feat(fuzz): add create logical table target

.github/workflows/develop.yml

@@ -130,7 +130,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        target: [ "fuzz_create_table", "fuzz_alter_table", "fuzz_create_database" ]
+        target: [ "fuzz_create_table", "fuzz_alter_table", "fuzz_create_database", "fuzz_create_logical_table" ]
     steps:
       - uses: actions/checkout@v4
       - uses: arduino/setup-protoc@v3

tests-fuzz/Cargo.toml

@@ -50,6 +50,13 @@ test = false
 bench = false
 doc = false
 
+[[bin]]
+name = "fuzz_create_logical_table"
+path = "targets/fuzz_create_logical_table.rs"
+test = false
+bench = false
+doc = false
+
 [[bin]]
 name = "fuzz_insert"
 path = "targets/fuzz_insert.rs"

tests-fuzz/src/generator/create_expr.rs

@@ -14,6 +14,7 @@
 
 use std::collections::HashMap;
 
+use datatypes::data_type::ConcreteDataType;
 use datatypes::value::Value;
 use derive_builder::Builder;
 use partition::partition::{PartitionBound, PartitionDef};
@@ -22,14 +23,19 @@ use rand::Rng;
 use snafu::{ensure, ResultExt};
 
 use super::Generator;
+use crate::context::TableContextRef;
 use crate::error::{self, Error, Result};
-use crate::fake::{random_capitalize_map, MappedGenerator, WordGenerator};
+use crate::fake::{
+    merge_two_word_map_fn, random_capitalize_map, uppercase_and_keyword_backtick_map,
+    MappedGenerator, WordGenerator,
+};
 use crate::generator::{ColumnOptionGenerator, ConcreteDataTypeGenerator, Random};
-use crate::ir::create_expr::{CreateDatabaseExprBuilder, CreateTableExprBuilder};
+use crate::ir::create_expr::{ColumnOption, CreateDatabaseExprBuilder, CreateTableExprBuilder};
 use crate::ir::{
     column_options_generator, generate_columns, generate_random_value,
-    partible_column_options_generator, ts_column_options_generator, ColumnTypeGenerator,
-    CreateDatabaseExpr, CreateTableExpr, Ident, PartibleColumnTypeGenerator, TsColumnTypeGenerator,
+    partible_column_options_generator, primary_key_column_options_generator,
+    ts_column_options_generator, Column, ColumnTypeGenerator, CreateDatabaseExpr, CreateTableExpr,
+    Ident, PartibleColumnTypeGenerator, StringColumnTypeGenerator, TsColumnTypeGenerator,
 };
 
 #[derive(Builder)]
@@ -200,6 +206,116 @@ impl<R: Rng + 'static> Generator<CreateTableExpr, R> for CreateTableExprGenerato
     }
 }
 
