Merge pull request #17986 from calixteman/fix_struct_tree

Allow to insert several annotations under the same parent in the structure tree

src/core/primitives.js

@@ -378,6 +378,12 @@ class RefSetCache {
   clear() {
     this._map.clear();
   }
+
+  *items() {
+    for (const [ref, value] of this._map) {
+      yield [Ref.fromString(ref), value];
+    }
+  }
 }
 
 function isName(v, name) {

src/core/struct_tree.js

@@ -119,19 +119,19 @@ class StructTreeRoot {
     newRefs,
   }) {
     const root = pdfManager.catalog.cloneDict();
+    const cache = new RefSetCache();
+    cache.put(catalogRef, root);
+
     const structTreeRootRef = xref.getNewTemporaryRef();
     root.set("StructTreeRoot", structTreeRootRef);
 
-    const buffer = [];
-    await writeObject(catalogRef, root, buffer, xref);
-    newRefs.push({ ref: catalogRef, data: buffer.join("") });
-
     const structTreeRoot = new Dict(xref);
     structTreeRoot.set("Type", Name.get("StructTreeRoot"));
     const parentTreeRef = xref.getNewTemporaryRef();
     structTreeRoot.set("ParentTree", parentTreeRef);
     const kids = [];
     structTreeRoot.set("K", kids);
+    cache.put(structTreeRootRef, structTreeRoot);
 
     const parentTree = new Dict(xref);
     const nums = [];
@@ -144,18 +144,18 @@ class StructTreeRoot {
       nums,
       xref,
       pdfManager,
-      newRefs,
-      buffer,
+      cache,
     });
     structTreeRoot.set("ParentTreeNextKey", nextKey);
 
-    buffer.length = 0;
-    await writeObject(parentTreeRef, parentTree, buffer, xref);
-    newRefs.push({ ref: parentTreeRef, data: buffer.join("") });
+    cache.put(parentTreeRef, parentTree);
 
-    buffer.length = 0;
-    await writeObject(structTreeRootRef, structTreeRoot, buffer, xref);
-    newRefs.push({ ref: structTreeRootRef, data: buffer.join("") });
+    const buffer = [];
+    for (const [ref, obj] of cache.items()) {
+      buffer.length = 0;
+      await writeObject(ref, obj, buffer, xref);
+      newRefs.push({ ref, data: buffer.join("") });
+    }
   }
 
   async canUpdateStructTree({ pdfManager, xref, newAnnotationsByPage }) {
@@ -232,6 +232,8 @@ class StructTreeRoot {
     const xref = this.dict.xref;
     const structTreeRoot = this.dict.clone();
     const structTreeRootRef = this.ref;
+    const cache = new RefSetCache();
+    cache.put(structTreeRootRef, structTreeRoot);
 
     let parentTreeRef = structTreeRoot.getRaw("ParentTree");
     let parentTree;
@@ -243,6 +245,7 @@ class StructTreeRoot {
       structTreeRoot.set("ParentTree", parentTreeRef);
     }
     parentTree = parentTree.clone();
+    cache.put(parentTreeRef, parentTree);
 
     let nums = parentTree.getRaw("Nums");
     let numsRef = null;
@@ -255,47 +258,27 @@ class StructTreeRoot {
       parentTree.set("Nums", nums);
     }
 
-    let kids = structTreeRoot.getRaw("K");
-    let kidsRef = null;
-    if (kids instanceof Ref) {
-      kidsRef = kids;
-      kids = xref.fetch(kidsRef);
-    } else {
-      kidsRef = xref.getNewTemporaryRef();
-      structTreeRoot.set("K", kidsRef);
-    }
-    kids = Array.isArray(kids) ? kids.slice() : [kids];
-
-    const buffer = [];
     const newNextkey = await StructTreeRoot.#writeKids({
       newAnnotationsByPage,
       structTreeRootRef,
-      kids,
+      kids: null,
       nums,
       xref,
       pdfManager,
-      newRefs,
-      buffer,
+      cache,
     });
     structTreeRoot.set("ParentTreeNextKey", newNextkey);
 
-    buffer.length = 0;
-    await writeObject(kidsRef, kids, buffer, xref);
-    newRefs.push({ ref: kidsRef, data: buffer.join("") });
-
     if (numsRef) {
-      buffer.length = 0;
-      await writeObject(numsRef, nums, buffer, xref);
-      newRefs.push({ ref: numsRef, data: buffer.join("") });
+      cache.put(numsRef, nums);
     }
 
