(feat): Add “Copy text from screen” OCR action - #333
Merged
Conversation
Adds `Copy text from screen`: drag a region of the desktop image Wayscriber is already displaying, recognize its text with a local Tesseract, and put the result on the clipboard. It is a one-shot action rather than a drawing tool, so the active tool, the drawing history, boards, and sessions are all untouched. The region is cropped from the frozen or zoomed capture the renderer is showing, never from annotations or Wayscriber chrome, and a freeze OCR creates is a freeze OCR releases — a user-owned freeze or an active zoom survives untouched. Recognition runs off the event loop through a capacity-one controller; a second request reports busy rather than queueing a region the user has moved on from. Recognized text never reaches application state, a log line, or a `Debug` rendering: the worker owns encoding, the engine call, and the clipboard write in one stack frame, and only a character count comes back. Two shared seams come out of this rather than being copied: - `state/screen_image.rs` now owns displayed-source resolution, the checked logical-to-image mapping, and the rectangle crop, for the eyedropper, the canvas renderer, and OCR alike. - `clipboard_text.rs` holds the generic `wl-copy`/`wl-paste` commands that used to live in the color-picker module. Input ownership is explicit throughout. A region records the device that started it, so a hovering pen cannot drag a mouse-owned selection and a stray release cannot submit one. Both screen-region modals now share one keyboard boundary (engaged, including the capture wait) and one pen boundary (active, once the selector is on screen), and a pen already down when a modal opens is disowned until it lifts so it cannot resize the drawing tool behind the selector. Discoverability is opt-in: the action is in the command palette and the keybinding configurator with no default shortcut, since `O` is already the orange quick color, and the `top.utility.ocr` toolbar button ships hidden like Screenshot. `capture.ocr_languages` validates the Tesseract language argument so an authored value can never carry a path, an option, or shell syntax into the argument vector; the engine is invoked through the process broker under a new `HelperKind::Tesseract` allowlist with an explicit argument vector and a self-deleting temporary PNG. Tesseract is packaged as an optional dependency (optdepends on Arch, recommends on deb/rpm) because the feature is opt-in at every level. Also fixes a latent bug in `ToolbarItemsConfig::set_visibility_setting`, which rewrote an item to the end of its list even when the setting was unchanged. It was invisible while only one item was hidden by default; a second one made a batch reset report "changed" forever.
The gate built and tested that configuration but never linted it, so code reachable only behind an optional feature could leave its callers dead without it and still pass. Building alone does not promote that to an error. Caught exactly that during OCR work: a predicate used only from `tablet-input` code was dead in the default build, and only a manual clippy run found it. Passes clean today, so it is a free tripwire.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Add an opt-in screen-region OCR action that recognizes text locally with Tesseract and copies it to the clipboard.
The implementation also extracts shared screen-image geometry used by OCR and the eyedropper, and prevents callback-thread launchers from stalling behind long broker operations.