8 @@ pub enum TokenKind {
     FUSE,
     #[token("GENERATED", ignore(ascii_case))]
     GENERATED,
+    #[token("GEOMETRY", ignore(ascii_case))]
+    GEOMETRY,
     #[token("GLOBAL", ignore(ascii_case))]
     GLOBAL,
     #[token("GRAPH", ignore(ascii_case))]

src/query/ast/tests/it/testdata/expr-error.txt

@@ -30,7 +30,7 @@ error:
   --> SQL:1:14
   |
 1 | CAST(col1 AS foo)
-  | ----         ^^^ unexpected `foo`, expecting `BOOL`, `FLOAT`, `BOOLEAN`, `FLOAT32`, `FLOAT64`, `BLOB`, `JSON`, `DOUBLE`, `LONGBLOB`, `UINT8`, `TINYINT`, `UINT16`, `SMALLINT`, `UINT32`, `INT`, `INTEGER`, `UINT64`, `UNSIGNED`, `BIGINT`, `INT8`, `INT16`, `INT32`, `INT64`, `SIGNED`, `DECIMAL`, `ARRAY`, `MAP`, `BITMAP`, `TUPLE`, `DATE`, `DATETIME`, `TIMESTAMP`, `BINARY`, `VARBINARY`, `MEDIUMBLOB`, `TINYBLOB`, `STRING`, `VARCHAR`, `CHAR`, `CHARACTER`, `TEXT`, `VARIANT`, or `NULLABLE`
+  | ----         ^^^ unexpected `foo`, expecting `BOOL`, `FLOAT`, `BOOLEAN`, `FLOAT32`, `FLOAT64`, `BLOB`, `JSON`, `DOUBLE`, `LONGBLOB`, `GEOMETRY`, `UINT8`, `TINYINT`, `UINT16`, `SMALLINT`, `UINT32`, `INT`, `INTEGER`, `UINT64`, `UNSIGNED`, `BIGINT`, `INT8`, `INT16`, `INT32`, `INT64`, `SIGNED`, `DECIMAL`, `ARRAY`, `MAP`, `BITMAP`, `TUPLE`, `DATE`, `DATETIME`, `TIMESTAMP`, `BINARY`, `VARBINARY`, `MEDIUMBLOB`, `TINYBLOB`, `STRING`, `VARCHAR`, `CHAR`, `CHARACTER`, `TEXT`, `VARIANT`, or `NULLABLE`
   | |             
   | while parsing `CAST(... AS ...)`
   | while parsing expression