+/// Generate a physical table with 2 columns: ts of TimestampType::Millisecond as time index and val of Float64Type.
+#[derive(Builder)]
+#[builder(pattern = "owned")]
+pub struct CreatePhysicalTableExprGenerator<R: Rng + 'static> {
+    #[builder(default = "Box::new(WordGenerator)")]
+    name_generator: Box<dyn Random<Ident, R>>,
+}
+
+impl<R: Rng + 'static> Generator<CreateTableExpr, R> for CreatePhysicalTableExprGenerator<R> {
+    type Error = Error;
+
+    fn generate(&self, rng: &mut R) -> Result<CreateTableExpr> {
+        let if_not_exists = rng.gen_bool(0.5);
+
+        Ok(CreateTableExpr {
+            table_name: self.name_generator.gen(rng),
+            columns: vec![
+                Column {
+                    name: Ident::new("ts"),
+                    column_type: ConcreteDataType::timestamp_millisecond_datatype(),
+                    options: vec![ColumnOption::TimeIndex],
+                },
+                Column {
+                    name: Ident::new("val"),
+                    column_type: ConcreteDataType::float64_datatype(),
+                    options: vec![],
+                },
+            ],
+            if_not_exists,
+            partition: None,
+            engine: "metric".to_string(),
+            options: [("physical_metric_table".to_string(), "".into())].into(),
+            primary_keys: vec![],
+        })
+    }
+}
+
+/// Generate a logical table based on an existing physical table.
+#[derive(Builder)]
+#[builder(pattern = "owned")]
+pub struct CreateLogicalTableExprGenerator<R: Rng + 'static> {
+    physical_table_ctx: TableContextRef,
+    #[builder(default = "Box::new(WordGenerator)")]
+    name_generator: Box<dyn Random<Ident, R>>,
+    labels: usize,
+}
+
+impl<R: Rng + 'static> Generator<CreateTableExpr, R> for CreateLogicalTableExprGenerator<R> {
+    type Error = Error;
+
+    fn generate(&self, rng: &mut R) -> Result<CreateTableExpr> {
+        // Currently we mock the usage of GreptimeDB as Prometheus' backend, the physical table must have two columns.
+        ensure!(
+            self.physical_table_ctx.columns.len() == 2,
+            error::UnexpectedSnafu {
+                violated: "The physical table must have two columns"
+            }
+        );
+
+        // Generates the logical table columns based on the physical table.
+        let physical_table_name = self.physical_table_ctx.name.to_string().replace('`', "");
+        let table_generator = CreateTableExprGeneratorBuilder::default()
+            .name_generator(Box::new(MappedGenerator::new(
+                WordGenerator,
+                merge_two_word_map_fn(random_capitalize_map, uppercase_and_keyword_backtick_map),
+            )))
+            .columns(self.labels)
+            .engine("metric")
+            .column_type_generator(Box::new(StringColumnTypeGenerator))
+            .column_options_generator(Box::new(primary_key_column_options_generator))
+            .with_clause([("on_physical_table".to_string(), physical_table_name)])
+            .build()
+            .unwrap();
+
+        let mut table = table_generator.generate(rng)?;
+        table.table_name = self
+            .physical_table_ctx
+            .generate_unique_table_name(rng, self.name_generator.as_ref());
+
+        let logical_ts = table.columns.iter().position(|column| {
+            column
+                .options
+                .iter()
+                .any(|option| option == &ColumnOption::TimeIndex)
+        });
+        table.columns.remove(logical_ts.unwrap());
+        table.columns.iter_mut().for_each(|column| {
+            if column.column_type == ConcreteDataType::string_datatype() {
+                column
+                    .options
+                    .retain(|option| option == &ColumnOption::PrimaryKey);
+            }
+        });
+        table
+            .columns
+            .extend(self.physical_table_ctx.columns.clone());
+
+        let mut primary_keys = vec![];
+        for (idx, column) in table.columns.iter().enumerate() {
+            if column.is_primary_key() {
+                primary_keys.push(idx);
+            }
+        }
+        primary_keys.shuffle(rng);
+        table.primary_keys = primary_keys;
+
+        Ok(table)
+    }
+}
+
 #[derive(Builder)]
 #[builder(default, pattern = "owned")]
 pub struct CreateDatabaseExprGenerator<R: Rng + 'static> {
@@ -236,10 +352,14 @@ impl<R: Rng + 'static> Generator<CreateDatabaseExpr, R> for CreateDatabaseExprGe
 
 #[cfg(test)]
 mod tests {
+    use std::sync::Arc;
+
+    use datatypes::data_type::ConcreteDataType;
     use datatypes::value::Value;
     use rand::SeedableRng;
 
     use super::*;
+    use crate::context::TableContext;
 
     #[test]
     fn test_float64() {
@@ -296,6 +416,91 @@ mod tests {
         assert_eq!(expected, serialized);
     }
 
