Correctly update the xref table when an annotation is deleted

src/core/document.js

@@ -46,7 +46,15 @@ import {
   XRefEntryException,
   XRefParseException,
 } from "./core_utils.js";
-import { Dict, isName, isRefsEqual, Name, Ref, RefSet } from "./primitives.js";
+import {
+  Dict,
+  isName,
+  isRefsEqual,
+  Name,
+  Ref,
+  RefSet,
+  RefSetCache,
+} from "./primitives.js";
 import { getXfaFontDict, getXfaFontName } from "./xfa_fonts.js";
 import { BaseStream } from "./base_stream.js";
 import { calculateMD5 } from "./crypto.js";
@@ -272,7 +280,7 @@ class Page {
           continue;
         }
         if (annotation.deleted) {
-          deletedAnnotations.put(ref);
+          deletedAnnotations.put(ref, ref);
           continue;
         }
         existingAnnotations?.put(ref);
@@ -300,7 +308,7 @@ class Page {
       options: this.evaluatorOptions,
     });
 
-    const deletedAnnotations = new RefSet();
+    const deletedAnnotations = new RefSetCache();
     const existingAnnotations = new RefSet();
     this.#replaceIdByRef(annotations, deletedAnnotations, existingAnnotations);
 
@@ -335,6 +343,9 @@ class Page {
       { ref: this.ref, data: buffer.join("") },
       ...newData.annotations
     );
+    for (const deletedRef of deletedAnnotations) {
+      objects.push({ ref: deletedRef, data: null });
+    }
 
     return objects;
   }

src/core/writer.js

@@ -293,10 +293,18 @@ async function getXRefTable(xrefInfo, baseOffset, newRefs, newXref, buffer) {
     }
     // The EOL is \r\n to make sure that every entry is exactly 20 bytes long.
     // (see 7.5.4 - Cross-Reference Table).
-    buffer.push(
-      `${baseOffset.toString().padStart(10, "0")} ${Math.min(ref.gen, 0xffff).toString().padStart(5, "0")} n\r\n`
-    );
-    baseOffset += data.length;
+    if (data !== null) {
+      buffer.push(
+        `${baseOffset.toString().padStart(10, "0")} ${Math.min(ref.gen, 0xffff).toString().padStart(5, "0")} n\r\n`
+      );
+      baseOffset += data.length;
+    } else {
+      buffer.push(
+        `0000000000 ${Math.min(ref.gen + 1, 0xffff)
+          .toString()
+          .padStart(5, "0")} f\r\n`
+      );
+    }
   }
   computeIDs(baseOffset, xrefInfo, newXref);
   buffer.push("trailer\n");
@@ -327,11 +335,17 @@ async function getXRefStreamTable(
   let maxOffset = 0;
   let maxGen = 0;
   for (const { ref, data } of newRefs) {
+    let gen;
     maxOffset = Math.max(maxOffset, baseOffset);
-    const gen = Math.min(ref.gen, 0xffff);
+    if (data !== null) {
+      gen = Math.min(ref.gen, 0xffff);
+      xrefTableData.push([1, baseOffset, gen]);
+      baseOffset += data.length;
+    } else {
+      gen = Math.min(ref.gen + 1, 0xffff);
+      xrefTableData.push([0, 0, gen]);
+    }
     maxGen = Math.max(maxGen, gen);
-    xrefTableData.push([1, baseOffset, gen]);
-    baseOffset += data.length;
   }
   newXref.set("Index", getIndexes(newRefs));
   const offsetSize = getSizeInBytes(maxOffset);
@@ -420,15 +434,13 @@ async function incrementalUpdate({
     });
   }
 
