svn commit: r1374406 - in /hadoop/common/branches/branch-0.23/hadoop-hdfs-project/hadoop-hdfs: CHANGES.txt src/main/java/org/apache/hadoop/hdfs/ByteRangeInputStream.java

[da...@apache.org]

Author: daryn
Date: Fri Aug 17 19:46:50 2012
New Revision: 1374406

URL: http://svn.apache.org/viewvc?rev=1374406&view=rev
Log:
svn merge -c 1374122. FIXES: HDFS-3788. ByteRangeInputStream should not expect HTTP Content-Length header when chunked transfer-encoding is used.

Modified:
    hadoop/common/branches/branch-0.23/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt
    hadoop/common/branches/branch-0.23/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/ByteRangeInputStream.java

Modified: hadoop/common/branches/branch-0.23/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt
URL: http://svn.apache.org/viewvc/hadoop/common/branches/branch-0.23/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt?rev=1374406&r1=1374405&r2=1374406&view=diff
==============================================================================
--- hadoop/common/branches/branch-0.23/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt (original)
+++ hadoop/common/branches/branch-0.23/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt Fri Aug 17 19:46:50 2012
@@ -102,6 +102,9 @@ Release 0.23.3 - UNRELEASED
     HDFS-3718. Datanode won't shutdown because of runaway DataBlockScanner
     thread (Kihwal Lee via daryn)
 
+    HDFS-3788. ByteRangeInputStream should not expect HTTP Content-Length header
+    when chunked transfer-encoding is used.  (szetszwo)
+
 Release 0.23.2 - UNRELEASED
 
   INCOMPATIBLE CHANGES

Modified: hadoop/common/branches/branch-0.23/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/ByteRangeInputStream.java
URL: http://svn.apache.org/viewvc/hadoop/common/branches/branch-0.23/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/ByteRangeInputStream.java?rev=1374406&r1=1374405&r2=1374406&view=diff
==============================================================================
--- hadoop/common/branches/branch-0.23/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/ByteRangeInputStream.java (original)
+++ hadoop/common/branches/branch-0.23/hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/ByteRangeInputStream.java Fri Aug 17 19:46:50 2012
@@ -22,12 +22,15 @@ import java.io.IOException;
 import java.io.InputStream;
 import java.net.HttpURLConnection;
 import java.net.URL;
+import java.util.List;
+import java.util.Map;
+import java.util.StringTokenizer;
 
 import org.apache.commons.io.input.BoundedInputStream;
 import org.apache.hadoop.fs.FSInputStream;
-import org.apache.hadoop.hdfs.server.namenode.StreamFile;
 
 import com.google.common.annotations.VisibleForTesting;
+import com.google.common.net.HttpHeaders;
 
 /**
  * To support HTTP byte streams, a new connection to an HTTP server needs to be
@@ -70,7 +73,7 @@ public abstract class ByteRangeInputStre
   protected URLOpener resolvedURL;
   protected long startPos = 0;
   protected long currentPos = 0;
-  protected long filelength;
+  protected Long fileLength = null;
 
   StreamStatus status = StreamStatus.SEEK;
 
@@ -120,28 +123,60 @@ public abstract class ByteRangeInputStre
     connection.connect();
     checkResponseCode(connection);
 
-    final String cl = connection.getHeaderField(StreamFile.CONTENT_LENGTH);
-    if (cl == null) {
-      throw new IOException(StreamFile.CONTENT_LENGTH+" header is missing");
-    }
-    final long streamlength = Long.parseLong(cl);
-    filelength = startPos + streamlength;
-    // Java has a bug with >2GB request streams.  It won't bounds check
-    // the reads so the transfer blocks until the server times out
-    InputStream is =
-        new BoundedInputStream(connection.getInputStream(), streamlength);
-
     resolvedURL.setURL(getResolvedUrl(connection));
-    
-    return is;
+
+    InputStream in = connection.getInputStream();
+    final Map<String, List<String>> headers = connection.getHeaderFields();
+    if (isChunkedTransferEncoding(headers)) {
+      // file length is not known
+      fileLength = null;
+    } else {
+      // for non-chunked transfer-encoding, get content-length
+      final String cl = connection.getHeaderField(HttpHeaders.CONTENT_LENGTH);
+      if (cl == null) {
+        throw new IOException(HttpHeaders.CONTENT_LENGTH + " is missing: "
+            + headers);
+      }
+      final long streamlength = Long.parseLong(cl);
+      fileLength = startPos + streamlength;
+
+      // Java has a bug with >2GB request streams.  It won't bounds check
+      // the reads so the transfer blocks until the server times out
+      in = new BoundedInputStream(in, streamlength);
+    }
+
+    return in;
   }
   
+  private static boolean isChunkedTransferEncoding(
+      final Map<String, List<String>> headers) {
+    return contains(headers, HttpHeaders.TRANSFER_ENCODING, "chunked")
+        || contains(headers, HttpHeaders.TE, "chunked");
+  }
+
+  /** Does the HTTP header map contain the given key, value pair? */
+  private static boolean contains(final Map<String, List<String>> headers,
+      final String key, final String value) {
+    final List<String> values = headers.get(key);
+    if (values != null) {
+      for(String v : values) {
+        for(final StringTokenizer t = new StringTokenizer(v, ",");
+            t.hasMoreTokens(); ) {
+          if (value.equalsIgnoreCase(t.nextToken())) {
+            return true;
+          }
+        }
+      }
+    }
+    return false;
+  }
+
   private int update(final int n) throws IOException {
     if (n != -1) {
       currentPos += n;
-    } else if (currentPos < filelength) {
+    } else if (fileLength != null && currentPos < fileLength) {
       throw new IOException("Got EOF but currentPos = " + currentPos
-          + " < filelength = " + filelength);
+          + " < filelength = " + fileLength);
     }
     return n;
   }