feat: schema lazy eval for config attr (#797)

* feat: support schema override attribute lazy eval and reduce unnecessary calculations and improve performance. Signed-off-by: peefy <[email protected]> * feat: add without insert index check for dict attributes Signed-off-by: peefy <[email protected]> --------- Signed-off-by: peefy <[email protected]>
kcl-lang · Oct 23, 2023 · 2b751aa · 2b751aa
1 parent d28ea21
commit 2b751aa
Show file tree

Hide file tree

Showing 13 changed files with 152 additions and 33 deletions.
diff --git a/kclvm/compiler/src/codegen/llvm/node.rs b/kclvm/compiler/src/codegen/llvm/node.rs
@@ -1391,10 +1391,6 @@ impl<'ctx> TypedResultWalker<'ctx> for LLVMCodeGenContext<'ctx> {
             self.walk_decorator_with_name(&decorator.node, Some(name), false)
                 .expect(kcl_error::COMPILE_ERROR_MSG);
         }
-        let value = match &schema_attr.value {
-            Some(value) => self.walk_expr(value).expect(kcl_error::COMPILE_ERROR_MSG),
-            None => self.undefined_value(),
-        };
         let config_value = self
             .get_variable(value::SCHEMA_CONFIG_NAME)
             .expect(kcl_error::INTERNAL_ERROR_MSG);
@@ -1409,22 +1405,6 @@ impl<'ctx> TypedResultWalker<'ctx> for LLVMCodeGenContext<'ctx> {
             &ApiFunc::kclvm_config_attr_map.name(),
             &[schema_value, string_ptr_value, type_str_ptr_value],
         );
-        if let Some(op) = &schema_attr.op {
-            match op {
-                // Union
-                ast::BinOrAugOp::Aug(ast::AugOp::BitOr) => {
-                    let org_value = self.build_call(
-                        &ApiFunc::kclvm_dict_get_value.name(),
-                        &[schema_value, string_ptr_value],
-                    );
-                    let fn_name = ApiFunc::kclvm_value_op_bit_or;
-                    let value = self.build_call(&fn_name.name(), &[org_value, value]);
-                    self.dict_merge(schema_value, name, value, 1, -1);
-                }
-                // Assign
-                _ => self.dict_merge(schema_value, name, value, 1, -1),
-            }
-        }
         let has_key = self
             .build_call(
                 &ApiFunc::kclvm_dict_has_value.name(),
@@ -1436,13 +1416,58 @@ impl<'ctx> TypedResultWalker<'ctx> for LLVMCodeGenContext<'ctx> {
                 .build_int_compare(IntPredicate::NE, has_key, self.native_i8_zero(), "");
         let then_block = self.append_block("");
         let else_block = self.append_block("");
+        let end_block = self.append_block("");
         self.builder
             .build_conditional_branch(has_key, then_block, else_block);
         self.builder.position_at_end(then_block);
         let config_attr_value = self.build_call(
             &ApiFunc::kclvm_dict_get_entry.name(),
             &[config_value, string_ptr_value],
         );
+        // If the attribute operator is not `=`, eval the schema attribute value.
+        // if is_not_override:
+        let is_override_attr = self
+            .build_call(
+                &ApiFunc::kclvm_dict_is_override_attr.name(),
+                &[config_value, string_ptr_value],
+            )
+            .into_int_value();
+        let is_not_override_attr = self.builder.build_int_compare(
+            IntPredicate::EQ,
+            is_override_attr,
+            self.native_i8_zero(),
+            "",
+        );
+        let is_not_override_then_block = self.append_block("");
+        let is_not_override_else_block = self.append_block("");
+        self.builder.build_conditional_branch(
+            is_not_override_attr,
+            is_not_override_then_block,
+            is_not_override_else_block,
+        );
+        self.builder.position_at_end(is_not_override_then_block);
+        let value = match &schema_attr.value {
+            Some(value) => self.walk_expr(value).expect(kcl_error::COMPILE_ERROR_MSG),
+            None => self.undefined_value(),
+        };
+        if let Some(op) = &schema_attr.op {
+            match op {
+                // Union
+                ast::BinOrAugOp::Aug(ast::AugOp::BitOr) => {
+                    let org_value = self.build_call(
+                        &ApiFunc::kclvm_dict_get_value.name(),
+                        &[schema_value, string_ptr_value],
+                    );
+                    let fn_name = ApiFunc::kclvm_value_op_bit_or;
+                    let value = self.build_call(&fn_name.name(), &[org_value, value]);
+                    self.dict_merge(schema_value, name, value, 1, -1);
+                }
+                // Assign
+                _ => self.dict_merge(schema_value, name, value, 1, -1),
+            }
+        }
+        self.br(is_not_override_else_block);
+        self.builder.position_at_end(is_not_override_else_block);
         self.value_union(schema_value, config_attr_value);
         let cal_map = self
             .get_variable(value::SCHEMA_CAL_MAP)
@@ -1476,8 +1501,31 @@ impl<'ctx> TypedResultWalker<'ctx> for LLVMCodeGenContext<'ctx> {
                 }
             }
         }
-        self.br(else_block);
+        self.br(end_block);
         self.builder.position_at_end(else_block);
+        // Lazy eval for the schema attribute.
+        let value = match &schema_attr.value {
+            Some(value) => self.walk_expr(value).expect(kcl_error::COMPILE_ERROR_MSG),
+            None => self.undefined_value(),
+        };
+        if let Some(op) = &schema_attr.op {
+            match op {
+                // Union
+                ast::BinOrAugOp::Aug(ast::AugOp::BitOr) => {
+                    let org_value = self.build_call(
+                        &ApiFunc::kclvm_dict_get_value.name(),
+                        &[schema_value, string_ptr_value],
+                    );
+                    let fn_name = ApiFunc::kclvm_value_op_bit_or;
+                    let value = self.build_call(&fn_name.name(), &[org_value, value]);
+                    self.dict_merge(schema_value, name, value, 1, -1);
+                }
+                // Assign
+                _ => self.dict_merge(schema_value, name, value, 1, -1),
+            }
+        }
+        self.br(end_block);
+        self.builder.position_at_end(end_block);
         Ok(schema_value)
     }
 