-  let buffer, baseOffset;
+  const buffer = [];
+  let baseOffset = originalData.length;
   const lastByte = originalData.at(-1);
-  if (lastByte === /* \n */ 0x0a || lastByte === /* \r */ 0x0d) {
-    buffer = [];
-    baseOffset = originalData.length;
-  } else {
+  if (lastByte !== /* \n */ 0x0a && lastByte !== /* \r */ 0x0d) {
     // Avoid to concatenate %%EOF with an object definition
-    buffer = ["\n"];
-    baseOffset = originalData.length + 1;
+    buffer.push("\n");
+    baseOffset += 1;
   }
 
   const newXref = getTrailerDict(xrefInfo, newRefs, useXrefStream);
@@ -436,7 +448,9 @@ async function incrementalUpdate({
     (a, b) => /* compare the refs */ a.ref.num - b.ref.num
   );
   for (const { data } of newRefs) {
-    buffer.push(data);
+    if (data !== null) {
+      buffer.push(data);
+    }
   }
 
   await (useXrefStream

test/test_manifest.json

@@ -9919,5 +9919,26 @@
     "rounds": 1,
     "link": true,
     "type": "eq"
+  },
+  {
+    "id": "delete_freetexts",
+    "file": "pdfs/freetexts.pdf",
+    "md5": "da1310a25ab796c1201810070d5032a3",
+    "rounds": 1,
+    "type": "eq",
+    "save": true,
+    "print": true,
+    "annotationStorage": {
+      "pdfjs_internal_editor_0": {
+        "deleted": true,
+        "pageIndex": 0,
+        "id": "26R"
+      },
+      "pdfjs_internal_editor_3": {
+        "deleted": true,
+        "pageIndex": 0,
+        "id": "51R"
+      }
+    }
   }
 ]

test/unit/writer_spec.js

@@ -247,4 +247,71 @@ describe("Writer", function () {
       expect(data).toEqual(expected);
     });
   });
+
+  it("should update a file with a deleted object", async function () {
+    const originalData = new Uint8Array();
+    const newRefs = [
+      { ref: Ref.get(123, 0x2d), data: null },
+      { ref: Ref.get(456, 0x4e), data: "abc\n" },
+    ];
+    const xrefInfo = {
+      newRef: Ref.get(789, 0),
+      startXRef: 314,
+      fileIds: ["id", ""],
+      rootRef: null,
+      infoRef: null,
+      encryptRef: null,
+      filename: "foo.pdf",
+      info: {},
+    };
+
+    let data = await incrementalUpdate({
+      originalData,
+      xrefInfo,
+      newRefs,
+      useXrefStream: true,
+    });
+    data = bytesToString(data);
+
+    let expected =
+      "\nabc\n" +
+      "789 0 obj\n" +
+      "<< /Prev 314 /Size 790 /Type /XRef /Index [123 1 456 1 789 1] " +
+      "/W [1 1 1] /ID [(id) (\x01#Eg\x89\xab\xcd\xef\xfe\xdc\xba\x98vT2\x10)] " +
+      "/Length 9>> stream\n" +
+      "\x00\x00\x2e" +
+      "\x01\x01\x4e" +
+      "\x01\x05\x00\n" +
+      "endstream\n" +
+      "endobj\n" +
+      "startxref\n" +
+      "5\n" +
+      "%%EOF\n";
+    expect(data).toEqual(expected);
+
+    data = await incrementalUpdate({
+      originalData,
+      xrefInfo,
+      newRefs,
+      useXrefStream: false,
+    });
+    data = bytesToString(data);
+
+    expected =
+      "\nabc\n" +
+      "xref\n" +
+      "123 1\n" +
+      "0000000000 00046 f\r\n" +
+      "456 1\n" +
+      "0000000001 00078 n\r\n" +
+      "789 1\n" +
+      "0000000005 00000 n\r\n" +
+      "trailer\n" +
+      "<< /Prev 314 /Size 789 " +
+      "/ID [(id) (\x01#Eg\x89\xab\xcd\xef\xfe\xdc\xba\x98vT2\x10)]>>\n" +
+      "startxref\n" +
+      "5\n" +
+      "%%EOF\n";
+    expect(data).toEqual(expected);
+  });
 });