chore[fuzzer]: add projection and filter expressions to the fuzzer (#3586)

I will refactor the fuzz into two [file and array] in a future PR

Signed-off-by: Joe Isaacs <[email protected]>

Author: joseph-isaacs

Cargo.lock

@@ -30,6 +30,7 @@ vortex-btrblocks = { workspace = true }
 vortex-buffer = { workspace = true }
 vortex-dtype = { workspace = true, features = ["arbitrary"] }
 vortex-error = { workspace = true }
+vortex-expr = { workspace = true, features = ["arbitrary"] }
 vortex-file = { workspace = true }
 vortex-mask = { workspace = true }
 vortex-scalar = { workspace = true, features = ["arbitrary"] }

fuzz/Cargo.toml

@@ -6,19 +6,25 @@ use arrow_ord::sort::SortOptions;
 use futures_util::TryStreamExt;
 use libfuzzer_sys::{Corpus, fuzz_target};
 use vortex_array::arrays::ChunkedArray;
-use vortex_array::arrays::arbitrary::ArbitraryArray;
 use vortex_array::arrow::IntoArrowArray;
 use vortex_array::compute::{Operator, compare};
 use vortex_array::{Array, ArrayRef, Canonical, IntoArray, ToCanonical};
 use vortex_buffer::ByteBufferMut;
 use vortex_dtype::{DType, StructFields};
 use vortex_error::{VortexExpect, VortexUnwrap, vortex_panic};
+use vortex_expr::root;
 use vortex_file::{VortexOpenOptions, VortexWriteOptions};
+use vortex_fuzz::FuzzFileAction;
 use vortex_utils::aliases::DefaultHashBuilder;
 use vortex_utils::aliases::hash_set::HashSet;
 
-fuzz_target!(|array_data: ArbitraryArray| -> Corpus {
-    let array_data = array_data.0;
+fuzz_target!(|fuzz: FuzzFileAction| -> Corpus {
+    let FuzzFileAction {
+        array,
+        projection,
+        filter,
+    } = fuzz;
+    let array_data = array;
 
     if has_nullable_struct(array_data.dtype()) || has_duplicate_field_names(array_data.dtype()) {
         return Corpus::Reject;
@@ -41,6 +47,8 @@ fuzz_target!(|array_data: ArbitraryArray| -> Corpus {
             .vortex_unwrap()
             .scan()
             .vortex_unwrap()
+            .with_projection(projection.unwrap_or_else(|| root()))
+            .with_some_filter(filter)
             .into_array_stream()
             .vortex_unwrap()
             .try_collect::<Vec<_>>()

fuzz/fuzz_targets/file_io.rs

@@ -25,6 +25,8 @@ use vortex_array::{Array, ArrayRef, IntoArray};
 use vortex_btrblocks::BtrBlocksCompressor;
 use vortex_dtype::{DType, Nullability};
 use vortex_error::{VortexExpect, VortexUnwrap, vortex_panic};
+use vortex_expr::ExprRef;
+use vortex_expr::arbitrary::{filter_expr, projection_expr};
 use vortex_mask::Mask;
 use vortex_scalar::Scalar;
 use vortex_scalar::arbitrary::random_scalar;
@@ -253,3 +255,22 @@ fn actions_for_dtype(dtype: &DType) -> HashSet<usize> {
         _ => ALL_ACTIONS.collect(),
     }
 }
+
+#[derive(Debug)]
+pub struct FuzzFileAction {
+    pub array: ArrayRef,
+    pub projection: Option<ExprRef>,
+    pub filter: Option<ExprRef>,
+}
+
+impl<'a> Arbitrary<'a> for FuzzFileAction {
+    fn arbitrary(u: &mut Unstructured<'a>) -> Result<Self> {
+        let array = ArbitraryArray::arbitrary(u)?.0;
+        let dtype = array.dtype().clone();
+        Ok(FuzzFileAction {
+            array,
+            projection: projection_expr(u, &dtype)?,
+            filter: filter_expr(u, &dtype)?,
+        })
+    }
+}

fuzz/src/lib.rs

@@ -17,6 +17,7 @@ version = { workspace = true }
 workspace = true
 
 [dependencies]
+arbitrary = { workspace = true, optional = true }
 dyn-hash = { workspace = true }
 itertools = { workspace = true }
 prost = { workspace = true, optional = true }
@@ -34,6 +35,11 @@ vortex-utils = { workspace = true }
 vortex-expr = { path = ".", features = ["test-harness"] }
 
