Move json_type_def to separate module

vectordotdev · Dec 13, 2024 · dd269f4 · dd269f4
1 parent 2f37d18
commit dd269f4
Show file tree

Hide file tree

Showing 6 changed files with 84 additions and 103 deletions.
diff --git a/src/stdlib/json_utils/json_type_def.rs b/src/stdlib/json_utils/json_type_def.rs
@@ -0,0 +1,24 @@
+use crate::prelude::{Collection, TypeDef};
+use crate::value::Kind;
+
+pub(crate) fn json_inner_kind() -> Kind {
+    Kind::null()
+        | Kind::bytes()
+        | Kind::integer()
+        | Kind::float()
+        | Kind::boolean()
+        | Kind::array(Collection::any())
+        | Kind::object(Collection::any())
+}
+
+pub(crate) fn json_type_def() -> TypeDef {
+    TypeDef::bytes()
+        .fallible()
+        .or_boolean()
+        .or_integer()
+        .or_float()
+        .add_null()
+        .or_null()
+        .or_array(Collection::from_unknown(json_inner_kind()))
+        .or_object(Collection::from_unknown(json_inner_kind()))
+}
diff --git a/src/stdlib/json_utils/mod.rs b/src/stdlib/json_utils/mod.rs
@@ -0,0 +1 @@
+pub(crate) mod json_type_def;
diff --git a/src/stdlib/mod.rs b/src/stdlib/mod.rs
@@ -31,6 +31,7 @@ pub use wasm_unsupported_function::WasmUnsupportedFunction;
 
 use crate::compiler::Function;
 
+mod json_utils;
 mod string_utils;
 mod util;
 mod wasm_unsupported_function;

diff --git a/src/stdlib/parse_cbor.rs b/src/stdlib/parse_cbor.rs
@@ -1,4 +1,5 @@
 use crate::compiler::prelude::*;
+use crate::stdlib::json_utils::json_type_def::json_type_def;
 use ciborium::de::from_reader;
 use zstd::zstd_safe::WriteBuf;
 
@@ -27,11 +28,38 @@ impl Function for ParseCbor {
     }
 
     fn examples(&self) -> &'static [Example] {
-        &[Example {
-            title: "object",
-            source: r#"parse_cbor!(decode_base64!("oWVmaWVsZGV2YWx1ZQ=="))"#,
-            result: Ok(r#"{ "field": "value" }"#),
-        }]
+        &[
+            Example {
+                title: "object",
+                source: r#"parse_cbor!(decode_base64!("oWVmaWVsZGV2YWx1ZQ=="))"#,
+                result: Ok(r#"{ "field": "value" }"#),
+            },
+            Example {
+                title: "array",
+                source: r#"parse_cbor!(decode_base64!("gvUA"))"#,
+                result: Ok("[true, 0]"),
+            },
+            Example {
+                title: "string",
+                source: r#"parse_cbor!(decode_base64!("ZWhlbGxv"))"#,
+                result: Ok("hello"),
+            },
+            Example {
+                title: "integer",
+                source: r#"parse_cbor!(decode_base64!("GCo="))"#,
+                result: Ok("42"),
+            },
+            Example {
+                title: "float",
+                source: r#"parse_cbor!(decode_base64!("+0BFEKPXCj1x"))"#,
+                result: Ok("42.13"),
+            },
+            Example {
+                title: "boolean",
+                source: r#"parse_cbor!(decode_base64!("9A=="))"#,
+                result: Ok("false"),
+            },
+        ]
     }
 
     fn compile(
@@ -65,31 +93,10 @@ impl FunctionExpression for ParseCborFn {
     }
 
     fn type_def(&self, _: &state::TypeState) -> TypeDef {
-        type_def()
+        json_type_def()
     }
 }
 
