Avi Drissman | 4e1b7bc | 2022-09-15 14:03:50 | [diff] [blame] | 1 | // Copyright 2022 The Chromium Authors |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 2 | // Use of this source code is governed by a BSD-style license that can be |
| 3 | // found in the LICENSE file. |
| 4 | |
Abigail Klein | 1ed2224 | 2022-12-16 16:48:43 | [diff] [blame] | 5 | #include "chrome/renderer/accessibility/ax_tree_distiller.h" |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 6 | |
Abigail Klein | e8c97f2 | 2022-04-04 21:39:11 | [diff] [blame] | 7 | #include <memory> |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 8 | #include <queue> |
Abigail Klein | e9c8e3c | 2022-07-22 15:47:25 | [diff] [blame] | 9 | #include <utility> |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 10 | #include <vector> |
| 11 | |
Abigail Klein | 277fe7c2 | 2022-04-11 21:38:45 | [diff] [blame] | 12 | #include "base/containers/contains.h" |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 13 | #include "base/strings/utf_string_conversions.h" |
Abigail Klein | 1ed2224 | 2022-12-16 16:48:43 | [diff] [blame] | 14 | #include "content/public/renderer/render_frame.h" |
Jocelyn Tran | dab4ca60 | 2023-06-05 18:32:30 | [diff] [blame] | 15 | #include "content/public/renderer/render_thread.h" |
| 16 | #include "services/metrics/public/cpp/mojo_ukm_recorder.h" |
| 17 | #include "services/metrics/public/cpp/ukm_builders.h" |
Abigail Klein | 1ed2224 | 2022-12-16 16:48:43 | [diff] [blame] | 18 | #include "third_party/blink/public/common/browser_interface_broker_proxy.h" |
Ramin Halavati | eddadb6 | 2022-05-04 17:29:49 | [diff] [blame] | 19 | #include "ui/accessibility/accessibility_features.h" |
Jacob Francis | e11099b | 2024-02-17 04:34:04 | [diff] [blame] | 20 | #include "ui/accessibility/ax_enums.mojom-shared.h" |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 21 | #include "ui/accessibility/ax_node.h" |
| 22 | #include "ui/accessibility/ax_tree.h" |
| 23 | |
| 24 | namespace { |
| 25 | |
Abigail Klein | 277fe7c2 | 2022-04-11 21:38:45 | [diff] [blame] | 26 | // TODO: Consider moving this to AXNodeProperties. |
| 27 | static const ax::mojom::Role kContentRoles[]{ |
Jacob Francis | e11099b | 2024-02-17 04:34:04 | [diff] [blame] | 28 | ax::mojom::Role::kHeading, ax::mojom::Role::kParagraph, |
| 29 | ax::mojom::Role::kNote, ax::mojom::Role::kImage, |
| 30 | ax::mojom::Role::kFigcaption}; |
Abigail Klein | 277fe7c2 | 2022-04-11 21:38:45 | [diff] [blame] | 31 | |
| 32 | // TODO: Consider moving this to AXNodeProperties. |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 33 | static const ax::mojom::Role kRolesToSkip[]{ |
| 34 | ax::mojom::Role::kAudio, |
| 35 | ax::mojom::Role::kBanner, |
| 36 | ax::mojom::Role::kButton, |
| 37 | ax::mojom::Role::kComplementary, |
| 38 | ax::mojom::Role::kContentInfo, |
| 39 | ax::mojom::Role::kFooter, |
| 40 | ax::mojom::Role::kFooterAsNonLandmark, |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 41 | ax::mojom::Role::kLabelText, |
| 42 | ax::mojom::Role::kNavigation, |
| 43 | }; |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 44 | |
Abigail Klein | ffd37a2 | 2023-11-29 21:52:14 | [diff] [blame] | 45 | // Find all of the main and article nodes. Also, include unignored heading nodes |
| 46 | // which lie outside of the main and article node. |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 47 | // TODO(crbug.com/1266555): Replace this with a call to |
| 48 | // OneShotAccessibilityTreeSearch. |
Abigail Klein | 41d20e5 | 2022-11-28 19:24:06 | [diff] [blame] | 49 | void GetContentRootNodes(const ui::AXNode* root, |
