[HADOOP-18562]: S3A: support custom S3 and STS headers.

hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/Constants.java

@@ -775,6 +775,23 @@ private Constants() {
       "fs.s3a." + Constants.AWS_SERVICE_IDENTIFIER_STS.toLowerCase()
           + ".signing-algorithm";
 
+  /**
+   * List of custom headers to be set on the service client.
+   * Multiple parameters can be used to specify custom headers.
+   * fs.s3a.s3.custom.headers - headers to add on all the s3 requests.
+   * fs.s3a.sts.custom.headers - headers to add on all the sts requests.
+   * Examples
+   * CustomHeader {@literal ->} 'Header1:Value1'
+   * CustomHeaders {@literal ->} 'Header1=Value1:Value2,Header2=Value1'
+   */
+  public static final String CUSTOM_HEADERS_STS =
+      "fs.s3a." + Constants.AWS_SERVICE_IDENTIFIER_STS.toLowerCase()
+          + ".custom.headers";
+
+  public static final String CUSTOM_HEADERS_S3 =
+          "fs.s3a." + Constants.AWS_SERVICE_IDENTIFIER_S3.toLowerCase()
+                  + ".custom.headers";
+
   public static final String S3N_FOLDER_SUFFIX = "_$folder$";
   public static final String FS_S3A_BLOCK_SIZE = "fs.s3a.block.size";
   public static final String FS_S3A = "s3a";

hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/impl/AWSClientConfig.java

@@ -22,6 +22,8 @@
 import java.net.URI;
 import java.net.URISyntaxException;
 import java.time.Duration;
+import java.util.Arrays;
+import java.util.List;
 import java.util.concurrent.TimeUnit;
 
 import org.slf4j.Logger;
@@ -72,6 +74,8 @@
 import static org.apache.hadoop.fs.s3a.Constants.SIGNING_ALGORITHM_STS;
 import static org.apache.hadoop.fs.s3a.Constants.SOCKET_TIMEOUT;
 import static org.apache.hadoop.fs.s3a.Constants.USER_AGENT_PREFIX;
+import static org.apache.hadoop.fs.s3a.Constants.CUSTOM_HEADERS_S3;
+import static org.apache.hadoop.fs.s3a.Constants.CUSTOM_HEADERS_STS;
 import static org.apache.hadoop.fs.s3a.impl.ConfigurationHelper.enforceMinimumDuration;
 import static org.apache.hadoop.fs.s3a.impl.ConfigurationHelper.getDuration;
 import static org.apache.hadoop.util.Preconditions.checkArgument;
@@ -116,6 +120,8 @@ public static ClientOverrideConfiguration.Builder createClientConfigBuilder(Conf
 
     initUserAgent(conf, overrideConfigBuilder);
 
+    initRequestHeaders(conf, overrideConfigBuilder, awsServiceIdentifier);
+
     String signer = conf.getTrimmed(SIGNING_ALGORITHM, "");
     if (!signer.isEmpty()) {
       LOG.debug("Signer override = {}", signer);
@@ -407,6 +413,48 @@ private static void initSigner(Configuration conf,
     }
   }
 
+  /**
+   *
+   * @param conf hadoop configuration
+   * @param clientConfig client configuration to update
+   * @param awsServiceIdentifier service name
+   */
+  private static void initRequestHeaders(Configuration conf,
+      ClientOverrideConfiguration.Builder clientConfig, String awsServiceIdentifier) {
+    String configKey = null;
+    switch (awsServiceIdentifier) {
+      case AWS_SERVICE_IDENTIFIER_S3:
+        configKey = CUSTOM_HEADERS_S3;
+        break;
+      case AWS_SERVICE_IDENTIFIER_STS:
+        configKey = CUSTOM_HEADERS_STS;
+        break;
+      default:
+        // Nothing to do. The original signer override is already setup
+    }
+    if (configKey != null) {
+      String[] customHeaders = conf.getTrimmedStrings(configKey);
+      if (customHeaders == null || customHeaders.length == 0) {
+        LOG.debug("No custom headers specified");
+        return;
+      }
+
+      for (String customHeader : customHeaders) {
+        String[] parts = customHeader.split("=");
+        if (parts.length != 2) {
+          String message = "Invalid format (Expected header1=value1:value2,header2=value1) for Header: ["
+                          + customHeader
+                          + "]";
+          LOG.error(message);
+          throw new IllegalArgumentException(message);
+        }
+
+        List<String> values = Arrays.asList(parts[1].split(":"));
+        clientConfig.putHeader(parts[0], values);
+      }
+    }
+  }
+
   /**
    * Configures request timeout in the client configuration.
    * This is independent of the timeouts set in the sync and async HTTP clients;