RUST-1992 Make serde an optional feature (#554)

Co-authored-by: Isabel Atkinson <[email protected]>

Author: abr-egn

.evergreen/run-fuzzer.sh

@@ -30,11 +30,12 @@ run_fuzzer() {
 }
 
 # Run existing targets
-run_fuzzer "deserialize"
+run_fuzzer "decode"
 run_fuzzer "raw_deserialize"
+run_fuzzer "raw_deserialize_utf8_lossy"
 run_fuzzer "iterate"
 
 # Run new security-focused targets
 run_fuzzer "type_markers"
 run_fuzzer "string_handling"
-run_fuzzer "serialization"
+run_fuzzer "encoding"

.evergreen/run-tests.sh

@@ -4,7 +4,9 @@ set -o errexit
 
 . ~/.cargo/env
 
-RUST_BACKTRACE=1 cargo test
+# Test with default features and excluding doctests (some of which require the 'serde' feature)
+RUST_BACKTRACE=1 cargo test --all-targets
+# Test with all features and including doctests
 RUST_BACKTRACE=1 cargo test --all-features
 
 cd serde-tests

Cargo.toml

@@ -47,6 +47,7 @@ serde_path_to_error = ["dep:serde_path_to_error"]
 # if enabled, include serde_with interop.
 # should be used in conjunction with chrono-0_4 or uuid-0_8.
 serde_with-3 = ["dep:serde_with"]
+serde = ["dep:serde"]
 
 [lib]
 name = "bson"
@@ -55,7 +56,7 @@ name = "bson"
 ahash = "0.8.0"
 chrono = { version = "0.4.15", features = ["std"], default-features = false, optional = true }
 rand = "0.9"
-serde = { version = "1.0", features = ["derive"] }
+serde = { version = "1.0", features = ["derive"], optional = true }
 serde_json = { version = "1.0", features = ["preserve_order"] }
 indexmap = "2.1.0"
 hex = "0.4.2"

examples/deserialize.rs

@@ -2,3 +2,4 @@
 target
 corpus
 artifacts
+Cargo.lock

examples/serialize.rs

@@ -10,6 +10,7 @@ cargo-fuzz = true
 
 [dependencies.bson]
 path = ".."
+features = ["serde"]
 
 [dependencies.libfuzzer-sys]
 version = "0.4.0"
@@ -24,8 +25,8 @@ version = "1.0"
 members = ["."]
 
 [[bin]]
-name = "deserialize"
-path = "fuzz_targets/deserialize.rs"
+name = "decode"
+path = "fuzz_targets/decode.rs"
 
 [[bin]]
 name = "iterate"
@@ -48,8 +49,8 @@ name = "string_handling"
 path = "fuzz_targets/string_handling.rs"
 
 [[bin]]
-name = "serialization"
-path = "fuzz_targets/serialization.rs"
+name = "encoding"
+path = "fuzz_targets/encoding.rs"
 
 [[bin]]
 name = "generate_corpus"

examples/test.bson

@@ -7,8 +7,8 @@ use bson::Document;
 use std::io::Cursor;
 
 fuzz_target!(|buf: &[u8]| {
-    if let Ok(doc) = Document::from_reader(&mut Cursor::new(&buf[..])) {
+    if let Ok(doc) = Document::decode_from_reader(&mut Cursor::new(&buf[..])) {
         let mut vec = Vec::with_capacity(buf.len());
-        let _ = doc.to_writer(&mut vec);
+        let _ = doc.encode_to_writer(&mut vec);
     }
 });