diff --git a/kclvm/runtime/src/_kclvm.bc b/kclvm/runtime/src/_kclvm.bc
diff --git a/kclvm/runtime/src/_kclvm.h b/kclvm/runtime/src/_kclvm.h
@@ -304,6 +304,8 @@ void kclvm_dict_insert_unpack(kclvm_value_ref_t* p, kclvm_value_ref_t* v);
 
 void kclvm_dict_insert_value(kclvm_value_ref_t* p, kclvm_value_ref_t* key, kclvm_value_ref_t* v, kclvm_size_t op, kclvm_size_t insert_index);
 
+kclvm_bool_t kclvm_dict_is_override_attr(kclvm_value_ref_t* p, kclvm_char_t* key);
+
 kclvm_value_ref_t* kclvm_dict_keys(kclvm_value_ref_t* p);
 
 kclvm_size_t kclvm_dict_len(kclvm_value_ref_t* p);
@@ -336,7 +338,7 @@ kclvm_value_ref_t* kclvm_iterator_next_value(kclvm_iterator_t* p, kclvm_value_re
 
 kclvm_value_ref_t* kclvm_json_decode(kclvm_context_t* _ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* _kwargs);
 
-kclvm_value_ref_t* kclvm_json_dump_to_file(kclvm_context_t* _ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* _kwargs);
+kclvm_value_ref_t* kclvm_json_dump_to_file(kclvm_context_t* _ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* kwargs);
 
 kclvm_value_ref_t* kclvm_json_encode(kclvm_context_t* _ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* kwargs);
 
