Re-land: Support adding attributes to PDF document structure nodes.

Originally landed: https://skia-review.googlesource.com/c/skia/+/268878
Reverted: https://skia-review.googlesource.com/c/skia/+/271858

The issue was with compilation when PDF support is disabled. See
the diff between patchsets 1 and 2.

This is an important part of writing a tagged PDF. Many of the nodes
in the document structure tree need additional attributes, just like
in HTML.

This change aims to add support for a few useful attributes, not to
be comprehensive.

Bug: chromium:1039816
Change-Id: I15f8b6c41d4fdaa4b6e21775ab6d26ec57eb0f5d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/271916
Commit-Queue: Dominic Mazzoni <dmazzoni@chromium.org>
Reviewed-by: Mike Reed <reed@google.com>

Author: minorninth

RELEASE_NOTES.txt

@@ -23,6 +23,9 @@ Milestone 82
     related calls is treated as a request to do no color correction at decode
     time.
 
+  * Add new APIs to add attributes to document structure node when
+    creating a tagged PDF.
+
   * Remove CGFontRef parameter from SkCreateTypefaceFromCTFont.
     Use CTFontManagerCreateFontDescriptorFromData instead of
     CGFontCreateWithDataProvider to create CTFonts to avoid memory use issues.

gn/tests.gni

@@ -169,6 +169,7 @@ tests_sources = [
   "$_tests/PDFMetadataAttributeTest.cpp",
   "$_tests/PDFOpaqueSrcModeToSrcOverTest.cpp",
   "$_tests/PDFPrimitivesTest.cpp",
+  "$_tests/PDFTaggedTableTest.cpp",
   "$_tests/PDFTaggedTest.cpp",
   "$_tests/PackBitsTest.cpp",
   "$_tests/PackedConfigsTextureTest.cpp",

include/docs/SkPDFDocument.h

@@ -5,15 +5,21 @@
 
 #include "include/core/SkDocument.h"
 
+#include <vector>
+
+#include "include/core/SkColor.h"
 #include "include/core/SkMilestone.h"
 #include "include/core/SkScalar.h"
 #include "include/core/SkString.h"
 #include "include/core/SkTime.h"
+#include "include/private/SkNoncopyable.h"
 
 #define SKPDF_STRING(X) SKPDF_STRING_IMPL(X)
 #define SKPDF_STRING_IMPL(X) #X
 
 class SkExecutor;
+class SkPDFArray;
+class SkPDFTagTree;
 
 namespace SkPDF {
 
@@ -71,16 +77,42 @@ enum class DocumentStructureType {
     kForm,        //!< Form control (not like an HTML FORM element)
 };
 
+/** Attributes for nodes in the PDF tree. */
+class SK_API AttributeList : SkNoncopyable {
+public:
+    AttributeList();
+    ~AttributeList();
+
+    // Each attribute must have an owner (e.g. "Layout", "List", "Table", etc)
+    // and an attribute name (e.g. "BBox", "RowSpan", etc.) from PDF32000_2008 14.8.5,
+    // and then a value of the proper type according to the spec.
+    void appendInt(const char* owner, const char* name, int value);
+    void appendFloat(const char* owner, const char* name, float value);
+    void appendString(const char* owner, const char* name, const char* value);
+    void appendFloatArray(const char* owner,
+                          const char* name,
+                          const std::vector<float>& value);
+    void appendStringArray(const char* owner,
+                           const char* name,
+                           const std::vector<SkString>& value);
+
+private:
+    friend class ::SkPDFTagTree;
+
+    std::unique_ptr<SkPDFArray> fAttrs;
+};
+
 /** A node in a PDF structure tree, giving a semantic representation
     of the content.  Each node ID is associated with content
     by passing the SkCanvas and node ID to SkPDF::SetNodeId() when drawing.
     NodeIDs should be unique within each tree.
 */
 struct StructureElementNode {
-    const StructureElementNode* fChildren = nullptr;
-    size_t fChildCount;
-    int fNodeId;
-    DocumentStructureType fType;
+    StructureElementNode* fChildren = nullptr;
+    size_t fChildCount = 0;
+    int fNodeId = 0;
+    DocumentStructureType fType = DocumentStructureType::kNonStruct;
+    AttributeList fAttributes;
 };
 
 /** Optional metadata to be passed into the PDF factory function.
@@ -149,7 +181,7 @@ struct Metadata {
         a semantic representation of the content. The caller
         should retain ownership.
     */
-    const StructureElementNode* fStructureElementTreeRoot = nullptr;
+    StructureElementNode* fStructureElementTreeRoot = nullptr;
 