fuzz/.gitignore

@@ -45,20 +45,19 @@ fn compare_values(val1: &Bson, val2: &Bson) -> bool {
 }
 
 fuzz_target!(|input: &[u8]| {
-    if let Ok(rawdoc) = RawDocument::from_bytes(&input) {
+    if let Ok(rawdoc) = RawDocument::decode_from_bytes(&input) {
         if let Ok(doc) = Document::try_from(rawdoc) {
             let out = RawDocumentBuf::try_from(&doc).unwrap();
             let out_bytes = out.as_bytes();
             if input != out_bytes {
-                let reserialized = RawDocument::from_bytes(&out_bytes).unwrap();
-                let reserialized_doc = Document::try_from(reserialized).unwrap();
-                // Ensure that the reserialized document is the same as the original document, the
+                let reencoded = RawDocument::decode_from_bytes(&out_bytes).unwrap();
+                let reencoded_doc = Document::try_from(reencoded).unwrap();
+                // Ensure that the re-encoded document is the same as the original document, the
                 // bytes can differ while still resulting in the same Document.
-                if !compare_docs(&doc, &reserialized_doc) {
+                if !compare_docs(&doc, &reencoded_doc) {
                     panic!(
-                        "Reserialized document is not the same as the original document: {:?} != \
-                         {:?}",
-                        doc, reserialized_doc
+                        "Reencoded document is not the same as the original document: {:?} != {:?}",
+                        doc, reencoded_doc
                     );
                 }
             }

fuzz/Cargo.toml

@@ -5,7 +5,7 @@ extern crate bson;
 use bson::RawDocument;
 
 fuzz_target!(|buf: &[u8]| {
-    if let Ok(doc) = RawDocument::from_bytes(buf) {
+    if let Ok(doc) = RawDocument::decode_from_bytes(buf) {
         for _ in doc {}
     }
 });

fuzz/fuzz_targets/deserialize.rs renamed to fuzz/fuzz_targets/decode.rs

@@ -5,8 +5,7 @@ extern crate bson;
 use bson::Document;
 
 fuzz_target!(|buf: &[u8]| {
-    if let Ok(doc) = bson::from_slice::<Document>(buf) {
-        let mut vec = Vec::with_capacity(buf.len());
-        let _ = doc.to_writer(&mut vec);
+    if let Ok(doc) = bson::deserialize_from_slice::<Document>(buf) {
+        let _ = bson::serialize_to_vec(&doc);
     }
 });

fuzz/fuzz_targets/serialization.rs renamed to fuzz/fuzz_targets/encoding.rs

@@ -1,11 +1,11 @@
 #![no_main]
-#[macro_use] extern crate libfuzzer_sys;
+#[macro_use]
+extern crate libfuzzer_sys;
 extern crate bson;
-use bson::Document;
+use bson::{serde_helpers::Utf8LossyDeserialization, Document};
 
 fuzz_target!(|buf: &[u8]| {
-    if let Ok(doc) = bson::from_slice_utf8_lossy::<Document>(buf) {
-        let mut vec = Vec::with_capacity(buf.len());
-        let _ = doc.to_writer(&mut vec);
+    if let Ok(doc) = bson::deserialize_from_slice::<Utf8LossyDeserialization<Document>>(buf) {
+        let _ = bson::serialize_to_vec(&doc.0);
     }
 });

fuzz/fuzz_targets/iterate.rs

@@ -6,7 +6,7 @@ use bson::{RawBsonRef, RawDocument};
 use std::convert::TryInto;
 
 fuzz_target!(|buf: &[u8]| {
-    if let Ok(doc) = RawDocument::from_bytes(buf) {
+    if let Ok(doc) = RawDocument::decode_from_bytes(buf) {
         for elem in doc.iter_elements().flatten() {
             // Convert to RawBsonRef and check string-related types
             if let Ok(bson) = elem.try_into() {

fuzz/fuzz_targets/raw_deserialize.rs

@@ -6,7 +6,7 @@ use bson::{RawBsonRef, RawDocument};
 use std::convert::TryInto;
 
 fuzz_target!(|buf: &[u8]| {
-    if let Ok(doc) = RawDocument::from_bytes(buf) {
+    if let Ok(doc) = RawDocument::decode_from_bytes(buf) {
         for elem in doc.iter_elements().flatten() {
             let _: Result<RawBsonRef, _> = elem.try_into();
         }

fuzz/fuzz_targets/raw_deserialize_utf8_lossy.rs

@@ -29,14 +29,18 @@ fn generate_length_edge_cases(dir: &Path) -> std::io::Result<()> {
     let min_doc = doc! {};
     fs::write(
         target_dir.join("min_doc"),
-        bson::to_vec(&min_doc).map_err(|e| Error::new(ErrorKind::Other, e.to_string()))?,
+        min_doc
+            .encode_to_vec()
+            .map_err(|e| Error::new(ErrorKind::Other, e.to_string()))?,
     )?;
 
