Merge pull request #20623 from calixteman/bug2014080

In tagged pdfs, TH can be either a column header or a row header (bug 2014080)

Author: calixteman

test/integration/accessibility_spec.mjs

@@ -544,4 +544,71 @@ describe("accessibility", () => {
       );
     });
   });
+
+  describe("A TH in a TR itself in a TBody is rowheader", () => {
+    let pages;
+
+    beforeEach(async () => {
+      pages = await loadAndWait("bug2014080.pdf", ".textLayer");
+    });
+
+    afterEach(async () => {
+      await closePages(pages);
+    });
+
+    it("must check that the table has the right structure", async () => {
+      await Promise.all(
+        pages.map(async ([browserName, page]) => {
+          let elementRole = await page.evaluate(() =>
+            Array.from(
+              document.querySelector(".structTree [role='table']").children
+            ).map(child => child.getAttribute("role"))
+          );
+
+          // THeader and TBody must be rowgroup.
+          expect(elementRole)
+            .withContext(`In ${browserName}`)
+            .toEqual(["rowgroup", "rowgroup"]);
+
+          elementRole = await page.evaluate(() =>
+            Array.from(
+              document.querySelector(
+                ".structTree [role='table'] > [role='rowgroup'] > [role='row']"
+              ).children
+            ).map(child => child.getAttribute("role"))
+          );
+
+          // THeader has 3 columnheader.
+          expect(elementRole)
+            .withContext(`In ${browserName}`)
+            .toEqual(["columnheader", "columnheader", "columnheader"]);
+
+          elementRole = await page.evaluate(() =>
+            Array.from(
+              document.querySelector(
+                ".structTree [role='table'] > [role='rowgroup']:nth-child(2)"
+              ).children
+            ).map(child => child.getAttribute("role"))
+          );
+
+          // TBody has 5 rows.
+          expect(elementRole)
+            .withContext(`In ${browserName}`)
+            .toEqual(["row", "row", "row", "row", "row"]);
+
+          elementRole = await page.evaluate(() =>
+            Array.from(
+              document.querySelector(
+                ".structTree [role='table'] > [role='rowgroup']:nth-child(2) > [role='row']:first-child"
+              ).children
+            ).map(child => child.getAttribute("role"))
+          );
+          // First row has a rowheader and 2 cells.
+          expect(elementRole)
+            .withContext(`In ${browserName}`)
+            .toEqual(["rowheader", "cell", "cell"]);
+        })
+      );
+    });
+  });
 });

test/pdfs/.gitignore

@@ -872,3 +872,4 @@
 !page_with_number_and_link.pdf
 !Brotli-Prototype-FileA.pdf
 !bug2013793.pdf
+!bug2014080.pdf

test/pdfs/bug2014080.pdf

@@ -61,8 +61,8 @@ const PDF_ROLE_TO_HTML_ROLE = {
   TR: "row",
   TH: "columnheader",
   TD: "cell",
-  THead: "columnheader",
-  TBody: null,
+  THead: "rowgroup",
+  TBody: "rowgroup",
   TFoot: null,
   // Standard structure type Caption
   Caption: null,
@@ -353,7 +353,7 @@ class StructTreeLayerBuilder {
     }
   }
 
-  #walk(node) {
+  #walk(node, parentNodes = []) {
     if (!node) {
       return null;
     }
@@ -378,7 +378,14 @@ class StructTreeLayerBuilder {
         element.setAttribute("role", "heading");
         element.setAttribute("aria-level", match[1]);
       } else if (PDF_ROLE_TO_HTML_ROLE[role]) {
-        element.setAttribute("role", PDF_ROLE_TO_HTML_ROLE[role]);
+        element.setAttribute(
+          "role",
+          role === "TH" &&
+            parentNodes.at(-1)?.role === "TR" &&
+            parentNodes.at(-2)?.role === "TBody"
+            ? "rowheader" // TH inside TR itself in TBody is a rowheader.
+            : PDF_ROLE_TO_HTML_ROLE[role]
+        );
       }
       if (role === "Figure" && this.#addImageInTextLayer(node, element)) {
         return element;
@@ -423,9 +430,11 @@ class StructTreeLayerBuilder {
         // parent node to avoid creating an extra span.
         this.#setAttributes(node.children[0], element);
       } else {
+        parentNodes.push(node);
         for (const kid of node.children) {
-          element.append(this.#walk(kid));
+          element.append(this.#walk(kid, parentNodes));
         }
+        parentNodes.pop();
       }
     }
     return element;