-    buffer.length = 0;
-    await writeObject(parentTreeRef, parentTree, buffer, xref);
-    newRefs.push({ ref: parentTreeRef, data: buffer.join("") });
-
-    buffer.length = 0;
-    await writeObject(structTreeRootRef, structTreeRoot, buffer, xref);
-    newRefs.push({ ref: structTreeRootRef, data: buffer.join("") });
+    const buffer = [];
+    for (const [ref, obj] of cache.items()) {
+      buffer.length = 0;
+      await writeObject(ref, obj, buffer, xref);
+      newRefs.push({ ref, data: buffer.join("") });
+    }
   }
 
   static async #writeKids({
@@ -305,8 +288,7 @@ class StructTreeRoot {
     nums,
     xref,
     pdfManager,
-    newRefs,
-    buffer,
+    cache,
   }) {
     const objr = Name.get("OBJR");
     let nextKey = -Infinity;
@@ -349,19 +331,15 @@ class StructTreeRoot {
           tagDict.set("ActualText", actualText);
         }
 
-        if (structTreeParent) {
-          await this.#updateParentTag({
-            structTreeParent,
-            tagDict,
-            newTagRef: tagRef,
-            fallbackRef: structTreeRootRef,
-            xref,
-            newRefs,
-            buffer,
-          });
-        } else {
-          tagDict.set("P", structTreeRootRef);
-        }
+        await this.#updateParentTag({
+          structTreeParent,
+          tagDict,
+          newTagRef: tagRef,
+          structTreeRootRef,
+          fallbackKids: kids,
+          xref,
+          cache,
+        });
 
         const objDict = new Dict(xref);
         tagDict.set("K", objDict);
@@ -372,23 +350,24 @@ class StructTreeRoot {
         }
         objDict.set("Obj", ref);
 
-        buffer.length = 0;
-        await writeObject(tagRef, tagDict, buffer, xref);
-        newRefs.push({ ref: tagRef, data: buffer.join("") });
-
+        cache.put(tagRef, tagDict);
         nums.push(parentTreeId, tagRef);
-        kids.push(tagRef);
       }
     }
     return nextKey + 1;
   }
 
   static #collectParents({ elements, xref, pageDict, numberTree }) {
-    const idToElement = new Map();
+    const idToElements = new Map();
     for (const element of elements) {
       if (element.structTreeParentId) {
         const id = parseInt(element.structTreeParentId.split("_mc")[1], 10);
-        idToElement.set(id, element);
+        let elems = idToElements.get(id);
+        if (!elems) {
+          elems = [];
+          idToElements.set(id, elems);
+        }
+        elems.push(element);
       }
     }
 
@@ -400,13 +379,16 @@ class StructTreeRoot {
     const parentArray = numberTree.get(id);
 
     const updateElement = (kid, pageKid, kidRef) => {
-      const element = idToElement.get(kid);
-      if (element) {
+      const elems = idToElements.get(kid);
+      if (elems) {
         const parentRef = pageKid.getRaw("P");
         const parentDict = xref.fetchIfRef(parentRef);
         if (parentRef instanceof Ref && parentDict instanceof Dict) {
           // It should always the case, but we check just in case.
-          element.structTreeParent = { ref: kidRef, dict: pageKid };
+          const params = { ref: kidRef, dict: pageKid };
+          for (const element of elems) {
+            element.structTreeParent = params;
+          }
         }
         return true;
       }
@@ -431,67 +413,73 @@ class StructTreeRoot {
         if (Number.isInteger(kid) && updateElement(kid, pageKid, kidRef)) {
           break;
         }
+        if (!(kid instanceof Dict)) {
+          continue;
+        }
+        if (!isName(kid.get("Type"), "MCR")) {
+          break;
+        }
+        const mcid = kid.get("MCID");
+        if (Number.isInteger(mcid) && updateElement(mcid, pageKid, kidRef)) {
+          break;
+        }
       }
     }
   }
 
   static async #updateParentTag({
-    structTreeParent: { ref, dict },
+    structTreeParent,
     tagDict,
     newTagRef,
-    fallbackRef,
+    structTreeRootRef,
+    fallbackKids,
     xref,
-    newRefs,
-    buffer,
+    cache,
   }) {
-    // We get the parent of the tag.
-    const parentRef = dict.getRaw("P");
-    let parentDict = xref.fetchIfRef(parentRef);
+    let ref = null;
+    let parentRef;
+    if (structTreeParent) {
+      ({ ref } = structTreeParent);
+
+      // We get the parent of the tag.
+      parentRef = structTreeParent.dict.getRaw("P") || structTreeRootRef;
+    } else {
+      parentRef = structTreeRootRef;
+    }
 
     tagDict.set("P", parentRef);
 
