Add tagged annotations in the structure tree (bug 1850797)

Author: calixteman

src/core/annotation.js

@@ -77,10 +77,11 @@ class AnnotationFactory {
    * @param {PDFManager} pdfManager
    * @param {Object} idFactory
    * @param {boolean} collectFields
+   * @param {Object} [pageRef]
    * @returns {Promise} A promise that is resolved with an {Annotation}
    *   instance.
    */
-  static create(xref, ref, pdfManager, idFactory, collectFields) {
+  static create(xref, ref, pdfManager, idFactory, collectFields, pageRef) {
     return Promise.all([
       pdfManager.ensureCatalog("acroForm"),
       // Only necessary to prevent the `pdfManager.docBaseUrl`-getter, used
@@ -91,18 +92,29 @@ class AnnotationFactory {
       pdfManager.ensureCatalog("attachments"),
       pdfManager.ensureDoc("xfaDatasets"),
       collectFields ? this._getPageIndex(xref, ref, pdfManager) : -1,
-    ]).then(([acroForm, baseUrl, attachments, xfaDatasets, pageIndex]) =>
-      pdfManager.ensure(this, "_create", [
-        xref,
-        ref,
-        pdfManager,
-        idFactory,
+      pageRef ? pdfManager.ensureCatalog("structTreeRoot") : null,
+    ]).then(
+      ([
         acroForm,
+        baseUrl,
         attachments,
         xfaDatasets,
-        collectFields,
         pageIndex,
-      ])
+        structTreeRoot,
+      ]) =>
+        pdfManager.ensure(this, "_create", [
+          xref,
+          ref,
+          pdfManager,
+          idFactory,
+          acroForm,
+          attachments,
+          xfaDatasets,
+          collectFields,
+          pageIndex,
+          structTreeRoot,
+          pageRef,
+        ])
     );
   }
 
@@ -118,7 +130,9 @@ class AnnotationFactory {
     attachments = null,
     xfaDatasets,
     collectFields,
-    pageIndex = -1
+    pageIndex = -1,
+    structTreeRoot = null,
+    pageRef = null
   ) {
     const dict = xref.fetchIfRef(ref);
     if (!(dict instanceof Dict)) {
@@ -150,6 +164,8 @@ class AnnotationFactory {
         !collectFields && acroFormDict.get("NeedAppearances") === true,
       pageIndex,
       evaluatorOptions: pdfManager.evaluatorOptions,
+      structTreeRoot,
+      pageRef,
     };
 
     switch (subtype) {
@@ -594,6 +610,13 @@ class Annotation {
     const isLocked = !!(this.flags & AnnotationFlag.LOCKED);
     const isContentLocked = !!(this.flags & AnnotationFlag.LOCKEDCONTENTS);
 
+    if (params.structTreeRoot) {
+      let structParent = dict.get("StructParent");
+      structParent =
+        Number.isInteger(structParent) && structParent >= 0 ? structParent : -1;
+      params.structTreeRoot.addAnnotationIdToPage(params.pageRef, structParent);
+    }
+
     // Expose public properties using a data object.
     this.data = {
       annotationFlags: this.flags,

src/core/document.js

@@ -651,6 +651,9 @@ class Page {
     if (!structTreeRoot) {
       return null;
     }
+    // Ensure that the structTree will contain the page's annotations.
+    await this._parsedAnnotations;
+
     const structTree = await this.pdfManager.ensure(this, "_parseStructTree", [
       structTreeRoot,
     ]);
@@ -662,7 +665,7 @@ class Page {
    */
   _parseStructTree(structTreeRoot) {
     const tree = new StructTreePage(structTreeRoot, this.pageDict);
-    tree.parse();
+    tree.parse(this.ref);
     return tree;
   }
 
