feat(fuzz): add insert target (#3499)

* fix(common-time): allow building nanos timestamp from parts split from i64::MIN Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * feat(fuzz): add insert target Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> * chore: cleanup cargo.toml and polish comments Signed-off-by: Zhenchi <zhongzc_arch@outlook.com> --------- Signed-off-by: Zhenchi <zhongzc_arch@outlook.com>
GreptimeTeam · Mar 13, 2024 · b55905c · b55905c
1 parent fb4da05
commit b55905c
Show file tree

Hide file tree

Showing 8 changed files with 360 additions and 46 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/tests-fuzz/Cargo.toml b/tests-fuzz/Cargo.toml
@@ -18,6 +18,7 @@ common-macro = { workspace = true }
 common-query = { workspace = true }
 common-runtime = { workspace = true }
 common-telemetry = { workspace = true }
+common-time = { workspace = true }
 datatypes = { workspace = true }
 derive_builder = { workspace = true }
 dotenv = "0.15"
@@ -39,13 +40,6 @@ sqlx = { version = "0.6", features = [
 ] }
 
 [dev-dependencies]
-dotenv = "0.15"
-sqlx = { version = "0.6", features = [
-    "runtime-tokio-rustls",
-    "mysql",
-    "postgres",
-    "chrono",
-] }
 tokio = { workspace = true }
 
 [[bin]]
@@ -54,3 +48,10 @@ path = "targets/fuzz_create_table.rs"
 test = false
 bench = false
 doc = false
+
+[[bin]]
+name = "fuzz_insert"
+path = "targets/fuzz_insert.rs"
+test = false
+bench = false
+doc = false
diff --git a/tests-fuzz/src/generator/insert_expr.rs b/tests-fuzz/src/generator/insert_expr.rs
@@ -14,6 +14,7 @@
 
 use std::marker::PhantomData;
 
+use datatypes::value::Value;
 use derive_builder::Builder;
 use rand::seq::SliceRandom;
 use rand::Rng;
@@ -22,7 +23,7 @@ use crate::context::TableContextRef;
 use crate::error::{Error, Result};
 use crate::fake::WordGenerator;
 use crate::generator::{Generator, Random};
-use crate::ir::insert_expr::InsertIntoExpr;
+use crate::ir::insert_expr::{InsertIntoExpr, RowValue};
 use crate::ir::{generate_random_value, Ident};
 
 /// Generates [InsertIntoExpr].
@@ -41,30 +42,64 @@ pub struct InsertExprGenerator<R: Rng + 'static> {
 impl<R: Rng + 'static> Generator<InsertIntoExpr, R> for InsertExprGenerator<R> {
     type Error = Error;
 
-    /// Generates the [CreateTableExpr].
+    /// Generates the [InsertIntoExpr].
     fn generate(&self, rng: &mut R) -> Result<InsertIntoExpr> {
-        let mut columns = self.table_ctx.columns.clone();
-        columns.shuffle(rng);
+        // Whether to omit all columns, i.e. INSERT INTO table_name VALUES (...)
+        let omit_column_list = rng.gen_bool(0.2);
 
-        let mut rows = Vec::with_capacity(self.rows);
+        let mut values_columns = vec![];
+        if omit_column_list {
+            // If omit column list, then all columns are required in the values list
+            values_columns = self.table_ctx.columns.clone();
+        } else {
+            for column in &self.table_ctx.columns {
+                let can_omit = column.is_nullable() || column.has_default_value();
+
+                // 50% chance to omit a column if it's not required
+                if !can_omit || rng.gen_bool(0.5) {
+                    values_columns.push(column.clone());
+                }
+            }
+            values_columns.shuffle(rng);
+
+            // If all columns are omitted, pick a random column
+            if values_columns.is_empty() {
+                values_columns.push(self.table_ctx.columns.choose(rng).unwrap().clone());
+            }
+        }
+
+        let mut values_list = Vec::with_capacity(self.rows);
         for _ in 0..self.rows {
-            let mut row = Vec::with_capacity(columns.len());
-            for column in &columns {
-                // TODO(weny): generates the special cases
-                row.push(generate_random_value(
+            let mut row = Vec::with_capacity(values_columns.len());
+            for column in &values_columns {
+                if column.is_nullable() && rng.gen_bool(0.2) {
+                    row.push(RowValue::Value(Value::Null));
+                    continue;
+                }
+
+                if column.has_default_value() && rng.gen_bool(0.2) {
+                    row.push(RowValue::Default);
+                    continue;
+                }
+
+                row.push(RowValue::Value(generate_random_value(
                     rng,
                     &column.column_type,
                     Some(self.word_generator.as_ref()),
-                ));
+                )));
             }
 