src/query/ast/tests/it/testdata/statement-error.txt

@@ -29,7 +29,7 @@ error:
   --> SQL:1:19
   |
 1 | create table a (c varch)
-  | ------          - ^^^^^ unexpected `varch`, expecting `VARCHAR`, `CHAR`, `VARIANT`, `CHARACTER`, `VARBINARY`, `ARRAY`, `BINARY`, `MAP`, `DATE`, `STRING`, `FLOAT32`, `FLOAT64`, `DECIMAL`, `SMALLINT`, `DATETIME`, `NULLABLE`, `BOOLEAN`, `BOOL`, `UINT8`, `TINYINT`, `UINT16`, `UINT32`, `INT`, `INTEGER`, `UINT64`, `UNSIGNED`, `BIGINT`, `INT8`, `INT16`, `INT32`, `INT64`, `SIGNED`, `FLOAT`, `DOUBLE`, `BITMAP`, `TUPLE`, `TIMESTAMP`, `LONGBLOB`, `MEDIUMBLOB`, `TINYBLOB`, `BLOB`, `TEXT`, or `JSON`
+  | ------          - ^^^^^ unexpected `varch`, expecting `VARCHAR`, `CHAR`, `VARIANT`, `CHARACTER`, `VARBINARY`, `ARRAY`, `BINARY`, `MAP`, `DATE`, `STRING`, `FLOAT32`, `FLOAT64`, `DECIMAL`, `SMALLINT`, `DATETIME`, `NULLABLE`, `BOOLEAN`, `BOOL`, `UINT8`, `TINYINT`, `UINT16`, `UINT32`, `INT`, `INTEGER`, `UINT64`, `UNSIGNED`, `BIGINT`, `INT8`, `INT16`, `INT32`, `INT64`, `SIGNED`, `FLOAT`, `DOUBLE`, `BITMAP`, `TUPLE`, `TIMESTAMP`, `LONGBLOB`, `MEDIUMBLOB`, `TINYBLOB`, `BLOB`, `TEXT`, `JSON`, or `GEOMETRY`
   | |               |  
   | |               while parsing `<column name> <type> [DEFAULT <expr>] [AS (<expr>) VIRTUAL] [AS (<expr>) STORED] [COMMENT '<comment>']`
   | while parsing `CREATE [OR REPLACE] TABLE [IF NOT EXISTS] [<database>.]<table> [<source>] [<table_options>]`
@@ -42,7 +42,7 @@ error:
   --> SQL:1:25
   |
 1 | create table a (c tuple())
-  | ------          - ----- ^ unexpected `)`, expecting `BOOLEAN`, `BOOL`, `UINT8`, `TINYINT`, `UINT16`, `SMALLINT`, `UINT32`, `INT`, `INTEGER`, `UINT64`, `UNSIGNED`, `BIGINT`, `INT8`, `INT16`, `INT32`, `INT64`, `SIGNED`, `FLOAT32`, `FLOAT`, `FLOAT64`, `DOUBLE`, `DECIMAL`, `ARRAY`, `MAP`, `BITMAP`, `TUPLE`, `DATE`, `DATETIME`, `TIMESTAMP`, `BINARY`, `VARBINARY`, `LONGBLOB`, `MEDIUMBLOB`, `TINYBLOB`, `BLOB`, `STRING`, `VARCHAR`, `CHAR`, `CHARACTER`, `TEXT`, `VARIANT`, `JSON`, `NULLABLE`, <Ident>, or <QuotedString>
+  | ------          - ----- ^ unexpected `)`, expecting `BOOLEAN`, `BOOL`, `UINT8`, `TINYINT`, `UINT16`, `SMALLINT`, `UINT32`, `INT`, `INTEGER`, `UINT64`, `UNSIGNED`, `BIGINT`, `INT8`, `INT16`, `INT32`, `INT64`, `SIGNED`, `FLOAT32`, `FLOAT`, `FLOAT64`, `DOUBLE`, `DECIMAL`, `ARRAY`, `MAP`, `BITMAP`, `TUPLE`, `DATE`, `DATETIME`, `TIMESTAMP`, `BINARY`, `VARBINARY`, `LONGBLOB`, `MEDIUMBLOB`, `TINYBLOB`, `BLOB`, `STRING`, `VARCHAR`, `CHAR`, `CHARACTER`, `TEXT`, `VARIANT`, `JSON`, `GEOMETRY`, `NULLABLE`, <Ident>, or <QuotedString>
   | |               | |      
   | |               | while parsing type name
   | |               while parsing `<column name> <type> [DEFAULT <expr>] [AS (<expr>) VIRTUAL] [AS (<expr>) STORED] [COMMENT '<comment>']`
@@ -70,7 +70,7 @@ error:
   --> SQL:1:38
   |
 1 | create table a (b tuple(c int, uint64));
-  | ------          - -----              ^ unexpected `)`, expecting `BOOLEAN`, `BOOL`, `UINT8`, `TINYINT`, `UINT16`, `SMALLINT`, `UINT32`, `INT`, `INTEGER`, `UINT64`, `UNSIGNED`, `BIGINT`, `INT8`, `INT16`, `INT32`, `INT64`, `SIGNED`, `FLOAT32`, `FLOAT`, `FLOAT64`, `DOUBLE`, `DECIMAL`, `ARRAY`, `MAP`, `BITMAP`, `TUPLE`, `DATE`, `DATETIME`, `TIMESTAMP`, `BINARY`, `VARBINARY`, `LONGBLOB`, `MEDIUMBLOB`, `TINYBLOB`, `BLOB`, `STRING`, `VARCHAR`, `CHAR`, `CHARACTER`, `TEXT`, `VARIANT`, `JSON`, or `NULLABLE`
+  | ------          - -----              ^ unexpected `)`, expecting `BOOLEAN`, `BOOL`, `UINT8`, `TINYINT`, `UINT16`, `SMALLINT`, `UINT32`, `INT`, `INTEGER`, `UINT64`, `UNSIGNED`, `BIGINT`, `INT8`, `INT16`, `INT32`, `INT64`, `SIGNED`, `FLOAT32`, `FLOAT`, `FLOAT64`, `DOUBLE`, `DECIMAL`, `ARRAY`, `MAP`, `BITMAP`, `TUPLE`, `DATE`, `DATETIME`, `TIMESTAMP`, `BINARY`, `VARBINARY`, `LONGBLOB`, `MEDIUMBLOB`, `TINYBLOB`, `BLOB`, `STRING`, `VARCHAR`, `CHAR`, `CHARACTER`, `TEXT`, `VARIANT`, `JSON`, `GEOMETRY`, or `NULLABLE`
   | |               | |                   
   | |               | while parsing TUPLE(<name> <type>, ...)
   | |               | while parsing type name

