feat: introduce SKIPPING index (part 1) (#5155)

* skip index parser Signed-off-by: Ruihang Xia <waynestxia@gmail.com> * wip: sqlness Signed-off-by: Ruihang Xia <waynestxia@gmail.com> * impl show create part Signed-off-by: Ruihang Xia <waynestxia@gmail.com> * add empty line Signed-off-by: Ruihang Xia <waynestxia@gmail.com> * change keyword to SKIPPING INDEX Signed-off-by: Ruihang Xia <waynestxia@gmail.com> * rename local variables Signed-off-by: Ruihang Xia <waynestxia@gmail.com> --------- Signed-off-by: Ruihang Xia <waynestxia@gmail.com>
GreptimeTeam · Dec 16, 2024 · 5469832 · 5469832
1 parent 5ffda7e
commit 5469832
Show file tree

Hide file tree

Showing 14 changed files with 371 additions and 21 deletions.
diff --git a/src/api/src/v1/column_def.rs b/src/api/src/v1/column_def.rs
@@ -16,7 +16,7 @@ use std::collections::HashMap;
 
 use datatypes::schema::{
     ColumnDefaultConstraint, ColumnSchema, FulltextAnalyzer, FulltextOptions, COMMENT_KEY,
-    FULLTEXT_KEY, INVERTED_INDEX_KEY,
+    FULLTEXT_KEY, INVERTED_INDEX_KEY, SKIPPING_INDEX_KEY,
 };
 use greptime_proto::v1::Analyzer;
 use snafu::ResultExt;
@@ -29,6 +29,8 @@ use crate::v1::{ColumnDef, ColumnOptions, SemanticType};
 const FULLTEXT_GRPC_KEY: &str = "fulltext";
 /// Key used to store inverted index options in gRPC column options.
 const INVERTED_INDEX_GRPC_KEY: &str = "inverted_index";
+/// Key used to store skip index options in gRPC column options.
+const SKIPPING_INDEX_GRPC_KEY: &str = "skipping_index";
 
 /// Tries to construct a `ColumnSchema` from the given  `ColumnDef`.
 pub fn try_as_column_schema(column_def: &ColumnDef) -> Result<ColumnSchema> {
@@ -60,6 +62,9 @@ pub fn try_as_column_schema(column_def: &ColumnDef) -> Result<ColumnSchema> {
         if let Some(inverted_index) = options.options.get(INVERTED_INDEX_GRPC_KEY) {
             metadata.insert(INVERTED_INDEX_KEY.to_string(), inverted_index.clone());
         }
+        if let Some(skipping_index) = options.options.get(SKIPPING_INDEX_GRPC_KEY) {
+            metadata.insert(SKIPPING_INDEX_KEY.to_string(), skipping_index.clone());
+        }
     }
 
     ColumnSchema::new(&column_def.name, data_type.into(), column_def.is_nullable)
@@ -84,6 +89,11 @@ pub fn options_from_column_schema(column_schema: &ColumnSchema) -> Option<Column
             .options
             .insert(INVERTED_INDEX_GRPC_KEY.to_string(), inverted_index.clone());
     }
+    if let Some(skipping_index) = column_schema.metadata().get(SKIPPING_INDEX_KEY) {
+        options
+            .options
+            .insert(SKIPPING_INDEX_GRPC_KEY.to_string(), skipping_index.clone());
+    }
 
     (!options.options.is_empty()).then_some(options)
 }

diff --git a/src/datatypes/src/error.rs b/src/datatypes/src/error.rs
@@ -232,6 +232,12 @@ pub enum Error {
         #[snafu(implicit)]
         location: Location,
     },
+    #[snafu(display("Invalid skipping index option: {}", msg))]
+    InvalidSkippingIndexOption {
+        msg: String,
+        #[snafu(implicit)]
+        location: Location,
+    },
 }
 
 impl ErrorExt for Error {
@@ -252,7 +258,8 @@ impl ErrorExt for Error {
             | InvalidPrecisionOrScale { .. }
             | InvalidJson { .. }
             | InvalidVector { .. }
-            | InvalidFulltextOption { .. } => StatusCode::InvalidArguments,
+            | InvalidFulltextOption { .. }
+            | InvalidSkippingIndexOption { .. } => StatusCode::InvalidArguments,
 
