From 2be98febd935d942a94013ecd0375d0c0db95ab0 Mon Sep 17 00:00:00 2001
From: Asger Hautop Drewsen <asger@tyilo.com>
Date: Tue, 20 Aug 2024 15:32:20 +0200
Subject: [PATCH] Add deserialize-utf8-lossy feature to always deserialize
 using lossy UTF-8 conversion

---
 Cargo.toml                   |  2 ++
 src/change_stream.rs         |  2 +-
 src/change_stream/session.rs |  2 +-
 src/cursor.rs                |  2 +-
 src/cursor/common.rs         |  2 +-
 src/cursor/session.rs        |  2 +-
 src/lib.rs                   | 10 ++++++++++
 7 files changed, 17 insertions(+), 5 deletions(-)

diff --git a/Cargo.toml b/Cargo.toml
index bc1a979b1..8c2471ab2 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -61,6 +61,8 @@ snappy-compression = ["dep:snap"]
 # The In Use Encryption API is unstable and may have backwards-incompatible changes in minor version updates.
 in-use-encryption-unstable = ["dep:mongocrypt", "dep:rayon", "dep:num_cpus"]
 
+deserialize-utf8-lossy = []
+
 # Enables support for emitting tracing events.
 # The tracing API is unstable and may have backwards-incompatible changes in minor version updates.
 # TODO: pending https://github.com/tokio-rs/tracing/issues/2036 stop depending directly on log.
diff --git a/src/change_stream.rs b/src/change_stream.rs
index 86c2bb0b7..1e8fab534 100644
--- a/src/change_stream.rs
+++ b/src/change_stream.rs
@@ -159,7 +159,7 @@ where
     /// ```
     pub async fn next_if_any(&mut self) -> Result<Option<T>> {
         Ok(match NextInBatchFuture::new(self).await? {
-            BatchValue::Some { doc, .. } => Some(bson::from_slice(doc.as_bytes())?),
+            BatchValue::Some { doc, .. } => Some(crate::de::from_slice(doc.as_bytes())?),
             BatchValue::Empty | BatchValue::Exhausted => None,
         })
     }
diff --git a/src/change_stream/session.rs b/src/change_stream/session.rs
index 0af9e2fa4..fda3c2282 100644
--- a/src/change_stream/session.rs
+++ b/src/change_stream/session.rs
@@ -149,7 +149,7 @@ where
                     match bv {
                         BatchValue::Some { doc, .. } => {
                             self.data.document_returned = true;
-                            return Ok(Some(bson::from_slice(doc.as_bytes())?));
+                            return Ok(Some(crate::de::from_slice(doc.as_bytes())?));
                         }
                         BatchValue::Empty | BatchValue::Exhausted => return Ok(None),
                     }
diff --git a/src/cursor.rs b/src/cursor.rs
index d22474441..8df5b305a 100644
--- a/src/cursor.rs
+++ b/src/cursor.rs
@@ -270,7 +270,7 @@ impl<T> Cursor<T> {
     where
         T: Deserialize<'a>,
     {
-        bson::from_slice(self.current().as_bytes()).map_err(Error::from)
+        crate::de::from_slice(self.current().as_bytes()).map_err(Error::from)
     }
 
     /// Update the type streamed values will be parsed as.
diff --git a/src/cursor/common.rs b/src/cursor/common.rs
index 4c8bc3b5a..30e03ef62 100644
--- a/src/cursor/common.rs
+++ b/src/cursor/common.rs
@@ -278,7 +278,7 @@ where
             Poll::Pending => return Poll::Pending,
             Poll::Ready(bv) => match bv? {
                 BatchValue::Some { doc, .. } => {
-                    return Poll::Ready(Some(Ok(bson::from_slice(doc.as_bytes())?)))
+                    return Poll::Ready(Some(Ok(crate::de::from_slice(doc.as_bytes())?)))
                 }
                 BatchValue::Empty => continue,
                 BatchValue::Exhausted => return Poll::Ready(None),
diff --git a/src/cursor/session.rs b/src/cursor/session.rs
index 6f460ac63..404603758 100644
--- a/src/cursor/session.rs
+++ b/src/cursor/session.rs
@@ -306,7 +306,7 @@ impl<T> SessionCursor<T> {
     where
         T: Deserialize<'a>,
     {
-        bson::from_slice(self.current().as_bytes()).map_err(Error::from)
+        crate::de::from_slice(self.current().as_bytes()).map_err(Error::from)
     }
 
     /// Update the type streamed values will be parsed as.
diff --git a/src/lib.rs b/src/lib.rs
index 3ae64b154..e4f33ba06 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -81,3 +81,13 @@ compile_error!(
     "The feature 'compat-3-0-0' must be enabled to ensure forward compatibility with future \
      versions of this crate."
 );
+
+#[cfg(not(feature = "deserialize-utf8-lossy"))]
+pub(crate) mod de {
+    pub(crate) use ::bson::from_slice;
+}
+
+#[cfg(feature = "deserialize-utf8-lossy")]
+pub(crate) mod de {
+    pub(crate) use ::bson::from_slice_utf8_lossy as from_slice;
+}