From 7cde1f585bc00e42c459a463956b6bc260d2eaaa Mon Sep 17 00:00:00 2001
From: frankwang <73262844+Frank-III@users.noreply.github.com>
Date: Sun, 8 Dec 2024 00:51:38 -0600
Subject: [PATCH] feat(core): Add `content_encoding` to `MetaData` (#5400)

---
 core/src/lib.rs                  |  5 ++---
 core/src/raw/http_util/header.rs |  4 ++++
 core/src/services/s3/backend.rs  |  1 +
 core/src/types/capability.rs     |  2 ++
 core/src/types/metadata.rs       | 13 +++++++++++++
 5 files changed, 22 insertions(+), 3 deletions(-)

diff --git a/core/src/lib.rs b/core/src/lib.rs
index 7a7f6eae18ff..8f989adf89a5 100644
--- a/core/src/lib.rs
+++ b/core/src/lib.rs
@@ -144,7 +144,6 @@ mod tests {
     use std::mem::size_of;
 
     use super::*;
-
     /// This is not a real test case.
     ///
     /// We assert our public structs here to make sure we don't introduce
@@ -152,8 +151,8 @@ mod tests {
     #[test]
     fn assert_size() {
         assert_eq!(32, size_of::<Operator>());
-        assert_eq!(296, size_of::<Entry>());
-        assert_eq!(272, size_of::<Metadata>());
+        assert_eq!(320, size_of::<Entry>());
+        assert_eq!(296, size_of::<Metadata>());
         assert_eq!(1, size_of::<EntryMode>());
         assert_eq!(24, size_of::<Scheme>());
     }
diff --git a/core/src/raw/http_util/header.rs b/core/src/raw/http_util/header.rs
index 6c8ba65dbab9..2da77a4b08dc 100644
--- a/core/src/raw/http_util/header.rs
+++ b/core/src/raw/http_util/header.rs
@@ -173,6 +173,10 @@ pub fn parse_into_metadata(path: &str, headers: &HeaderMap) -> Result<Metadata>
         m.set_content_type(v);
     }
 
+    if let Some(v) = parse_content_encoding(headers)? {
+        m.set_content_encoding(v);
+    }
+
     if let Some(v) = parse_content_range(headers)? {
         m.set_content_range(v);
     }
diff --git a/core/src/services/s3/backend.rs b/core/src/services/s3/backend.rs
index 52f311415489..8ffd6b7ab7b7 100644
--- a/core/src/services/s3/backend.rs
+++ b/core/src/services/s3/backend.rs
@@ -923,6 +923,7 @@ impl Access for S3Backend {
             .set_name(&self.core.bucket)
             .set_native_capability(Capability {
                 stat: true,
+                stat_has_content_encoding: true,
                 stat_with_if_match: true,
                 stat_with_if_none_match: true,
                 stat_with_override_cache_control: !self.core.disable_stat_with_override,
diff --git a/core/src/types/capability.rs b/core/src/types/capability.rs
index 2fdf56e87ce3..42c064368180 100644
--- a/core/src/types/capability.rs
+++ b/core/src/types/capability.rs
@@ -90,6 +90,8 @@ pub struct Capability {
     pub stat_has_content_range: bool,
     /// Indicates whether content type information is available in stat response
     pub stat_has_content_type: bool,
+    /// Indicates whether content encoding information is available in stat response
+    pub stat_has_content_encoding: bool,
     /// Indicates whether entity tag is available in stat response
     pub stat_has_etag: bool,
     /// Indicates whether last modified timestamp is available in stat response
diff --git a/core/src/types/metadata.rs b/core/src/types/metadata.rs
index 7314f25db5fd..c1d4155e1971 100644
--- a/core/src/types/metadata.rs
+++ b/core/src/types/metadata.rs
@@ -39,6 +39,7 @@ pub struct Metadata {
     content_md5: Option<String>,
     content_range: Option<BytesContentRange>,
     content_type: Option<String>,
+    content_encoding: Option<String>,
     etag: Option<String>,
     last_modified: Option<DateTime<Utc>>,
     version: Option<String>,
@@ -56,6 +57,7 @@ impl Metadata {
             content_length: None,
             content_md5: None,
             content_type: None,
+            content_encoding: None,
             content_range: None,
             last_modified: None,
             etag: None,
@@ -202,6 +204,17 @@ impl Metadata {
         self
     }
 
+    /// Content Encoding of this entry.
+    pub fn content_encoding(&self) -> Option<&str> {
+        self.content_encoding.as_deref()
+    }
+
+    /// Set Content Encoding of this entry.
+    pub fn set_content_encoding(&mut self, v: &str) -> &mut Self {
+        self.content_encoding = Some(v.to_string());
+        self
+    }
+
     /// Content Range of this entry.
     ///
     /// Content Range is defined by [RFC 9110](https://httpwg.org/specs/rfc9110.html#field.content-range).