| 50 | std::vector<const ui::AXNode*>* content_root_nodes) { |
Kristi Saney | 414dd79b | 2023-04-20 20:45:44 | [diff] [blame] | 51 | if (!root) { |
| 52 | return; |
| 53 | } |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 54 | std::queue<const ui::AXNode*> queue; |
Abigail Klein | 930f02c | 2022-11-07 15:03:50 | [diff] [blame] | 55 | queue.push(root); |
Abigail Klein | ffd37a2 | 2023-11-29 21:52:14 | [diff] [blame] | 56 | bool has_main_or_heading = false; |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 57 | while (!queue.empty()) { |
Abigail Klein | 930f02c | 2022-11-07 15:03:50 | [diff] [blame] | 58 | const ui::AXNode* node = queue.front(); |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 59 | queue.pop(); |
Abigail Klein | 41d20e5 | 2022-11-28 19:24:06 | [diff] [blame] | 60 | // If a main or article node is found, add it to the list of content root |
| 61 | // nodes and continue. Do not explore children for nested article nodes. |
| 62 | if (node->GetRole() == ax::mojom::Role::kMain || |
| 63 | node->GetRole() == ax::mojom::Role::kArticle) { |
| 64 | content_root_nodes->push_back(node); |
Abigail Klein | ffd37a2 | 2023-11-29 21:52:14 | [diff] [blame] | 65 | has_main_or_heading = true; |
| 66 | continue; |
| 67 | } |
| 68 | // If a heading node is found, add it to the list of content root nodes, |
| 69 | // too. It may be removed later if the tree doesn't contain a main or |
| 70 | // article node. |
| 71 | if (node->GetRole() == ax::mojom::Role::kHeading) { |
| 72 | content_root_nodes->push_back(node); |
Abigail Klein | 41d20e5 | 2022-11-28 19:24:06 | [diff] [blame] | 73 | continue; |
| 74 | } |
Abigail Klein | 930f02c | 2022-11-07 15:03:50 | [diff] [blame] | 75 | for (auto iter = node->UnignoredChildrenBegin(); |
| 76 | iter != node->UnignoredChildrenEnd(); ++iter) { |
| 77 | queue.push(iter.get()); |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 78 | } |
| 79 | } |
Abigail Klein | ffd37a2 | 2023-11-29 21:52:14 | [diff] [blame] | 80 | if (!has_main_or_heading) { |
| 81 | content_root_nodes->clear(); |
| 82 | } |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 83 | } |
| 84 | |
Abigail Klein | 277fe7c2 | 2022-04-11 21:38:45 | [diff] [blame] | 85 | // Recurse through the root node, searching for content nodes (any node whose |
| 86 | // role is in kContentRoles). Skip branches which begin with a node with role |
| 87 | // in kRolesToSkip. Once a content node is identified, add it to the vector |
| 88 | // |content_node_ids|, whose pointer is passed through the recursion. |
| 89 | void AddContentNodesToVector(const ui::AXNode* node, |
| 90 | std::vector<ui::AXNodeID>* content_node_ids) { |
Jacob Francis | e11099b | 2024-02-17 04:34:04 | [diff] [blame] | 91 | const auto& role = node->GetRole(); |
| 92 | if (base::Contains(kContentRoles, role)) { |
| 93 | // TODO(1464340): Remove when flag is no longer necessary. Skip these roles |
| 94 | // if the flag is not enabled. |
| 95 | if (!features::IsReadAnythingImagesViaAlgorithmEnabled() && |
| 96 | (role == ax::mojom::Role::kFigcaption || |
| 97 | role == ax::mojom::Role::kImage)) { |
| 98 | return; |
| 99 | } |
Abigail Klein | 277fe7c2 | 2022-04-11 21:38:45 | [diff] [blame] | 100 | content_node_ids->emplace_back(node->id()); |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 101 | return; |
| 102 | } |
Abigail Klein | 277fe7c2 | 2022-04-11 21:38:45 | [diff] [blame] | 103 | if (base::Contains(kRolesToSkip, node->GetRole())) |
| 104 | return; |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 105 | for (auto iter = node->UnignoredChildrenBegin(); |
