[WIP] Update the page labels tree when a pdf is extracted (bug 1997379)

Author: calixteman

src/core/catalog.js

@@ -735,6 +735,16 @@ class Catalog {
     return rawDests;
   }
 
+  get rawPageLabels() {
+    const obj = this.#catDict.getRaw("PageLabels");
+    if (!obj) {
+      return null;
+    }
+
+    const numberTree = new NumberTree(obj, this.xref);
+    return numberTree.getAll();
+  }
+
   get pageLabels() {
     let obj = null;
     try {
@@ -749,17 +759,15 @@ class Catalog {
   }
 
   #readPageLabels() {
-    const obj = this.#catDict.getRaw("PageLabels");
-    if (!obj) {
+    const nums = this.rawPageLabels;
+    if (!nums) {
       return null;
     }
 
     const pageLabels = new Array(this.numPages);
     let style = null,
       prefix = "";
 
-    const numberTree = new NumberTree(obj, this.xref);
-    const nums = numberTree.getAll();
     let currentLabel = "",
       currentIndex = 1;
 

src/core/editor/pdf_editor.js

@@ -25,6 +25,7 @@ import { StringStream } from "../stream.js";
 import { stringToAsciiOrUTF16BE } from "../core_utils.js";
 
 const MAX_LEAVES_PER_PAGES_NODE = 16;
+const MAX_IN_NAME_TREE_NODE = 64;
 
 class PageData {
   constructor(page, documentData) {
@@ -39,6 +40,7 @@ class PageData {
 class DocumentData {
   constructor(document) {
     this.document = document;
+    this.pageLabels = null;
     this.pagesMap = new RefSetCache();
     this.oldRefMapping = new RefSetCache();
   }
@@ -61,6 +63,7 @@ class PDFEditor {
     this.version = "1.7";
     this.title = title;
     this.author = author;
+    this.pageLabels = null;
   }
 
   /**
@@ -253,6 +256,8 @@ class PDFEditor {
     await Promise.all(promises);
     promises.length = 0;
 
+    this.#collectPageLabels();
+
     for (const page of this.oldPages) {
       promises.push(this.#postCollectPageData(page));
     }
@@ -270,7 +275,12 @@ class PDFEditor {
    * @param {DocumentData} documentData
    * @return {Promise<void>}
    */
-  async #collectDocumentData(documentData) {}
+  async #collectDocumentData(documentData) {
+    const { document } = documentData;
+    await document.pdfManager
+      .ensureCatalog("rawPageLabels")
+      .then(pageLabels => (documentData.pageLabels = pageLabels));
+  }
 
   /**
    * Post process the collected page data.
@@ -306,6 +316,56 @@ class PDFEditor {
     pageData.annotations = newAnnotations.length > 0 ? newAnnotations : null;
   }
 
+  async #collectPageLabels() {
+    // We can only preserve page labels when editing a single PDF file.
+    // This is consistent with behavior in Adobe Acrobat.
+    if (!this.hasSingleFile) {
+      return;
+    }
+    const {
+      documentData: { document, pageLabels },
+    } = this.oldPages[0];
+    if (!pageLabels) {
+      return;
+    }
+    const numPages = document.numPages;
+    const oldPageLabels = [];
+    const oldPageIndices = new Set(
+      this.oldPages.map(({ page: { pageIndex } }) => pageIndex)
+    );
+    let currentLabel = null;
+    let stFirstIndex = -1;
+    for (let i = 0; i < numPages; i++) {
+      const newLabel = pageLabels.get(i);
+      if (newLabel) {
+        currentLabel = newLabel;
+        stFirstIndex = currentLabel.has("St") ? i : -1;
+      }
+      if (!oldPageIndices.has(i)) {
+        continue;
+      }
+      if (stFirstIndex !== -1) {
+        const st = currentLabel.get("St");
+        currentLabel = currentLabel.clone();
+        currentLabel.set("St", st + (i - stFirstIndex));
+        stFirstIndex = -1;
+      }
+      oldPageLabels.push(currentLabel);
+    }
+    currentLabel = oldPageLabels[0];
+    let currentIndex = 0;
+    const newPageLabels = (this.pageLabels = [[0, currentLabel]]);
+    for (let i = 0, ii = oldPageLabels.length; i < ii; i++) {
+      const label = oldPageLabels[i];
+      if (label === currentLabel) {
+        continue;
+      }
+      currentIndex = i;
+      currentLabel = label;
+      newPageLabels.push([currentIndex, currentLabel]);
+    }
+  }
+
   /**
    * Create a copy of a page.
    * @param {number} pageIndex
@@ -423,6 +483,63 @@ class PDFEditor {
     }
   }
 