src/query/formats/src/field_decoder/fast_values.rs

@@ -490,7 +490,7 @@ impl FastFieldDecoderValues {
     ) -> Result<()> {
         let mut buf = Vec::new();
         self.read_string_inner(reader, &mut buf, positions)?;
-        let geom = parse_to_ewkb(&buf)?;
+        let geom = parse_to_ewkb(&buf, None)?;
         column.put_slice(geom.as_bytes());
         column.commit_row();
         Ok(())

src/query/formats/src/field_decoder/json_ast.rs

@@ -330,7 +330,7 @@ impl FieldJsonAstDecoder {
     fn read_geometry(&self, column: &mut BinaryColumnBuilder, value: &Value) -> Result<()> {
         match value {
             Value::String(v) => {
-                let geom = parse_to_ewkb(v.as_bytes())?;
+                let geom = parse_to_ewkb(v.as_bytes(), None)?;
                 column.put_slice(&geom);
                 column.commit_row();
                 Ok(())

src/query/formats/src/field_decoder/nested.rs

@@ -327,7 +327,7 @@ impl NestedValues {
     ) -> Result<()> {
         let mut buf = Vec::new();
         self.read_string_inner(reader, &mut buf)?;
-        let geom = parse_to_ewkb(&buf)?;
+        let geom = parse_to_ewkb(&buf, None)?;
         column.put_slice(geom.as_bytes());
         column.commit_row();
         Ok(())

src/query/formats/src/field_decoder/separated_text.rs

@@ -345,7 +345,7 @@ impl SeparatedTextDecoder {
     }
 
     fn read_geometry(&self, column: &mut BinaryColumnBuilder, data: &[u8]) -> Result<()> {
-        let geom = parse_to_ewkb(data)?;
+        let geom = parse_to_ewkb(data, None)?;
         column.put_slice(geom.as_bytes());
         column.commit_row();
         Ok(())

src/query/formats/src/field_encoder/values.rs

@@ -32,9 +32,9 @@ use databend_common_io::constants::NAN_BYTES_LOWER;
 use databend_common_io::constants::NAN_BYTES_SNAKE;
 use databend_common_io::constants::NULL_BYTES_UPPER;
 use databend_common_io::constants::TRUE_BYTES_NUM;
-use geozero::wkb::Ewkb;
-use geozero::CoordDimensions;
-use geozero::ToWkb;
+use geozero::wkb::FromWkb;
+use geozero::wkb::WkbDialect;
+use geozero::wkt::Ewkt;
 use lexical_core::ToLexical;
 use micromarshal::Marshal;
 use micromarshal::Unmarshal;
@@ -296,10 +296,9 @@ impl FieldEncoderValues {
         in_nested: bool,
     ) {
         let v = unsafe { column.index_unchecked(row_index) };
-        let s = Ewkb(v.to_vec())
-            .to_ewkb(CoordDimensions::xy(), None)
-            .unwrap();
-        self.write_string_inner(&s, out_buf, in_nested);
+        let mut data_cursor = std::io::Cursor::new(v);
+        let s = Ewkt::from_wkb(&mut data_cursor, WkbDialect::Ewkb).unwrap();
+        self.write_string_inner(s.0.as_bytes(), out_buf, in_nested);
     }
 
     fn write_array<T: ValueType>(

src/query/functions/Cargo.toml

@@ -51,9 +51,11 @@ ordered-float = { workspace = true, features = [
     "serde",
     "rand",
 ] }
+proj = { version = "0.27.2", features = ["geo-types", "bundled_proj"] }
 rand = { workspace = true }
 regex = { workspace = true }
 roaring = "0.10.1"
+scroll = "0.12.0"
 sha1 = "0.10.5"
 sha2 = "0.10.6"
 simdutf8 = "0.1.4"