minor: raw BSON perf and API improvements (#335)

Author: patrickfreed

src/bson.rs

@@ -1050,7 +1050,10 @@ impl Display for Regex {
 /// Represents a BSON code with scope value.
 #[derive(Debug, Clone, PartialEq)]
 pub struct JavaScriptCodeWithScope {
+    /// The JavaScript code.
     pub code: String,
+
+    /// The scope document containing variable bindings.
     pub scope: Document,
 }
 

src/raw/array.rs

@@ -212,6 +212,11 @@ impl RawArray {
     pub fn as_bytes(&self) -> &[u8] {
         self.doc.as_bytes()
     }
+
+    /// Whether this array contains any elements or not.
+    pub fn is_empty(&self) -> bool {
+        self.doc.is_empty()
+    }
 }
 
 impl std::fmt::Debug for RawArray {

src/raw/bson_ref.rs

@@ -533,20 +533,17 @@ impl<'a> From<&'a Binary> for RawBsonRef<'a> {
 /// A BSON regex referencing raw bytes stored elsewhere.
 #[derive(Clone, Copy, Debug, PartialEq)]
 pub struct RawRegexRef<'a> {
-    pub(crate) pattern: &'a str,
-    pub(crate) options: &'a str,
-}
-
-impl<'a> RawRegexRef<'a> {
-    /// Gets the pattern portion of the regex.
-    pub fn pattern(self) -> &'a str {
-        self.pattern
-    }
+    /// The regex pattern to match.
+    pub pattern: &'a str,
 
-    /// Gets the options portion of the regex.
-    pub fn options(self) -> &'a str {
-        self.options
-    }
+    /// The options for the regex.
+    ///
+    /// Options are identified by characters, which must be stored in
+    /// alphabetical order. Valid options are 'i' for case insensitive matching, 'm' for
+    /// multiline matching, 'x' for verbose mode, 'l' to make \w, \W, etc. locale dependent,
+    /// 's' for dotall mode ('.' matches everything), and 'u' to make \w, \W, etc. match
+    /// unicode.
+    pub options: &'a str,
 }
 
 impl<'de: 'a, 'a> Deserialize<'de> for RawRegexRef<'a> {
@@ -594,22 +591,14 @@ impl<'a> From<RawRegexRef<'a>> for RawBsonRef<'a> {
 /// A BSON "code with scope" value referencing raw bytes stored elsewhere.
 #[derive(Clone, Copy, Debug, PartialEq)]
 pub struct RawJavaScriptCodeWithScopeRef<'a> {
-    pub(crate) code: &'a str,
+    /// The JavaScript code.
+    pub code: &'a str,
 
-    pub(crate) scope: &'a RawDocument,
+    /// The scope document containing variable bindings.
+    pub scope: &'a RawDocument,
 }
 
 impl<'a> RawJavaScriptCodeWithScopeRef<'a> {
-    /// Gets the code in the value.
-    pub fn code(self) -> &'a str {
-        self.code
-    }
-
-    /// Gets the scope in the value.
-    pub fn scope(self) -> &'a RawDocument {
-        self.scope
-    }
-
     pub(crate) fn len(self) -> i32 {
         4 + 4 + self.code.len() as i32 + 1 + self.scope.as_bytes().len() as i32
     }

src/raw/document.rs

@@ -6,6 +6,7 @@ use std::{
 use serde::{ser::SerializeMap, Deserialize, Serialize};
 
 use crate::{
+    de::MIN_BSON_DOCUMENT_SIZE,
     raw::{error::ErrorKind, serde::OwnedOrBorrowedRawDocument, RAW_DOCUMENT_NEWTYPE},
     DateTime,
     Timestamp,
@@ -398,8 +399,8 @@ impl RawDocument {
     ///     "bool": true,
     /// };
     ///
-    /// assert_eq!(doc.get_regex("regex")?.pattern(), r"end\s*$");
-    /// assert_eq!(doc.get_regex("regex")?.options(), "i");
+    /// assert_eq!(doc.get_regex("regex")?.pattern, r"end\s*$");
+    /// assert_eq!(doc.get_regex("regex")?.options, "i");
     /// assert!(matches!(doc.get_regex("bool").unwrap_err().kind, ValueAccessErrorKind::UnexpectedType { .. }));
     /// assert!(matches!(doc.get_regex("unknown").unwrap_err().kind, ValueAccessErrorKind::NotPresent));
     /// # Ok::<(), Box<dyn std::error::Error>>(())
@@ -486,6 +487,11 @@ impl RawDocument {
     pub fn as_bytes(&self) -> &[u8] {
         &self.data
     }
+
+    /// Returns whether this document contains any elements or not.
+    pub fn is_empty(&self) -> bool {
+        self.as_bytes().len() == MIN_BSON_DOCUMENT_SIZE as usize
+    }
 }
 
 impl<'de: 'a, 'a> Deserialize<'de> for &'a RawDocument {

src/raw/document_buf.rs

@@ -70,7 +70,8 @@ pub struct RawDocumentBuf {
 impl RawDocumentBuf {
     /// Creates a new, empty [`RawDocumentBuf`].
     pub fn new() -> RawDocumentBuf {
-        let mut data: Vec<u8> = MIN_BSON_DOCUMENT_SIZE.to_le_bytes().to_vec();
+        let mut data = Vec::new();
+        data.extend(&MIN_BSON_DOCUMENT_SIZE.to_le_bytes());
         data.push(0);
         Self { data }
     }
@@ -192,27 +193,27 @@ impl RawDocumentBuf {
     /// assert_eq!(doc.to_document()?, expected);
     /// # Ok::<(), Error>(())
     /// ```
-    pub fn append(&mut self, key: impl Into<String>, value: impl Into<RawBson>) {
-        fn append_string(doc: &mut RawDocumentBuf, value: String) {
+    pub fn append(&mut self, key: impl AsRef<str>, value: impl Into<RawBson>) {
+        fn append_string(doc: &mut RawDocumentBuf, value: &str) {
             doc.data
                 .extend(&((value.as_bytes().len() + 1) as i32).to_le_bytes());
-            doc.data.extend(value.into_bytes());
+            doc.data.extend(value.as_bytes());
             doc.data.push(0);
         }
 