+  /**
+   * Create a name or number tree from the given map.
+   * @param {Array<[string, any]>} map
+   * @returns {Ref}
+   */
+  #makeNameNumTree(map, areNames) {
+    const allEntries = map.sort(
+      areNames
+        ? ([keyA], [keyB]) => keyA.localeCompare(keyB)
+        : ([keyA], [keyB]) => keyA - keyB
+    );
+    const maxLeaves =
+      MAX_IN_NAME_TREE_NODE <= 1 ? allEntries.length : MAX_IN_NAME_TREE_NODE;
+    const [treeRef, treeDict] = this.newDict;
+    const stack = [{ dict: treeDict, entries: allEntries }];
+    const valueType = areNames ? "Names" : "Nums";
+
+    while (stack.length > 0) {
+      const { dict, entries } = stack.pop();
+      if (entries.length <= maxLeaves) {
+        dict.set("Limits", [entries[0][0], entries.at(-1)[0]]);
+        dict.set(valueType, entries.flat());
+        continue;
+      }
+      const entriesChunks = [];
+      const chunkSize = Math.ceil(entries.length / maxLeaves);
+      for (let i = 0; i < entries.length; i += chunkSize) {
+        entriesChunks.push(entries.slice(i, i + chunkSize));
+      }
+      const entriesRefs = [];
+      dict.set("Kids", entriesRefs);
+      for (const chunk of entriesChunks) {
+        const [entriesRef, entriesDict] = this.newDict;
+        entriesRefs.push(entriesRef);
+        entriesDict.set("Limits", [chunk[0][0], chunk.at(-1)[0]]);
+        stack.push({ dict: entriesDict, entries: chunk });
+      }
+    }
+    return treeRef;
+  }
+
+  /**
+   * Create the page labels tree if it exists.
+   */
+  #makePageLabelsTree() {
+    const { pageLabels } = this;
+    if (!pageLabels || pageLabels.length === 0) {
+      return;
+    }
+    const { rootDict } = this;
+    const pageLabelsRef = this.#makeNameNumTree(
+      this.pageLabels,
+      /* areNames = */ false
+    );
+    rootDict.set("PageLabels", pageLabelsRef);
+  }
+
   /**
    * Create the root dictionary.
    * @returns {Promise<void>}
@@ -432,6 +549,7 @@ class PDFEditor {
     rootDict.setIfName("Type", "Catalog");
     rootDict.set("Version", this.version);
     this.#makePageTree();
+    this.#makePageLabelsTree();
   }
 
   /**

test/pdfs/.gitignore

@@ -757,3 +757,4 @@
 !doc_1_3_pages.pdf
 !doc_2_3_pages.pdf
 !doc_3_3_pages.pdf
+!labelled_pages.pdf

test/pdfs/labelled_pages.pdf

@@ -5495,5 +5495,39 @@ small scripts as well as for`);
         await loadingTask.destroy();
       });
     });
+
+    describe("Page labels", function () {
+      it("extract page and check labels", async function () {
+        let loadingTask = getDocument(
+          buildGetDocumentParams("labelled_pages.pdf")
+        );
+        const pdfDoc = await loadingTask.promise;
+        let labels = await pdfDoc.getPageLabels();
+        expect(labels).toEqual([
+          "i" /* Page 0 */,
+          "ii" /* Page 1 */,
+          "iii" /* Page 2 */,
+          "iv" /* Page 3 */,
+          "1" /* Page 4 */,
+          "2" /* Page 5 */,
+          "3" /* Page 6 */,
+          "a" /* Page 7 */,
+          "b" /* Page 8 */,
+          "4" /* Page 9 */,
+          "5" /* Page 10 */,
+        ]);
+
+        const data = await pdfDoc.extractPages({
+          document: null,
+          includePages: [0, 1, 5, 7, 10],
+        });
+        await loadingTask.destroy();
+        loadingTask = getDocument(data);
+        const newPdfDoc = await loadingTask.promise;
+        labels = await newPdfDoc.getPageLabels();
+        expect(labels).toEqual(["i", "ii", "1", "a", "5"]);
+        await loadingTask.destroy();
+      });
+    });
   });
 });