RUST-1687 Add human_readable_serialization option to Collection (#902)

Co-authored-by: Isabel Atkinson <isabel.atkinson@mongodb.com>

Author: maiconpavi

src/coll.rs

@@ -143,6 +143,7 @@ struct CollectionInner {
     selection_criteria: Option<SelectionCriteria>,
     read_concern: Option<ReadConcern>,
     write_concern: Option<WriteConcern>,
+    human_readable_serialization: bool,
 }
 
 impl<T> Collection<T> {
@@ -157,6 +158,7 @@ impl<T> Collection<T> {
         let write_concern = options
             .write_concern
             .or_else(|| db.write_concern().cloned());
+        let human_readable_serialization = options.human_readable_serialization.unwrap_or_default();
 
         Self {
             inner: Arc::new(CollectionInner {
@@ -166,6 +168,7 @@ impl<T> Collection<T> {
                 selection_criteria,
                 read_concern,
                 write_concern,
+                human_readable_serialization,
             }),
             _phantom: Default::default(),
         }
@@ -1119,14 +1122,16 @@ where
         options: impl Into<Option<FindOneAndReplaceOptions>>,
         session: impl Into<Option<&mut ClientSession>>,
     ) -> Result<Option<T>> {
+        let mut options = options.into();
         let replacement = to_document_with_options(
             replacement.borrow(),
-            SerializerOptions::builder().human_readable(false).build(),
+            SerializerOptions::builder()
+                .human_readable(self.inner.human_readable_serialization)
+                .build(),
         )?;
 
         let session = session.into();
 
-        let mut options = options.into();
         resolve_write_concern_with_session!(self, options, session.as_ref())?;
 
         let op = FindAndModify::<T>::with_replace(self.namespace(), filter, replacement, options)?;
@@ -1205,7 +1210,13 @@ where
 
         while n_attempted < ds.len() {
             let docs: Vec<&T> = ds.iter().skip(n_attempted).map(Borrow::borrow).collect();
-            let insert = Insert::new_encrypted(self.namespace(), docs, options.clone(), encrypted);
+            let insert = Insert::new_encrypted(
+                self.namespace(),
+                docs,
+                options.clone(),
+                encrypted,
+                self.inner.human_readable_serialization,
+            );
 
             match self
                 .client()
@@ -1332,6 +1343,7 @@ where
             self.namespace(),
             vec![doc],
             options.map(InsertManyOptions::from_insert_one_options),
+            self.inner.human_readable_serialization,
         );
         self.client()
             .execute_operation(insert, session)
@@ -1382,16 +1394,18 @@ where
         options: impl Into<Option<ReplaceOptions>>,
         session: impl Into<Option<&mut ClientSession>>,
     ) -> Result<UpdateResult> {
+        let mut options = options.into();
         let replacement = to_document_with_options(
             replacement.borrow(),
-            SerializerOptions::builder().human_readable(false).build(),
+            SerializerOptions::builder()
+                .human_readable(self.inner.human_readable_serialization)
+                .build(),
         )?;
 
         bson_util::replacement_document_check(&replacement)?;
 
         let session = session.into();
 
-        let mut options = options.into();
         resolve_write_concern_with_session!(self, options, session.as_ref())?;
 
         let update = Update::new(

src/coll/options.rs

@@ -28,6 +28,11 @@ pub struct CollectionOptions {
 
     /// The default write concern for operations.
     pub write_concern: Option<WriteConcern>,
+
+    /// Sets the [`bson::SerializerOptions::human_readable`] option for the [`Bson`] serializer.
+    /// The default value is `false`.
+    /// Note: Specifying `true` for this value will decrease the performance of insert operations.
+    pub human_readable_serialization: Option<bool>,
 }
 