     /** Executor to handle threaded work within PDF Backend. If this is nullptr,
         then all work will be done serially on the main thread. To have worker

src/pdf/SkDocument_PDF_None.cpp

@@ -8,8 +8,14 @@
 #include "include/core/SkCanvas.h"
 #include "include/docs/SkPDFDocument.h"
 
+class SkPDFArray {};
+
 sk_sp<SkDocument> SkPDF::MakeDocument(SkWStream*, const SkPDF::Metadata&) { return nullptr; }
 
 void SkPDF::SetNodeId(SkCanvas* c, int n) {
     c->drawAnnotation({0, 0, 0, 0}, "PDF_Node_Key", SkData::MakeWithCopy(&n, sizeof(n)).get());
 }
+
+SkPDF::AttributeList::AttributeList() = default;
+
+SkPDF::AttributeList::~AttributeList() = default;

src/pdf/SkPDFTag.cpp

@@ -66,6 +66,70 @@ static const char* tag_name_from_type(SkPDF::DocumentStructureType type) {
     SK_ABORT("bad tag");
 }
 
+SkPDF::AttributeList::AttributeList() = default;
+
+SkPDF::AttributeList::~AttributeList() = default;
+
+void SkPDF::AttributeList::appendInt(
+        const char* owner, const char* name, int value) {
+    if (!fAttrs)
+        fAttrs = SkPDFMakeArray();
+    std::unique_ptr<SkPDFDict> attrDict = SkPDFMakeDict();
+    attrDict->insertName("O", owner);
+    attrDict->insertInt(name, value);
+    fAttrs->appendObject(std::move(attrDict));
+}
+
+void SkPDF::AttributeList::appendFloat(
+        const char* owner, const char* name, float value) {
+    if (!fAttrs)
+        fAttrs = SkPDFMakeArray();
+    std::unique_ptr<SkPDFDict> attrDict = SkPDFMakeDict();
+    attrDict->insertName("O", owner);
+    attrDict->insertScalar(name, value);
+    fAttrs->appendObject(std::move(attrDict));
+}
+
+void SkPDF::AttributeList::appendString(
+        const char* owner, const char* name, const char* value) {
+    if (!fAttrs)
+        fAttrs = SkPDFMakeArray();
+    std::unique_ptr<SkPDFDict> attrDict = SkPDFMakeDict();
+    attrDict->insertName("O", owner);
+    attrDict->insertName(name, value);
+    fAttrs->appendObject(std::move(attrDict));
+}
+
+void SkPDF::AttributeList::appendFloatArray(
+        const char* owner, const char* name, const std::vector<float>& value) {
+    if (!fAttrs)
+        fAttrs = SkPDFMakeArray();
+    std::unique_ptr<SkPDFDict> attrDict = SkPDFMakeDict();
+    attrDict->insertName("O", owner);
+    std::unique_ptr<SkPDFArray> pdfArray = SkPDFMakeArray();
+    for (float element : value) {
+        pdfArray->appendScalar(element);
+    }
+    attrDict->insertObject(name, std::move(pdfArray));
+    fAttrs->appendObject(std::move(attrDict));
+}
+
+void SkPDF::AttributeList::appendStringArray(
+        const char* owner,
+        const char* name,
+        const std::vector<SkString>& value) {
+    if (!fAttrs)
+        fAttrs = SkPDFMakeArray();
+    std::unique_ptr<SkPDFDict> attrDict = SkPDFMakeDict();
+    attrDict->insertName("O", owner);
+    std::unique_ptr<SkPDFArray> pdfArray = SkPDFMakeArray();
+    for (SkString element : value) {
+        pdfArray->appendName(element);
+    }
+    attrDict->insertObject(name, std::move(pdfArray));
+    fAttrs->appendObject(std::move(attrDict));
+}
+
 struct SkPDFTagNode {
     SkPDFTagNode* fChildren = nullptr;
     size_t fChildCount = 0;
@@ -82,16 +146,18 @@ struct SkPDFTagNode {
         kYes,
         kNo,
     } fCanDiscard = kUnknown;
+    std::unique_ptr<SkPDFArray> fAttributes;
 };
 
 SkPDFTagTree::SkPDFTagTree() : fArena(4 * sizeof(SkPDFTagNode)) {}
 
 SkPDFTagTree::~SkPDFTagTree() = default;
 