     // We get the kids in order to insert a new tag at the right position.
-    let saveParentDict = false;
-    let parentKids;
-    let parentKidsRef = parentDict.getRaw("K");
-    if (!(parentKidsRef instanceof Ref)) {
-      parentKids = parentKidsRef;
-      parentKidsRef = xref.getNewTemporaryRef();
-      parentDict = parentDict.clone();
-      parentDict.set("K", parentKidsRef);
-      saveParentDict = true;
-    } else {
-      parentKids = xref.fetch(parentKidsRef);
-    }
-
-    if (Array.isArray(parentKids)) {
-      const index = parentKids.indexOf(ref);
-      if (index >= 0) {
-        parentKids = parentKids.slice();
-        parentKids.splice(index + 1, 0, newTagRef);
-      } else {
-        warn("Cannot update the struct tree: parent kid not found.");
-        tagDict.set("P", fallbackRef);
-        return;
-      }
-    } else if (parentKids instanceof Dict) {
-      parentKids = [parentKidsRef, newTagRef];
-      parentKidsRef = xref.getNewTemporaryRef();
-      parentDict.set("K", parentKidsRef);
-      saveParentDict = true;
-    }
-
-    buffer.length = 0;
-    await writeObject(parentKidsRef, parentKids, buffer, xref);
-    newRefs.push({ ref: parentKidsRef, data: buffer.join("") });
-
-    if (!saveParentDict) {
+    const parentDict = xref.fetchIfRef(parentRef);
+    if (!parentDict) {
+      fallbackKids.push(newTagRef);
       return;
     }
 
-    buffer.length = 0;
-    await writeObject(parentRef, parentDict, buffer, xref);
-    newRefs.push({ ref: parentRef, data: buffer.join("") });
+    let cachedParentDict = cache.get(parentRef);
+    if (!cachedParentDict) {
+      cachedParentDict = parentDict.clone();
+      cache.put(parentRef, cachedParentDict);
+    }
+    const parentKidsRaw = cachedParentDict.getRaw("K");
+    let cachedParentKids =
+      parentKidsRaw instanceof Ref ? cache.get(parentKidsRaw) : null;
+    if (!cachedParentKids) {
+      cachedParentKids = xref.fetchIfRef(parentKidsRaw);
+      cachedParentKids = Array.isArray(cachedParentKids)
+        ? cachedParentKids.slice()
+        : [parentKidsRaw];
+      const parentKidsRef = xref.getNewTemporaryRef();
+      cachedParentDict.set("K", parentKidsRef);
+      cache.put(parentKidsRef, cachedParentKids);
+    }
+
+    const index = cachedParentKids.indexOf(ref);
+    cachedParentKids.splice(
+      index >= 0 ? index + 1 : cachedParentKids.length,
+      0,
+      newTagRef
+    );
   }
 }
 

test/pdfs/.gitignore

@@ -645,3 +645,4 @@
 !issue12213.pdf
 !tracemonkey_freetext.pdf
 !issue17998.pdf
+!pdfjs_wikipedia.pdf

test/unit/api_spec.js

@@ -1030,6 +1030,20 @@ describe("api", function () {
       await pdfLoadingTask.destroy();
     });
 
