| Scott Violet | 26f9aa6 | 2023-10-25 23:29:59 | [diff] [blame^] | 1 | // Copyright 2023 The Chromium Authors |
| 2 | // Use of this source code is governed by a BSD-style license that can be |
| 3 | // found in the LICENSE file. |
| 4 | |
| 5 | #ifndef CHROME_BROWSER_CONTENT_EXTRACTION_INNER_TEXT_H_ |
| 6 | #define CHROME_BROWSER_CONTENT_EXTRACTION_INNER_TEXT_H_ |
| 7 | |
| 8 | #include <memory> |
| 9 | #include <string> |
| 10 | |
| 11 | #include "base/functional/callback_forward.h" |
| 12 | #include "third_party/abseil-cpp/absl/types/optional.h" |
| 13 | #include "third_party/blink/public/mojom/content_extraction/inner_text.mojom-forward.h" |
| 14 | |
| 15 | namespace content { |
| 16 | class RenderFrameHost; |
| 17 | } |
| 18 | |
| 19 | namespace content_extraction { |
| 20 | |
| 21 | struct InnerTextResult { |
| 22 | // The combined inner-text. See comments in GetInnerText() for details. |
| 23 | std::string inner_text; |
| 24 | |
| 25 | // Offset of the supplied node in `inner_text`. Only set if a node-id is |
| 26 | // supplied to GetInnerText() and a matching node was found. |
| 27 | absl::optional<unsigned> node_offset; |
| 28 | }; |
| 29 | |
| 30 | using InnerTextCallback = |
| 31 | base::OnceCallback<void(std::unique_ptr<InnerTextResult> result)>; |
| 32 | |
| 33 | // Requests the inner-text for the specified `host` as well as all local |
| 34 | // same-origin iframes. The returned inner-text contains the combined inner-text |
| 35 | // of all suitable iframes. Only the inner-text of the first body or frameset |
| 36 | // is used. The text is combined as the iframes are encountered. For example, |
| 37 | // the following structure: |
| 38 | // <body> |
| 39 | // A <iframe src="a.html></iframe> |
| 40 | // B <iframe src="b.html></iframe> |
| 41 | // </body> |
| 42 | // results in the string "A <a-inner-text> B <b-inner-test>" where a-inner-text |
| 43 | // and b-inner-text are replaced with the inner-text of a.html and b.html. |
| 44 | // |
| 45 | // You may also supply a DomNodeId. The callback will be called with the offset |
| 46 | // of the start of the node in the text. |
| 47 | // |
| 48 | // If querying the inner-text fails (renderer crash, or page shutdown during |
| 49 | // request) then null is supplied to the callback. |
| 50 | // |
| 51 | // NOTE: This function services the request as soon as called, it does not wait |
| 52 | // for the page to finish loading. |
| 53 | void GetInnerText(content::RenderFrameHost& host, |
| 54 | absl::optional<int> node_id, |
| 55 | InnerTextCallback callback); |
| 56 | |
| 57 | // Exposed for testing. |
| 58 | namespace internal { |
| 59 | |
| 60 | // Returns true if `frame` frame valid. |
| 61 | bool IsInnerTextFrameValid(const blink::mojom::InnerTextFramePtr& frame); |
| 62 | |
| 63 | // Converts `frame` into an InnerTextResult. |
| 64 | std::unique_ptr<InnerTextResult> CreateInnerTextResult( |
| 65 | const blink::mojom::InnerTextFrame& frame); |
| 66 | |
| 67 | } // namespace internal |
| 68 | } // namespace content_extraction |
| 69 | |
| 70 | #endif // CHROME_BROWSER_CONTENT_EXTRACTION_INNER_TEXT_H_ |
