Support partial serialization & deserialization

This feature allows writing structs which contain `Document` nodes.

- When deserializing, everything under the `Document` node is simply
  attached to the struct.
- When serializing, the `Document` node is passed through unmodified.

Signed-off-by: Chris Frantz <cfrantz@google.com>

Author: Chris Frantz

src/annotate.rs

@@ -1,6 +1,6 @@
 use std::fmt;
 
-use crate::{AnnotatedSerializer, Document, Error};
+use crate::{AnnotatedSerializer, Deserializer, Document, Error};
 
 /// Specifies the formatting options to use when serializing.
 pub enum Format {
@@ -61,18 +61,43 @@ impl<T: ?Sized + serde::Serialize> Annotate for T {
 }
 
 // We use a private trait to identify whether the Serializer passed to
-// serde::Serialize for dyn Annotate is AnnotatedSerializer.
-unsafe trait IsAnnotatedSerializer {
-    fn is_annotated_serializer(&self) -> bool;
+// various functions is our Serializer.
+pub(crate) unsafe trait IsSerializer {
+    fn is_serde_annotate(&self) -> bool;
 }
 
-unsafe impl<T: serde::Serializer> IsAnnotatedSerializer for T {
-    default fn is_annotated_serializer(&self) -> bool {
+unsafe impl<T: serde::Serializer> IsSerializer for T {
+    default fn is_serde_annotate(&self) -> bool {
         false
     }
 }
-unsafe impl<'a> IsAnnotatedSerializer for &mut AnnotatedSerializer<'a> {
-    fn is_annotated_serializer(&self) -> bool {
+
+unsafe impl<'a> IsSerializer for &mut AnnotatedSerializer<'a> {
+    fn is_serde_annotate(&self) -> bool {
+        true
+    }
+}
+
+// This marker trait is to avoid specifying lifetimes in the default
+// implementation.  When I specify lifetimes in the default impl, the
+// compiler complains that the specialized impl repeats parameter `'de`.
+trait _IsDeserializer {}
+impl<'de, T: serde::Deserializer<'de>> _IsDeserializer for T {}
+
+// We use a private trait to identify whether the Deserializer passed to
+// various functions is our Deserializer.
+pub(crate) unsafe trait IsDeserializer {
+    fn is_serde_annotate(&self) -> bool;
+}
+
+unsafe impl<T: _IsDeserializer> IsDeserializer for T {
+    default fn is_serde_annotate(&self) -> bool {
+        false
+    }
+}
+
+unsafe impl<'de> IsDeserializer for &mut Deserializer<'de> {
+    fn is_serde_annotate(&self) -> bool {
         true
     }
 }
@@ -89,7 +114,7 @@ unsafe impl<'a> IsAnnotatedSerializer for &mut AnnotatedSerializer<'a> {
 // AnnotatedSerializer and just force the types with `transmute`.
 impl serde::Serialize for dyn Annotate {
     fn serialize<S: serde::Serializer>(&self, serializer: S) -> Result<S::Ok, S::Error> {
-        if !serializer.is_annotated_serializer() {
+        if !serializer.is_serde_annotate() {
             panic!(
                 "Expected to be called by AnnotatedSerializer, not {:?}",
                 std::any::type_name::<S>()

src/de.rs

@@ -59,7 +59,7 @@ impl Deserialize {
 
 /// A `Deserializer` deserializes a parsed document.
 pub struct Deserializer<'de> {
-    doc: &'de Document,
+    pub(crate) doc: &'de Document,
 }
 
 impl<'de> Deserializer<'de> {

src/lib.rs

@@ -9,6 +9,7 @@ mod error;
 mod hexdump;
 mod integer;
 mod json;
+mod partial;
 mod relax;
 mod ser;
 mod yaml;
@@ -18,7 +19,7 @@ pub use annotate_derive::*;
 pub use color::ColorProfile;
 pub use de::{from_str, Deserialize, Deserializer};
 pub use doc_iter::DocPath;
-pub use document::Document;
+pub use document::{BytesFormat, CommentFormat, Document, StrFormat};
 pub use error::Error;
 pub use integer::{Int, IntValue};
 pub use json::Json;

