RUST-536 Eliminate redundant clones (#377)

Author: patrickfreed

benchmarks/Cargo.toml

@@ -6,7 +6,7 @@ edition = "2018"
 
 [features]
 default = ["tokio-runtime"]
-tokio-runtime = ["tokio/fs", "tokio/macros", "tokio/rt-core", "tokio/rt-threaded"]
+tokio-runtime = ["tokio/fs", "tokio/macros", "tokio/rt", "tokio/rt-multi-thread", "tokio-stream"]
 async-std-runtime = ["async-std"]
 
 [dependencies]
@@ -16,8 +16,9 @@ lazy_static = "1.4.0"
 clap = "2.33.3"
 indicatif = "0.15.0"
 async-trait = "0.1.41"
-tokio = { version = "0.2.23", features = ["sync"] }
+tokio = { version = "1.6", features = ["sync"] }
+tokio-stream = { version = "0.1.6", features = ["io-util"], optional = true }
 # "unstable" feature is needed for `spawn_blocking`, which is only used in task setup
-async-std = { version = "=1.6.2", optional = true, features = ["attributes", "unstable"] }
+async-std = { version = "1.9.0", optional = true, features = ["attributes", "unstable"] }
 futures = "0.3.8"
 anyhow = "1.0.34"

benchmarks/src/bench/find_many.rs

@@ -2,7 +2,7 @@ use std::{convert::TryInto, path::PathBuf};
 
 use anyhow::{bail, Result};
 use futures::stream::StreamExt;
-use mongodb::{bson::Bson, Client, Collection, Database};
+use mongodb::{Client, Collection, Database, bson::{Bson, Document}};
 use serde_json::Value;
 
 use crate::{
@@ -12,7 +12,7 @@ use crate::{
 
 pub struct FindManyBenchmark {
     db: Database,
-    coll: Collection,
+    coll: Collection<Document>,
 }
 
 // Specifies the options to `FindManyBenchmark::setup` operation.

benchmarks/src/bench/find_one.rs

@@ -1,12 +1,7 @@
 use std::{convert::TryInto, path::PathBuf};
 
 use anyhow::{bail, Result};
-use mongodb::{
-    bson::{doc, Bson},
-    Client,
-    Collection,
-    Database,
-};
+use mongodb::{Client, Collection, Database, bson::{Bson, Document, doc}};
 use serde_json::Value;
 
 use crate::{
@@ -17,7 +12,7 @@ use crate::{
 pub struct FindOneBenchmark {
     db: Database,
     num_iter: usize,
-    coll: Collection,
+    coll: Collection<Document>,
 }
 
 // Specifies the options to a `FindOneBenchmark::setup` operation.

benchmarks/src/bench/insert_many.rs

@@ -14,7 +14,7 @@ use crate::bench::{Benchmark, COLL_NAME, DATABASE_NAME};
 pub struct InsertManyBenchmark {
     db: Database,
     num_copies: usize,
-    coll: Collection,
+    coll: Collection<Document>,
     doc: Document,
 }
 
@@ -45,7 +45,7 @@ impl Benchmark for InsertManyBenchmark {
         let mut file = spawn_blocking_and_await!(File::open(options.path))?;
         let json: Value = spawn_blocking_and_await!(serde_json::from_reader(&mut file))?;
 
-        let coll = db.collection(&COLL_NAME);
+        let coll = db.collection(COLL_NAME.as_str());
 
         Ok(InsertManyBenchmark {
             db,
@@ -60,13 +60,13 @@ impl Benchmark for InsertManyBenchmark {
 
     async fn before_task(&mut self) -> Result<()> {
         self.coll.drop(None).await?;
-        self.db.create_collection(&COLL_NAME, None).await?;
+        self.db.create_collection(COLL_NAME.as_str(), None).await?;
 
         Ok(())
     }
 
     async fn do_task(&self) -> Result<()> {
-        let insertions = vec![self.doc.clone(); self.num_copies];
+        let insertions = vec![&self.doc; self.num_copies];
         self.coll.insert_many(insertions, None).await?;
 
         Ok(())

benchmarks/src/bench/insert_one.rs

@@ -14,7 +14,7 @@ use crate::bench::{Benchmark, COLL_NAME, DATABASE_NAME};
 pub struct InsertOneBenchmark {
     db: Database,
     num_iter: usize,
-    coll: Collection,
+    coll: Collection<Document>,
     doc: Document,
 }
 
@@ -60,14 +60,14 @@ impl Benchmark for InsertOneBenchmark {
 
     async fn before_task(&mut self) -> Result<()> {
         self.coll.drop(None).await?;
-        self.db.create_collection(&COLL_NAME, None).await?;
+        self.db.create_collection(COLL_NAME.as_str(), None).await?;
 
         Ok(())
     }
 
     async fn do_task(&self) -> Result<()> {
         for _ in 0..self.num_iter {
-            self.coll.insert_one(self.doc.clone(), None).await?;
+            self.coll.insert_one(&self.doc, None).await?;
         }
 
