Implement additional IAccessibleText methods to return the nearest

character, word, line, sentence, or paragraph within a text field.
These methods are needed by SAToGo.

BUG=36217
TEST=Added new unit test.

Review URL: http://codereview.chromium.org/3550017

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@61979 0039d316-1c4b-4281-b951-d872f2087c98

3 files changed, 304 insertions, 65 deletions

diff --git a/chrome/browser/accessibility/browser_accessibility_win.cc b/chrome/browser/accessibility/browser_accessibility_win.cc
index 432eb1a..8817dfa 100644
--- a/chrome/browser/accessibility/browser_accessibility_win.cc
+++ b/chrome/browser/accessibility/browser_accessibility_win.cc
@@ -54,7 +54,7 @@ HRESULT BrowserAccessibilityWin::accDoDefaultAction(VARIANT var_id) {
 }
 
 STDMETHODIMP BrowserAccessibilityWin::accHitTest(LONG x_left, LONG y_top,
-                                              VARIANT* child) {
+                                                 VARIANT* child) {
   if (!instance_active_)
     return E_FAIL;
 
@@ -65,8 +65,8 @@ STDMETHODIMP BrowserAccessibilityWin::accHitTest(LONG x_left, LONG y_top,
 }
 
 STDMETHODIMP BrowserAccessibilityWin::accLocation(LONG* x_left, LONG* y_top,
-                                               LONG* width, LONG* height,
-                                               VARIANT var_id) {
+                                                  LONG* width, LONG* height,
+                                                  VARIANT var_id) {
   if (!instance_active_)
     return E_FAIL;
 
@@ -534,45 +534,6 @@ STDMETHODIMP BrowserAccessibilityWin::get_nCharacters(LONG* n_characters) {
   return S_OK;
 }
 
-STDMETHODIMP BrowserAccessibilityWin::get_text(
-    LONG start_offset, LONG end_offset, BSTR* text) {
-  if (!instance_active_)
-    return E_FAIL;
-
-  if (!text)
-    return E_INVALIDARG;
-
-  string16 text_str;
-  if (role_ == WebAccessibility::ROLE_TEXT_FIELD) {
-    text_str = value_;
-  } else {
-    text_str = name_;
-  }
-
-  // The spec allows the arguments to be reversed.
-  if (start_offset > end_offset) {
-    LONG tmp = start_offset;
-    start_offset = end_offset;
-    end_offset = tmp;
-  }
-
-  // The spec does not allow the start or end offsets to be out or range;
-  // we must return an error if so.
-  LONG len = text_str.length();
-  if (start_offset < 0)
-    return E_INVALIDARG;
-  if (end_offset > len)
-    return E_INVALIDARG;
-
-  string16 substr = text_str.substr(start_offset, end_offset - start_offset);
-  if (substr.empty())
-    return S_FALSE;
-
-  *text = SysAllocString(substr.c_str());
-  DCHECK(*text);
-  return S_OK;
-}
-
 STDMETHODIMP BrowserAccessibilityWin::get_caretOffset(LONG* offset) {
   if (!instance_active_)
     return E_FAIL;
@@ -585,7 +546,7 @@ STDMETHODIMP BrowserAccessibilityWin::get_caretOffset(LONG* offset) {
     if (GetAttributeAsInt(WebAccessibility::ATTR_TEXT_SEL_START, &sel_start)) {
       *offset = sel_start;
     } else {
-  *offset = 0;
+      *offset = 0;
     }
   } else {
     *offset = 0;
@@ -646,6 +607,121 @@ STDMETHODIMP BrowserAccessibilityWin::get_selection(LONG selection_index,
   return S_OK;
 }
 