| 106 | iter != node->UnignoredChildrenEnd(); ++iter) { |
Abigail Klein | 277fe7c2 | 2022-04-11 21:38:45 | [diff] [blame] | 107 | AddContentNodesToVector(iter.get(), content_node_ids); |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 108 | } |
| 109 | } |
| 110 | |
| 111 | } // namespace |
| 112 | |
Abigail Klein | 1ed2224 | 2022-12-16 16:48:43 | [diff] [blame] | 113 | AXTreeDistiller::AXTreeDistiller( |
Abigail Klein | 1ed2224 | 2022-12-16 16:48:43 | [diff] [blame] | 114 | OnAXTreeDistilledCallback on_ax_tree_distilled_callback) |
Jocelyn Tran | 2696ed2 | 2024-01-17 16:47:14 | [diff] [blame] | 115 | : on_ax_tree_distilled_callback_(on_ax_tree_distilled_callback) { |
Jocelyn Tran | dab4ca60 | 2023-06-05 18:32:30 | [diff] [blame] | 116 | // TODO(crbug.com/1450930): Use a global ukm recorder instance instead. |
| 117 | mojo::Remote<ukm::mojom::UkmRecorderFactory> factory; |
| 118 | content::RenderThread::Get()->BindHostReceiver( |
| 119 | factory.BindNewPipeAndPassReceiver()); |
| 120 | ukm_recorder_ = ukm::MojoUkmRecorder::Create(*factory); |
| 121 | } |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 122 | |
| 123 | AXTreeDistiller::~AXTreeDistiller() = default; |
| 124 | |
Abigail Klein | fdeb3d7 | 2023-01-19 15:20:19 | [diff] [blame] | 125 | void AXTreeDistiller::Distill(const ui::AXTree& tree, |
Abigail Klein | dfdde35 | 2023-01-27 21:03:10 | [diff] [blame] | 126 | const ui::AXTreeUpdate& snapshot, |
Jocelyn Tran | dab4ca60 | 2023-06-05 18:32:30 | [diff] [blame] | 127 | const ukm::SourceId ukm_source_id) { |
Jocelyn Tran | dab4ca60 | 2023-06-05 18:32:30 | [diff] [blame] | 128 | base::TimeTicks start_time = base::TimeTicks::Now(); |
Jacob Francis | 1ba3a62 | 2023-11-21 22:18:21 | [diff] [blame] | 129 | |
Kristi Saney | 4e5438b | 2023-05-16 21:00:58 | [diff] [blame] | 130 | std::vector<ui::AXNodeID> content_node_ids; |
Jacob Francis | 1ba3a62 | 2023-11-21 22:18:21 | [diff] [blame] | 131 | if (features::IsReadAnythingWithAlgorithmEnabled()) { |
| 132 | // Try with the algorithm first. |
| 133 | DistillViaAlgorithm(tree, ukm_source_id, &content_node_ids); |
| 134 | } |
Kristi Saney | 4e5438b | 2023-05-16 21:00:58 | [diff] [blame] | 135 | |
Abigail Klein | 59872624 | 2023-02-22 18:44:14 | [diff] [blame] | 136 | // If Read Anything with Screen 2x is enabled and the main content extractor |
Kristi Saney | 4e5438b | 2023-05-16 21:00:58 | [diff] [blame] | 137 | // is bound, kick off Screen 2x run, which distills the AXTree in the |
| 138 | // utility process using ML. |
Abigail Klein | 59872624 | 2023-02-22 18:44:14 | [diff] [blame] | 139 | if (features::IsReadAnythingWithScreen2xEnabled() && |
| 140 | main_content_extractor_.is_bound()) { |
Jocelyn Tran | dab4ca60 | 2023-06-05 18:32:30 | [diff] [blame] | 141 | DistillViaScreen2x(tree, snapshot, ukm_source_id, start_time, |
| 142 | &content_node_ids); |
Abigail Klein | af1b516 | 2022-12-01 01:40:40 | [diff] [blame] | 143 | return; |
| 144 | } |
Abigail Klein | af1b516 | 2022-12-01 01:40:40 | [diff] [blame] | 145 | |
Kristi Saney | 4e5438b | 2023-05-16 21:00:58 | [diff] [blame] | 146 | // Ensure we still callback if Screen2x is not available. |
| 147 | on_ax_tree_distilled_callback_.Run(tree.GetAXTreeID(), content_node_ids); |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 148 | } |
| 149 | |
Kristi Saney | 4e5438b | 2023-05-16 21:00:58 | [diff] [blame] | 150 | void AXTreeDistiller::DistillViaAlgorithm( |
| 151 | const ui::AXTree& tree, |
Jocelyn Tran | dab4ca60 | 2023-06-05 18:32:30 | [diff] [blame] | 152 | const ukm::SourceId ukm_source_id, |