         Ok(())

benchmarks/src/bench/json_multi_export.rs

@@ -2,7 +2,7 @@ use std::path::PathBuf;
 
 use anyhow::Result;
 use futures::stream::{FuturesUnordered, StreamExt, TryStreamExt};
-use mongodb::{bson::doc, Client, Collection, Database};
+use mongodb::{Client, Collection, Database, bson::{Document, doc}};
 
 use crate::{
     bench::{parse_json_file_to_documents, Benchmark, COLL_NAME, DATABASE_NAME},
@@ -13,7 +13,7 @@ const TOTAL_FILES: usize = 100;
 
 pub struct JsonMultiExportBenchmark {
     db: Database,
-    coll: Collection,
+    coll: Collection<Document>,
 }
 
 // Specifies the options to a `JsonMultiExportBenchmark::setup` operation.

benchmarks/src/bench/json_multi_import.rs

@@ -2,7 +2,7 @@ use std::path::PathBuf;
 
 use anyhow::Result;
 use futures::stream::{FuturesUnordered, StreamExt};
-use mongodb::{options::InsertManyOptions, Client, Collection, Database};
+use mongodb::{Client, Collection, Database, bson::Document, options::InsertManyOptions};
 
 use crate::{
     bench::{parse_json_file_to_documents, Benchmark, COLL_NAME, DATABASE_NAME},
@@ -13,7 +13,7 @@ const TOTAL_FILES: usize = 100;
 
 pub struct JsonMultiImportBenchmark {
     db: Database,
-    coll: Collection,
+    coll: Collection<Document>,
     path: PathBuf,
 }
 
@@ -43,7 +43,7 @@ impl Benchmark for JsonMultiImportBenchmark {
 
     async fn before_task(&mut self) -> Result<()> {
         self.coll.drop(None).await?;
-        self.db.create_collection(&COLL_NAME, None).await?;
+        self.db.create_collection(COLL_NAME.as_str(), None).await?;
 
         Ok(())
     }
@@ -68,7 +68,7 @@ impl Benchmark for JsonMultiImportBenchmark {
 
                 docs.append(&mut new_docs);
 
-                let opts = Some(InsertManyOptions::builder().ordered(Some(false)).build());
+                let opts = Some(InsertManyOptions::builder().ordered(false).build());
                 coll_ref.insert_many(docs, opts).await.unwrap();
             });
         }

benchmarks/src/fs.rs

@@ -59,6 +59,14 @@ impl BufReader {
     }
 
     pub(crate) fn lines(self) -> impl Stream<Item = std::io::Result<String>> {
-        self.inner.lines()
+        #[cfg(feature = "tokio-runtime")]
+        {
+            tokio_stream::wrappers::LinesStream::new(self.inner.lines())
+        }
+
+        #[cfg(feature = "async-std-runtime")]
+        {
+            self.inner.lines()
+        }
     }
 }

benchmarks/src/main.rs

@@ -37,7 +37,6 @@ use crate::bench::{
     json_multi_export::JsonMultiExportBenchmark,
     json_multi_import::JsonMultiImportBenchmark,
     run_command::RunCommandBenchmark,
-    TARGET_ITERATION_COUNT,
 };
 
 lazy_static! {
@@ -52,7 +51,12 @@ fn get_nth_percentile(durations: &[Duration], n: f64) -> Duration {
 fn score_test(durations: Vec<Duration>, name: &str, task_size: f64, more_info: bool) -> f64 {
     let median = get_nth_percentile(&durations, 50.0);
     let score = task_size / (median.as_millis() as f64 / 1000.0);
-    println!("TEST: {} -- Score: {}\n", name, score);
+    println!(
+        "TEST: {} -- Score: {}, Median Iteration Time: {:.3}ms\n",
+        name,
+        score,
+        median.as_secs_f64()
+    );
 
     if more_info {
         println!(

src/bson_util/mod.rs

@@ -5,7 +5,7 @@ use std::{convert::TryFrom, time::Duration};
 use serde::{de::Error, ser, Deserialize, Deserializer, Serialize, Serializer};
 
 use crate::{
-    bson::{doc, oid::ObjectId, Binary, Bson, Document, JavaScriptCodeWithScope, Regex},
+    bson::{doc, Binary, Bson, Document, JavaScriptCodeWithScope, Regex},
     error::{ErrorKind, Result},
 };
 
@@ -31,11 +31,6 @@ pub(crate) fn get_u64(val: &Bson) -> Option<u64> {
     }
 }
 
-pub(crate) fn add_id(doc: &mut Document) {
-    doc.entry("_id".to_string())
-        .or_insert_with(|| Bson::ObjectId(ObjectId::new()));
-}
-
 pub(crate) fn to_bson_array(docs: &[Document]) -> Bson {
     Bson::Array(docs.iter().map(|doc| Bson::Document(doc.clone())).collect())
 }
@@ -171,7 +166,7 @@ where
         .ok_or_else(|| D::Error::custom(format!("could not deserialize u64 from {:?}", bson)))
 }
 