+STDMETHODIMP BrowserAccessibilityWin::get_text(
+    LONG start_offset, LONG end_offset, BSTR* text) {
+  if (!instance_active_)
+    return E_FAIL;
+
+  if (!text)
+    return E_INVALIDARG;
+
+  const string16& text_str = TextForIAccessibleText();
+
+  // The spec allows the arguments to be reversed.
+  if (start_offset > end_offset) {
+    LONG tmp = start_offset;
+    start_offset = end_offset;
+    end_offset = tmp;
+  }
+
+  // The spec does not allow the start or end offsets to be out or range;
+  // we must return an error if so.
+  LONG len = text_str.length();
+  if (start_offset < 0)
+    return E_INVALIDARG;
+  if (end_offset > len)
+    return E_INVALIDARG;
+
+  string16 substr = text_str.substr(start_offset, end_offset - start_offset);
+  if (substr.empty())
+    return S_FALSE;
+
+  *text = SysAllocString(substr.c_str());
+  DCHECK(*text);
+  return S_OK;
+}
+
+STDMETHODIMP BrowserAccessibilityWin::get_textAtOffset(
+    LONG offset,
+    enum IA2TextBoundaryType boundary_type,
+    LONG* start_offset, LONG* end_offset,
+    BSTR* text) {
+  if (!instance_active_)
+    return E_FAIL;
+
+  if (!start_offset || !end_offset || !text)
+    return E_INVALIDARG;
+
+  // The IAccessible2 spec says we don't have to implement the "sentence"
+  // boundary type, we can just let the screenreader handle it.
+  if (boundary_type == IA2_TEXT_BOUNDARY_SENTENCE) {
+    *start_offset = 0;
+    *end_offset = 0;
+    *text = NULL;
+    return S_FALSE;
+  }
+
+  const string16& text_str = TextForIAccessibleText();
+
+  *start_offset = FindBoundary(text_str, boundary_type, offset, -1);
+  *end_offset = FindBoundary(text_str, boundary_type, offset, 1);
+  return get_text(*start_offset, *end_offset, text);
+}
+
+STDMETHODIMP BrowserAccessibilityWin::get_textBeforeOffset(
+    LONG offset,
+    enum IA2TextBoundaryType boundary_type,
+    LONG* start_offset, LONG* end_offset,
+    BSTR* text) {
+  if (!instance_active_)
+    return E_FAIL;
+
+  if (!start_offset || !end_offset || !text)
+    return E_INVALIDARG;
+
+  // The IAccessible2 spec says we don't have to implement the "sentence"
+  // boundary type, we can just let the screenreader handle it.
+  if (boundary_type == IA2_TEXT_BOUNDARY_SENTENCE) {
+    *start_offset = 0;
+    *end_offset = 0;
+    *text = NULL;
+    return S_FALSE;
+  }
+
+  const string16& text_str = TextForIAccessibleText();
+
+  *start_offset = FindBoundary(text_str, boundary_type, offset, -1);
+  *end_offset = offset;
+  return get_text(*start_offset, *end_offset, text);
+}
+
+STDMETHODIMP BrowserAccessibilityWin::get_textAfterOffset(
+    LONG offset,
+    enum IA2TextBoundaryType boundary_type,
+    LONG* start_offset, LONG* end_offset,
+    BSTR* text) {
+  if (!instance_active_)
+    return E_FAIL;
+
+  if (!start_offset || !end_offset || !text)
+    return E_INVALIDARG;
+
+  // The IAccessible2 spec says we don't have to implement the "sentence"
+  // boundary type, we can just let the screenreader handle it.
+  if (boundary_type == IA2_TEXT_BOUNDARY_SENTENCE) {
+    *start_offset = 0;
+    *end_offset = 0;
+    *text = NULL;
+    return S_FALSE;
+  }
+
+  const string16& text_str = TextForIAccessibleText();
+
+  *start_offset = offset;
+  *end_offset = FindBoundary(text_str, boundary_type, offset, 1);
+  return get_text(*start_offset, *end_offset, text);
+}
+
 //
 // ISimpleDOMDocument methods.
 //
@@ -1129,6 +1205,76 @@ string16 BrowserAccessibilityWin::Escape(string16 str) {
   return EscapeQueryParamValueUTF8(str, false);
 }
 
+const string16& BrowserAccessibilityWin::TextForIAccessibleText() {
+  if (role_ == WebAccessibility::ROLE_TEXT_FIELD) {
+    return value_;
+  } else {
+    return name_;
+  }
+}
+
+LONG BrowserAccessibilityWin::FindBoundary(
+    const string16& text,
+    IA2TextBoundaryType boundary,
+    LONG start_offset,
+    LONG direction) {
+  LONG text_size = static_cast<LONG>(text.size());
+  DCHECK(start_offset >= 0 && start_offset <= text_size);
+  DCHECK(direction == 1 || direction == -1);
+
+  if (boundary == IA2_TEXT_BOUNDARY_CHAR) {
+    if (direction == 1 && start_offset < text_size)
+      return start_offset + 1;
+    else
+      return start_offset;
+  }
+
+  LONG result = start_offset;
+  for (;;) {
+    LONG pos;
+    if (direction == 1) {
+      if (result >= text_size)
+        return text_size;
+      pos = result;
+    } else {
+      if (result <= 0)
+        return 0;
+      pos = result - 1;
+    }
+
+    switch (boundary) {
+      case IA2_TEXT_BOUNDARY_WORD:
+        if (IsWhitespace(text[pos]))
+          return result;
+        break;
+      case IA2_TEXT_BOUNDARY_LINE:
+      case IA2_TEXT_BOUNDARY_PARAGRAPH:
+        if (text[pos] == '\n')
+          return result;
+      case IA2_TEXT_BOUNDARY_SENTENCE:
+        // Note that we don't actually have to implement sentence support;
+        // currently IAccessibleText functions return S_FALSE so that
+        // screenreaders will handle it on their own.
+        if ((text[pos] == '.' || text[pos] == '!' || text[pos] == '?') &&
+            (pos == text_size - 1 || IsWhitespace(text[pos + 1]))) {
+          return result;
+        }
+      case IA2_TEXT_BOUNDARY_ALL:
+      default:
+        break;
+    }
+
+    if (direction > 0) {
+      result++;
+    } else if (direction < 0) {
+      result--;
+    } else {
+      NOTREACHED();
+      return result;
+    }
+  }
+}
+
 void BrowserAccessibilityWin::InitRoleAndState() {
   ia_state_ = 0;
   ia2_state_ = IA2_STATE_OPAQUE;
diff --git a/chrome/browser/accessibility/browser_accessibility_win.h b/chrome/browser/accessibility/browser_accessibility_win.h
index dbd014f..16ea136 100644
--- a/chrome/browser/accessibility/browser_accessibility_win.h
+++ b/chrome/browser/accessibility/browser_accessibility_win.h
@@ -230,8 +230,6 @@ class BrowserAccessibilityWin
 