Kristi Saney | 4e5438b | 2023-05-16 21:00:58 | [diff] [blame] | 153 | std::vector<ui::AXNodeID>* content_node_ids) { |
Jocelyn Tran | dab4ca60 | 2023-06-05 18:32:30 | [diff] [blame] | 154 | base::TimeTicks start_time = base::TimeTicks::Now(); |
Abigail Klein | 41d20e5 | 2022-11-28 19:24:06 | [diff] [blame] | 155 | std::vector<const ui::AXNode*> content_root_nodes; |
Abigail Klein | fdeb3d7 | 2023-01-19 15:20:19 | [diff] [blame] | 156 | GetContentRootNodes(tree.root(), &content_root_nodes); |
Abigail Klein | 41d20e5 | 2022-11-28 19:24:06 | [diff] [blame] | 157 | for (const ui::AXNode* content_root_node : content_root_nodes) { |
Kristi Saney | 4e5438b | 2023-05-16 21:00:58 | [diff] [blame] | 158 | AddContentNodesToVector(content_root_node, content_node_ids); |
Abigail Klein | 41d20e5 | 2022-11-28 19:24:06 | [diff] [blame] | 159 | } |
Jocelyn Tran | dab4ca60 | 2023-06-05 18:32:30 | [diff] [blame] | 160 | RecordRulesMetrics(ukm_source_id, base::TimeTicks::Now() - start_time, |
| 161 | !content_node_ids->empty()); |
| 162 | } |
| 163 | |
| 164 | void AXTreeDistiller::RecordRulesMetrics(ukm::SourceId ukm_source_id, |
| 165 | base::TimeDelta elapsed_time, |
| 166 | bool success) { |
| 167 | if (success) { |
| 168 | base::UmaHistogramTimes( |
| 169 | "Accessibility.ReadAnything.RulesDistillationTime.Success", |
| 170 | elapsed_time); |
| 171 | ukm::builders::Accessibility_ReadAnything(ukm_source_id) |
| 172 | .SetRulesDistillationTime_Success(elapsed_time.InMilliseconds()) |
| 173 | .Record(ukm_recorder_.get()); |
| 174 | } else { |
| 175 | base::UmaHistogramTimes( |
| 176 | "Accessibility.ReadAnything.RulesDistillationTime.Failure", |
| 177 | elapsed_time); |
| 178 | ukm::builders::Accessibility_ReadAnything(ukm_source_id) |
| 179 | .SetRulesDistillationTime_Failure(elapsed_time.InMilliseconds()) |
| 180 | .Record(ukm_recorder_.get()); |
| 181 | } |
Abigail Klein | a303f27 | 2022-03-31 22:32:57 | [diff] [blame] | 182 | } |
| 183 | |
Abigail Klein | ad097706 | 2023-05-19 18:21:08 | [diff] [blame] | 184 | void AXTreeDistiller::DistillViaScreen2x( |
| 185 | const ui::AXTree& tree, |
| 186 | const ui::AXTreeUpdate& snapshot, |
Jocelyn Tran | dab4ca60 | 2023-06-05 18:32:30 | [diff] [blame] | 187 | const ukm::SourceId ukm_source_id, |
| 188 | base::TimeTicks start_time, |
Abigail Klein | ad097706 | 2023-05-19 18:21:08 | [diff] [blame] | 189 | std::vector<ui::AXNodeID>* content_node_ids_algorithm) { |
Ramin Halavati | eddadb6 | 2022-05-04 17:29:49 | [diff] [blame] | 190 | DCHECK(main_content_extractor_.is_bound()); |
Abigail Klein | ad097706 | 2023-05-19 18:21:08 | [diff] [blame] | 191 | // Make a copy of |content_node_ids_algorithm| rather than sending a pointer. |
Ramin Halavati | eddadb6 | 2022-05-04 17:29:49 | [diff] [blame] | 192 | main_content_extractor_->ExtractMainContent( |
Abigail Klein | dfdde35 | 2023-01-27 21:03:10 | [diff] [blame] | 193 | snapshot, ukm_source_id, |
Abigail Klein | 34db57e | 2023-05-01 22:37:37 | [diff] [blame] | 194 | base::BindOnce(&AXTreeDistiller::ProcessScreen2xResult, |
Abigail Klein | ad097706 | 2023-05-19 18:21:08 | [diff] [blame] | 195 | weak_ptr_factory_.GetWeakPtr(), tree.GetAXTreeID(), |
Jocelyn Tran | dab4ca60 | 2023-06-05 18:32:30 | [diff] [blame] | 196 | ukm_source_id, start_time, *content_node_ids_algorithm)); |
Ramin Halavati | eddadb6 | 2022-05-04 17:29:49 | [diff] [blame] | 197 | } |
| 198 | |
| 199 | void AXTreeDistiller::ProcessScreen2xResult( |
Abigail Klein | 34db57e | 2023-05-01 22:37:37 | [diff] [blame] | 200 | const ui::AXTreeID& tree_id, |
