Simplify saving added/modified annotations.

Having this map to collect the different changes will allow to know if some objects have already been modified.

src/core/document.js

@@ -70,7 +70,6 @@ import { OperatorList } from "./operator_list.js";
 import { PartialEvaluator } from "./evaluator.js";
 import { StreamsSequenceStream } from "./decode_stream.js";
 import { StructTreePage } from "./struct_tree.js";
-import { writeObject } from "./writer.js";
 import { XFAFactory } from "./xfa/factory.js";
 import { XRef } from "./xref.js";
 
@@ -314,7 +313,7 @@ class Page {
     await Promise.all(promises);
   }
 
-  async saveNewAnnotations(handler, task, annotations, imagePromises) {
+  async saveNewAnnotations(handler, task, annotations, imagePromises, changes) {
     if (this.xfaFactory) {
       throw new Error("XFA: Cannot save new annotations.");
     }
@@ -348,7 +347,8 @@ class Page {
       partialEvaluator,
       task,
       annotations,
-      imagePromises
+      imagePromises,
+      changes
     );
 
     for (const { ref } of newData.annotations) {
@@ -358,27 +358,20 @@ class Page {
       }
     }
 
-    const savedDict = pageDict.get("Annots");
-    pageDict.set("Annots", annotationsArray);
-    const buffer = [];
-    await writeObject(this.ref, pageDict, buffer, this.xref);
-    if (savedDict) {
-      pageDict.set("Annots", savedDict);
-    }
+    const dict = pageDict.clone();
+    dict.set("Annots", annotationsArray);
+    changes.put(this.ref, {
+      data: dict,
+    });
 
-    const objects = newData.dependencies;
-    objects.push(
-      { ref: this.ref, data: buffer.join("") },
-      ...newData.annotations
-    );
     for (const deletedRef of deletedAnnotations) {
-      objects.push({ ref: deletedRef, data: null });
+      changes.put(deletedRef, {
+        data: null,
+      });
     }
-
-    return objects;
   }
 
-  save(handler, task, annotationStorage) {
+  save(handler, task, annotationStorage, changes) {
     const partialEvaluator = new PartialEvaluator({
       xref: this.xref,
       handler,
@@ -395,11 +388,11 @@ class Page {
     // Fetch the page's annotations and save the content
     // in case of interactive form fields.
     return this._parsedAnnotations.then(function (annotations) {
-      const newRefsPromises = [];
+      const promises = [];
       for (const annotation of annotations) {
-        newRefsPromises.push(
+        promises.push(
           annotation
-            .save(partialEvaluator, task, annotationStorage)
+            .save(partialEvaluator, task, annotationStorage, changes)
             .catch(function (reason) {
               warn(
                 "save - ignoring annotation data during " +
@@ -410,9 +403,7 @@ class Page {
         );
       }
 
-      return Promise.all(newRefsPromises).then(function (newRefs) {
-        return newRefs.filter(newRef => !!newRef);
-      });
+      return Promise.all(promises);
     });
   }
 

src/core/primitives.js

@@ -383,6 +383,14 @@ class RefSetCache {
     this._map.clear();
   }
 
+  isEmpty() {
+    return this._map.size === 0;
+  }
+
+  *values() {
+    yield* this._map.values();
+  }
+
   *items() {
     for (const [ref, value] of this._map) {
       yield [Ref.fromString(ref), value];

src/core/struct_tree.js

@@ -17,7 +17,6 @@ import { AnnotationPrefix, stringToPDFString, warn } from "../shared/util.js";
 import { Dict, isName, Name, Ref, RefSetCache } from "./primitives.js";
 import { lookupNormalRect, stringToAsciiOrUTF16BE } from "./core_utils.js";
 import { NumberTree } from "./name_number_tree.js";
-import { writeObject } from "./writer.js";
 
 const MAX_DEPTH = 40;
 
@@ -117,7 +116,7 @@ class StructTreeRoot {
     xref,
     catalogRef,
     pdfManager,
-    newRefs,
+    changes,
   }) {
     const root = pdfManager.catalog.cloneDict();
     const cache = new RefSetCache();
@@ -146,18 +145,17 @@ class StructTreeRoot {
       nums,
       xref,
       pdfManager,
-      newRefs,
+      changes,
       cache,
     });
     structTreeRoot.set("ParentTreeNextKey", nextKey);
 
     cache.put(parentTreeRef, parentTree);
 
-    const buffer = [];
     for (const [ref, obj] of cache.items()) {
-      buffer.length = 0;
-      await writeObject(ref, obj, buffer, xref);
-      newRefs.push({ ref, data: buffer.join("") });
+      changes.put(ref, {
+        data: obj,
+      });
     }
   }
 
@@ -235,7 +233,7 @@ class StructTreeRoot {
     return true;
   }
 
-  async updateStructureTree({ newAnnotationsByPage, pdfManager, newRefs }) {
+  async updateStructureTree({ newAnnotationsByPage, pdfManager, changes }) {
     const xref = this.dict.xref;
     const structTreeRoot = this.dict.clone();
     const structTreeRootRef = this.ref;
@@ -273,7 +271,7 @@ class StructTreeRoot {
       nums,
       xref,
       pdfManager,
-      newRefs,
+      changes,
       cache,
     });
 