 /// Specifies whether a

src/operation/insert.rs

@@ -31,29 +31,33 @@ pub(crate) struct Insert<'a, T> {
     inserted_ids: Vec<Bson>,
     options: Option<InsertManyOptions>,
     encrypted: bool,
+    human_readable_serialization: bool,
 }
 
 impl<'a, T> Insert<'a, T> {
     pub(crate) fn new(
         ns: Namespace,
         documents: Vec<&'a T>,
         options: Option<InsertManyOptions>,
+        human_readable_serialization: bool,
     ) -> Self {
-        Self::new_encrypted(ns, documents, options, false)
+        Self::new_encrypted(ns, documents, options, false, human_readable_serialization)
     }
 
     pub(crate) fn new_encrypted(
         ns: Namespace,
         documents: Vec<&'a T>,
         options: Option<InsertManyOptions>,
         encrypted: bool,
+        human_readable_serialization: bool,
     ) -> Self {
         Self {
             ns,
             options,
             documents,
             inserted_ids: vec![],
             encrypted,
+            human_readable_serialization,
         }
     }
 
@@ -82,7 +86,17 @@ impl<'a, T: Serialize> OperationWithDefaults for Insert<'a, T> {
             .take(description.max_write_batch_size as usize)
             .enumerate()
         {
-            let mut doc = bson::to_raw_document_buf(d)?;
+            let mut doc = if self.human_readable_serialization {
+                let serializer_options = bson::SerializerOptions::builder()
+                    .human_readable(true)
+                    .build();
+                bson::RawDocumentBuf::from_document(&bson::to_document_with_options(
+                    d,
+                    serializer_options,
+                )?)?
+            } else {
+                bson::to_raw_document_buf(d)?
+            };
             let id = match doc.get("_id")? {
                 Some(b) => b.try_into()?,
                 None => {

src/operation/insert/test.rs

@@ -49,6 +49,7 @@ fn fixtures(opts: Option<InsertManyOptions>) -> TestFixtures {
         },
         DOCUMENTS.iter().collect(),
         Some(options.clone()),
+        false,
     );
 
     TestFixtures {
@@ -116,7 +117,7 @@ fn build() {
 #[test]
 fn build_ordered() {
     let docs = vec![Document::new()];
-    let mut insert = Insert::new(Namespace::empty(), docs.iter().collect(), None);
+    let mut insert = Insert::new(Namespace::empty(), docs.iter().collect(), None, false);
     let cmd = insert
         .build(&StreamDescription::new_testing())
         .expect("should succeed");
@@ -128,6 +129,7 @@ fn build_ordered() {
         Namespace::empty(),
         docs.iter().collect(),
         Some(InsertManyOptions::builder().ordered(false).build()),
+        false,
     );
     let cmd = insert
         .build(&StreamDescription::new_testing())
@@ -140,6 +142,7 @@ fn build_ordered() {
         Namespace::empty(),
         docs.iter().collect(),
         Some(InsertManyOptions::builder().ordered(true).build()),
+        false,
     );
     let cmd = insert
         .build(&StreamDescription::new_testing())
@@ -152,6 +155,7 @@ fn build_ordered() {
         Namespace::empty(),
         docs.iter().collect(),
         Some(InsertManyOptions::builder().build()),
+        false,
     );
     let cmd = insert
         .build(&StreamDescription::new_testing())
@@ -170,7 +174,7 @@ struct Documents<D> {
 fn generate_ids() {
     let docs = vec![doc! { "x": 1 }, doc! { "_id": 1_i32, "x": 2 }];
 
-    let mut insert = Insert::new(Namespace::empty(), docs.iter().collect(), None);
+    let mut insert = Insert::new(Namespace::empty(), docs.iter().collect(), None, false);
     let cmd = insert.build(&StreamDescription::new_testing()).unwrap();
     let serialized = insert.serialize_command(cmd).unwrap();
 
@@ -253,7 +257,7 @@ fn serialize_all_types() {
         "_id": ObjectId::new(),
     }];
 
