Fix opened connections outside ResourceRetriever

DEVSIX-5037

Author: Egor Martsynkovsky

io/src/main/java/com/itextpdf/io/util/UrlUtil.java

@@ -103,21 +103,21 @@ public static InputStream openStream(URL url) throws IOException {
 
     /**
      * This method gets the last redirected url.
-     * @param initialUrl an initial URL
-     * @return the last redirected url
-     * @throws IOException
+     *
+     * @param initialUrl an initial URL.
+     *
+     * @return the last redirected url.
+     *
+     * @throws IOException signals that an I/O exception has occurred.
+     *
+     * @deprecated {@link UrlUtil#getInputStreamOfFinalConnection(URL)} can be used to get input stream from final
+     * connection.
      */
+    @Deprecated
     public static URL getFinalURL(URL initialUrl) throws IOException {
-        URL finalUrl = null;
-        URL nextUrl = initialUrl;
-        while (nextUrl != null) {
-            finalUrl = nextUrl;
-            URLConnection connection = finalUrl.openConnection();
-            String location = connection.getHeaderField("location");
-            // Close input stream deliberately to close the handle which is created during getHeaderField invocation
-            connection.getInputStream().close();
-            nextUrl = location != null ? new URL(location) : null;
-        }
+        final URLConnection finalConnection = getFinalConnection(initialUrl);
+        final URL finalUrl = finalConnection.getURL();
+        finalConnection.getInputStream().close();
         return finalUrl;
     }
 
@@ -138,4 +138,44 @@ public static String getFileUriString(String filename) throws MalformedURLExcept
     public static String getNormalizedFileUriString(String filename) {
         return "file://" + UrlUtil.toNormalizedURI(filename).getPath();
     }
+
+    /**
+     * Gets the input stream of connection related to last redirected url. You should manually close input stream after
+     * calling this method to not hold any open resources.
+     *
+     * @param initialUrl an initial URL.
+     *
+     * @return an input stream of connection related to the last redirected url.
+     *
+     * @throws IOException signals that an I/O exception has occurred.
+     */
+    public static InputStream getInputStreamOfFinalConnection(URL initialUrl) throws IOException {
+        final URLConnection finalConnection = getFinalConnection(initialUrl);
+        return finalConnection.getInputStream();
+    }
+
+    /**
+     * Gets the connection related to the last redirected url. You should close connection manually after calling
+     * this method, to not hold any open resources.
+     *
+     * @param initialUrl an initial URL.
+     *
+     * @return connection related to the last redirected url.
+     *
+     * @throws IOException signals that an I/O exception has occurred.
+     */
+     static URLConnection getFinalConnection(URL initialUrl) throws IOException {
+        URL nextUrl = initialUrl;
+        URLConnection connection = null;
+        while (nextUrl != null) {
+            connection = nextUrl.openConnection();
+            final String location = connection.getHeaderField("location");
+            nextUrl = location == null ? null : new URL(location);
+            if (nextUrl != null) {
+                // close input stream deliberately to close the handle which is created during getHeaderField invocation
+                connection.getInputStream().close();
+            }
+        }
+        return connection;
+    }
 }

io/src/test/java/com/itextpdf/io/util/UrlUtilTest.java

@@ -45,9 +45,13 @@ This file is part of the iText (R) project.
 import com.itextpdf.commons.utils.FileUtil;
 import com.itextpdf.test.ExtendedITextTest;
 import com.itextpdf.test.annotations.type.UnitTest;
+
 import java.io.File;
 import java.io.IOException;
 import java.io.InputStream;
+import java.net.URL;
+import java.net.URLConnection;
+import java.net.UnknownHostException;
 import java.nio.charset.StandardCharsets;
 import java.nio.file.Paths;
 import org.junit.Assert;
@@ -65,7 +69,9 @@ public static void beforeClass() {
         createDestinationFolder(destinationFolder);
     }
 
-    // Tests that after invocation of the getFinalURL method for local files, no handles are left open and the file is free to be removed
+
+    // Tests that after invocation of the getFinalURL method for local files, no handles are left open and the file
+    // is free to be removed.
     @Test
     public void getFinalURLDoesNotLockFileTest() throws IOException {
         File tempFile = FileUtil.createTempFile(destinationFolder);
@@ -75,6 +81,44 @@ public void getFinalURLDoesNotLockFileTest() throws IOException {
         Assert.assertTrue(FileUtil.deleteFile(tempFile));
     }
 