Jocelyn Tran | dab4ca60 | 2023-06-05 18:32:30 | [diff] [blame] | 201 | const ukm::SourceId ukm_source_id, |
| 202 | base::TimeTicks start_time, |
Abigail Klein | ad097706 | 2023-05-19 18:21:08 | [diff] [blame] | 203 | std::vector<ui::AXNodeID> content_node_ids_algorithm, |
| 204 | const std::vector<ui::AXNodeID>& content_node_ids_screen2x) { |
| 205 | // Merge the results from the algorithm and from screen2x. |
| 206 | for (ui::AXNodeID content_node_id_screen2x : content_node_ids_screen2x) { |
| 207 | if (!base::Contains(content_node_ids_algorithm, content_node_id_screen2x)) { |
| 208 | content_node_ids_algorithm.push_back(content_node_id_screen2x); |
| 209 | } |
| 210 | } |
Jocelyn Tran | dab4ca60 | 2023-06-05 18:32:30 | [diff] [blame] | 211 | RecordMergedMetrics(ukm_source_id, base::TimeTicks::Now() - start_time, |
| 212 | !content_node_ids_algorithm.empty()); |
Abigail Klein | ad097706 | 2023-05-19 18:21:08 | [diff] [blame] | 213 | on_ax_tree_distilled_callback_.Run(tree_id, content_node_ids_algorithm); |
Abigail Klein | af1b516 | 2022-12-01 01:40:40 | [diff] [blame] | 214 | |
Kristi Saney | 84a7864 | 2023-05-10 18:41:54 | [diff] [blame] | 215 | // TODO(crbug.com/1266555): If no content nodes were identified, and |
Abigail Klein | d5fef9a | 2022-11-07 15:03:50 | [diff] [blame] | 216 | // there is a selection, try sending Screen2x a partial tree just containing |
| 217 | // the selected nodes. |
Ramin Halavati | eddadb6 | 2022-05-04 17:29:49 | [diff] [blame] | 218 | } |
Abigail Klein | 7849d1a | 2023-01-19 15:20:19 | [diff] [blame] | 219 | |
Jocelyn Tran | 2696ed2 | 2024-01-17 16:47:14 | [diff] [blame] | 220 | void AXTreeDistiller::ScreenAIServiceReady(content::RenderFrame* render_frame) { |
| 221 | if (main_content_extractor_.is_bound() || !render_frame) { |
Abigail Klein | 59872624 | 2023-02-22 18:44:14 | [diff] [blame] | 222 | return; |
| 223 | } |
Jocelyn Tran | 2696ed2 | 2024-01-17 16:47:14 | [diff] [blame] | 224 | render_frame->GetBrowserInterfaceBroker()->GetInterface( |
Abigail Klein | 59872624 | 2023-02-22 18:44:14 | [diff] [blame] | 225 | main_content_extractor_.BindNewPipeAndPassReceiver()); |
| 226 | main_content_extractor_.set_disconnect_handler( |
| 227 | base::BindOnce(&AXTreeDistiller::OnMainContentExtractorDisconnected, |
| 228 | weak_ptr_factory_.GetWeakPtr())); |
| 229 | } |
| 230 | |
Abigail Klein | 7849d1a | 2023-01-19 15:20:19 | [diff] [blame] | 231 | void AXTreeDistiller::OnMainContentExtractorDisconnected() { |
Abigail Klein | 3bd99e0 | 2023-01-24 00:29:18 | [diff] [blame] | 232 | on_ax_tree_distilled_callback_.Run(ui::AXTreeIDUnknown(), |
| 233 | std::vector<ui::AXNodeID>()); |
Abigail Klein | 7849d1a | 2023-01-19 15:20:19 | [diff] [blame] | 234 | } |
Jocelyn Tran | dab4ca60 | 2023-06-05 18:32:30 | [diff] [blame] | 235 | |
| 236 | void AXTreeDistiller::RecordMergedMetrics(ukm::SourceId ukm_source_id, |
| 237 | base::TimeDelta elapsed_time, |
| 238 | bool success) { |
| 239 | if (success) { |
| 240 | base::UmaHistogramTimes( |
| 241 | "Accessibility.ReadAnything.MergedDistillationTime.Success", |
| 242 | elapsed_time); |
| 243 | ukm::builders::Accessibility_ReadAnything(ukm_source_id) |
| 244 | .SetMergedDistillationTime_Success(elapsed_time.InMilliseconds()) |
| 245 | .Record(ukm_recorder_.get()); |
| 246 | } else { |
| 247 | base::UmaHistogramTimes( |
| 248 | "Accessibility.ReadAnything.MergedDistillationTime.Failure", |
| 249 | elapsed_time); |
| 250 | ukm::builders::Accessibility_ReadAnything(ukm_source_id) |
| 251 | .SetMergedDistillationTime_Failure(elapsed_time.InMilliseconds()) |
| 252 | .Record(ukm_recorder_.get()); |
| 253 | } |
| 254 | } |