-static void copy(const SkPDF::StructureElementNode& node,
-                 SkPDFTagNode* dst,
-                 SkArenaAlloc* arena,
-                 SkTHashMap<int, SkPDFTagNode*>* nodeMap) {
+// static
+void SkPDFTagTree::Copy(SkPDF::StructureElementNode& node,
+                        SkPDFTagNode* dst,
+                        SkArenaAlloc* arena,
+                        SkTHashMap<int, SkPDFTagNode*>* nodeMap) {
     nodeMap->set(node.fNodeId, dst);
     size_t childCount = node.fChildCount;
     SkPDFTagNode* children = arena->makeArray<SkPDFTagNode>(childCount);
@@ -100,14 +166,15 @@ static void copy(const SkPDF::StructureElementNode& node,
     dst->fType = node.fType;
     dst->fChildren = children;
     for (size_t i = 0; i < childCount; ++i) {
-        copy(node.fChildren[i], &children[i], arena, nodeMap);
+        Copy(node.fChildren[i], &children[i], arena, nodeMap);
     }
+    dst->fAttributes = std::move(node.fAttributes.fAttrs);
 }
 
-void SkPDFTagTree::init(const SkPDF::StructureElementNode* node) {
+void SkPDFTagTree::init(SkPDF::StructureElementNode* node) {
     if (node) {
         fRoot = fArena.make<SkPDFTagNode>();
-        copy(*node, fRoot, &fArena, &fNodeMap);
+        Copy(*node, fRoot, &fArena, &fNodeMap);
     }
 }
 
@@ -184,6 +251,13 @@ SkPDFIndirectReference prepare_tag_tree_to_emit(SkPDFIndirectReference parent,
     dict.insertName("S", tag_name_from_type(node->fType));
     dict.insertRef("P", parent);
     dict.insertObject("K", std::move(kids));
+    SkString idString;
+    idString.printf("%d", node->fNodeId);
+    dict.insertName("ID", idString.c_str());
+    if (node->fAttributes) {
+        dict.insertObject("A", std::move(node->fAttributes));
+    }
+
     return doc->emit(dict, ref);
 }
 
@@ -223,4 +297,3 @@ SkPDFIndirectReference SkPDFTagTree::makeStructTreeRoot(SkPDFDocument* doc) {
     structTreeRoot.insertRef("ParentTree", doc->emit(parentTree));
     return doc->emit(structTreeRoot, ref);
 }
-

src/pdf/SkPDFTag.h

@@ -21,12 +21,17 @@ class SkPDFTagTree {
 public:
     SkPDFTagTree();
     ~SkPDFTagTree();
-    void init(const SkPDF::StructureElementNode*);
+    void init(SkPDF::StructureElementNode*);
     void reset();
     int getMarkIdForNodeId(int nodeId, unsigned pageIndex);
     SkPDFIndirectReference makeStructTreeRoot(SkPDFDocument* doc);
 
 private:
+    static void Copy(SkPDF::StructureElementNode& node,
+                     SkPDFTagNode* dst,
+                     SkArenaAlloc* arena,
+                     SkTHashMap<int, SkPDFTagNode*>* nodeMap);
+
     SkArenaAlloc fArena;
     SkTHashMap<int, SkPDFTagNode*> fNodeMap;
     SkPDFTagNode* fRoot = nullptr;

