rgw/cloud-restore: strip quotes from ETag on cloud tier fetch

When objects are restored from a cloud endpoint, the ETag value read
from the HTTP response includes surrounding double-quotes per RFC 7232.
RGW stores ETags unquoted internally, and dump_etag() adds its own
quotes when serving responses. The mismatch results in double-quoted
ETags like ""abc123-6"" on restored objects.

Strip the quotes from both the etag output parameter and the
RGW_ATTR_ETAG attribute after fetching from the cloud endpoint,
matching the unquoted format RGW uses everywhere else.

Signed-off-by: Matthew N. Heler <matthew.heler@hotmail.com>

diff --git a/src/rgw/driver/rados/rgw_lc_tier.cc b/src/rgw/driver/rados/rgw_lc_tier.cc
index 15aa5653652ee604d893e0cdbcf4d73e1cdc49ba..c483ae4b7a06382e8345e5092a906fd052dcd414 100644 (file)
--- a/src/rgw/driver/rados/rgw_lc_tier.cc
+++ b/src/rgw/driver/rados/rgw_lc_tier.cc
@@ -405,6 +405,19 @@ int rgw_cloud_tier_get_object(RGWLCCloudTierCtx& tier_ctx, bool head,
     }
   }
 
+  /*
+   * The HTTP ETag header value is a quoted-string per RFC 7232.
+   * Strip the surrounding quotes so we do not double quote them.
+   */
+  etag = rgw_string_unquote(etag);
+
+  if (auto i = attrs.find(RGW_ATTR_ETAG); i != attrs.end()) {
+    const string unquoted = rgw_string_unquote(i->second.to_str());
+    bufferlist bl;
+    bl.append(unquoted);
+    i->second = std::move(bl);
+  }
+
   ldpp_dout(tier_ctx.dpp, 20) << __func__ << "(): Successfully fetched object from cloud bucket:" << dest_bucket << ", object: " << target_obj_name << dendl;
   return ret;
 }