-pub fn doc_size_bytes(doc: &Document) -> usize {
+pub fn doc_size_bytes(doc: &Document) -> u64 {
     // 
     // * i32 length prefix (4 bytes)
     // * for each element:
@@ -182,19 +177,19 @@ pub fn doc_size_bytes(doc: &Document) -> usize {
     // * null terminator (1 byte)
     4 + doc
         .into_iter()
-        .map(|(key, val)| 1 + key.len() + 1 + size_bytes(val))
-        .sum::<usize>()
+        .map(|(key, val)| 1 + key.len() as u64 + 1 + size_bytes(val))
+        .sum::<u64>()
         + 1
 }
 
-pub fn size_bytes(val: &Bson) -> usize {
+pub fn size_bytes(val: &Bson) -> u64 {
     match val {
         Bson::Double(_) => 8,
         // 
         // * length prefix (4 bytes)
         // * number of UTF-8 bytes
         // * null terminator (1 byte)
-        Bson::String(s) => 4 + s.len() + 1,
+        Bson::String(s) => 4 + s.len() as u64 + 1,
         // An array is serialized as a document with the keys "0", "1", "2", etc., so the size of
         // an array is:
         //
@@ -210,7 +205,7 @@ pub fn size_bytes(val: &Bson) -> usize {
                 .iter()
                 .enumerate()
                 .map(|(i, val)| 1 + num_decimal_digits(i) + 1 + size_bytes(val))
-                .sum::<usize>()
+                .sum::<u64>()
                 + 1
         }
         Bson::Document(doc) => doc_size_bytes(doc),
@@ -220,21 +215,21 @@ pub fn size_bytes(val: &Bson) -> usize {
         //   * number of UTF-8 bytes
         //   * null terminator (1 byte)
         Bson::RegularExpression(Regex { pattern, options }) => {
-            pattern.len() + 1 + options.len() + 1
+            pattern.len() as u64 + 1 + options.len() as u64 + 1
         }
         // 
         // * length prefix (4 bytes)
         // * number of UTF-8 bytes
         // * null terminator (1 byte)
-        Bson::JavaScriptCode(code) => 4 + code.len() + 1,
+        Bson::JavaScriptCode(code) => 4 + code.len() as u64 + 1,
         // 
         // * i32 length prefix (4 bytes)
         // * i32 length prefix for code (4 bytes)
         // * number of UTF-8 bytes in code
         // * null terminator for code (1 byte)
         // * length of document
         Bson::JavaScriptCodeWithScope(JavaScriptCodeWithScope { code, scope }) => {
-            4 + 4 + code.len() + 1 + doc_size_bytes(scope)
+            4 + 4 + code.len() as u64 + 1 + doc_size_bytes(scope)
         }
         Bson::Int32(_) => 4,
         Bson::Int64(_) => 8,
@@ -243,14 +238,14 @@ pub fn size_bytes(val: &Bson) -> usize {
         // * i32 length prefix (4 bytes)
         // * subtype (1 byte)
         // * number of bytes
-        Bson::Binary(Binary { bytes, .. }) => 4 + 1 + bytes.len(),
+        Bson::Binary(Binary { bytes, .. }) => 4 + 1 + bytes.len() as u64,
         Bson::ObjectId(_) => 12,
         Bson::DateTime(_) => 8,
         // 
         // * i32 length prefix (4 bytes)
         // * subtype (1 byte)
         // * number of UTF-8 bytes
-        Bson::Symbol(s) => 4 + 1 + s.len(),
+        Bson::Symbol(s) => 4 + 1 + s.len() as u64,
         Bson::Decimal128(..) => 128 / 8,
         Bson::Undefined | Bson::MaxKey | Bson::MinKey => 0,
         // DbPointer doesn't have public details exposed by the BSON library, but it comprises of a
@@ -267,7 +262,19 @@ pub fn size_bytes(val: &Bson) -> usize {
     }
 }
 
-fn num_decimal_digits(n: usize) -> usize {
+/// The size in bytes of the provided document's entry in a BSON array at the given index.
+pub(crate) fn array_entry_size_bytes(index: usize, doc: &Document) -> u64 {
+    // 
+    //   * type (1 byte)
+    //   * number of decimal digits in key
+    //   * null terminator for the key (1 byte)
+    //   * size of value
+    1 + num_decimal_digits(index) + 1 + doc_size_bytes(&doc)
+}
+
+/// The number of digits in `n` in base 10.
+/// Useful for calculating the size of an array entry in BSON.
+fn num_decimal_digits(n: usize) -> u64 {
     let mut digits = 1;
     let mut curr = 10;
 
@@ -332,6 +339,6 @@ mod test {
         let mut serialized_bytes = Vec::new();
         doc.to_writer(&mut serialized_bytes).unwrap();
 
-        assert_eq!(size_bytes, serialized_bytes.len());
+        assert_eq!(size_bytes, serialized_bytes.len() as u64);
     }
 }