修复HDFS上的readBytesFromOtherInputStream：#536 (comment)

src/main/java/com/hankcs/hanlp/corpus/io/IOUtil.java

@@ -255,18 +255,27 @@ private static byte[] readBytesFromFileInputStream(FileInputStream fis) throws I
     }
 
     /**
-     * 将InputStream中的数据读入到字节数组中
+     * 将非FileInputStream的某InputStream中的全部数据读入到字节数组中
      *
      * @param is
      * @return
      * @throws IOException
      */
     public static byte[] readBytesFromOtherInputStream(InputStream is) throws IOException
     {
-        byte[] targetArray = new byte[is.available()];
-        readBytesFromOtherInputStream(is, targetArray);
-        is.close();
-        return targetArray;
+        ByteArrayOutputStream data = new ByteArrayOutputStream();
+
+        int readBytes;
+        byte[] buffer = new byte[Math.max(is.available(), 4096)]; // 最低4KB的缓冲区
+
+        while ((readBytes = is.read(buffer, 0, buffer.length)) != -1)
+        {
+            data.write(buffer, 0, readBytes);
+        }
+
+        data.flush();
+
+        return data.toByteArray();
     }
 
     /**

src/test/java/com/hankcs/hanlp/corpus/io/IOUtilTest.java

@@ -0,0 +1,34 @@
+package com.hankcs.hanlp.corpus.io;
+
+import junit.framework.TestCase;
+
+import java.io.ByteArrayInputStream;
+import java.util.Random;
+
+public class IOUtilTest extends TestCase
+{
+    public void testReadBytesFromOtherInputStream() throws Exception
+    {
+        Random random = new Random(System.currentTimeMillis());
+        byte[] originalData = new byte[1024 * 1024]; // 1MB
+        random.nextBytes(originalData);
+        ByteArrayInputStream is = new ByteArrayInputStream(originalData){
+            @Override
+            public synchronized int available()
+            {
+                int realAvailable = super.available();
+                if (realAvailable > 0)
+                {
+                    return 2048; // 模拟某些网络InputStream
+                }
+                return realAvailable;
+            }
+        };
+        byte[] readData = IOUtil.readBytesFromOtherInputStream(is);
+        assertEquals(originalData.length, readData.length);
+        for (int i = 0; i < originalData.length; i++)
+        {
+            assertEquals(originalData[i], readData[i]);
+        }
+    }
+}