Populate Buttons in PDFiumPage

This CL introduces a method PDFiumPage::PopulateButton() which reads
button form fields (radio button, checkbox and push button) from the PDF
document and stores relevant information in a vector within PDFiumPage.

The CL also includes a new test file with sample radio buttons, checkbox
and push button. A unit test is also added to validate the new method.

Bug: 1030242
Change-Id: I1f5102f8c20027abe785fa1833c2ecc9cbce8734
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2249285
Commit-Queue: Mansi Awasthi <maawas@microsoft.com>
Reviewed-by: Lei Zhang <thestig@chromium.org>
Reviewed-by: Ankit Kumar 🌪️ <ankk@microsoft.com>
Cr-Commit-Position: refs/heads/master@{#784392}

pdf/pdfium/pdfium_page.cc

@@ -244,6 +244,11 @@ uint32_t CountInternalTextOverlaps(const std::vector<T>& text_objects) {
   return overlaps;
 }
 
+bool IsRadioButtonOrCheckBox(int button_type) {
+  return button_type == FPDF_FORMFIELD_CHECKBOX ||
+         button_type == FPDF_FORMFIELD_RADIOBUTTON;
+}
+
 }  // namespace
 
 PDFiumPage::LinkTarget::LinkTarget() : page(-1) {}
@@ -1268,12 +1273,45 @@ void PDFiumPage::PopulateChoiceField(FPDF_ANNOTATION annot) {
   choice_fields_.push_back(std::move(choice_field));
 }
 
+void PDFiumPage::PopulateButton(FPDF_ANNOTATION annot) {
+  DCHECK(annot);
+  FPDF_FORMHANDLE form_handle = engine_->form();
+  int button_type = FPDFAnnot_GetFormFieldType(form_handle, annot);
+  DCHECK(button_type == FPDF_FORMFIELD_PUSHBUTTON ||
+         IsRadioButtonOrCheckBox(button_type));
+
+  Button button;
+  if (!PopulateFormFieldProperties(annot, &button))
+    return;
+
+  button.type = button_type;
+  if (IsRadioButtonOrCheckBox(button_type)) {
+    button.control_count = FPDFAnnot_GetFormControlCount(form_handle, annot);
+    if (button.control_count <= 0)
+      return;
+
+    button.control_index = FPDFAnnot_GetFormControlIndex(form_handle, annot);
+    button.value = base::UTF16ToUTF8(CallPDFiumWideStringBufferApi(
+        base::BindRepeating(&FPDFAnnot_GetFormFieldExportValue, form_handle,
+                            annot),
+        /*check_expected_size=*/true));
+    button.is_checked = FPDFAnnot_IsChecked(form_handle, annot);
+  }
+  buttons_.push_back(std::move(button));
+}
+
 void PDFiumPage::PopulateFormField(FPDF_ANNOTATION annot) {
   DCHECK_EQ(FPDFAnnot_GetSubtype(annot), FPDF_ANNOT_WIDGET);
   int form_field_type = FPDFAnnot_GetFormFieldType(engine_->form(), annot);
 
   // TODO(crbug.com/1030242): Populate other types of form fields too.
   switch (form_field_type) {
+    case FPDF_FORMFIELD_PUSHBUTTON:
+    case FPDF_FORMFIELD_CHECKBOX:
+    case FPDF_FORMFIELD_RADIOBUTTON: {
+      PopulateButton(annot);
+      break;
+    }
     case FPDF_FORMFIELD_COMBOBOX:
     case FPDF_FORMFIELD_LISTBOX: {
       PopulateChoiceField(annot);
@@ -1461,6 +1499,12 @@ PDFiumPage::ChoiceField::ChoiceField(const ChoiceField& that) = default;
 
 PDFiumPage::ChoiceField::~ChoiceField() = default;
 