+    function findNode(parent, node, index, check) {
+      if (check(node)) {
+        return [parent.children[index - 1], node];
+      }
+      for (let i = 0; i < node.children?.length ?? 0; i++) {
+        const child = node.children[i];
+        const elements = findNode(node, child, i, check);
+        if (elements) {
+          return elements;
+        }
+      }
+      return null;
+    }
+
     it("gets number of pages", function () {
       expect(pdfDocument.numPages).toEqual(3);
     });
@@ -2396,7 +2410,22 @@ describe("api", function () {
       pdfDoc = await loadingTask.promise;
       const page = await pdfDoc.getPage(1);
       const tree = await page.getStructTree();
-      const leaf = tree.children[0].children[6].children[1];
+      const [predecessor, leaf] = findNode(
+        null,
+        tree,
+        0,
+        node => node.role === "Figure"
+      );
+
+      expect(predecessor).toEqual({
+        role: "Span",
+        children: [
+          {
+            type: "content",
+            id: "p3R_mc12",
+          },
+        ],
+      });
 
       expect(leaf).toEqual({
         role: "Figure",
@@ -2412,6 +2441,104 @@ describe("api", function () {
       await loadingTask.destroy();
     });
 
+    it("write a new stamp annotation in a tagged pdf (with some MCIDs), save and check the structure tree", async function () {
+      if (isNodeJS) {
+        pending("Cannot create a bitmap from Node.js.");
+      }
+
+      const TEST_IMAGES_PATH = "../images/";
+      const filename = "firefox_logo.png";
+      const path = new URL(TEST_IMAGES_PATH + filename, window.location).href;
+
+      const response = await fetch(path);
+      const blob = await response.blob();
+      const bitmap = await createImageBitmap(blob);
+
+      let loadingTask = getDocument(
+        buildGetDocumentParams("pdfjs_wikipedia.pdf")
+      );
+      let pdfDoc = await loadingTask.promise;
+      for (let i = 0; i < 2; i++) {
+        pdfDoc.annotationStorage.setValue(`pdfjs_internal_editor_${i}`, {
+          annotationType: AnnotationEditorType.STAMP,
+          bitmapId: `im${i}`,
+          pageIndex: 0,
+          rect: [257 + i, 572 + i, 286 + i, 603 + i],
+          rotation: 0,
+          isSvg: false,
+          structTreeParentId: "p2R_mc155",
+          accessibilityData: {
+            type: "Figure",
+            alt: `Firefox logo ${i}`,
+          },
+          bitmap: structuredClone(bitmap),
+        });
+      }
+
+      const data = await pdfDoc.saveDocument();
+      await loadingTask.destroy();
+
+      loadingTask = getDocument(data);
+      pdfDoc = await loadingTask.promise;
+      const page = await pdfDoc.getPage(1);
+      const tree = await page.getStructTree();
+
+      let [predecessor, figure] = findNode(
+        null,
+        tree,
+        0,
+        node => node.role === "Figure" && node.alt === "Firefox logo 1"
+      );
+      expect(predecessor).toEqual({
+        role: "NonStruct",
+        children: [
+          {
+            type: "content",
+            id: "p2R_mc155",
+          },
+        ],
+      });
+      expect(figure).toEqual({
+        role: "Figure",
+        children: [
+          {
+            type: "annotation",
+            id: "pdfjs_internal_id_420R",
+          },
+        ],
+        alt: "Firefox logo 1",
+      });
+
+      [predecessor, figure] = findNode(
+        null,
+        tree,
+        0,
+        node => node.role === "Figure" && node.alt === "Firefox logo 0"
+      );
+      expect(predecessor).toEqual({
+        role: "Figure",
+        children: [
+          {
+            type: "annotation",
+            id: "pdfjs_internal_id_420R",
+          },
+        ],
+        alt: "Firefox logo 1",
+      });
+      expect(figure).toEqual({
+        role: "Figure",
+        children: [
+          {
+            type: "annotation",
+            id: "pdfjs_internal_id_416R",
+          },
+        ],
+        alt: "Firefox logo 0",
+      });
+
+      await loadingTask.destroy();
+    });
+
     it("write a new stamp annotation in a tagged pdf, save, repeat and check the structure tree", async function () {
       if (isNodeJS) {
         pending("Cannot create a bitmap from Node.js.");

test/unit/primitives_spec.js

@@ -498,6 +498,15 @@ describe("primitives", function () {
       cache.put(ref2, obj2);
       expect([...cache]).toEqual([obj1, obj2]);
     });
+
+    it("should support iteration over key-value pairs", function () {
+      cache.put(ref1, obj1);
+      cache.put(ref2, obj2);
+      expect([...cache.items()]).toEqual([
+        [ref1, obj1],
+        [ref2, obj2],
+      ]);
+    });
   });
 
   describe("isName", function () {