-fn inner_kind() -> Kind {
-    Kind::null()
-        | Kind::bytes()
-        | Kind::integer()
-        | Kind::float()
-        | Kind::boolean()
-        | Kind::array(Collection::any())
-        | Kind::object(Collection::any())
-}
-
-fn type_def() -> TypeDef {
-    TypeDef::bytes()
-        .fallible()
-        .or_boolean()
-        .or_integer()
-        .or_float()
-        .add_null()
-        .or_array(Collection::from_unknown(inner_kind()))
-        .or_object(Collection::from_unknown(inner_kind()))
-}
-
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -113,13 +120,13 @@ mod tests {
         parses {
             args: func_args![ value: value!(read_cbor_file("simple.cbor").as_bytes()) ],
             want: Ok(value!({ field: "value" })),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
 
         complex_cbor {
             args: func_args![ value: value!(read_cbor_file("complex.cbor").as_bytes()) ],
             want: Ok(value!({ object: {string: "value", number: 42, array: ["hello", "world"], boolean: false} })),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
     ];
 }
diff --git a/src/stdlib/parse_json.rs b/src/stdlib/parse_json.rs
@@ -6,6 +6,7 @@ use serde_json::{
 };
 
 use crate::compiler::prelude::*;
+use crate::stdlib::json_utils::json_type_def::json_type_def;
 
 fn parse_json(value: Value, lossy: Option<Value>) -> Resolved {
     let lossy = lossy.map(Value::try_boolean).transpose()?.unwrap_or(true);
@@ -226,7 +227,7 @@ impl FunctionExpression for ParseJsonFn {
     }
 
     fn type_def(&self, _: &state::TypeState) -> TypeDef {
-        type_def()
+        json_type_def()
     }
 }
 
@@ -250,31 +251,10 @@ impl FunctionExpression for ParseJsonMaxDepthFn {
     }
 
     fn type_def(&self, _: &state::TypeState) -> TypeDef {
-        type_def()
+        json_type_def()
     }
 }
 
-fn inner_kind() -> Kind {
-    Kind::null()
-        | Kind::bytes()
-        | Kind::integer()
-        | Kind::float()
-        | Kind::boolean()
-        | Kind::array(Collection::any())
-        | Kind::object(Collection::any())
-}
-
-fn type_def() -> TypeDef {
-    TypeDef::bytes()
-        .fallible()
-        .or_boolean()
-        .or_integer()
-        .or_float()
-        .add_null()
-        .or_array(Collection::from_unknown(inner_kind()))
-        .or_object(Collection::from_unknown(inner_kind()))
-}
-
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -286,74 +266,62 @@ mod tests {
         parses {
             args: func_args![ value: r#"{"field": "value"}"# ],
             want: Ok(value!({ field: "value" })),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
 
         complex_json {
             args: func_args![ value: r#"{"object": {"string":"value","number":42,"array":["hello","world"],"boolean":false}}"# ],
             want: Ok(value!({ object: {string: "value", number: 42, array: ["hello", "world"], boolean: false} })),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
 
         invalid_json_errors {
             args: func_args![ value: r#"{"field": "value"# ],
             want: Err("unable to parse json: EOF while parsing a string at line 1 column 16"),
-            tdef: TypeDef::bytes().fallible()
-                .or_boolean()
-                .or_integer()
-                .or_float()
-                .or_null()
-                .or_array(Collection::from_unknown(inner_kind()))
-                .or_object(Collection::from_unknown(inner_kind())),
+            tdef: json_type_def(),
         }
 
         max_depth {
             args: func_args![ value: r#"{"top_layer": {"layer_one": "finish", "layer_two": 2}}"#, max_depth: 1],
             want: Ok(value!({ top_layer: r#"{"layer_one": "finish", "layer_two": 2}"# })),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
 
         max_depth_array {
             args: func_args![ value: r#"[{"top_layer": {"next_layer": ["finish"]}}]"#, max_depth: 2],
             want: Ok(value!([{ top_layer: r#"{"next_layer": ["finish"]}"# }])),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
 
         max_depth_exceeds_layers {
             args: func_args![ value: r#"{"top_layer": {"layer_one": "finish", "layer_two": 2}}"#, max_depth: 10],
             want: Ok(value!({ top_layer: {layer_one: "finish", layer_two: 2} })),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
 
         invalid_json_with_max_depth {
             args: func_args![ value: r#"{"field": "value"#, max_depth: 3 ],
             want: Err("unable to read json: EOF while parsing a string at line 1 column 16"),
-            tdef: TypeDef::bytes().fallible()
-                .or_boolean()
-                .or_integer()
-                .or_float()
-                .or_null()
-                .or_array(Collection::from_unknown(inner_kind()))
-                .or_object(Collection::from_unknown(inner_kind())),
+            tdef: json_type_def(),
         }
 
         invalid_input_max_depth {
             args: func_args![ value: r#"{"top_layer": "finish"}"#, max_depth: 129],
             want: Err("max_depth value should be greater than 0 and less than 128, got 129"),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
 