             ValueExceedsPrecision { .. }
             | CastType { .. }

diff --git a/src/datatypes/src/schema.rs b/src/datatypes/src/schema.rs
@@ -28,10 +28,11 @@ use snafu::{ensure, ResultExt};
 use crate::error::{self, DuplicateColumnSnafu, Error, ProjectArrowSchemaSnafu, Result};
 use crate::prelude::ConcreteDataType;
 pub use crate::schema::column_schema::{
-    ColumnSchema, FulltextAnalyzer, FulltextOptions, Metadata,
+    ColumnSchema, FulltextAnalyzer, FulltextOptions, Metadata, SkippingIndexOptions,
     COLUMN_FULLTEXT_CHANGE_OPT_KEY_ENABLE, COLUMN_FULLTEXT_OPT_KEY_ANALYZER,
-    COLUMN_FULLTEXT_OPT_KEY_CASE_SENSITIVE, COMMENT_KEY, FULLTEXT_KEY, INVERTED_INDEX_KEY,
-    TIME_INDEX_KEY,
+    COLUMN_FULLTEXT_OPT_KEY_CASE_SENSITIVE, COLUMN_SKIPPING_INDEX_OPT_KEY_GRANULARITY,
+    COLUMN_SKIPPING_INDEX_OPT_KEY_TYPE, COMMENT_KEY, FULLTEXT_KEY, INVERTED_INDEX_KEY,
+    SKIPPING_INDEX_KEY, TIME_INDEX_KEY,
 };
 pub use crate::schema::constraint::ColumnDefaultConstraint;
 pub use crate::schema::raw::RawSchema;

diff --git a/src/datatypes/src/schema/column_schema.rs b/src/datatypes/src/schema/column_schema.rs
@@ -39,12 +39,20 @@ const DEFAULT_CONSTRAINT_KEY: &str = "greptime:default_constraint";
 pub const FULLTEXT_KEY: &str = "greptime:fulltext";
 /// Key used to store whether the column has inverted index in arrow field's metadata.
 pub const INVERTED_INDEX_KEY: &str = "greptime:inverted_index";
+/// Key used to store skip options in arrow field's metadata.
+pub const SKIPPING_INDEX_KEY: &str = "greptime:skipping_index";
 
 /// Keys used in fulltext options
 pub const COLUMN_FULLTEXT_CHANGE_OPT_KEY_ENABLE: &str = "enable";
 pub const COLUMN_FULLTEXT_OPT_KEY_ANALYZER: &str = "analyzer";
 pub const COLUMN_FULLTEXT_OPT_KEY_CASE_SENSITIVE: &str = "case_sensitive";
 
+/// Keys used in SKIPPING index options
+pub const COLUMN_SKIPPING_INDEX_OPT_KEY_GRANULARITY: &str = "granularity";
+pub const COLUMN_SKIPPING_INDEX_OPT_KEY_TYPE: &str = "type";
+
+pub const DEFAULT_GRANULARITY: u32 = 10240;
+
 /// Schema of a column, used as an immutable struct.
 #[derive(Clone, PartialEq, Eq, Serialize, Deserialize)]
 pub struct ColumnSchema {
@@ -298,6 +306,34 @@ impl ColumnSchema {
         );
         Ok(())
     }
+
+    /// Retrieves the skipping index options for the column.
+    pub fn skipping_index_options(&self) -> Result<Option<SkippingIndexOptions>> {
+        match self.metadata.get(SKIPPING_INDEX_KEY) {
+            None => Ok(None),
+            Some(json) => {
+                let options =
+                    serde_json::from_str(json).context(error::DeserializeSnafu { json })?;
+                Ok(Some(options))
+            }
+        }
+    }
+
+    pub fn with_skipping_options(mut self, options: SkippingIndexOptions) -> Result<Self> {
+        self.metadata.insert(
+            SKIPPING_INDEX_KEY.to_string(),
+            serde_json::to_string(&options).context(error::SerializeSnafu)?,
+        );
+        Ok(self)
+    }
+
+    pub fn set_skipping_options(&mut self, options: &SkippingIndexOptions) -> Result<()> {
+        self.metadata.insert(
+            SKIPPING_INDEX_KEY.to_string(),
+            serde_json::to_string(options).context(error::SerializeSnafu)?,
+        );
+        Ok(())
+    }
 }
 
 /// Column extended type set in column schema's metadata.
@@ -495,6 +531,76 @@ impl fmt::Display for FulltextAnalyzer {
     }
 }
 