+    #[test]
+    fn test_create_logical_table_expr_generator() {
+        let mut rng = rand::thread_rng();
+
+        let physical_table_expr = CreatePhysicalTableExprGeneratorBuilder::default()
+            .build()
+            .unwrap()
+            .generate(&mut rng)
+            .unwrap();
+        assert_eq!(physical_table_expr.engine, "metric");
+        assert_eq!(physical_table_expr.columns.len(), 2);
+
+        let physical_ts = physical_table_expr.columns.iter().position(|column| {
+            column
+                .options
+                .iter()
+                .any(|option| option == &ColumnOption::TimeIndex)
+        });
+        let physical_ts_name = physical_table_expr.columns[physical_ts.unwrap()]
+            .name
+            .value
+            .to_string();
+
+        let physical_table_ctx = Arc::new(TableContext::from(&physical_table_expr));
+
+        let logical_table_expr = CreateLogicalTableExprGeneratorBuilder::default()
+            .physical_table_ctx(physical_table_ctx)
+            .labels(5)
+            .build()
+            .unwrap()
+            .generate(&mut rng)
+            .unwrap();
+        let logical_ts = logical_table_expr.columns.iter().position(|column| {
+            column
+                .options
+                .iter()
+                .any(|option| option == &ColumnOption::TimeIndex)
+        });
+        let logical_ts_name = logical_table_expr.columns[logical_ts.unwrap()]
+            .name
+            .value
+            .to_string();
+
+        assert_eq!(logical_table_expr.engine, "metric");
+        assert_eq!(logical_table_expr.columns.len(), 6);
+        assert_eq!(logical_ts_name, physical_ts_name);
+        assert!(logical_table_expr
+            .columns
+            .iter()
+            .all(
+                |column| column.column_type != ConcreteDataType::string_datatype()
+                    || column
+                        .options
+                        .iter()
+                        .any(|option| option == &ColumnOption::PrimaryKey)
+            ));
+    }
+
+    #[test]
+    fn test_create_logical_table_expr_generator_deterministic() {
+        let mut rng = rand_chacha::ChaCha8Rng::seed_from_u64(0);
+        let physical_table_expr = CreatePhysicalTableExprGeneratorBuilder::default()
+            .build()
+            .unwrap()
+            .generate(&mut rng)
+            .unwrap();
+        let physical_table_serialized = serde_json::to_string(&physical_table_expr).unwrap();
+        let physical_table_expected = r#"{"table_name":{"value":"qui","quote_style":null},"columns":[{"name":{"value":"ts","quote_style":null},"column_type":{"Timestamp":{"Millisecond":null}},"options":["TimeIndex"]},{"name":{"value":"val","quote_style":null},"column_type":{"Float64":{}},"options":[]}],"if_not_exists":false,"partition":null,"engine":"metric","options":{"physical_metric_table":{"String":""}},"primary_keys":[]}"#;
+        assert_eq!(physical_table_expected, physical_table_serialized);
+
+        let physical_table_ctx = Arc::new(TableContext::from(&physical_table_expr));
+
+        let logical_table_expr = CreateLogicalTableExprGeneratorBuilder::default()
+            .physical_table_ctx(physical_table_ctx)
+            .labels(5)
+            .build()
+            .unwrap()
+            .generate(&mut rng)
+            .unwrap();
+
+        let logical_table_serialized = serde_json::to_string(&logical_table_expr).unwrap();
+        let logical_table_expected = r#"{"table_name":{"value":"nostrum","quote_style":null},"columns":[{"name":{"value":"NATUs","quote_style":"`"},"column_type":{"String":null},"options":["PrimaryKey"]},{"name":{"value":"adipisci","quote_style":null},"column_type":{"String":null},"options":["PrimaryKey"]},{"name":{"value":"cUMqUe","quote_style":"`"},"column_type":{"String":null},"options":["PrimaryKey"]},{"name":{"value":"MOLestIAS","quote_style":"`"},"column_type":{"String":null},"options":["PrimaryKey"]},{"name":{"value":"ts","quote_style":null},"column_type":{"Timestamp":{"Millisecond":null}},"options":["TimeIndex"]},{"name":{"value":"val","quote_style":null},"column_type":{"Float64":{}},"options":[]}],"if_not_exists":false,"partition":null,"engine":"metric","options":{"on_physical_table":{"String":"qui"}},"primary_keys":[0,2,3,1]}"#;
+        assert_eq!(logical_table_expected, logical_table_serialized);
+    }
+
     #[test]
     fn test_create_database_expr_generator() {
         let mut rng = rand::thread_rng();

tests-fuzz/src/ir.rs

@@ -63,6 +63,8 @@ lazy_static! {
         ConcreteDataType::date_datatype(),
         ConcreteDataType::datetime_datatype(),
     ];
+    pub static ref STRING_DATA_TYPES: Vec<ConcreteDataType> =
+        vec![ConcreteDataType::string_datatype(),];
 }
 
 impl_random!(ConcreteDataType, ColumnTypeGenerator, DATA_TYPES);
@@ -72,10 +74,16 @@ impl_random!(
     PartibleColumnTypeGenerator,
     PARTIBLE_DATA_TYPES
 );
+impl_random!(
+    ConcreteDataType,
+    StringColumnTypeGenerator,
+    STRING_DATA_TYPES
+);
 
 pub struct ColumnTypeGenerator;
 pub struct TsColumnTypeGenerator;
 pub struct PartibleColumnTypeGenerator;
+pub struct StringColumnTypeGenerator;
 
 /// Generates a random [Value].
 pub fn generate_random_value<R: Rng>(
@@ -318,6 +326,13 @@ pub fn ts_column_options_generator<R: Rng + 'static>(
     vec![ColumnOption::TimeIndex]
 }
 
+pub fn primary_key_column_options_generator<R: Rng + 'static>(
+    _: &mut R,
+    _: &ConcreteDataType,
+) -> Vec<ColumnOption> {
+    vec![ColumnOption::PrimaryKey, ColumnOption::NotNull]
+}
+
 /// Generates columns with given `names`.
 pub fn generate_columns<R: Rng + 'static>(
     rng: &mut R,