♻️ harmonize getting page count from a local input source

Author: ianardee

src/main/java/com/mindee/input/LocalInputSource.java

@@ -1,6 +1,7 @@
 package com.mindee.input;
 
 import com.mindee.image.ImageCompressor;
+import com.mindee.pdf.PDFUtils;
 import com.mindee.pdf.PdfBoxApi;
 import com.mindee.pdf.PdfCompressor;
 import com.mindee.pdf.PdfOperation;
@@ -48,6 +49,14 @@ public LocalInputSource(String fileAsBase64, String filename) {
     this.filename = filename;
   }
 
+  /**
+   * Get the number of pages in the document.
+   * @return the number of pages in the current file.
+   * @throws IOException If an I/O error occurs during the PDF operation.
+   */
+  public int getPageCount() throws IOException {
+    return PDFUtils.getNumberOfPages(this.file);
+  }
 
   /**
    * Applies PDF-specific operations on the current file based on the specified {@code PageOptions}.

src/main/java/com/mindee/pdf/PDFUtils.java

@@ -39,7 +39,16 @@ private PDFUtils() {
    * @param inputSource The PDF file.
    */
   public static int getNumberOfPages(LocalInputSource inputSource) throws IOException {
-    PDDocument document = Loader.loadPDF(inputSource.getFile());
+    return getNumberOfPages(inputSource.getFile());
+  }
+
+  /**
+   * Get the number of pages in the PDF.
+   *
+   * @param pdfBytes The PDF file as a byte array.
+   */
+  public static int getNumberOfPages(byte[] pdfBytes) throws IOException {
+    PDDocument document = Loader.loadPDF(pdfBytes);
     int pageCount = document.getNumberOfPages();
     document.close();
     return pageCount;

src/test/java/com/mindee/input/LocalInputSourceTest.java

@@ -21,58 +21,51 @@
 import org.junit.jupiter.api.Test;
 
 public class LocalInputSourceTest {
+  void assertMultipagePDF(LocalInputSource localInputSource, File file) throws IOException {
+    Assertions.assertNotNull(localInputSource);
+    Assertions.assertTrue(localInputSource.isPdf());
+    Assertions.assertTrue(localInputSource.hasSourceText());
+    Assertions.assertEquals(3, localInputSource.getPageCount());
+    Assertions.assertEquals("multipage_cut-3.pdf", localInputSource.getFilename());
+    Assertions.assertArrayEquals(localInputSource.getFile(), Files.readAllBytes(file.toPath()));
+  }
+
   @Test
   void loadDocument_withFile_mustReturnAValidLocalInputSource() throws IOException {
-    File file = new File("src/test/resources/file_types/pdf/multipage.pdf");
+    File file = new File("src/test/resources/file_types/pdf/multipage_cut-3.pdf");
     LocalInputSource localInputSource = new LocalInputSource(file);
-    Assertions.assertNotNull(localInputSource);
-    Assertions.assertArrayEquals(localInputSource.getFile(), Files.readAllBytes(file.toPath()));
+    assertMultipagePDF(localInputSource, file);
   }
 
   @Test
   void loadDocument_withInputStream_mustReturnAValidLocalInputSource() throws IOException {
-    File file = new File("src/test/resources/file_types/pdf/multipage.pdf");
+    File file = new File("src/test/resources/file_types/pdf/multipage_cut-3.pdf");
     LocalInputSource localInputSource = new LocalInputSource(
         Files.newInputStream(file.toPath()),
-        "multipage.pdf"
+        "multipage_cut-3.pdf"
     );
-    Assertions.assertNotNull(localInputSource);
-    Assertions.assertArrayEquals(localInputSource.getFile(), Files.readAllBytes(file.toPath()));
+    assertMultipagePDF(localInputSource, file);
   }
 
   @Test
   void loadDocument_withByteArray_mustReturnAValidLocalInputSource() throws IOException {
-    File file = new File("src/test/resources/file_types/pdf/multipage.pdf");
+    File file = new File("src/test/resources/file_types/pdf/multipage_cut-3.pdf");
     LocalInputSource localInputSource = new LocalInputSource(
         Files.readAllBytes(file.toPath()),
-        "multipage.pdf"
+        "multipage_cut-3.pdf"
     );
-    Assertions.assertNotNull(localInputSource);
-    Assertions.assertArrayEquals(localInputSource.getFile(), Files.readAllBytes(file.toPath()));
+    assertMultipagePDF(localInputSource, file);
   }
 
   @Test
   void loadDocument_withBase64Encoded_mustReturnAValidLocalInputSource() throws IOException {
-    File file = new File("src/test/resources/file_types/pdf/multipage.pdf");
+    File file = new File("src/test/resources/file_types/pdf/multipage_cut-3.pdf");
     String encodedFile = Base64.encodeBase64String(Files.readAllBytes(file.toPath()));
     LocalInputSource localInputSource = new LocalInputSource(
         encodedFile,
-        "multipage.pdf"
+        "multipage_cut-3.pdf"
     );
-    Assertions.assertNotNull(localInputSource);
-    Assertions.assertArrayEquals(localInputSource.getFile(), Files.readAllBytes(file.toPath()));
-  }
-
-  @Test
-  void pdf_inputSource_withText_mustDetectSourceText() throws MindeeException, IOException {
-    File file = new File("src/test/resources/file_types/pdf/multipage.pdf");
-    String encodedFile = Base64.encodeBase64String(Files.readAllBytes(file.toPath()));
-    LocalInputSource localInputSource = new LocalInputSource(
-        encodedFile,
-        "multipage.pdf"
-    );
-    Assertions.assertNotNull(localInputSource);
-    Assertions.assertTrue(localInputSource.hasSourceText());
+    assertMultipagePDF(localInputSource, file);
   }
 
   @Test
@@ -84,6 +77,7 @@ void pdf_inputSource_withoutText_mustNotDetectSourceText() throws MindeeExceptio
         "default_sample.pdf"
     );
     Assertions.assertNotNull(localInputSource);
+    Assertions.assertTrue(localInputSource.isPdf());
     Assertions.assertFalse(localInputSource.hasSourceText());
   }
 
@@ -96,6 +90,7 @@ void image_inputSource_mustNotDetectSourceText() throws MindeeException, IOExcep
         "default_sample.jpg"
     );
     Assertions.assertNotNull(localInputSource);
+    Assertions.assertFalse(localInputSource.isPdf());
     Assertions.assertFalse(localInputSource.hasSourceText());
   }
 

src/test/java/com/mindee/input/URLInputSourceTest.java

@@ -16,7 +16,6 @@
 public class URLInputSourceTest {
 
   private static final String TEST_URL = "https://example.com/testfile.pdf";
-  private static final String TEST_LOCAL_FILENAME = "testfile.pdf";
   private TestableURLInputSource urlInputSource;
 
   @BeforeEach