@@ -288,11 +286,10 @@ class StructTreeRoot {
       cache.put(numsRef, nums);
     }
 
-    const buffer = [];
     for (const [ref, obj] of cache.items()) {
-      buffer.length = 0;
-      await writeObject(ref, obj, buffer, xref);
-      newRefs.push({ ref, data: buffer.join("") });
+      changes.put(ref, {
+        data: obj,
+      });
     }
   }
 
@@ -304,13 +301,12 @@ class StructTreeRoot {
     nums,
     xref,
     pdfManager,
-    newRefs,
+    changes,
     cache,
   }) {
     const objr = Name.get("OBJR");
     let nextKey = -1;
     let structTreePageObjs;
-    const buffer = [];
 
     for (const [pageIndex, elements] of newAnnotationsByPage) {
       const page = await pdfManager.getPage(pageIndex);
@@ -350,9 +346,9 @@ class StructTreeRoot {
             // We update the existing tag.
             const tagDict = xref.fetch(objRef).clone();
             StructTreeRoot.#writeProperties(tagDict, accessibilityData);
-            buffer.length = 0;
-            await writeObject(objRef, tagDict, buffer, xref);
-            newRefs.push({ ref: objRef, data: buffer.join("") });
+            changes.put(objRef, {
+              data: tagDict,
+            });
             continue;
           }
         }

src/core/worker.js

@@ -34,7 +34,7 @@ import {
   getNewAnnotationsMap,
   XRefParseException,
 } from "./core_utils.js";
-import { Dict, isDict, Ref } from "./primitives.js";
+import { Dict, isDict, Ref, RefSetCache } from "./primitives.js";
 import { LocalPdfManager, NetworkPdfManager } from "./pdf_manager.js";
 import { AnnotationFactory } from "./annotation.js";
 import { clearGlobalCaches } from "./cleanup_helper.js";
@@ -540,6 +540,7 @@ class WorkerMessageHandler {
           pdfManager.ensureDoc("linearization"),
           pdfManager.ensureCatalog("structTreeRoot"),
         ];
+        const changes = new RefSetCache();
         const promises = [];
 
         const newAnnotationsByPage = !isPureXfa
@@ -590,7 +591,13 @@ class WorkerMessageHandler {
               pdfManager.getPage(pageIndex).then(page => {
                 const task = new WorkerTask(`Save (editor): page ${pageIndex}`);
                 return page
-                  .saveNewAnnotations(handler, task, annotations, imagePromises)
+                  .saveNewAnnotations(
+                    handler,
+                    task,
+                    annotations,
+                    imagePromises,
+                    changes
+                  )
                   .finally(function () {
                     finishWorkerTask(task);
                   });
@@ -600,26 +607,24 @@ class WorkerMessageHandler {
           if (structTreeRoot === null) {
             // No structTreeRoot exists, so we need to create one.
             promises.push(
-              Promise.all(newAnnotationPromises).then(async newRefs => {
+              Promise.all(newAnnotationPromises).then(async () => {
                 await StructTreeRoot.createStructureTree({
                   newAnnotationsByPage,
                   xref,
                   catalogRef,
                   pdfManager,
-                  newRefs,
+                  changes,
                 });
-                return newRefs;
               })
             );
           } else if (structTreeRoot) {
             promises.push(
-              Promise.all(newAnnotationPromises).then(async newRefs => {
+              Promise.all(newAnnotationPromises).then(async () => {
                 await structTreeRoot.updateStructureTree({
                   newAnnotationsByPage,
                   pdfManager,
-                  newRefs,
+                  changes,
                 });
-                return newRefs;
               })
             );
           }
@@ -633,7 +638,7 @@ class WorkerMessageHandler {
               pdfManager.getPage(pageIndex).then(function (page) {
                 const task = new WorkerTask(`Save: page ${pageIndex}`);
                 return page
-                  .save(handler, task, annotationStorage)
+                  .save(handler, task, annotationStorage, changes)
                   .finally(function () {
                     finishWorkerTask(task);
                   });
@@ -643,26 +648,21 @@ class WorkerMessageHandler {
         }
         const refs = await Promise.all(promises);
 
-        let newRefs = [];
         let xfaData = null;
         if (isPureXfa) {
           xfaData = refs[0];
           if (!xfaData) {
             return stream.bytes;
           }
-        } else {
-          newRefs = refs.flat(2);
-
-          if (newRefs.length === 0) {
-            // No new refs so just return the initial bytes
-            return stream.bytes;
-          }
+        } else if (changes.isEmpty()) {
+          // No new refs so just return the initial bytes
+          return stream.bytes;
         }
 
         const needAppearances =
           acroFormRef &&
           acroForm instanceof Dict &&
-          newRefs.some(ref => ref.needAppearances);
+          changes.values().some(ref => ref.needAppearances);
 
         const xfa = (acroForm instanceof Dict && acroForm.get("XFA")) || null;
         let xfaDatasetsRef = null;
@@ -712,7 +712,7 @@ class WorkerMessageHandler {
         return incrementalUpdate({
           originalData: stream.bytes,
           xrefInfo: newXrefInfo,
-          newRefs,
+          changes,
           xref,
           hasXfa: !!xfa,
           xfaDatasetsRef,