+/// Skipping options for a column.
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize, Default, Visit, VisitMut)]
+#[serde(rename_all = "kebab-case")]
+pub struct SkippingIndexOptions {
+    /// The granularity of the skip index.
+    pub granularity: u32,
+    /// The type of the skip index.
+    #[serde(default)]
+    pub index_type: SkipIndexType,
+}
+
+impl fmt::Display for SkippingIndexOptions {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        write!(f, "granularity={}", self.granularity)?;
+        write!(f, ", index_type={}", self.index_type)?;
+        Ok(())
+    }
+}
+
+/// Skip index types.
+#[derive(Debug, Default, Clone, PartialEq, Eq, Serialize, Deserialize, Visit, VisitMut)]
+pub enum SkipIndexType {
+    #[default]
+    BloomFilter,
+}
+
+impl fmt::Display for SkipIndexType {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        match self {
+            SkipIndexType::BloomFilter => write!(f, "BLOOM"),
+        }
+    }
+}
+
+impl TryFrom<HashMap<String, String>> for SkippingIndexOptions {
+    type Error = Error;
+
+    fn try_from(options: HashMap<String, String>) -> Result<Self> {
+        // Parse granularity with default value 1
+        let granularity = match options.get(COLUMN_SKIPPING_INDEX_OPT_KEY_GRANULARITY) {
+            Some(value) => value.parse::<u32>().map_err(|_| {
+                error::InvalidSkippingIndexOptionSnafu {
+                    msg: format!("Invalid granularity: {value}, expected: positive integer"),
+                }
+                .build()
+            })?,
+            None => DEFAULT_GRANULARITY,
+        };
+
+        // Parse index type with default value BloomFilter
+        let index_type = match options.get(COLUMN_SKIPPING_INDEX_OPT_KEY_TYPE) {
+            Some(typ) => match typ.to_ascii_uppercase().as_str() {
+                "BLOOM" => SkipIndexType::BloomFilter,
+                _ => {
+                    return error::InvalidSkippingIndexOptionSnafu {
+                        msg: format!("Invalid index type: {typ}, expected: 'BLOOM'"),
+                    }
+                    .fail();
+                }
+            },
+            None => SkipIndexType::default(),
+        };
+
+        Ok(SkippingIndexOptions {
+            granularity,
+            index_type,
+        })
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use std::sync::Arc;

diff --git a/src/operator/src/statement/ddl.rs b/src/operator/src/statement/ddl.rs
@@ -271,7 +271,8 @@ impl StatementExecutor {
 
         table_info.ident.table_id = table_id;
 
-        let table_info = Arc::new(table_info.try_into().context(CreateTableInfoSnafu)?);
+        let table_info: Arc<TableInfo> =
+            Arc::new(table_info.try_into().context(CreateTableInfoSnafu)?);
         create_table.table_id = Some(api::v1::TableId { id: table_id });
 
         let table = DistTable::table(table_info);

diff --git a/src/query/src/error.rs b/src/query/src/error.rs
@@ -316,6 +316,13 @@ pub enum Error {
         #[snafu(implicit)]
         location: Location,
     },
+
+    #[snafu(display("Failed to get SKIPPING index options"))]
+    GetSkippingIndexOptions {
+        source: datatypes::error::Error,
+        #[snafu(implicit)]
+        location: Location,
+    },
 }
 
 impl ErrorExt for Error {
@@ -366,7 +373,9 @@ impl ErrorExt for Error {
             MissingTableMutationHandler { .. } => StatusCode::Unexpected,
             GetRegionMetadata { .. } => StatusCode::RegionNotReady,
             TableReadOnly { .. } => StatusCode::Unsupported,
-            GetFulltextOptions { source, .. } => source.status_code(),
+            GetFulltextOptions { source, .. } | GetSkippingIndexOptions { source, .. } => {
+                source.status_code()
+            }
         }
     }
 

diff --git a/src/query/src/sql/show_create_table.rs b/src/query/src/sql/show_create_table.rs
@@ -19,7 +19,8 @@ use std::collections::HashMap;
 use common_meta::SchemaOptions;
 use datatypes::schema::{
     ColumnDefaultConstraint, ColumnSchema, SchemaRef, COLUMN_FULLTEXT_OPT_KEY_ANALYZER,
-    COLUMN_FULLTEXT_OPT_KEY_CASE_SENSITIVE, COMMENT_KEY,
+    COLUMN_FULLTEXT_OPT_KEY_CASE_SENSITIVE, COLUMN_SKIPPING_INDEX_OPT_KEY_GRANULARITY,
+    COLUMN_SKIPPING_INDEX_OPT_KEY_TYPE, COMMENT_KEY,
 };
 use snafu::ResultExt;
 use sql::ast::{ColumnDef, ColumnOption, ColumnOptionDef, Expr, Ident, ObjectName};
@@ -32,7 +33,8 @@ use table::metadata::{TableInfoRef, TableMeta};
 use table::requests::{FILE_TABLE_META_KEY, TTL_KEY, WRITE_BUFFER_SIZE_KEY};
 