@@ -702,7 +704,7 @@ kclvm_value_ref_t* kclvm_value_union_all(kclvm_context_t* _ctx, kclvm_value_ref_
 
 kclvm_value_ref_t* kclvm_yaml_decode(kclvm_context_t* _ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* _kwargs);
 
-kclvm_value_ref_t* kclvm_yaml_dump_to_file(kclvm_context_t* _ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* _kwargs);
+kclvm_value_ref_t* kclvm_yaml_dump_to_file(kclvm_context_t* _ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* kwargs);
 
 kclvm_value_ref_t* kclvm_yaml_encode(kclvm_context_t* _ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* kwargs);
 

diff --git a/kclvm/runtime/src/_kclvm.ll b/kclvm/runtime/src/_kclvm.ll
@@ -252,6 +252,8 @@ declare void @kclvm_dict_insert_unpack(%kclvm_value_ref_t* %p, %kclvm_value_ref_
 
 declare void @kclvm_dict_insert_value(%kclvm_value_ref_t* %p, %kclvm_value_ref_t* %key, %kclvm_value_ref_t* %v, %kclvm_size_t %op, %kclvm_size_t %insert_index);
 
+declare %kclvm_bool_t @kclvm_dict_is_override_attr(%kclvm_value_ref_t* %p, %kclvm_char_t* %key);
+
 declare %kclvm_value_ref_t* @kclvm_dict_keys(%kclvm_value_ref_t* %p);
 
 declare %kclvm_size_t @kclvm_dict_len(%kclvm_value_ref_t* %p);
@@ -284,7 +286,7 @@ declare %kclvm_value_ref_t* @kclvm_iterator_next_value(%kclvm_iterator_t* %p, %k
 
 declare %kclvm_value_ref_t* @kclvm_json_decode(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %_kwargs);
 
-declare %kclvm_value_ref_t* @kclvm_json_dump_to_file(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %_kwargs);
+declare %kclvm_value_ref_t* @kclvm_json_dump_to_file(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %kwargs);
 
 declare %kclvm_value_ref_t* @kclvm_json_encode(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %kwargs);
 
@@ -430,10 +432,6 @@ declare %kclvm_value_ref_t* @kclvm_schema_value_new(%kclvm_context_t* %ctx, %kcl
 
 declare %kclvm_size_t @kclvm_strlen(i8* %ptr);
 
-declare void @kclvm_testing_arguments(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %_args, %kclvm_value_ref_t* %_kwargs);
-
-declare void @kclvm_testing_setting_file(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %_args, %kclvm_value_ref_t* %_kwargs);
-
 declare %kclvm_value_ref_t* @kclvm_units_to_G(%kclvm_context_t* %ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %kwargs);
 
 declare %kclvm_value_ref_t* @kclvm_units_to_Gi(%kclvm_context_t* %ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %kwargs);
@@ -654,7 +652,7 @@ declare %kclvm_value_ref_t* @kclvm_value_union_all(%kclvm_context_t* %_ctx, %kcl
 
 declare %kclvm_value_ref_t* @kclvm_yaml_decode(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %_kwargs);
 
-declare %kclvm_value_ref_t* @kclvm_yaml_dump_to_file(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %_kwargs);
+declare %kclvm_value_ref_t* @kclvm_yaml_dump_to_file(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %kwargs);
 
 declare %kclvm_value_ref_t* @kclvm_yaml_encode(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %kwargs);
 

diff --git a/kclvm/runtime/src/_kclvm.rs b/kclvm/runtime/src/_kclvm.rs
@@ -138,6 +138,7 @@ pub enum ApiFunc {
     kclvm_dict_insert,
     kclvm_dict_insert_unpack,
     kclvm_dict_insert_value,
+    kclvm_dict_is_override_attr,
     kclvm_dict_keys,
     kclvm_dict_len,
     kclvm_dict_merge,