 [features]
+arbitrary = [
+    "dep:arbitrary",
+    "vortex-scalar/arbitrary",
+    "vortex-dtype/arbitrary",
+]
 serde = ["dep:serde", "vortex-dtype/serde", "vortex-error/serde"]
 proto = ["vortex-proto/expr", "vortex-error/prost", "dep:prost", "serde"]
 test-harness = []

vortex-expr/Cargo.toml

@@ -0,0 +1,63 @@
+use std::cmp::max;
+
+use arbitrary::{Result as AResult, Unstructured};
+use vortex_dtype::{DType, FieldName};
+use vortex_scalar::arbitrary::random_scalar;
+
+use crate::{BinaryExpr, ExprRef, Operator, and_collect, get_item_scope, lit, pack};
+
+pub fn projection_expr(u: &mut Unstructured<'_>, dtype: &DType) -> AResult<Option<ExprRef>> {
+    let Some(struct_dtype) = dtype.as_struct() else {
+        return Ok(None);
+    };
+
+    let column_count = u.int_in_range::<usize>(0..=max(struct_dtype.nfields(), 10))?;
+
+    let cols = (0..column_count)
+        .map(|_| {
+            let get_item = u.choose(struct_dtype.names().iter().as_slice())?;
+            Ok((get_item.clone(), get_item_scope(get_item.clone())))
+        })
+        .collect::<AResult<Vec<_>>>()?;
+
+    Ok(Some(pack(cols, u.arbitrary()?)))
+}
+
+pub fn filter_expr(u: &mut Unstructured<'_>, dtype: &DType) -> AResult<Option<ExprRef>> {
+    let Some(struct_dtype) = dtype.as_struct() else {
+        return Ok(None);
+    };
+
+    let filter_count = u.int_in_range::<usize>(0..=max(struct_dtype.nfields(), 10))?;
+
+    let filters = (0..filter_count)
+        .map(|_| {
+            let (col, dtype) =
+                u.choose_iter(struct_dtype.names().iter().zip(struct_dtype.fields()))?;
+            random_comparison(u, col, &dtype)
+        })
+        .collect::<AResult<Vec<_>>>()?;
+
+    Ok(and_collect(filters))
+}
+
+fn random_comparison(u: &mut Unstructured<'_>, col: &FieldName, dtype: &DType) -> AResult<ExprRef> {
+    let scalar = random_scalar(u, dtype)?;
+    Ok(BinaryExpr::new_expr(
+        get_item_scope(col.clone()),
+        arbitrary_comparison_operator(u)?,
+        lit(scalar),
+    ))
+}
+
+fn arbitrary_comparison_operator(u: &mut Unstructured<'_>) -> AResult<Operator> {
+    Ok(match u.int_in_range(0..=5)? {
+        0 => Operator::Eq,
+        1 => Operator::NotEq,
+        2 => Operator::Gt,
+        3 => Operator::Gte,
+        4 => Operator::Lt,
+        5 => Operator::Lte,
+        _ => unreachable!("range 0..=5"),
+    })
+}

vortex-expr/src/arbitrary.rs

@@ -8,6 +8,8 @@ use dyn_hash::DynHash;
 mod binary;
 
 mod analysis;
+#[cfg(feature = "arbitrary")]
+pub mod arbitrary;
 mod between;
 mod cast;
 mod field;