   STDMETHODIMP get_nCharacters(LONG* n_characters);
 
-  STDMETHODIMP get_text(LONG start_offset, LONG end_offset, BSTR* text);
-
   STDMETHODIMP get_caretOffset(LONG* offset);
 
   STDMETHODIMP get_nSelections(LONG* n_selections);
@@ -240,6 +238,22 @@ class BrowserAccessibilityWin
                              LONG* start_offset,
                              LONG* end_offset);
 
+  STDMETHODIMP get_text(LONG start_offset, LONG end_offset, BSTR* text);
+
+  STDMETHODIMP get_textAtOffset(LONG offset,
+                                enum IA2TextBoundaryType boundary_type,
+                                LONG* start_offset, LONG* end_offset,
+                                BSTR* text);
+
+  STDMETHODIMP get_textBeforeOffset(LONG offset,
+                                    enum IA2TextBoundaryType boundary_type,
+                                    LONG* start_offset, LONG* end_offset,
+                                    BSTR* text);
+
+  STDMETHODIMP get_textAfterOffset(LONG offset,
+                                   enum IA2TextBoundaryType boundary_type,
+                                   LONG* start_offset, LONG* end_offset,
+                                   BSTR* text);
 
   // IAccessibleText methods not implemented.
   STDMETHODIMP addSelection(LONG start_offset, LONG end_offset) {
@@ -256,26 +270,8 @@ class BrowserAccessibilityWin
     return E_NOTIMPL;
   }
   STDMETHODIMP get_offsetAtPoint(LONG x, LONG y,
-                             enum IA2CoordinateType coord_type,
-                             LONG* offset) {
-    return E_NOTIMPL;
-  }
-  STDMETHODIMP get_textBeforeOffset(LONG offset,
-                                enum IA2TextBoundaryType boundary_type,
-                                LONG* start_offset, LONG* end_offset,
-                                BSTR* text) {
-    return E_NOTIMPL;
-  }
-  STDMETHODIMP get_textAfterOffset(LONG offset,
-                               enum IA2TextBoundaryType boundary_type,
-                               LONG* start_offset, LONG* end_offset,
-                               BSTR* text) {
-    return E_NOTIMPL;
-  }
-  STDMETHODIMP get_textAtOffset(LONG offset,
-                            enum IA2TextBoundaryType boundary_type,
-                            LONG* start_offset, LONG* end_offset,
-                            BSTR* text) {
+                                 enum IA2CoordinateType coord_type,
+                                 LONG* offset) {
     return E_NOTIMPL;
   }
   STDMETHODIMP removeSelection(LONG selection_index) {
@@ -484,6 +480,18 @@ class BrowserAccessibilityWin
   // Escape a string like it would be escaped for a URL or HTML form.
   string16 Escape(string16 str);
 