tests/PDFTaggedTableTest.cpp

@@ -0,0 +1,135 @@
+/*
+ * Copyright 2020 Google Inc.
+ *
+ * Use of this source code is governed by a BSD-style license that can be
+ * found in the LICENSE file.
+ */
+#include "tests/Test.h"
+
+#include "include/core/SkCanvas.h"
+#include "include/core/SkFont.h"
+#include "include/core/SkStream.h"
+#include "include/docs/SkPDFDocument.h"
+
+using PDFTag = SkPDF::StructureElementNode;
+
+// Test building a tagged PDF containing a table.
+// Add this to args.gn to output the PDF to a file:
+//   extra_cflags = [ "-DSK_PDF_TEST_TAGS_OUTPUT_PATH=\"/tmp/table.pdf\"" ]
+DEF_TEST(SkPDF_tagged_table, r) {
+    REQUIRE_PDF_DOCUMENT(SkPDF_tagged, r);
+#ifdef SK_PDF_TEST_TAGS_OUTPUT_PATH
+    SkFILEWStream outputStream(SK_PDF_TEST_TAGS_OUTPUT_PATH);
+#else
+    SkDynamicMemoryWStream outputStream;
+#endif
+
+    SkSize pageSize = SkSize::Make(612, 792);  // U.S. Letter
+
+    SkPDF::Metadata metadata;
+    metadata.fTitle = "Example Tagged Table PDF";
+    metadata.fCreator = "Skia";
+    SkTime::DateTime now;
+    SkTime::GetDateTime(&now);
+    metadata.fCreation = now;
+    metadata.fModified = now;
+
+    constexpr int kRowCount = 5;
+    constexpr int kColCount = 4;
+    const char* cellData[kRowCount * kColCount] = {
+        "Car",                  "Engine",   "City MPG", "Highway MPG",
+        "Mitsubishi Mirage ES", "Gas",      "28",       "47",
+        "Toyota Prius Three",   "Hybrid",   "43",       "59",
+        "Nissan Leaf SL",       "Electric", "N/A",      nullptr,
+        "Tesla Model 3",        nullptr,    "N/A",      nullptr
+    };
+
+    // The document tag.
+    PDFTag root;
+    root.fNodeId = 1;
+    root.fType = SkPDF::DocumentStructureType::kDocument;
+    root.fChildCount = 2;
+    PDFTag rootChildren[2];
+
+    // Heading.
+    PDFTag& h1 = rootChildren[0];
+    h1.fNodeId = 2;
+    h1.fType = SkPDF::DocumentStructureType::kH1;
+    h1.fChildCount = 0;
+
+    // Table.
+    PDFTag& table = rootChildren[1];
+    table.fNodeId = 3;
+    table.fType = SkPDF::DocumentStructureType::kTable;
+    table.fChildCount = 5;
+    table.fAttributes.appendFloatArray("Layout", "BBox", {72, 72, 360, 360});
+
+    PDFTag rows[kRowCount];
+    PDFTag all_cells[kRowCount * kColCount];
+    for (int rowIndex = 0; rowIndex < kRowCount; rowIndex++) {
+        PDFTag& row = rows[rowIndex];
+        row.fNodeId = 4 + rowIndex;
+        row.fType = SkPDF::DocumentStructureType::kTR;
+        row.fChildCount = kColCount;
+        PDFTag* cells = &all_cells[rowIndex * kColCount];
+
+        for (int colIndex = 0; colIndex < kColCount; colIndex++) {
+            int cellIndex = rowIndex * kColCount + colIndex;
+            PDFTag& cell = cells[colIndex];
+            cell.fNodeId = 10 + cellIndex;
+            if (!cellData[cellIndex])
+                cell.fType = SkPDF::DocumentStructureType::kNonStruct;
+            else if (rowIndex == 0 || colIndex == 0)
+                cell.fType = SkPDF::DocumentStructureType::kTH;
+            else
+                cell.fType = SkPDF::DocumentStructureType::kTD;
+            cell.fChildCount = 0;
+
+            if (cellIndex == 13) {
+                cell.fAttributes.appendInt("Table", "RowSpan", 2);
+            } else if (cellIndex == 14 || cellIndex == 18) {
+                cell.fAttributes.appendInt("Table", "ColSpan", 2);
+            } else if (cell.fType == SkPDF::DocumentStructureType::kTH) {
+                cell.fAttributes.appendString(
+                    "Table", "Scope", rowIndex == 0 ? "Column" : "Row");
+            }
+        }
+        row.fChildren = cells;
+    }
+    table.fChildren = rows;
+    root.fChildren = rootChildren;
+
+    metadata.fStructureElementTreeRoot = &root;
+    sk_sp<SkDocument> document = SkPDF::MakeDocument(
+        &outputStream, metadata);
+
+    SkPaint paint;
+    paint.setColor(SK_ColorBLACK);
+
+    SkCanvas* canvas =
+            document->beginPage(pageSize.width(),
+                                pageSize.height());
+    SkPDF::SetNodeId(canvas, 2);
+    SkFont font(nullptr, 36);
+    canvas->drawString("Tagged PDF Table", 72, 72, font, paint);
+
+    font.setSize(14);
+    for (int rowIndex = 0; rowIndex < kRowCount; rowIndex++) {
+        for (int colIndex = 0; colIndex < kColCount; colIndex++) {
+            int cellIndex = rowIndex * kColCount + colIndex;
+            const char* str = cellData[cellIndex];
+            if (!str)
+                continue;
+
+            int x = 72 + colIndex * 108 + (colIndex > 0 ? 72 : 0);
+            int y = 144 + rowIndex * 48;
+
+            SkPDF::SetNodeId(canvas, 10 + cellIndex);
+            canvas->drawString(str, x, y, font, paint);
+        }
+    }
+
+    document->endPage();
+    document->close();
+    outputStream.flush();
+}