         // // TODO: provide a function version of the `test_function!` macro.
         max_int {
             args: func_args![ value: format!("{{\"num\": {}}}", i64::MAX - 1)],
             want: Ok(value!({"num": 9_223_372_036_854_775_806_i64})),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
 
         lossy_float_conversion {
             args: func_args![ value: r#"{"num": 9223372036854775808}"#],
             want: Ok(value!({"num": 9.223_372_036_854_776e18})),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
 
         // Checks that the parsing uses the default lossy argument value
@@ -362,21 +330,21 @@ mod tests {
             // 0xf5 is out of the range of valid UTF-8 bytes
             args: func_args![ value: Bytes::from_static(&[0x22,0xf5,0x22])],
             want: Ok(value!(std::char::REPLACEMENT_CHARACTER.to_string())),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
 
         parse_invalid_utf8_lossy_arg_true {
             // 0xf5 is out of the range of valid UTF-8 bytes
             args: func_args![ value: Bytes::from_static(&[0x22,0xf5,0x22]), lossy: true],
             // U+FFFD is the replacement character for invalid UTF-8
             want: Ok(value!(std::char::REPLACEMENT_CHARACTER.to_string())),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
 
         invalid_utf8_json_lossy_arg_false {
             args: func_args![ value: Bytes::from_static(&[0x22,0xf5,0x22]), lossy: false],
             want: Err("unable to parse json: invalid unicode code point at line 1 column 3"),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
     ];
 
@@ -387,7 +355,7 @@ mod tests {
         no_roundtrip_float_conversion {
             args: func_args![ value: r#"{"num": 1626175065.5934923}"#],
             want: Ok(value!({"num": 1_626_175_065.593_492_5})),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
     ];
 
@@ -398,7 +366,7 @@ mod tests {
         roundtrip_float_conversion {
             args: func_args![ value: r#"{"num": 1626175065.5934923}"#],
             want: Ok(value!({"num": 1_626_175_065.593_492_3})),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
     ];
 }
diff --git a/src/stdlib/parse_proto.rs b/src/stdlib/parse_proto.rs
@@ -1,6 +1,7 @@
 use crate::compiler::prelude::*;
 use crate::protobuf::get_message_descriptor;
 use crate::protobuf::parse_proto;
+use crate::stdlib::json_utils::json_type_def::json_type_def;
 use once_cell::sync::Lazy;
 use prost_reflect::MessageDescriptor;
 use std::env;
@@ -112,31 +113,10 @@ impl FunctionExpression for ParseProtoFn {
     }
 
     fn type_def(&self, _: &state::TypeState) -> TypeDef {
-        type_def()
+        json_type_def()
     }
 }
 
-fn inner_kind() -> Kind {
-    Kind::null()
-        | Kind::bytes()
-        | Kind::integer()
-        | Kind::float()
-        | Kind::boolean()
-        | Kind::array(Collection::any())
-        | Kind::object(Collection::any())
-}
-
-fn type_def() -> TypeDef {
-    TypeDef::bytes()
-        .fallible()
-        .or_boolean()
-        .or_integer()
-        .or_float()
-        .add_null()
-        .or_array(Collection::from_unknown(inner_kind()))
-        .or_object(Collection::from_unknown(inner_kind()))
-}
-
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -159,15 +139,15 @@ mod tests {
                 desc_file: test_data_dir().join("test_protobuf.desc").to_str().unwrap().to_owned(),
                 message_type: "test_protobuf.Person"],
             want: Ok(value!({ name: "someone", phones: [{number: "123456"}] })),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
 
         parses_proto3 {
             args: func_args![ value: read_pb_file("person_someone3.pb"),
                 desc_file: test_data_dir().join("test_protobuf3.desc").to_str().unwrap().to_owned(),
                 message_type: "test_protobuf3.Person"],
             want: Ok(value!({ data: {data_phone: "HOME"}, name: "someone", phones: [{number: "1234", type: "MOBILE"}] })),
-            tdef: type_def(),
+            tdef: json_type_def(),
         }
     ];
 }