+    // Tests, that getFinalConnection will be redirected some times for other urls, and initialUrl will be different
+    // from final url.
+    @Test
+    public void getFinalConnectionWhileRedirectingTest() throws IOException {
+        URL initialUrl = new URL("http://itextpdf.com");
+        URL expectedURL = new URL("https://itextpdf.com/en");
+        URLConnection finalConnection = null;
+
+        try {
+            finalConnection = UrlUtil.getFinalConnection(initialUrl);
+
+            Assert.assertNotNull(finalConnection);
+            Assert.assertNotEquals(initialUrl, finalConnection.getURL());
+            Assert.assertEquals(expectedURL, finalConnection.getURL());
+        } finally {
+            finalConnection.getInputStream().close();
+        }
+    }
+
+    // This test checks that when we pass invalid url and trying get stream related to final redirected url,exception
+    // would be thrown.
+    @Test
+    public void getInputStreamOfFinalConnectionThrowExceptionTest() throws IOException {
+        URL invalidUrl = new URL("http://itextpdf");
+
+        Assert.assertThrows(UnknownHostException.class, () -> UrlUtil.getInputStreamOfFinalConnection(invalidUrl));
+    }
+
+    // This test checks that when we pass valid url and trying get stream related to final redirected url, it would
+    // not be null.
+    @Test
+    public void getInputStreamOfFinalConnectionTest() throws IOException {
+        URL initialUrl = new URL("http://itextpdf.com");
+        InputStream streamOfFinalConnectionOfInvalidUrl = UrlUtil.getInputStreamOfFinalConnection(initialUrl);
+
+        Assert.assertNotNull(streamOfFinalConnectionOfInvalidUrl);
+    }
+
     @Test
     public void getBaseUriTest() throws IOException {
         String absolutePathRoot = Paths.get("").toAbsolutePath().toUri().toURL().toExternalForm();

styled-xml-parser/src/main/java/com/itextpdf/styledxmlparser/resolver/resource/DefaultResourceRetriever.java

@@ -24,6 +24,7 @@ This file is part of the iText (R) project.
 
 import com.itextpdf.commons.utils.MessageFormatUtil;
 import com.itextpdf.io.util.StreamUtil;
+import com.itextpdf.io.util.UrlUtil;
 import com.itextpdf.styledxmlparser.logs.StyledXmlParserLogMessageConstant;
 import com.itextpdf.styledxmlparser.exceptions.ReadingByteLimitException;
 
@@ -90,7 +91,7 @@ public InputStream getInputStreamByUrl(URL url) throws IOException {
                             url));
             return null;
         }
-        return new LimitedInputStream(url.openStream(), resourceSizeByteLimit);
+        return new LimitedInputStream(UrlUtil.getInputStreamOfFinalConnection(url), resourceSizeByteLimit);
     }
 
     /**

styled-xml-parser/src/main/java/com/itextpdf/styledxmlparser/resolver/resource/ResourceResolver.java

@@ -45,7 +45,6 @@ This file is part of the iText (R) project.
 import com.itextpdf.commons.utils.Base64;
 import com.itextpdf.commons.utils.MessageFormatUtil;
 import com.itextpdf.io.image.ImageDataFactory;
-import com.itextpdf.io.util.UrlUtil;
 import com.itextpdf.kernel.pdf.xobject.PdfImageXObject;
 import com.itextpdf.kernel.pdf.xobject.PdfXObject;
 import com.itextpdf.styledxmlparser.logs.StyledXmlParserLogMessageConstant;
@@ -277,7 +276,6 @@ protected PdfXObject tryResolveBase64ImageSource(String src) {
     protected PdfXObject tryResolveUrlImageSource(String uri) {
         try {
             URL url = uriResolver.resolveAgainstBaseUri(uri);
-            url = UrlUtil.getFinalURL(url);
             String imageResolvedSrc = url.toExternalForm();
             PdfXObject imageXObject = imageCache.getImage(imageResolvedSrc);
             if (imageXObject == null) {
@@ -295,9 +293,9 @@ protected PdfXObject tryResolveUrlImageSource(String uri) {
     /**
      * Create a iText XObject based on the image stored at the passed location.
      *
-     * @param url location of the Image file
-     * @return {@link PdfXObject} containing the Image loaded in
-     * @throws Exception thrown if error occurred during fetching or constructing the image
+     * @param url location of the Image file.
+     * @return {@link PdfXObject} containing the Image loaded in.
+     * @throws Exception thrown if error occurred during fetching or constructing the image.
      */
     protected PdfXObject createImageByUrl(URL url) throws Exception {
         byte[] bytes = retriever.getByteArrayByUrl(url);