@@ -740,7 +743,8 @@ class Page {
               annotationRef,
               this.pdfManager,
               this._localIdFactory,
-              /* collectFields */ false
+              /* collectFields */ false,
+              this.ref
             ).catch(function (reason) {
               warn(`_parsedAnnotations: "${reason}".`);
               return null;
@@ -1719,7 +1723,8 @@ class PDFDocument {
         fieldRef,
         this.pdfManager,
         this._localIdFactory,
-        /* collectFields */ true
+        /* collectFields */ true,
+        /* pageRef */ null
       )
         .then(annotation => annotation?.getFieldObject())
         .catch(function (reason) {

src/core/struct_tree.js

@@ -13,29 +13,48 @@
  * limitations under the License.
  */
 
-import { Dict, isName, Name, Ref } from "./primitives.js";
-import { stringToPDFString, warn } from "../shared/util.js";
+import { AnnotationPrefix, stringToPDFString, warn } from "../shared/util.js";
+import { Dict, isName, Name, Ref, RefSetCache } from "./primitives.js";
 import { NumberTree } from "./name_number_tree.js";
 
 const MAX_DEPTH = 40;
 
 const StructElementType = {
-  PAGE_CONTENT: "PAGE_CONTENT",
-  STREAM_CONTENT: "STREAM_CONTENT",
-  OBJECT: "OBJECT",
-  ELEMENT: "ELEMENT",
+  PAGE_CONTENT: 1,
+  STREAM_CONTENT: 2,
+  OBJECT: 3,
+  ANNOTATION: 4,
+  ELEMENT: 5,
 };
 
 class StructTreeRoot {
   constructor(rootDict) {
     this.dict = rootDict;
     this.roleMap = new Map();
+    this.structParentIds = null;
   }
 
   init() {
     this.readRoleMap();
   }
 
+  #addIdToPage(pageRef, id, type) {
+    if (!(pageRef instanceof Ref) || id < 0) {
+      return;
+    }
+    this.structParentIds ||= new RefSetCache();
+    let ids = this.structParentIds.get(pageRef);
+    if (!ids) {
+      ids = [];
+      this.structParentIds.put(pageRef, ids);
+    }
+    ids.push([id, type]);
+  }
+
+  addAnnotationIdToPage(pageRef, id) {
+    this.#addIdToPage(pageRef, id, StructElementType.ANNOTATION);
+  }
+
   readRoleMap() {
     const roleMapDict = this.dict.get("RoleMap");
     if (!(roleMapDict instanceof Dict)) {
@@ -129,12 +148,10 @@ class StructElementNode {
       if (this.tree.pageDict.objId !== pageObjId) {
         return null;
       }
+      const kidRef = kidDict.getRaw("Stm");
       return new StructElement({
         type: StructElementType.STREAM_CONTENT,
-        refObjId:
-          kidDict.getRaw("Stm") instanceof Ref
-            ? kidDict.getRaw("Stm").toString()
-            : null,
+        refObjId: kidRef instanceof Ref ? kidRef.toString() : null,
         pageObjId,
         mcid: kidDict.get("MCID"),
       });
@@ -144,12 +161,10 @@ class StructElementNode {
       if (this.tree.pageDict.objId !== pageObjId) {
         return null;
       }
+      const kidRef = kidDict.getRaw("Obj");
       return new StructElement({
         type: StructElementType.OBJECT,
-        refObjId:
-          kidDict.getRaw("Obj") instanceof Ref
-            ? kidDict.getRaw("Obj").toString()
-            : null,
+        refObjId: kidRef instanceof Ref ? kidRef.toString() : null,
         pageObjId,
       });
     }
@@ -186,7 +201,7 @@ class StructTreePage {
     this.nodes = [];
   }
 
-  parse() {
+  parse(pageRef) {
     if (!this.root || !this.rootDict) {
       return;
     }
@@ -196,18 +211,42 @@ class StructTreePage {
       return;
     }
     const id = this.pageDict.get("StructParents");
-    if (!Number.isInteger(id)) {
+    const ids =
+      pageRef instanceof Ref && this.root.structParentIds?.get(pageRef);
+    if (!Number.isInteger(id) && !ids) {
       return;
     }
+
+    const map = new Map();
     const numberTree = new NumberTree(parentTree, this.rootDict.xref);
-    const parentArray = numberTree.get(id);
-    if (!Array.isArray(parentArray)) {
+
+    if (Number.isInteger(id)) {
+      const parentArray = numberTree.get(id);
+      if (Array.isArray(parentArray)) {
+        for (const ref of parentArray) {
+          if (ref instanceof Ref) {
+            this.addNode(this.rootDict.xref.fetch(ref), map);
+          }
+        }
+      }
+    }
+
+    if (!ids) {
       return;
     }
-    const map = new Map();
-    for (const ref of parentArray) {
-      if (ref instanceof Ref) {
-        this.addNode(this.rootDict.xref.fetch(ref), map);
+    for (const [elemId, type] of ids) {
+      const obj = numberTree.get(elemId);
+      if (obj) {
+        const elem = this.addNode(this.rootDict.xref.fetchIfRef(obj), map);
+        if (
+          elem?.kids?.length === 1 &&
+          elem.kids[0].type === StructElementType.OBJECT
+        ) {
+          // The node in the struct tree is wrapping an object (annotation
+          // or xobject), so we need to update the type of the node to match
+          // the type of the object.
+          elem.kids[0].type = type;
+        }
       }
     }
   }
@@ -322,6 +361,11 @@ class StructTreePage {
             type: "object",
             id: kid.refObjId,
           });
+        } else if (kid.type === StructElementType.ANNOTATION) {
+          obj.children.push({
+            type: "annotation",
+            id: `${AnnotationPrefix}${kid.refObjId}`,
+          });
         }
       }
     }