-            rows.push(row);
+            values_list.push(row);
         }
 
         Ok(InsertIntoExpr {
             table_name: self.table_ctx.name.to_string(),
-            columns,
-            rows,
+            columns: if omit_column_list {
+                vec![]
+            } else {
+                values_columns
+            },
+            values_list,
         })
     }
 }
diff --git a/tests-fuzz/src/ir.rs b/tests-fuzz/src/ir.rs
@@ -22,10 +22,13 @@ pub(crate) mod select_expr;
 use core::fmt;
 
 pub use alter_expr::AlterTableExpr;
+use common_time::{Date, DateTime, Timestamp};
 pub use create_expr::CreateTableExpr;
 use datatypes::data_type::ConcreteDataType;
+use datatypes::types::TimestampType;
 use datatypes::value::Value;
 use derive_builder::Builder;
+pub use insert_expr::InsertIntoExpr;
 use lazy_static::lazy_static;
 use rand::seq::SliceRandom;
 use rand::Rng;
@@ -91,14 +94,62 @@ pub fn generate_random_value<R: Rng>(
             Some(random) => Value::from(random.gen(rng).value),
             None => Value::from(rng.gen::<char>().to_string()),
         },
-        ConcreteDataType::Date(_) => Value::from(rng.gen::<i32>()),
-        ConcreteDataType::DateTime(_) => Value::from(rng.gen::<i64>()),
-        &ConcreteDataType::Timestamp(_) => Value::from(rng.gen::<u64>()),
+        ConcreteDataType::Date(_) => generate_random_date(rng),
+        ConcreteDataType::DateTime(_) => generate_random_datetime(rng),
+        &ConcreteDataType::Timestamp(ts_type) => generate_random_timestamp(rng, ts_type),
 
         _ => unimplemented!("unsupported type: {datatype}"),
     }
 }
 
+fn generate_random_timestamp<R: Rng>(rng: &mut R, ts_type: TimestampType) -> Value {
+    let v = match ts_type {
+        TimestampType::Second(_) => {
+            let min = i64::from(Timestamp::MIN_SECOND);
+            let max = i64::from(Timestamp::MAX_SECOND);
+            let value = rng.gen_range(min..=max);
+            Timestamp::new_second(value)
+        }
+        TimestampType::Millisecond(_) => {
+            let min = i64::from(Timestamp::MIN_MILLISECOND);
+            let max = i64::from(Timestamp::MAX_MILLISECOND);
+            let value = rng.gen_range(min..=max);
+            Timestamp::new_millisecond(value)
+        }
+        TimestampType::Microsecond(_) => {
+            let min = i64::from(Timestamp::MIN_MICROSECOND);
+            let max = i64::from(Timestamp::MAX_MICROSECOND);
+            let value = rng.gen_range(min..=max);
+            Timestamp::new_microsecond(value)
+        }
+        TimestampType::Nanosecond(_) => {
+            let min = i64::from(Timestamp::MIN_NANOSECOND);
+            let max = i64::from(Timestamp::MAX_NANOSECOND);
+            let value = rng.gen_range(min..=max);
+            Timestamp::new_nanosecond(value)
+        }
+    };
+    Value::from(v)
+}
+
+fn generate_random_datetime<R: Rng>(rng: &mut R) -> Value {
+    let min = i64::from(Timestamp::MIN_MILLISECOND);
+    let max = i64::from(Timestamp::MAX_MILLISECOND);
+    let value = rng.gen_range(min..=max);
+    let datetime = Timestamp::new_millisecond(value)
+        .to_chrono_datetime()
+        .unwrap();
+    Value::from(DateTime::from(datetime))
+}
+
+fn generate_random_date<R: Rng>(rng: &mut R) -> Value {
+    let min = i64::from(Timestamp::MIN_MILLISECOND);
+    let max = i64::from(Timestamp::MAX_MILLISECOND);
+    let value = rng.gen_range(min..=max);
+    let date = Timestamp::new_millisecond(value).to_chrono_date().unwrap();
+    Value::from(Date::from(date))
+}
+
 /// An identifier.
 #[derive(Debug, Clone, Serialize, Deserialize, PartialEq, PartialOrd, Eq, Ord, Hash)]
 pub struct Ident {
@@ -181,6 +232,24 @@ impl Column {
             .iter()
             .any(|opt| opt == &ColumnOption::PrimaryKey)
     }
+
+    /// Returns true if it's nullable.
+    pub fn is_nullable(&self) -> bool {
+        !self
+            .options
+            .iter()
+            .any(|opt| matches!(opt, ColumnOption::NotNull | ColumnOption::TimeIndex))
+    }
+
+    // Returns true if it has default value.
+    pub fn has_default_value(&self) -> bool {
+        self.options.iter().any(|opt| {
+            matches!(
+                opt,
+                ColumnOption::DefaultValue(_) | ColumnOption::DefaultFn(_)
+            )
+        })
+    }
 }
 
 /// Returns droppable columns. i.e., non-primary key columns, non-ts columns.

diff --git a/tests-fuzz/src/ir/insert_expr.rs b/tests-fuzz/src/ir/insert_expr.rs
@@ -12,14 +12,37 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+use std::fmt::Display;
+
 use datatypes::value::Value;
 
 use crate::ir::Column;
 
