devmobasa · devmobasa · Aug 12, 2026 · Aug 11, 2026 · Aug 11, 2026 · Aug 12, 2026
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -42,6 +42,8 @@ serde = { version = "1.0", features = ["derive"] }
 serde_ignored = "0.1"
 schemars = { version = "1.2", features = ["derive"], optional = true }
 libc = "0.2"
+# Securely created, self-deleting temporary files for the OCR engine handoff.
+tempfile = "3"
 flate2 = "1.0"
 zune-jpeg = "0.5"
 # Grapheme/word segmentation for the text-editor caret and selection.

diff --git a/README.md b/README.md
@@ -140,6 +140,7 @@ The v0.9.23+ prebuilt `wayscriber` packages require glibc 2.39 and GTK 4.12 —
 - Full-screen saves, active-window grabs, region capture
 - Copy to clipboard or save to file
 - Uses `grim`, `slurp`, `wl-clipboard` (installed automatically by deb/rpm/AUR packages; fallback: xdg-desktop-portal)
+- Copy text from screen (OCR): drag a region of the shown desktop and get its text on the clipboard (needs `tesseract`; no default shortcut)
 
 ### Sessions and persistence
 - Session persistence is enabled by default for boards, undo/redo history, and tool state
@@ -329,6 +330,9 @@ If you are on a stable NixOS channel and want the newest release, use [the proje
     grim
     slurp
     wl-clipboard
+
+    # Optional: copy text from screen (OCR)
+    (tesseract.override { enableLanguages = [ "eng" ]; })
   ];
 }
 ```
@@ -538,6 +542,19 @@ sudo apt-get install wl-clipboard grim slurp   # Debian/Ubuntu
 sudo dnf install wl-clipboard grim slurp       # Fedora
 ```
 
+### Copy text from screen (OCR)
+
+`Copy text from screen` recognizes the text in a dragged screen region and copies
+it to the clipboard. It is optional: the action has no default shortcut and its
+toolbar button is hidden until you turn it on. Install Tesseract and the language
+data you configure in `[capture].ocr_languages` (default `eng`):
+
+```bash
+sudo pacman -S tesseract tesseract-data-eng       # Arch/Omarchy
+sudo apt-get install tesseract-ocr                # Debian/Ubuntu
+sudo dnf install tesseract tesseract-langpack-eng # Fedora
+```
+
 ---
 
 ## First launch

diff --git a/config.example.toml b/config.example.toml
@@ -275,6 +275,10 @@ export_all_boards_pdf_file = []
 # Open the most recent capture folder
 open_capture_folder = ["Ctrl+Alt+O"]
 
+# Select a screen region and copy the text recognized in it (needs Tesseract).
+# Unbound by default: "O" is already the orange quick color.
+copy_text_from_screen = []
+
 # Toggle frozen mode
 toggle_frozen_mode = ["Ctrl+Shift+F"]
 
@@ -628,6 +632,9 @@ rebind_modifier = "ctrl_shift"
 [ui.toolbar.items]
 # Checked in the configurator means shown; IDs listed here are hidden.
 # The screenshot toolbar button is hidden by default.
+# Copy text (OCR) is hidden by a built-in baseline instead of by this list,
+# so upgrading a config written before it existed cannot reveal it; enable it
+# through toolbar customization, which records an entry in `shown` below.
 # Section-level ids (side.group.*) act as explicit overrides that beat the
 # layout-mode baseline and survive mode switches.
 hidden = [
@@ -660,6 +667,7 @@ top_controls = [
   "top.utility.text",
   "top.utility.sticky-note",
   "top.utility.screenshot",
+  "top.utility.ocr",
   "top.utility.clear-canvas",
   "top.utility.highlight",
 ]
@@ -1384,6 +1392,11 @@ copy_to_clipboard = true
 # Use --no-exit-after-capture to keep the overlay open for a run.
 exit_after_capture = false
 
+# Languages used by "Copy text from screen" (OCR), in Tesseract's
+# plus-separated form, for example "eng" or "eng+deu". The matching Tesseract
+# language packages must be installed. OCR also needs `enabled = true` above.
+ocr_languages = "eng"
+
 # ═══════════════════════════════════════════════════════════════════════════════
 # EXPORT SETTINGS
 # ═══════════════════════════════════════════════════════════════════════════════

diff --git a/configurator/src/app/pages/capture.rs b/configurator/src/app/pages/capture.rs
@@ -15,6 +15,8 @@ use crate::models::{
     PdfTransparentBackgroundOption, TabId, TextField, ToggleField,
 };
 