src/display/annotation_layer.js

@@ -21,6 +21,7 @@
 import {
   AnnotationBorderStyleType,
   AnnotationEditorType,
+  AnnotationPrefix,
   AnnotationType,
   FeatureTest,
   LINE_FACTOR,
@@ -30,7 +31,6 @@ import {
   warn,
 } from "../shared/util.js";
 import {
-  AnnotationPrefix,
   DOMSVGFactory,
   getFilenameFromUrl,
   PDFDateString,

src/display/display_utils.js

@@ -31,8 +31,6 @@ import {
 
 const SVG_NS = "http://www.w3.org/2000/svg";
 
-const AnnotationPrefix = "pdfjs_internal_id_";
-
 class PixelsPerInch {
   static CSS = 96.0;
 
@@ -1005,7 +1003,6 @@ function setLayerDimensions(
 }
 
 export {
-  AnnotationPrefix,
   deprecated,
   DOMCanvasFactory,
   DOMCMapReaderFactory,

src/shared/util.js

@@ -1047,6 +1047,8 @@ function getUuid() {
   return bytesToString(buf);
 }
 
+const AnnotationPrefix = "pdfjs_internal_id_";
+
 export {
   AbortException,
   AnnotationActionEventType,
@@ -1057,6 +1059,7 @@ export {
   AnnotationFieldFlag,
   AnnotationFlag,
   AnnotationMode,
+  AnnotationPrefix,
   AnnotationReplyType,
   AnnotationType,
   assert,

test/integration/accessibility_spec.js

@@ -139,4 +139,35 @@ describe("accessibility", () => {
       );
     });
   });
+
+  describe("Stamp annotation accessibility", () => {
+    let pages;
+
+    beforeAll(async () => {
+      pages = await loadAndWait("tagged_stamp.pdf", ".annotationLayer");
+    });
+
+    afterAll(async () => {
+      await closePages(pages);
+    });
+
+    it("must check that the stamp annotation is linked to the struct tree", async () => {
+      await Promise.all(
+        pages.map(async ([browserName, page]) => {
+          await page.waitForSelector(".structTree");
+
+          const isLinkedToStampAnnotation = await page.$eval(
+            ".structTree [role='figure']",
+            el =>
+              document
+                .getElementById(el.getAttribute("aria-owns"))
+                .classList.contains("stampAnnotation")
+          );
+          expect(isLinkedToStampAnnotation)
+            .withContext(`In ${browserName}`)
+            .toEqual(true);
+        })
+      );
+    });
+  });
 });

test/pdfs/.gitignore

@@ -610,3 +610,4 @@
 !annotation_hidden_noview.pdf
 !widget_hidden_print.pdf
 !empty_protected.pdf
+!tagged_stamp.pdf

test/pdfs/tagged_stamp.pdf

@@ -100,14 +100,15 @@ class StructTreeLayerBuilder {
   }
 
   #setAttributes(structElement, htmlElement) {
-    if (structElement.alt !== undefined) {
-      htmlElement.setAttribute("aria-label", structElement.alt);
+    const { alt, id, lang } = structElement;
+    if (alt !== undefined) {
+      htmlElement.setAttribute("aria-label", alt);
     }
-    if (structElement.id !== undefined) {
-      htmlElement.setAttribute("aria-owns", structElement.id);
+    if (id !== undefined) {
+      htmlElement.setAttribute("aria-owns", id);
     }
-    if (structElement.lang !== undefined) {
-      htmlElement.setAttribute("lang", structElement.lang);
+    if (lang !== undefined) {
+      htmlElement.setAttribute("lang", lang);
     }
   }