add CIDSet for font subset to ensure PDF/A compliance

(cherry picked from commit 4873173b0edcf83d871e6fb86b96b81cec28653e)

openpdf/src/main/java/com/lowagie/text/pdf/TrueTypeFontUnicode.java

@@ -49,6 +49,10 @@
 
 package com.lowagie.text.pdf;
 
+import com.lowagie.text.DocumentException;
+import com.lowagie.text.Utilities;
+import com.lowagie.text.error_messages.MessageLocalization;
+
 import java.io.IOException;
 import java.util.ArrayList;
 import java.util.Arrays;
@@ -57,10 +61,6 @@
 import java.util.List;
 import java.util.Map;
 
-import com.lowagie.text.DocumentException;
-import com.lowagie.text.Utilities;
-import com.lowagie.text.error_messages.MessageLocalization;
-
 /** Represents a True Type font with Unicode encoding. All the character
  * in the font can be used directly by using the encoding Identity-H or
  * Identity-V. This is the only way to represent some character sets such
@@ -396,7 +396,7 @@ void writeFont(PdfWriter writer, PdfIndirectReference ref, Object[] params) thro
         PdfObject pobj = null;
         PdfIndirectObject obj = null;
         PdfIndirectReference cidset = null;
-        if (writer.getPDFXConformance() == PdfWriter.PDFA1A || writer.getPDFXConformance() == PdfWriter.PDFA1B) {
+        if (subset || writer.getPDFXConformance() == PdfWriter.PDFA1A || writer.getPDFXConformance() == PdfWriter.PDFA1B) {
             PdfStream stream;
             if (metrics.length == 0) {
                 stream = new PdfStream(new byte[]{(byte)0x80});

openpdf/src/test/java/com/lowagie/text/pdf/FontSubsetTest.java

@@ -0,0 +1,103 @@
+package com.lowagie.text.pdf;
+
+import com.lowagie.text.Document;
+import com.lowagie.text.Font;
+import com.lowagie.text.Paragraph;
+import org.apache.commons.io.IOUtils;
+import org.bouncycastle.crypto.prng.FixedSecureRandom;
+import org.junit.jupiter.api.Test;
+
+import java.io.ByteArrayOutputStream;
+import java.io.IOException;
+import java.io.InputStream;
+import java.security.SecureRandom;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+import static org.junit.jupiter.api.Assertions.assertNotEquals;
+import static org.junit.jupiter.api.Assertions.assertTrue;
+
+public class FontSubsetTest {
+
+    /*
+     * See : https://github.com/LibrePDF/OpenPDF/issues/623
+     */
+    @Test
+    public void createSubsetPrefixTest() throws Exception {
+        BaseFont font = BaseFont.createFont("LiberationSerif-Regular.ttf", BaseFont.IDENTITY_H,
+                BaseFont.EMBEDDED,true, getFontByte("fonts/liberation-serif/LiberationSerif-Regular.ttf"), null);
+        assertNotEquals(font.createSubsetPrefix(), font.createSubsetPrefix());
+
+        byte[] baseSeed = new SecureRandom().generateSeed(512);
+        // init deterministic SecureRandom with a custom base seed
+        SecureRandom secureRandom = new FixedSecureRandom(baseSeed);
+        font.setSecureRandom(secureRandom);
+        assertNotEquals(font.createSubsetPrefix(), font.createSubsetPrefix()); // still different, as FixedSecureRandom generates a new random on each step
+
+        SecureRandom secureRandomOne = new FixedSecureRandom(baseSeed);
+        font.setSecureRandom(secureRandomOne);
+
+        String subsetPrefixOne = font.createSubsetPrefix();
+
+        // re-init FixedSecureRandom for deterministic generation
+        SecureRandom secureRandomTwo = new FixedSecureRandom(baseSeed);
+        font.setSecureRandom(secureRandomTwo);
+
+        String subsetPrefixTwo = font.createSubsetPrefix();
+        assertEquals(subsetPrefixOne, subsetPrefixTwo); // the desired deterministic behavior
+    }
+
+    private byte[] getFontByte(String fileName) throws IOException {
+        try (InputStream stream = BaseFont.getResourceStream(fileName, null)) {
+            return IOUtils.toByteArray(stream);
+        }
+    }
+
+    /*
+     * This test is to ensure creation of CIDSet dictionary when using a font subset (required for PDF/A compliance)
+     */
+    @Test
+    public void subsetTest() throws Exception {
+        checkSubsetPresence(true);
+        checkSubsetPresence(false);
+    }
+
+    private void checkSubsetPresence(boolean subsetIncluded) throws Exception {
+        byte[] documentBytes;
+        try (ByteArrayOutputStream baos = new ByteArrayOutputStream()) {
+            Document document = new Document();
+            PdfWriter.getInstance(document, baos);
+            document.open();
+
+            BaseFont font = BaseFont.createFont("LiberationSerif-Regular.ttf", BaseFont.IDENTITY_H,
+                    BaseFont.EMBEDDED,true, getFontByte("fonts/liberation-serif/LiberationSerif-Regular.ttf"), null);
+            font.setSubset(subsetIncluded);
+            String text = "This is the test string.";
+            document.add(new Paragraph(text, new Font(font, 12)));
+            document.close();
+
+            documentBytes = baos.toByteArray();
+        }
+
+        boolean fontFound = false;
+        try (PdfReader reader = new PdfReader(documentBytes)) {
+            for (int k = 1; k < reader.getXrefSize(); ++k) {
+                PdfObject obj = reader.getPdfObjectRelease(k);
+                if (obj == null || !obj.isDictionary())
+                    continue;
+                PdfDictionary dic = (PdfDictionary) obj;
+                PdfObject type = PdfReader.getPdfObjectRelease(dic.get(PdfName.TYPE));
+                if (type == null || !type.isName())
+                    continue;
+                PdfDictionary fd = dic.getAsDict(PdfName.FONTDESCRIPTOR);
+                if (PdfName.FONT.equals(type) && fd != null) {
+                    PdfIndirectReference cidset = fd.getAsIndirectObject(PdfName.CIDSET);
+                    assertEquals(subsetIncluded, cidset != null);
+                    fontFound = true;
+                    break;
+                }
+            }
+        }
+        assertTrue(fontFound);
+    }
+
+}