+PDFiumPage::Button::Button() = default;
+
+PDFiumPage::Button::Button(const Button& that) = default;
+
+PDFiumPage::Button::~Button() = default;
+
 // static
 uint32_t PDFiumPage::CountLinkHighlightOverlaps(
     const std::vector<Link>& links,

pdf/pdfium/pdfium_page.h

@@ -179,6 +179,7 @@ class PDFiumPage {
   FRIEND_TEST_ALL_PREFIXES(PDFiumPageHighlightTest, TestPopulateHighlights);
   FRIEND_TEST_ALL_PREFIXES(PDFiumPageTextFieldTest, TestPopulateTextFields);
   FRIEND_TEST_ALL_PREFIXES(PDFiumPageChoiceFieldTest, TestPopulateChoiceFields);
+  FRIEND_TEST_ALL_PREFIXES(PDFiumPageButtonTest, TestPopulateButtons);
   FRIEND_TEST_ALL_PREFIXES(PDFiumPageOverlappingTest, CountPartialOverlaps);
   FRIEND_TEST_ALL_PREFIXES(PDFiumPageOverlappingTest, CountCompleteOverlaps);
 
@@ -277,6 +278,29 @@ class PDFiumPage {
     std::vector<ChoiceFieldOption> options;
   };
 
+  // Represents a button within the page.
+  struct Button : FormField {
+    Button();
+    Button(const Button& other);
+    ~Button();
+
+    std::string value;
+    // A button can be of type radio, checkbox or push button.
+    int type;
+    // Represents if the radio button or checkbox is checked.
+    bool is_checked = false;
+    // Represents count of controls in the control group. A group of
+    // interactive form annotations is collectively called a form control
+    // group. Here an interactive form annotation should be either a radio
+    // button or a checkbox.
+    uint32_t control_count = 0;
+    // Represents index of the control in the control group. A group of
+    // interactive form annotations is collectively called a form control
+    // group. Here an interactive form annotation should be either a radio
+    // button or a checkbox. Value of |control_index| is -1 for push button.
+    int control_index = -1;
+  };
+
   // Returns a link index if the given character index is over a link, or -1
   // otherwise.
   int GetLink(int char_index, LinkTarget* target);
@@ -296,7 +320,9 @@ class PDFiumPage {
   void PopulateTextField(FPDF_ANNOTATION annot);
   // Populate |choice_fields_| with |annot|.
   void PopulateChoiceField(FPDF_ANNOTATION annot);
-  // Populate form fields like text field and choice field on the page.
+  // Populate |buttons_| with |annot|.
+  void PopulateButton(FPDF_ANNOTATION annot);
+  // Populate form fields like text field, choice field and button on the page.
   void PopulateFormField(FPDF_ANNOTATION annot);
   // Returns link type and fills target associated with a destination. Returns
   // NONSELECTABLE_AREA if detection failed.
@@ -354,6 +380,7 @@ class PDFiumPage {
   std::vector<Highlight> highlights_;
   std::vector<TextField> text_fields_;
   std::vector<ChoiceField> choice_fields_;
+  std::vector<Button> buttons_;
   bool logged_overlapping_annotations_ = false;
   bool calculated_page_object_text_run_breaks_ = false;
   // The set of character indices on which text runs need to be broken for page

pdf/pdfium/pdfium_page_unittest.cc

@@ -17,6 +17,7 @@
 #include "pdf/test/test_utils.h"
 #include "ppapi/c/private/ppb_pdf.h"
 #include "testing/gtest/include/gtest/gtest.h"
+#include "third_party/pdfium/public/fpdf_formfill.h"
 #include "ui/gfx/range/range.h"
 
 namespace chrome_pdf {
@@ -601,6 +602,88 @@ TEST_F(PDFiumPageChoiceFieldTest, TestPopulateChoiceFields) {
   }
 }
 