-    let mut insert = Insert::new(Namespace::empty(), docs.iter().collect(), None);
+    let mut insert = Insert::new(Namespace::empty(), docs.iter().collect(), None, false);
     let cmd = insert.build(&StreamDescription::new_testing()).unwrap();
     let serialized = insert.serialize_command(cmd).unwrap();
     let cmd: Documents<Document> = bson::from_slice(serialized.as_slice()).unwrap();

src/test/coll.rs

@@ -1214,3 +1214,124 @@ fn test_namespace_fromstr() {
     assert_eq!(t.db, "something");
     assert_eq!(t.coll, "something.else");
 }
+
+#[cfg_attr(feature = "tokio-runtime", tokio::test)]
+#[cfg_attr(feature = "async-std-runtime", async_std::test)]
+async fn configure_human_readable_serialization() {
+    #[derive(Deserialize)]
+    struct StringOrBytes(String);
+
+    impl Serialize for StringOrBytes {
+        fn serialize<S>(&self, serializer: S) -> std::result::Result<S::Ok, S::Error>
+        where
+            S: serde::Serializer,
+        {
+            if serializer.is_human_readable() {
+                serializer.serialize_str(&self.0)
+            } else {
+                serializer.serialize_bytes(self.0.as_bytes())
+            }
+        }
+    }
+
+    #[derive(Deserialize, Serialize)]
+    struct Data {
+        id: u32,
+        s: StringOrBytes,
+    }
+
+    let client = TestClient::new().await;
+
+    let collection_options = CollectionOptions::builder()
+        .human_readable_serialization(false)
+        .build();
+
+    let non_human_readable_collection: Collection<Data> = client
+        .database("db")
+        .collection_with_options("nonhumanreadable", collection_options);
+    non_human_readable_collection.drop(None).await.unwrap();
+
+    non_human_readable_collection
+        .insert_one(
+            Data {
+                id: 0,
+                s: StringOrBytes("non human readable!".into()),
+            },
+            None,
+        )
+        .await
+        .unwrap();
+
+    // The inserted bytes will not deserialize to StringOrBytes properly, so find as a document
+    // instead.
+    let document_collection = non_human_readable_collection.clone_with_type::<Document>();
+    let doc = document_collection
+        .find_one(doc! { "id": 0 }, None)
+        .await
+        .unwrap()
+        .unwrap();
+    assert!(doc.get_binary_generic("s").is_ok());
+
+    non_human_readable_collection
+        .replace_one(
+            doc! { "id": 0 },
+            Data {
+                id: 1,
+                s: StringOrBytes("non human readable!".into()),
+            },
+            None,
+        )
+        .await
+        .unwrap();
+
+    let doc = document_collection
+        .find_one(doc! { "id": 1 }, None)
+        .await
+        .unwrap()
+        .unwrap();
+    assert!(doc.get_binary_generic("s").is_ok());
+
+    let collection_options = CollectionOptions::builder()
+        .human_readable_serialization(true)
+        .build();
+
+    let human_readable_collection: Collection<Data> = client
+        .database("db")
+        .collection_with_options("humanreadable", collection_options);
+    human_readable_collection.drop(None).await.unwrap();
+
+    human_readable_collection
+        .insert_one(
+            Data {
+                id: 0,
+                s: StringOrBytes("human readable!".into()),
+            },
+            None,
+        )
+        .await
+        .unwrap();
+
+    // Proper deserialization to a string demonstrates that the data was correctly serialized as a
+    // string.
+    human_readable_collection
+        .find_one(doc! { "id": 0 }, None)
+        .await
+        .unwrap();
+
+    human_readable_collection
+        .replace_one(
+            doc! { "id": 0 },
+            Data {
+                id: 1,
+                s: StringOrBytes("human readable!".into()),
+            },
+            None,
+        )
+        .await
+        .unwrap();
+
+    human_readable_collection
+        .find_one(doc! { "id": 1 }, None)
+        .await
+        .unwrap();
+}

src/test/spec/unified_runner/test_file.rs

@@ -378,6 +378,7 @@ impl CollectionOrDatabaseOptions {
             read_concern: self.read_concern.clone(),
             selection_criteria: self.selection_criteria.clone(),
             write_concern: self.write_concern.clone(),
+            human_readable_serialization: None,
         }
     }
 }