-        fn append_cstring(doc: &mut RawDocumentBuf, value: String) {
+        fn append_cstring(doc: &mut RawDocumentBuf, value: &str) {
             if value.contains('\0') {
                 panic!("cstr includes interior null byte: {}", value)
             }
-            doc.data.extend(value.into_bytes());
+            doc.data.extend(value.as_bytes());
             doc.data.push(0);
         }
 
         let original_len = self.data.len();
 
         // write the key for the next value to the end
         // the element type will replace the previous null byte terminator of the document
-        append_cstring(self, key.into());
+        append_cstring(self, key.as_ref());
 
         let value = value.into();
         let element_type = value.element_type();
@@ -222,7 +223,7 @@ impl RawDocumentBuf {
                 self.data.extend(&i.to_le_bytes());
             }
             RawBson::String(s) => {
-                append_string(self, s);
+                append_string(self, s.as_str());
             }
             RawBson::Document(d) => {
                 self.data.extend(d.into_bytes());
@@ -251,7 +252,7 @@ impl RawDocumentBuf {
                 self.data.extend(&dt.timestamp_millis().to_le_bytes());
             }
             RawBson::DbPointer(dbp) => {
-                append_string(self, dbp.namespace);
+                append_string(self, dbp.namespace.as_str());
                 self.data.extend(&dbp.id.bytes());
             }
             RawBson::Decimal128(d) => {
@@ -264,11 +265,11 @@ impl RawDocumentBuf {
                 self.data.extend(&i.to_le_bytes());
             }
             RawBson::RegularExpression(re) => {
-                append_cstring(self, re.pattern);
-                append_cstring(self, re.options);
+                append_cstring(self, re.pattern.as_str());
+                append_cstring(self, re.options.as_str());
             }
             RawBson::JavaScriptCode(js) => {
-                append_string(self, js);
+                append_string(self, js.as_str());
             }
             RawBson::JavaScriptCodeWithScope(code_w_scope) => {
                 let len = RawJavaScriptCodeWithScopeRef {
@@ -277,7 +278,7 @@ impl RawDocumentBuf {
                 }
                 .len();
                 self.data.extend(&len.to_le_bytes());
-                append_string(self, code_w_scope.code);
+                append_string(self, code_w_scope.code.as_str());
                 self.data.extend(code_w_scope.scope.into_bytes());
             }
             RawBson::Timestamp(ts) => {
@@ -287,7 +288,7 @@ impl RawDocumentBuf {
                 self.data.extend(&oid.bytes());
             }
             RawBson::Symbol(s) => {
-                append_string(self, s);
+                append_string(self, s.as_str());
             }
             RawBson::Null | RawBson::Undefined | RawBson::MinKey | RawBson::MaxKey => {}
         }
@@ -296,8 +297,8 @@ impl RawDocumentBuf {
         // append trailing null byte
         self.data.push(0);
         // update length
-        self.data
-            .splice(0..4, (self.data.len() as i32).to_le_bytes().iter().cloned());
+        let new_len = (self.data.len() as i32).to_le_bytes();
+        self.data[0..4].copy_from_slice(&new_len);
     }
 
     /// Convert this [`RawDocumentBuf`] to a [`Document`], returning an error
@@ -389,7 +390,7 @@ impl Borrow<RawDocument> for RawDocumentBuf {
     }
 }
 
-impl<S: Into<String>, T: Into<RawBson>> FromIterator<(S, T)> for RawDocumentBuf {
+impl<S: AsRef<str>, T: Into<RawBson>> FromIterator<(S, T)> for RawDocumentBuf {
     fn from_iter<I: IntoIterator<Item = (S, T)>>(iter: I) -> Self {
         let mut buf = RawDocumentBuf::new();
         for (k, v) in iter {

src/raw/serde.rs

@@ -349,8 +349,8 @@ impl<'de> Visitor<'de> for OwnedOrBorrowedRawBsonVisitor {
             let v: RawBson = map.next_value()?;
             doc.append(first_key, v);
 
-            while let Some((k, v)) = map.next_entry::<String, RawBson>()? {
-                doc.append(k, v);
+            while let Some((k, v)) = map.next_entry::<CowStr, RawBson>()? {
+                doc.append(k.0, v);
             }
 
             Ok(RawBson::Document(doc).into())

src/raw/test/mod.rs

@@ -310,9 +310,9 @@ fn javascript_with_scope() {
         .expect("no key javascript_with_scope")
         .as_javascript_with_scope()
         .expect("was not javascript with scope");
-    assert_eq!(js_with_scope.code(), "console.log(msg);");
+    assert_eq!(js_with_scope.code, "console.log(msg);");
     let (scope_key, scope_value_bson) = js_with_scope
-        .scope()
+        .scope
         .into_iter()
         .next()
         .expect("no next value in scope")

src/ser/raw/mod.rs

@@ -97,8 +97,8 @@ impl Serializer {
     /// Replace an i32 value at the given index with the given value.
     #[inline]
     fn replace_i32(&mut self, at: usize, with: i32) {
-        self.bytes
-            .splice(at..at + 4, with.to_le_bytes().iter().cloned());
+        let portion = &mut self.bytes[at..at + 4];
+        portion.copy_from_slice(&with.to_le_bytes());
     }
 }