fix: Correctly calculate Content-Length when reading from a stream (#1277)

Author: arjankowski

src/intTest/java/com/box/sdk/BoxFileIT.java

@@ -146,6 +146,27 @@ public void getRepresentationContentSucceeds() throws InterruptedException {
         }
     }
 
+    @Test
+    public void getRepresentationContentWithExtractedTextSucceeds() throws InterruptedException {
+        BoxAPIConnection api = jwtApiForServiceAccount();
+        String fileName = "text.pdf";
+        BoxFile file = null;
+        try {
+            file = uploadSampleFileToUniqueFolder(api, fileName);
+            final String fileId = file.getID();
+            String representationHint = "[extracted_text]";
+            ByteArrayOutputStream outputStream = new ByteArrayOutputStream();
+            Retry.retry(() -> {
+                new BoxFile(api, fileId).getRepresentationContent(representationHint, outputStream);
+                byte[] downloadedRepresentationContent = outputStream.toByteArray();
+                String text = new String(downloadedRepresentationContent, StandardCharsets.UTF_8);
+                assertTrue(text.contains("Lorem ipsum"));
+            }, 5, 100);
+        } finally {
+            deleteFile(file);
+        }
+    }
+
     @Test
     public void uploadFileStreamSucceeds() {
         BoxAPIConnection api = jwtApiForServiceAccount();
@@ -156,7 +177,7 @@ public void uploadFileStreamSucceeds() {
 
         BoxFile uploadedFile = null;
         try {
-            InputStream uploadStream =  new ByteArrayInputStream(fileContent);
+            InputStream uploadStream = new ByteArrayInputStream(fileContent);
             BoxFile.Info uploadedFileInfo = folder.uploadFile(uploadStream, BoxFileIT.generateString());
             uploadedFile = uploadedFileInfo.getResource();
 
@@ -552,11 +573,11 @@ public void canPaginateOverListOfVersions() {
 
             byte[] fileBytes = "Version 2".getBytes(StandardCharsets.UTF_8);
             uploadedFile.uploadNewVersion(
-                    new ByteArrayInputStream(fileBytes), null, fileBytes.length, mock(ProgressListener.class));
+                new ByteArrayInputStream(fileBytes), null, fileBytes.length, mock(ProgressListener.class));
 
             fileBytes = "Version 3".getBytes(StandardCharsets.UTF_8);
             uploadedFile.uploadNewVersion(
-                    new ByteArrayInputStream(fileBytes), null, fileBytes.length, mock(ProgressListener.class));
+                new ByteArrayInputStream(fileBytes), null, fileBytes.length, mock(ProgressListener.class));
 
             Collection<BoxFileVersion> versionsPart1 = uploadedFile.getVersionsRange(0, 1);
             assertThat(versionsPart1.size(), is(1));

src/main/java/com/box/sdk/BinaryBodyUtils.java

@@ -1,5 +1,6 @@
 package com.box.sdk;
 
+import com.box.sdk.http.HttpHeaders;
 import java.io.IOException;
 import java.io.InputStream;
 import java.io.OutputStream;
@@ -73,12 +74,43 @@ static void writeStreamWithContentLength(BoxAPIResponse response, OutputStream o
             } else {
                 input = response.getBody();
             }
-            writeStreamTo(input, output, response.getContentLength());
+            writeStreamTo(input, output, getContentLengthFromAPIResponse(response));
         } finally {
             response.close();
         }
     }
 
+    /**
+     * Get the content length from the API response.
+     * In some cases, the Content-Length is not provided in the response headers.
+     * This could happen when getting the content representation for a compressed data.
+     * In that case the API will switch to chunk mode and provide the length in the "X-Original-Content-Length" header.
+     *
+     * @param response API response.
+     * @return Content length.
+     */
+    private static long getContentLengthFromAPIResponse(BoxAPIResponse response) {
+        long length = response.getContentLength();
+        if (length == -1) {
+            String headerValue = null;
+            if (response.getHeaders().containsKey(HttpHeaders.CONTENT_LENGTH)) {
+                headerValue = response.getHeaders().get(HttpHeaders.CONTENT_LENGTH).get(0);
+            } else if (response.getHeaders().containsKey(HttpHeaders.X_ORIGINAL_CONTENT_LENGTH)) {
+                headerValue = response.getHeaders().get(HttpHeaders.X_ORIGINAL_CONTENT_LENGTH).get(0);
+            }
+
+            if (headerValue != null) {
+                try {
+                    length = Integer.parseInt(headerValue);
+                } catch (NumberFormatException e) {
+                    throw new RuntimeException("Invalid content length: " + headerValue);
+                }
+            }
+        }
+
+        return length;
+    }
+
     /**
      * Writes content of input stream to provided output.
      *
@@ -126,8 +158,8 @@ static void writeStreamTo(InputStream input, OutputStream output, long expectedL
                 totalBytesRead += n;  // Track the total bytes read
             }
             if (totalBytesRead != expectedLength) {
-                throw new IOException("Stream ended prematurely. Expected " + expectedLength
-                        + " bytes, but read " + totalBytesRead + " bytes.");
+                throw new IOException("Stream ended prematurely. Expected "
+                    + expectedLength + " bytes, but read " + totalBytesRead + " bytes.");
             }
         } catch (IOException e) {
             throw new RuntimeException("Error during streaming: " + e.getMessage(), e);

src/main/java/com/box/sdk/http/HttpHeaders.java

@@ -10,6 +10,11 @@ public final class HttpHeaders {
      */
     public static final String CONTENT_LENGTH = "Content-Length";
 
+    /**
+     * HTTP header key X-Original-Content-Length.
+     */
+    public static final String X_ORIGINAL_CONTENT_LENGTH = "X-Original-Content-Length";
+
     /**
      * HTTP header key Content-Type.
      */