     // Document with length near i32::MAX
     let large_doc = doc! { "a": "b".repeat(i32::MAX as usize / 2) };
     fs::write(
         target_dir.join("large_doc"),
-        bson::to_vec(&large_doc).map_err(|e| Error::new(ErrorKind::Other, e.to_string()))?,
+        large_doc
+            .encode_to_vec()
+            .map_err(|e| Error::new(ErrorKind::Other, e.to_string()))?,
     )?;
 
     Ok(())
@@ -73,7 +77,9 @@ fn generate_type_marker_cases(dir: &Path) -> std::io::Result<()> {
     };
     fs::write(
         target_dir.join("all_types"),
-        bson::to_vec(&all_types).map_err(|e| Error::new(ErrorKind::Other, e.to_string()))?,
+        all_types
+            .encode_to_vec()
+            .map_err(|e| Error::new(ErrorKind::Other, e.to_string()))?,
     )?;
 
     Ok(())
@@ -100,7 +106,9 @@ fn generate_string_edge_cases(dir: &Path) -> std::io::Result<()> {
     };
     fs::write(
         target_dir.join("utf8_cases"),
-        bson::to_vec(&utf8_cases).map_err(|e| Error::new(ErrorKind::Other, e.to_string()))?,
+        utf8_cases
+            .encode_to_vec()
+            .map_err(|e| Error::new(ErrorKind::Other, e.to_string()))?,
     )?;
 
     Ok(())
@@ -124,7 +132,9 @@ fn generate_serialization_cases(dir: &Path) -> std::io::Result<()> {
     }
     fs::write(
         target_dir.join("nested_doc"),
-        bson::to_vec(&nested_doc).map_err(|e| Error::new(ErrorKind::Other, e.to_string()))?,
+        nested_doc
+            .encode_to_vec()
+            .map_err(|e| Error::new(ErrorKind::Other, e.to_string()))?,
     )?;
 
     // Document with large binary data
@@ -136,7 +146,9 @@ fn generate_serialization_cases(dir: &Path) -> std::io::Result<()> {
     };
     fs::write(
         target_dir.join("large_binary"),
-        bson::to_vec(&large_binary).map_err(|e| Error::new(ErrorKind::Other, e.to_string()))?,
+        large_binary
+            .encode_to_vec()
+            .map_err(|e| Error::new(ErrorKind::Other, e.to_string()))?,
     )?;
 
     Ok(())

fuzz/fuzz_targets/string_handling.rs

@@ -8,7 +8,7 @@ edition = "2018"
 default = []
 
 [dependencies]
-bson = { path = "..", features = ["uuid-1", "chrono-0_4", "serde_with-3"] }
+bson = { path = "..", features = ["uuid-1", "chrono-0_4", "serde", "serde_with-3"] }
 serde = { version = "1.0", features = ["derive"] }
 pretty_assertions = "0.6.1"
 hex = "0.4.2"

fuzz/fuzz_targets/type_markers.rs

@@ -99,18 +99,18 @@ fn owned_raw_bson() {
     });
 
     let mut doc_buf = RawDocumentBuf::new();
-    doc_buf.append("a", "key");
-    doc_buf.append("number", 12);
-    doc_buf.append("bool", false);
-    doc_buf.append("nu", RawBson::Null);
+    doc_buf.append("a", "key").unwrap();
+    doc_buf.append("number", 12).unwrap();
+    doc_buf.append("bool", false).unwrap();
+    doc_buf.append("nu", RawBson::Null).unwrap();
 
     let mut array_buf = RawArrayBuf::new();
-    array_buf.push(1);
-    array_buf.push("string");
+    array_buf.push(1).unwrap();
+    array_buf.push("string").unwrap();
 
     let mut bson_doc = RawDocumentBuf::new();
-    bson_doc.append("first", true);
-    bson_doc.append("second", "string");
+    bson_doc.append("first", true).unwrap();
+    bson_doc.append("second", "string").unwrap();
 
     let expected = Foo {
         doc_buf,