+using PDFiumPageButtonTest = PDFiumTestBase;
+
+TEST_F(PDFiumPageButtonTest, TestPopulateButtons) {
+  struct ExpectedButton {
+    const char* name;
+    const char* value;
+    int type;
+    int flags;
+    bool is_checked;
+    uint32_t control_count;
+    int control_index;
+    pp::Rect bounding_rect;
+  };
+
+  static const ExpectedButton kExpectedButtons[] = {{"readOnlyCheckbox",
+                                                     "Yes",
+                                                     FPDF_FORMFIELD_CHECKBOX,
+                                                     1,
+                                                     true,
+                                                     1,
+                                                     0,
+                                                     {185, 43, 28, 28}},
+                                                    {"checkbox",
+                                                     "Yes",
+                                                     FPDF_FORMFIELD_CHECKBOX,
+                                                     2,
+                                                     false,
+                                                     1,
+                                                     0,
+                                                     {185, 96, 28, 28}},
+                                                    {"RadioButton",
+                                                     "value1",
+                                                     FPDF_FORMFIELD_RADIOBUTTON,
+                                                     49154,
+                                                     false,
+                                                     2,
+                                                     0,
+                                                     {185, 243, 28, 28}},
+                                                    {"RadioButton",
+                                                     "value2",
+                                                     FPDF_FORMFIELD_RADIOBUTTON,
+                                                     49154,
+                                                     true,
+                                                     2,
+                                                     1,
+                                                     {252, 243, 27, 28}},
+                                                    {"PushButton",
+                                                     "",
+                                                     FPDF_FORMFIELD_PUSHBUTTON,
+                                                     65536,
+                                                     false,
+                                                     0,
+                                                     -1,
+                                                     {118, 270, 55, 67}}};
+
+  TestClient client;
+  std::unique_ptr<PDFiumEngine> engine =
+      InitializeEngine(&client, FILE_PATH_LITERAL("form_buttons.pdf"));
+  ASSERT_TRUE(engine);
+  ASSERT_EQ(1, engine->GetNumberOfPages());
+
+  PDFiumPage* page = GetPDFiumPageForTest(engine.get(), 0);
+  ASSERT_TRUE(page);
+  page->PopulateAnnotations();
+  size_t buttons_count = page->buttons_.size();
+  ASSERT_EQ(base::size(kExpectedButtons), buttons_count);
+
+  for (size_t i = 0; i < buttons_count; ++i) {
+    EXPECT_EQ(kExpectedButtons[i].name, page->buttons_[i].name);
+    EXPECT_EQ(kExpectedButtons[i].value, page->buttons_[i].value);
+    EXPECT_EQ(kExpectedButtons[i].type, page->buttons_[i].type);
+    EXPECT_EQ(kExpectedButtons[i].flags, page->buttons_[i].flags);
+    EXPECT_EQ(kExpectedButtons[i].is_checked, page->buttons_[i].is_checked);
+    EXPECT_EQ(kExpectedButtons[i].control_count,
+              page->buttons_[i].control_count);
+    EXPECT_EQ(kExpectedButtons[i].control_index,
+              page->buttons_[i].control_index);
+    CompareRect(kExpectedButtons[i].bounding_rect,
+                page->buttons_[i].bounding_rect);
+  }
+}
+
 using PDFiumPageOverlappingTest = PDFiumTestBase;
 
 // The following scenarios are covered across both test cases:

pdf/test/data/form_buttons.in

+{{header}}
+{{object 1 0}} <<
+  /Type /Catalog
+  /Pages 2 0 R
+  /AcroForm <<
+    /Fields [5 0 R 6 0 R 7 0 R 10 0 R]
+  >>
+>>
+endobj
+{{object 2 0}} <<
+  /Type /Pages
+  /Count 1
+  /Kids [3 0 R]
+>>
+endobj
+{{object 3 0}} <<
+  /Type /Page
+  /Parent 2 0 R
+  /MediaBox [0 0 300 300]
+  /Resources <<
+    /ProcSet [/PDF /Text /ImageB /ImageC /ImageI]
+    /Font <<
+      /F1 4 0 R
+    >>
+  >>
+  /Annots [5 0 R 6 0 R 8 0 R 9 0 R 10 0 R]
+>>
+endobj
+{{object 4 0}} <<
+  /Type /Font
+  /Subtype /Type1
+  /Name /F1
+  /BaseFont /Courier
+>>
+endobj
+{{object 5 0}} <<
+  /Type /Annot
+  /Subtype /Widget
+  /FT /Btn
+  /F 4
+  /Ff 1
+  /P 3 0 R
+  /Rect [135 250 155 270]
+  /T (readOnlyCheckbox)
+  /TU (readOnlyCheckbox)
+  /AP <<
+    /N <<
+      /Yes 11 0 R
+      /Off 12 0 R
+    >>
+    /D <<
+      /Yes 11 0 R
+      /Off 12 0 R
+    >>
+    /R <<
+      /Yes 11 0 R
+      /Off 12 0 R
+    >>
+  >>
+  /AS /Yes
+  /BS <<
+    /S /S
+    /W 1
+  >>
+  /H /N
+  /MK <<
+    /BC [1 0 1]
+    /BG [1 .752941 .796078]
+    /CA (4)
+  >>
+  /V /Yes
+>>
+endobj
+{{object 6 0}} <<
+  /Type /Annot
+  /Subtype /Widget
+  /FT /Btn
+  /F 4
+  /Ff 2
+  /P 3 0 R
+  /Rect [135 210 155 230]
+  /T (checkbox)
+  /TU (checkbox)
+  /AP <<
+    /N <<
+      /Yes 11 0 R
+      /Off 12 0 R
+    >>
+    /D <<
+      /Yes 11 0 R
+      /Off 12 0 R
+    >>
+    /R <<
+      /Yes 11 0 R
+      /Off 12 0 R
+    >>
+  >>
+  /AS /Off
+  /BS <<
+    /S /S
+    /W 2
+  >>
+  /H /N
+  /MK <<
+    /BC [.1 .1 .1]
+    /BG [.8 .843 1]
+    /CA (5)
+  >>
+  /V /Off
+>>
+endobj
+{{object 7 0}} <<
+  /FT /Btn
+  /Ff 49154
+  /T (RadioButton)
+  /TU (RadioButton1)
+  /Kids [8 0 R 9 0 R]
+  /V /value2
+>>
+endobj
+{{object 8 0}} <<
+  /Type /Annot
+  /Subtype /Widget
+  /FT /Btn
+  /F 4
+  /P 3 0 R
+  /Parent 7 0 R
+  /Rect [135 100 155 120]
+  /AP <<
+    /N <<
+      /value1 13 0 R
+      /Off 14 0 R
+    >>
+    /D <<
+      /value1 13 0 R
+      /Off 14 0 R
+    >>
+    /R <<
+      /value1 13 0 R
+      /Off 14 0 R
+    >>
+  >>
+  /AS /Off
+  /BS <<
+    /S /S
+    /W 2
+  >>
+  /H /N
+  /MK <<
+    /BC [0 .501961 0]
+    /BG [0 0 1]
+    /CA (5)
+  >>
+>>
+endobj
+{{object 9 0}} <<
+  /Type /Annot
+  /Subtype /Widget
+  /FT /Btn
+  /F 4
+  /P 3 0 R
+  /Parent 7 0 R
+  /Rect [185 100 205 120]
+  /AP <<
+    /N <<
+      /value2 13 0 R
+      /Off 14 0 R
+    >>
+    /D <<
+      /value2 13 0 R
+      /Off 14 0 R
+    >>