src/partial.rs

@@ -0,0 +1,50 @@
+use serde::{Deserialize, Deserializer, Serialize, Serializer};
+
+use crate::annotate::{IsDeserializer, IsSerializer};
+use crate::Deserializer as AnnotatedDeserializer;
+use crate::{Document, Error};
+
+impl Serialize for Document {
+    fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
+    where
+        S: Serializer,
+    {
+        if serializer.is_serde_annotate() {
+            unsafe {
+                // If `serializer` is the correct type, then we can clone the
+                // Document node and return it.
+                let r: Result<Document, Error> = Ok(self.clone());
+                // We have to transmute because the we can't determine at compile
+                // time that `Result<Document, Error>` is the same type as
+                // `Result<S::Ok, S::Error>`.  If the serializer is
+                // `AnnotatedSerializer`, then it must be the same.
+                let result = std::mem::transmute_copy(&r);
+                std::mem::forget(r);
+                result
+            }
+        } else {
+            Err(serde::ser::Error::custom("Serializing document nodes is only supported with serde_annotate::AnnotatedSerializer"))
+        }
+    }
+}
+
+impl<'de> Deserialize<'de> for Document {
+    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
+    where
+        D: Deserializer<'de>,
+    {
+        if deserializer.is_serde_annotate() {
+            unsafe {
+                // If the deserializer is ours, then we can simply clone the
+                // deserializer's document node.
+                let dsz: &AnnotatedDeserializer = std::mem::transmute_copy(&deserializer);
+                std::mem::forget(deserializer);
+                Ok(dsz.doc.clone())
+            }
+        } else {
+            Err(serde::de::Error::custom(
+                "Deserializing document nodes is only supported with serde_annotate::Deserializer",
+            ))
+        }
+    }
+}

tests/BUILD.bazel

@@ -40,3 +40,15 @@ rust_test(
         "//third_party/rust/crates:serde",
     ],
 )
+
+rust_test(
+    name = "test_partial",
+    srcs = ["test_partial.rs"],
+    edition = "2021",
+    deps = [
+        "//:serde_annotate",
+        "//third_party/rust/crates:anyhow",
+        "//third_party/rust/crates:serde",
+        "//third_party/rust/crates:serde_json",
+    ],
+)

tests/test_partial.rs

@@ -0,0 +1,88 @@
+#![feature(min_specialization)]
+use anyhow::Result;
+use serde_annotate::serialize;
+use serde_annotate::{Document, StrFormat};
+
+#[derive(Debug, serde::Serialize, serde::Deserialize)]
+struct Partial {
+    n: i32,
+    doc: Document,
+}
+
+const SERIALIZE_RESULT: &str = r#"{
+  n: 5,
+  doc: [
+    "Hello",
+    "world"
+  ]
+}"#;
+
+#[test]
+fn test_partial_serialize() -> Result<()> {
+    let p = Partial {
+        n: 5,
+        doc: Document::Sequence(vec![
+            Document::String("Hello".into(), StrFormat::Standard),
+            Document::String("world".into(), StrFormat::Standard),
+        ]),
+    };
+    let s = serialize(&p)?.to_json5().to_string();
+    assert_eq!(s, SERIALIZE_RESULT);
+    Ok(())
+}
+
+#[test]
+fn test_partial_serialize_error() -> Result<()> {
+    let p = Partial {
+        n: 5,
+        doc: Document::Sequence(vec![
+            Document::String("Hello".into(), StrFormat::Standard),
+            Document::String("world".into(), StrFormat::Standard),
+        ]),
+    };
+    let s = serde_json::to_string_pretty(&p);
+    assert!(s.is_err());
+    assert_eq!(
+        s.unwrap_err().to_string(),
+        "Serializing document nodes is only supported with serde_annotate::AnnotatedSerializer"
+    );
+    Ok(())
+}
+
+#[test]
+fn test_partial_deserialize() -> Result<()> {
+    let doc = r#"{
+        n: 10,
+        doc: {
+            # A comment
+            key: "value",
+            i: 5,
+            j: 10,
+        }
+    }"#;
+    let p = serde_annotate::from_str::<Partial>(doc)?;
+    assert_eq!(p.n, 10);
+    if let Document::Mapping(m) = p.doc {
+        let (k, v) = m[0].as_kv()?;
+        assert_eq!(k.as_str()?, "key");
+        assert_eq!(v.as_str()?, "value");
+        let (k, v) = m[1].as_kv()?;
+        assert_eq!(k.as_str()?, "i");
+        assert_eq!(u32::try_from(v)?, 5);
+        let (k, v) = m[2].as_kv()?;
+        assert_eq!(k.as_str()?, "j");
+        assert_eq!(u32::try_from(v)?, 10);
+    } else {
+        assert!(false, "Expecting Document::Mapping");
+    }
+    Ok(())
+}
+
+#[test]
+fn test_partial_deserialize_error() -> Result<()> {
+    let p = serde_json::from_str::<Partial>(r#"{"n":5, "doc": []}"#);
+    assert!(p.is_err());
+    assert_eq!(p.unwrap_err().to_string(),
+        "Deserializing document nodes is only supported with serde_annotate::Deserializer at line 1 column 15");
+    Ok(())
+}