feat(fuzz): add create logical table target (#3756)

* feat(fuzz): add create logical table target * fix: drop physical table after fuzz test * fix: remove backticks of table name in with clause * fix: create physical and logical table properly * chore: update comments * chore(ci): add fuzz_create_logical_table ci cfg * fix: create one logical table once a time * fix: avoid possible duplicate table and column name * feat: use hard-code physical table * chore: remove useless phantom * refactor: create logical table with struct initialization * chore: suggested changes and corresponding test changes * chore: clean up
GreptimeTeam · Apr 26, 2024 · 934c7e3 · 934c7e3
1 parent d8ea7c5
commit 934c7e3
Show file tree

Hide file tree

Showing 5 changed files with 425 additions and 4 deletions.
diff --git a/.github/workflows/develop.yml b/.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

diff --git a/tests-fuzz/Cargo.toml b/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"

diff --git a/tests-fuzz/src/generator/create_expr.rs b/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 +23,16 @@ 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::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_options_generator, ts_column_options_generator,
+    Column, ColumnTypeGenerator, CreateDatabaseExpr, CreateTableExpr, Ident,
+    PartibleColumnTypeGenerator, StringColumnTypeGenerator, TsColumnTypeGenerator,
 };
 
 #[derive(Builder)]
@@ -200,6 +203,107 @@ 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>>,
+    #[builder(default = "false")]
+    if_not_exists: bool,
+}
+
+impl<R: Rng + 'static> Generator<CreateTableExpr, R> for CreatePhysicalTableExprGenerator<R> {
+    type Error = Error;
+
+    fn generate(&self, rng: &mut R) -> Result<CreateTableExpr> {
+        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: self.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,
+    labels: usize,
+    if_not_exists: bool,
+    #[builder(default = "Box::new(WordGenerator)")]
+    name_generator: Box<dyn Random<Ident, R>>,
+}
+
+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 logical_table_name = self
+            .physical_table_ctx
+            .generate_unique_table_name(rng, self.name_generator.as_ref());
+        let mut logical_table = CreateTableExpr {
+            table_name: logical_table_name,
+            columns: self.physical_table_ctx.columns.clone(),
+            if_not_exists: self.if_not_exists,
+            partition: None,
+            engine: "metric".to_string(),
+            options: [(
+                "on_physical_table".to_string(),
+                self.physical_table_ctx.name.value.clone().into(),
+            )]
+            .into(),
+            primary_keys: vec![],
+        };
+
+        let column_names = self.name_generator.choose(rng, self.labels);
+        logical_table.columns.extend(generate_columns(
+            rng,
+            column_names,
+            &StringColumnTypeGenerator,
+            Box::new(primary_key_options_generator),
+        ));
+
+        // Currently only the `primary key` option is kept in physical table,
+        // so we only keep the `primary key` option in the logical table for fuzz test.
+        let mut primary_keys = vec![];
+        for (idx, column) in logical_table.columns.iter().enumerate() {
+            if column.is_primary_key() {
+                primary_keys.push(idx);
+            }
+        }
+        primary_keys.shuffle(rng);
+        logical_table.primary_keys = primary_keys;
+
+        Ok(logical_table)
+    }
+}
+
 #[derive(Builder)]
 #[builder(default, pattern = "owned")]
 pub struct CreateDatabaseExprGenerator<R: Rng + 'static> {
@@ -236,10 +340,14 @@ 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 +404,95 @@ 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()
+            .if_not_exists(false)
+            .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)
+            .if_not_exists(false)
+            .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(), 7);
+        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()
+            .if_not_exists(false)
+            .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":"expedita","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)
+            .if_not_exists(false)
+            .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":"impedit","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":[]},{"name":{"value":"qui","quote_style":null},"column_type":{"String":null},"options":["PrimaryKey"]},{"name":{"value":"totam","quote_style":null},"column_type":{"String":null},"options":["PrimaryKey"]},{"name":{"value":"molestias","quote_style":null},"column_type":{"String":null},"options":["PrimaryKey"]},{"name":{"value":"natus","quote_style":null},"column_type":{"String":null},"options":["PrimaryKey"]},{"name":{"value":"cumque","quote_style":null},"column_type":{"String":null},"options":["PrimaryKey"]}],"if_not_exists":false,"partition":null,"engine":"metric","options":{"on_physical_table":{"String":"expedita"}},"primary_keys":[2,5,3,6,4]}"#;
+        assert_eq!(logical_table_expected, logical_table_serialized);
+    }
+
     #[test]
     fn test_create_database_expr_generator() {
         let mut rng = rand::thread_rng();

diff --git a/tests-fuzz/src/ir.rs b/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 +326,20 @@ pub fn ts_column_options_generator<R: Rng + 'static>(
     vec![ColumnOption::TimeIndex]
 }
 
+pub fn primary_key_and_not_null_column_options_generator<R: Rng + 'static>(
+    _: &mut R,
+    _: &ConcreteDataType,
+) -> Vec<ColumnOption> {
+    vec![ColumnOption::PrimaryKey, ColumnOption::NotNull]
+}
+
+pub fn primary_key_options_generator<R: Rng + 'static>(
+    _: &mut R,
+    _: &ConcreteDataType,
+) -> Vec<ColumnOption> {
+    vec![ColumnOption::PrimaryKey]
+}
+
 /// Generates columns with given `names`.
 pub fn generate_columns<R: Rng + 'static>(
     rng: &mut R,