 use crate::error::{
-    ConvertSqlTypeSnafu, ConvertSqlValueSnafu, GetFulltextOptionsSnafu, Result, SqlSnafu,
+    ConvertSqlTypeSnafu, ConvertSqlValueSnafu, GetFulltextOptionsSnafu,
+    GetSkippingIndexOptionsSnafu, Result, SqlSnafu,
 };
 
 /// Generates CREATE TABLE options from given table metadata and schema-level options.
@@ -115,6 +117,23 @@ fn create_column(column_schema: &ColumnSchema, quote_style: char) -> Result<Colu
         extensions.fulltext_options = Some(map.into());
     }
 
+    if let Some(opt) = column_schema
+        .skipping_index_options()
+        .context(GetSkippingIndexOptionsSnafu)?
+    {
+        let map = HashMap::from([
+            (
+                COLUMN_SKIPPING_INDEX_OPT_KEY_GRANULARITY.to_string(),
+                opt.granularity.to_string(),
+            ),
+            (
+                COLUMN_SKIPPING_INDEX_OPT_KEY_TYPE.to_string(),
+                opt.index_type.to_string(),
+            ),
+        ]);
+        extensions.skipping_index_options = Some(map.into());
+    }
+
     Ok(Column {
         column_def: ColumnDef {
             name: Ident::with_quote(quote_style, name),
@@ -219,7 +238,7 @@ mod tests {
 
     use common_time::timestamp::TimeUnit;
     use datatypes::prelude::ConcreteDataType;
-    use datatypes::schema::{FulltextOptions, Schema, SchemaRef};
+    use datatypes::schema::{FulltextOptions, Schema, SchemaRef, SkippingIndexOptions};
     use table::metadata::*;
     use table::requests::{
         TableOptions, FILE_TABLE_FORMAT_KEY, FILE_TABLE_LOCATION_KEY, FILE_TABLE_META_KEY,
@@ -230,7 +249,12 @@ mod tests {
     #[test]
     fn test_show_create_table_sql() {
         let schema = vec![
-            ColumnSchema::new("id", ConcreteDataType::uint32_datatype(), true),
+            ColumnSchema::new("id", ConcreteDataType::uint32_datatype(), true)
+                .with_skipping_options(SkippingIndexOptions {
+                    granularity: 4096,
+                    ..Default::default()
+                })
+                .unwrap(),
             ColumnSchema::new("host", ConcreteDataType::string_datatype(), true)
                 .set_inverted_index(true),
             ColumnSchema::new("cpu", ConcreteDataType::float64_datatype(), true),
@@ -300,7 +324,7 @@ mod tests {
         assert_eq!(
             r#"
 CREATE TABLE IF NOT EXISTS "system_metrics" (
-  "id" INT UNSIGNED NULL,
+  "id" INT UNSIGNED NULL SKIPPING INDEX WITH(granularity = '4096', type = 'BLOOM'),
   "host" STRING NULL,
   "cpu" DOUBLE NULL,
   "disk" FLOAT NULL,

diff --git a/src/sql/src/error.rs b/src/sql/src/error.rs
@@ -326,6 +326,13 @@ pub enum Error {
         location: Location,
     },
 
+    #[snafu(display("Failed to set SKIPPING index option"))]
+    SetSkippingIndexOption {
+        source: datatypes::error::Error,
+        #[snafu(implicit)]
+        location: Location,
+    },
+
     #[snafu(display("Datatype error: {}", source))]
     Datatype {
         source: datatypes::error::Error,
@@ -375,7 +382,7 @@ impl ErrorExt for Error {
             ConvertSqlValue { .. } | ConvertValue { .. } => StatusCode::Unsupported,
 
             PermissionDenied { .. } => StatusCode::PermissionDenied,
-            SetFulltextOption { .. } => StatusCode::Unexpected,
+            SetFulltextOption { .. } | SetSkippingIndexOption { .. } => StatusCode::Unexpected,
         }
     }