diff --git a/kclvm/runtime/src/_kclvm_addr.rs b/kclvm/runtime/src/_kclvm_addr.rs
@@ -145,6 +145,7 @@ pub fn _kclvm_get_fn_ptr_by_name(name: &str) -> u64 {
         "kclvm_dict_insert" => crate::kclvm_dict_insert as *const () as u64,
         "kclvm_dict_insert_unpack" => crate::kclvm_dict_insert_unpack as *const () as u64,
         "kclvm_dict_insert_value" => crate::kclvm_dict_insert_value as *const () as u64,
+        "kclvm_dict_is_override_attr" => crate::kclvm_dict_is_override_attr as *const () as u64,
         "kclvm_dict_keys" => crate::kclvm_dict_keys as *const () as u64,
         "kclvm_dict_len" => crate::kclvm_dict_len as *const () as u64,
         "kclvm_dict_merge" => crate::kclvm_dict_merge as *const () as u64,

diff --git a/kclvm/runtime/src/_kclvm_api_spec.rs b/kclvm/runtime/src/_kclvm_api_spec.rs
@@ -398,6 +398,10 @@
 // api-spec(c):    void kclvm_dict_clear(kclvm_value_ref_t* p);
 // api-spec(llvm): declare void @kclvm_dict_clear(%kclvm_value_ref_t* %p);
 
+// api-spec:       kclvm_dict_is_override_attr
+// api-spec(c):    kclvm_bool_t kclvm_dict_is_override_attr(kclvm_value_ref_t* p, kclvm_char_t* key);
+// api-spec(llvm): declare %kclvm_bool_t @kclvm_dict_is_override_attr(%kclvm_value_ref_t* %p, %kclvm_char_t* %key);
+
 // api-spec:       kclvm_dict_get
 // api-spec(c):    kclvm_value_ref_t* kclvm_dict_get(kclvm_value_ref_t* p, kclvm_value_ref_t* key);
 // api-spec(llvm): declare %kclvm_value_ref_t* @kclvm_dict_get(%kclvm_value_ref_t* %p, %kclvm_value_ref_t* %key);
@@ -899,8 +903,8 @@
 // api-spec(llvm): declare %kclvm_value_ref_t* @kclvm_json_decode(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %_kwargs);
 
 // api-spec:       kclvm_json_dump_to_file
-// api-spec(c):    kclvm_value_ref_t* kclvm_json_dump_to_file(kclvm_context_t* _ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* _kwargs);
-// api-spec(llvm): declare %kclvm_value_ref_t* @kclvm_json_dump_to_file(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %_kwargs);
+// api-spec(c):    kclvm_value_ref_t* kclvm_json_dump_to_file(kclvm_context_t* _ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* kwargs);
+// api-spec(llvm): declare %kclvm_value_ref_t* @kclvm_json_dump_to_file(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %kwargs);
 
 // api-spec:       kclvm_manifests_yaml_stream
 // api-spec(c):    void kclvm_manifests_yaml_stream(kclvm_context_t* ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* kwargs);
@@ -1251,6 +1255,6 @@
 // api-spec(llvm): declare %kclvm_value_ref_t* @kclvm_yaml_decode(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %_kwargs);
 
 // api-spec:       kclvm_yaml_dump_to_file
-// api-spec(c):    kclvm_value_ref_t* kclvm_yaml_dump_to_file(kclvm_context_t* _ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* _kwargs);
-// api-spec(llvm): declare %kclvm_value_ref_t* @kclvm_yaml_dump_to_file(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %_kwargs);
+// api-spec(c):    kclvm_value_ref_t* kclvm_yaml_dump_to_file(kclvm_context_t* _ctx, kclvm_value_ref_t* args, kclvm_value_ref_t* kwargs);
+// api-spec(llvm): declare %kclvm_value_ref_t* @kclvm_yaml_dump_to_file(%kclvm_context_t* %_ctx, %kclvm_value_ref_t* %args, %kclvm_value_ref_t* %kwargs);
 
diff --git a/kclvm/runtime/src/value/api.rs b/kclvm/runtime/src/value/api.rs
@@ -1050,6 +1050,22 @@ pub unsafe extern "C" fn kclvm_dict_clear(p: *mut kclvm_value_ref_t) {
     p.dict_clear();
 }
 
