Merge pull request #20826 from calixteman/outline_se

Handle outline with Structure Element (SE) destination

Author: timvandermeij

src/core/catalog.js

@@ -1542,6 +1542,105 @@ class Catalog {
    *   exist in most PDF documents).
    */
 
+  /**
+   * Derive a destination array from a Structure Element reference.
+   * Walks the SE dict to find its page (Pg) and optional bounding box (A.BBox),
+   * then returns an XYZ destination array that can be used for navigation.
+   * @param {XRef} xref
+   * @param {Ref} seRef
+   * @returns {Array|null}
+   */
+  static #getDestFromStructElement(xref, seRef) {
+    const seDict = xref.fetchIfRef(seRef);
+    if (!(seDict instanceof Dict)) {
+      return null;
+    }
+
+    // Try to find the page reference for this structure element.
+    // Search order: the element itself, its descendants down to leaf nodes,
+    // then ancestor elements via the P entry (up).
+    let pageRef = null;
+
+    // Check the element directly.
+    const directPg = seDict.getRaw("Pg");
+    if (directPg instanceof Ref) {
+      pageRef = directPg;
+    }
+
+    // Walk down into descendants (BFS) until a Pg is found or leaves are
+    // reached (e.g. integer MCIDs or MCR/OBJR dicts without further K).
+    if (!pageRef) {
+      const queue = [seDict];
+      while (queue.length > 0 && !pageRef) {
+        const node = queue.shift();
+        const kids = node.get("K");
+        let kidsArr;
+        if (Array.isArray(kids)) {
+          kidsArr = kids;
+        } else if (kids) {
+          kidsArr = [kids];
+        } else {
+          kidsArr = [];
+        }
+        for (const kid of kidsArr) {
+          const kidObj = xref.fetchIfRef(kid);
+          if (!(kidObj instanceof Dict)) {
+            continue; // integer MCID – leaf node, no Pg here
+          }
+          const pg = kidObj.getRaw("Pg");
+          if (pg instanceof Ref) {
+            pageRef = pg;
+            break;
+          }
+          queue.push(kidObj);
+        }
+      }
+    }
+
+    // Walk up the parent chain if still not found.
+    if (!pageRef) {
+      const MAX_DEPTH = 40;
+      let current = seDict;
+      for (let depth = 0; depth < MAX_DEPTH; depth++) {
+        const parentRaw = current.getRaw("P");
+        if (!(parentRaw instanceof Ref)) {
+          break;
+        }
+        const parentDict = xref.fetchIfRef(parentRaw);
+        if (!(parentDict instanceof Dict)) {
+          break;
+        }
+        if (isName(parentDict.get("Type"), "StructTreeRoot")) {
+          break;
+        }
+        const pg = parentDict.getRaw("Pg");
+        if (pg instanceof Ref) {
+          pageRef = pg;
+          break;
+        }
+        current = parentDict;
+      }
+    }
+
+    if (!pageRef) {
+      return null;
+    }
+
+    // Try to obtain precise coordinates from the element's attribute BBox.
+    let x = null,
+      y = null;
+    const attrs = seDict.get("A");
+    if (attrs instanceof Dict) {
+      const bboxArr = attrs.getArray("BBox");
+      if (isNumberArray(bboxArr, 4)) {
+        x = bboxArr[0];
+        y = bboxArr[3]; // top of the bbox in PDF page coordinates
+      }
+    }
+
+    return [pageRef, { name: "XYZ" }, x, y, null];
+  }
+
   /**
    * Helper function used to parse the contents of destination dictionaries.
    * @param {ParseDestDictionaryParameters} params
@@ -1773,6 +1872,35 @@ class Catalog {
         resultObj.dest = dest;
       }
     }
+
+    // Handle SE (Structure Element) entry: when no other destination has been
+    // found, derive one from the structure element's page and optional bbox.
+    if (
+      !resultObj.dest &&
+      !resultObj.url &&
+      !resultObj.action &&
+      !resultObj.attachment &&
+      !resultObj.setOCGState &&
+      !resultObj.resetForm
+    ) {
+      const seRef = destDict.getRaw("SE");
+      if (seRef instanceof Ref) {
+        try {
+          const seDest = Catalog.#getDestFromStructElement(
+            destDict.xref,
+            seRef
+          );
+          if (seDest) {
+            resultObj.dest = seDest;
+          }
+        } catch (ex) {
+          if (ex instanceof MissingDataException) {
+            throw ex;
+          }
+          info("SE parsing failed.");
+        }
+      }
+    }
   }
 }
 

test/integration/viewer_spec.mjs

@@ -1646,4 +1646,45 @@ describe("PDF viewer", () => {
       );
     });
   });
+
+  describe("Outline with SE (Structure Element) entries", () => {
+    let pages;
+
+    beforeEach(async () => {
+      pages = await loadAndWait(
+        "outlines_se.pdf",
+        `.page[data-page-number="1"] .endOfContent`
+      );
+    });
+
+    afterEach(async () => {
+      await closePages(pages);
+    });
+
+    it("should navigate to the correct page when clicking an outline item with an SE entry", async () => {
+      await Promise.all(
+        pages.map(async ([browserName, page]) => {
+          // Open the sidebar.
+          await showViewsManager(page);
+
+          // Switch to the outline view.
+          await page.click("#viewsManagerSelectorButton");
+          await page.waitForSelector("#outlinesViewMenu", { visible: true });
+          await page.click("#outlinesViewMenu");
+
+          for (let i = 2; i >= 1; i--) {
+            await waitAndClick(
+              page,
+              `#outlinesView .treeItem .treeItem:nth-child(${i}) a`
+            );
+            await page.waitForFunction(
+              pageNum => window.PDFViewerApplication.page === pageNum,
+              {},
+              i
+            );
+          }
+        })
+      );
+    });
+  });
 });

test/pdfs/.gitignore

@@ -879,3 +879,4 @@
 !sci-notation.pdf
 !nested_outline.pdf
 !form_two_pages.pdf
+!outlines_se.pdf