+  // Get the text of this node for the purposes of IAccessibleText - it may
+  // be the name, it may be the value, etc. depending on the role.
+  const string16& TextForIAccessibleText();
+
+  // Search forwards (direction == 1) or backwards (direction == -1) from
+  // the given offset until the given IAccessible2 boundary (like word,
+  // sentence) is found, and return its offset.
+  LONG FindBoundary(const string16& text,
+                    IA2TextBoundaryType boundary,
+                    LONG start_offset,
+                    LONG direction);
+
   // COM objects are reference-counted. When we're done with this object
   // and it's removed from our accessibility tree, a client may still be
   // holding onto a pointer to this object, so we mark it as inactive
diff --git a/chrome/browser/accessibility/browser_accessibility_win_unittest.cc b/chrome/browser/accessibility/browser_accessibility_win_unittest.cc
index 549700a..dd264a1 100644
--- a/chrome/browser/accessibility/browser_accessibility_win_unittest.cc
+++ b/chrome/browser/accessibility/browser_accessibility_win_unittest.cc
@@ -274,3 +274,88 @@ TEST_F(BrowserAccessibilityTest, TestChildrenChangeNoLeaks) {
   delete manager;
   ASSERT_EQ(0, CountedBrowserAccessibility::global_obj_count_);
 }
+
+TEST_F(BrowserAccessibilityTest, TestTextBoundaries) {
+  WebAccessibility text1;
+  text1.id = 11;
+  text1.role = WebAccessibility::ROLE_TEXT_FIELD;
+  text1.state = 0;
+  text1.value = L"One two three.\nFour five six.";
+
+  WebAccessibility root;
+  root.id = 1;
+  root.role = WebAccessibility::ROLE_DOCUMENT;
+  root.state = 0;
+  root.children.push_back(text1);
+
+  CountedBrowserAccessibility::global_obj_count_ = 0;
+  BrowserAccessibilityManager* manager = BrowserAccessibilityManager::Create(
+      GetDesktopWindow(), root, NULL,
+      new CountedBrowserAccessibilityFactory());
+  ASSERT_EQ(2, CountedBrowserAccessibility::global_obj_count_);
+
+  BrowserAccessibilityWin* root_obj =
+      manager->GetRoot()->toBrowserAccessibilityWin();
+  BrowserAccessibilityWin* text1_obj =
+      root_obj->GetChild(0)->toBrowserAccessibilityWin();
+
+  BSTR text;
+  long start;
+  long end;
+
+  long text1_len;
+  ASSERT_EQ(S_OK, text1_obj->get_nCharacters(&text1_len));
+
+  ASSERT_EQ(S_OK, text1_obj->get_text(0, text1_len, &text));
+  ASSERT_EQ(text, text1.value);
+  SysFreeString(text);
+
+  ASSERT_EQ(S_OK, text1_obj->get_text(0, 4, &text));
+  ASSERT_EQ(text, std::wstring(L"One "));
+  SysFreeString(text);
+
+  ASSERT_EQ(S_OK, text1_obj->get_textAtOffset(
+      1, IA2_TEXT_BOUNDARY_CHAR, &start, &end, &text));
+  ASSERT_EQ(start, 1);
+  ASSERT_EQ(end, 2);
+  ASSERT_EQ(text, std::wstring(L"n"));
+  SysFreeString(text);
+
+  ASSERT_EQ(S_FALSE, text1_obj->get_textAtOffset(
+      text1_len, IA2_TEXT_BOUNDARY_CHAR, &start, &end, &text));
+  ASSERT_EQ(start, text1_len);
+  ASSERT_EQ(end, text1_len);
+
+  ASSERT_EQ(S_OK, text1_obj->get_textAtOffset(
+      1, IA2_TEXT_BOUNDARY_WORD, &start, &end, &text));
+  ASSERT_EQ(start, 0);
+  ASSERT_EQ(end, 3);
+  ASSERT_EQ(text, std::wstring(L"One"));
+  SysFreeString(text);
+
+  ASSERT_EQ(S_OK, text1_obj->get_textAtOffset(
+      6, IA2_TEXT_BOUNDARY_WORD, &start, &end, &text));
+  ASSERT_EQ(start, 4);
+  ASSERT_EQ(end, 7);
+  ASSERT_EQ(text, std::wstring(L"two"));
+  SysFreeString(text);
+
+  ASSERT_EQ(S_OK, text1_obj->get_textAtOffset(
+      text1_len, IA2_TEXT_BOUNDARY_WORD, &start, &end, &text));
+  ASSERT_EQ(start, 25);
+  ASSERT_EQ(end, 29);
+  ASSERT_EQ(text, std::wstring(L"six."));
+  SysFreeString(text);
+
+  ASSERT_EQ(S_OK, text1_obj->get_textAtOffset(
+      1, IA2_TEXT_BOUNDARY_LINE, &start, &end, &text));
+  ASSERT_EQ(start, 0);
+  ASSERT_EQ(end, 13);
+  ASSERT_EQ(text, std::wstring(L"One two three"));
+  SysFreeString(text);
+
+  // Delete the manager and test that all BrowserAccessibility instances are
+  // deleted.
+  delete manager;
+  ASSERT_EQ(0, CountedBrowserAccessibility::global_obj_count_);
+}