+#[no_mangle]
+#[runtime_fn]
+pub unsafe extern "C" fn kclvm_dict_is_override_attr(
+    p: *const kclvm_value_ref_t,
+    key: *const kclvm_char_t,
+) -> kclvm_bool_t {
+    let p = ptr_as_ref(p);
+    let key = c2str(key);
+    let is_override_op = matches!(
+        p.dict_get_attr_operator(key),
+        Some(ConfigEntryOperationKind::Override)
+    );
+    let without_index = matches!(p.dict_get_insert_index(key), Some(-1) | None);
+    (is_override_op && without_index) as kclvm_bool_t
+}
+
 #[no_mangle]
 #[runtime_fn]
 pub unsafe extern "C" fn kclvm_dict_get(

diff --git a/kclvm/runtime/src/value/val_dict.rs b/kclvm/runtime/src/value/val_dict.rs
@@ -116,6 +116,26 @@ impl ValueRef {
         }
     }
 
+    /// Dict get value e.g., {k1 = v1, k2 = v2}.get_attr_operator(k1) == Some(ConfigEntryOperationKind::Override)
+    pub fn dict_get_attr_operator(&self, key: &str) -> Option<ConfigEntryOperationKind> {
+        match &*self.rc.borrow() {
+            Value::dict_value(ref dict) => dict.ops.get(key).cloned(),
+            Value::schema_value(ref schema) => schema.config.ops.get(key).cloned(),
+            _ => None,
+        }
+    }
+
+    /// Dict get value e.g., {k1 = v1, k2 = v2}.get_attr_operator(k1) == Some(ConfigEntryOperationKind::Override)
+    pub fn dict_get_insert_index(&self, key: &str) -> Option<i32> {
+        match &*self.rc.borrow() {
+            Value::dict_value(ref dict) => Some(*dict.insert_indexs.get(key).unwrap_or(&-1)),
+            Value::schema_value(ref schema) => {
+                Some(*schema.config.insert_indexs.get(key).unwrap_or(&-1))
+            }
+            _ => None,
+        }
+    }
+
     /// Dict get entry e.g., {k1: v1, k2, v2}.get_entry(k1) == {k1: v1}
     pub fn dict_get_entry(&self, key: &str) -> Option<ValueRef> {
         match &*self.rc.borrow() {

diff --git a/test/grammar/schema/config_op/override/override_3/main.k b/test/grammar/schema/config_op/override/override_3/main.k
@@ -0,0 +1,10 @@
+schema MySubSchema:
+    a: int
+    b: str
+    c: {str:str}
+    d: str = c[b]
+
+schema MyTestSchema:
+    field1?: [MySubSchema] = [{a = 1, b = "key", c = {key = "value"}}]
+
+instance = MyTestSchema {field1: None}
diff --git a/test/grammar/schema/config_op/override/override_3/stdout.golden b/test/grammar/schema/config_op/override/override_3/stdout.golden
@@ -0,0 +1,7 @@
+instance:
+  field1:
+  - a: 1
+    b: key
+    c:
+      key: value
+    d: value
diff --git a/test/grammar/schema/config_op/override/override_4/main.k b/test/grammar/schema/config_op/override/override_4/main.k
@@ -0,0 +1,10 @@
+schema MySubSchema:
+    a: int
+    b: str
+    c: {str:str}
+    d: str = c[b]
+
+schema MyTestSchema:
+    field1?: [MySubSchema] = [{a: 1}]
+
+instance = MyTestSchema {field1 = None}
diff --git a/test/grammar/schema/config_op/override/override_4/stdout.golden b/test/grammar/schema/config_op/override/override_4/stdout.golden
@@ -0,0 +1,2 @@
+instance:
+  field1: null