-pub type RowValue = Vec<Value>;
-
 pub struct InsertIntoExpr {
     pub table_name: String,
     pub columns: Vec<Column>,
-    pub rows: Vec<RowValue>,
+    pub values_list: Vec<RowValues>,
+}
+
+pub type RowValues = Vec<RowValue>;
+
+pub enum RowValue {
+    Value(Value),
+    Default,
+}
+
+impl Display for RowValue {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            RowValue::Value(v) => match v {
+                Value::Null => write!(f, "NULL"),
+                v @ (Value::String(_)
+                | Value::Timestamp(_)
+                | Value::DateTime(_)
+                | Value::Date(_)) => write!(f, "'{}'", v),
+                v => write!(f, "{}", v),
+            },
+            RowValue::Default => write!(f, "DEFAULT"),
+        }
+    }
 }
diff --git a/tests-fuzz/src/translator/mysql/insert_expr.rs b/tests-fuzz/src/translator/mysql/insert_expr.rs
@@ -22,33 +22,42 @@ impl DslTranslator<InsertIntoExpr, String> for InsertIntoExprTranslator {
     type Error = Error;
 
     fn translate(&self, input: &InsertIntoExpr) -> Result<String> {
-        let columns = input
-            .columns
-            .iter()
-            .map(|c| c.name.to_string())
-            .collect::<Vec<_>>()
-            .join(", ")
-            .to_string();
-
         Ok(format!(
-            "INSERT INTO {} ({})\nVALUES\n{};",
+            "INSERT INTO {} {} VALUES\n{};",
             input.table_name,
-            columns,
+            Self::format_columns(input),
             Self::format_values(input)
         ))
     }
 }
 
 impl InsertIntoExprTranslator {
+    fn format_columns(input: &InsertIntoExpr) -> String {
+        if input.columns.is_empty() {
+            "".to_string()
+        } else {
+            let list = input
+                .columns
+                .iter()
+                .map(|c| c.name.to_string())
+                .collect::<Vec<_>>()
+                .join(", ")
+                .to_string();
+
+            format!("({})", list)
+        }
+    }
+
     fn format_values(input: &InsertIntoExpr) -> String {
         input
-            .rows
+            .values_list
             .iter()
-            .map(|row| {
+            .map(|value| {
                 format!(
                     "({})",
-                    row.iter()
-                        .map(|v| format!("'{v}'"))
+                    value
+                        .iter()
+                        .map(|v| v.to_string())
                         .collect::<Vec<_>>()
                         .join(", ")
                 )
@@ -64,7 +73,7 @@ mod tests {
 
     use rand::SeedableRng;
 
-    use super::InsertIntoExprTranslator;
+    use super::*;
     use crate::generator::insert_expr::InsertExprGeneratorBuilder;
     use crate::generator::Generator;
     use crate::test_utils;
@@ -84,10 +93,23 @@ mod tests {
         let insert_expr = insert_expr_generator.generate(&mut rng).unwrap();
 
         let output = InsertIntoExprTranslator.translate(&insert_expr).unwrap();
-        let expected = r#"INSERT INTO test (host, idc, memory_util, ts, cpu_util, disk_util)
-VALUES
-('adipisci', 'debitis', '0.5495312687894465', '15292064470292927036', '0.9354265029131291', '0.8037816422279636'),
-('ut', 'sequi', '0.8807117723618908', '14214208091261382505', '0.5240550121500691', '0.350785883750684');"#;
+        let expected = r#"INSERT INTO test (ts, host, cpu_util) VALUES
+('+199601-11-07 21:32:56.695+0000', 'corrupti', 0.051130243193075464),
+('+40822-03-25 02:17:34.328+0000', NULL, 0.6552502332327004);"#;
+        assert_eq!(output, expected);
+
+        let insert_expr = insert_expr_generator.generate(&mut rng).unwrap();
+        let output = InsertIntoExprTranslator.translate(&insert_expr).unwrap();
+        let expected = r#"INSERT INTO test (cpu_util, disk_util, ts) VALUES
+(0.7074194466620976, 0.661288102315126, '-47252-05-08 07:33:49.567+0000'),
+(0.8266101224213618, 0.7947724277743285, '-224292-12-07 02:51:53.371+0000');"#;
+        assert_eq!(output, expected);
+
+        let insert_expr = insert_expr_generator.generate(&mut rng).unwrap();
+        let output = InsertIntoExprTranslator.translate(&insert_expr).unwrap();
+        let expected = r#"INSERT INTO test  VALUES
+('odio', NULL, 0.48809950435391647, 0.5228925709595407, 0.9091528874275897, '+241156-12-16 20:52:15.185+0000'),
+('dignissimos', 'labore', NULL, 0.12983559048685023, 0.6362040919831425, '-30691-06-17 23:41:09.938+0000');"#;
         assert_eq!(output, expected);
     }
 }
diff --git a/tests-fuzz/src/validator/column.rs b/tests-fuzz/src/validator/column.rs
@@ -98,7 +98,7 @@ impl PartialEq<Column> for ColumnEntry {
                 .iter()
                 .any(|opt| matches!(opt, ColumnOption::NotNull))
             {
-                debug!("ColumnOption::NotNull is not found");
+                debug!("ColumnOption::NotNull is found");
                 return false;
             }
         } else {