+    /R <<
+      /value2 13 0 R
+      /Off 14 0 R
+    >>
+  >>
+  /AS /value2
+  /BS <<
+    /S /S
+    /W 2
+  >>
+  /H /N
+  /MK <<
+    /BC [0 .501961 0]
+    /BG [0 0 1]
+    /CA (5)
+  >>
+>>
+endobj
+{{object 10 0}} <<
+  /Type /Annot
+  /Subtype /Widget
+  /F 4
+  /FT /Btn
+  /Rect [85 50 125 100]
+  /T (PushButton)
+  /Ff 65536
+  /H
+  /P
+  /MK <<
+    /BC [1 0 1]
+    /BG [1 .752941 .796078]
+    /CA (Submit)
+  >>
+>>
+endobj
+{{object 11 0}} <<
+  /Type /XObject
+  /Subtype /Form
+  /FormType 1
+  /BBox [0 0 20 20]
+  /Resources <<
+    /ProcSet [/PDF]
+  >>
+  {{streamlen}}
+>>
+stream
+q
+1 0 1 RG 1 w 0.5 0.5 19 19 re s
+Q
+q
+0 0 1 rg 0 0 1 RG
+11.90655 14.48997 m
+14.54638 17.62618 16.37396 18.07744 17.59234 18.07744 c
+6.807382 2.418942 6.672006 2.148189 6.446379 2.012813 c
+6.153064 1.877437 5.476184 1.877437 5.160306 1.877437 c
+4.438301 1.877437 4.280362 1.877437 4.122423 1.990251 c
+3.896797 2.148189 3.783983 2.464067 3.626045 2.779944 c
+3.129666 3.817827 2.249721 6.074095 2.249721 7.202228 c
+2.249721 7.585794 2.475348 7.766295 2.768663 7.969359 c
+3.242479 8.307799 3.919359 8.691365 4.505989 8.691365 c
+4.957242 8.691365 5.002368 8.307799 5.137744 7.946797 c
+f
+Q
+endstream
+endobj
+{{object 12 0}} <<
+  /Type /XObject
+  /Subtype /Form
+  /FormType 1
+  /BBox [0 0 20 20]
+  /Resources <<
+    /ProcSet [/PDF]
+  >>
+  {{streamlen}}
+>>
+stream
+q
+1 0 1 RG 1 w 0.5 0.5 19 19 re s
+Q
+endstream
+endobj
+{{object 13 0}} <<
+  /Type /XObject
+  /Subtype /Form
+  /FormType 1
+  /BBox [0 0 20 20]
+  /Resources <<
+    /ProcSet [/PDF]
+  >>
+  {{streamlen}}
+>>
+stream
+q
+0 .501961 0 RG 2 w
+1 0 0 1 10 10 cm
+9 0 m
+9 4.97079 4.97079 9 0 9 c
+-4.97079 9 -9 4.97079 -9 0 c
+-9 -4.97079 -4.97079 -9 0 -9 c
+4.97079 -9 9 -4.97079 9 0 c
+s
+Q
+q
+1 .752941 .796078 rg 1 .752941 .796078 RG
+17.2 15.95145 m
+11.20694 10 l
+17.2 4.027746 l
+15.97225 2.8 l
+10 8.793064 l
+4.027746 2.8 l
+2.8 4.027746 l
+8.813873 10 l
+2.8 15.97225 l
+4.027746 17.2 l
+10 11.20694 l
+15.97225 17.2 l
+17.2 15.95145 l
+h
+f
+Q
+endstream
+endobj
+{{object 14 0}} <<
+  /Type /XObject
+  /Subtype /Form
+  /FormType 1
+  /BBox [0 0 20 20]
+  /Resources <<
+    /ProcSet [/PDF]
+  >>
+  {{streamlen}}
+>>
+stream
+q
+0 .501961 0 RG 2 w
+1 0 0 1 10 10 cm
+9 0 m
+9 4.97079 4.97079 9 0 9 c
+-4.97079 9 -9 4.97079 -9 0 c
+-9 -4.97079 -4.97079 -9 0 -9 c
+4.97079 -9 9 -4.97079 9 0 c
+s
+Q
+endstream
+endobj
+{{xref}}
+{{trailer}}
+{{startxref}}
+%%EOF