+use wayscriber::config::validate_ocr_languages;
+
 use super::super::search::SearchArea;
 use super::super::state::ConfiguratorApp;
 use super::color_rows::{ResolvedColor, color_row};
@@ -56,6 +58,18 @@ pub(super) fn build(sender: &ComponentSender<ConfiguratorApp>) -> BuiltPage {
             "",
             |app| app.draft.capture_exit_after,
             |value| Message::ToggleChanged(ToggleField::CaptureExitAfter, value),
+        )
+        .entry_row_validated(
+            "OCR languages (e.g. eng, eng+deu \u{2014} needs the matching Tesseract packages)",
+            |app| app.draft.capture_ocr_languages.clone(),
+            |value| Message::TextChanged(TextField::CaptureOcrLanguages, value),
+            |app| {
+                validate_ocr_languages(&app.draft.capture_ocr_languages)
+                    .err()
+                    .map(|reason| {
+                        format!("{reason}. Use Tesseract codes joined by '+', such as eng+deu.")
+                    })
+            },
         );
 
     page.group_in_area("PDF export", SearchArea::CapturePdf)

diff --git a/configurator/src/models/config/draft/from_config.rs b/configurator/src/models/config/draft/from_config.rs
@@ -247,6 +247,7 @@ impl ConfigDraft {
             capture_format: config.capture.format.clone(),
             capture_copy_to_clipboard: config.capture.copy_to_clipboard,
             capture_exit_after: config.capture.exit_after_capture,
+            capture_ocr_languages: config.capture.ocr_languages.clone(),
             export_pdf_filename_template: config
                 .export
                 .pdf

diff --git a/configurator/src/models/config/draft/mod.rs b/configurator/src/models/config/draft/mod.rs
@@ -180,6 +180,7 @@ pub struct ConfigDraft {
     pub capture_format: String,
     pub capture_copy_to_clipboard: bool,
     pub capture_exit_after: bool,
+    pub capture_ocr_languages: String,
     pub export_pdf_filename_template: String,
     pub export_pdf_all_boards_filename_template: String,
     pub export_pdf_page_size: PdfPageSizeOption,

diff --git a/configurator/src/models/config/setters.rs b/configurator/src/models/config/setters.rs
@@ -366,6 +366,7 @@ impl ConfigDraft {
             TextField::CaptureSaveDirectory => self.capture_save_directory = value,
             TextField::CaptureFilename => self.capture_filename_template = value,
             TextField::CaptureFormat => self.capture_format = value,
+            TextField::CaptureOcrLanguages => self.capture_ocr_languages = value,
             TextField::ExportPdfFilenameTemplate => self.export_pdf_filename_template = value,
             TextField::ExportPdfAllBoardsFilenameTemplate => {
                 self.export_pdf_all_boards_filename_template = value

diff --git a/configurator/src/models/config/to_config/capture.rs b/configurator/src/models/config/to_config/capture.rs
@@ -1,14 +1,21 @@
 use super::super::draft::ConfigDraft;
 use crate::models::error::FormError;
-use wayscriber::config::Config;
+use wayscriber::config::{Config, validate_ocr_languages};
 
 impl ConfigDraft {
-    pub(super) fn apply_capture(&self, config: &mut Config, _errors: &mut Vec<FormError>) {
+    pub(super) fn apply_capture(&self, config: &mut Config, errors: &mut Vec<FormError>) {
         config.capture.enabled = self.capture_enabled;
         config.capture.save_directory = self.capture_save_directory.clone();
         config.capture.filename_template = self.capture_filename_template.clone();
         config.capture.format = self.capture_format.clone();
         config.capture.copy_to_clipboard = self.capture_copy_to_clipboard;
         config.capture.exit_after_capture = self.capture_exit_after;
+        match validate_ocr_languages(&self.capture_ocr_languages) {
+            Ok(languages) => config.capture.ocr_languages = languages,
+            Err(reason) => errors.push(FormError::new(
+                "capture.ocr_languages",
+                format!("OCR languages: {reason}."),
+            )),
+        }
     }
 }
diff --git a/configurator/src/models/fields/toggles.rs b/configurator/src/models/fields/toggles.rs
@@ -136,6 +136,7 @@ pub enum TextField {
     CaptureSaveDirectory,
     CaptureFilename,
     CaptureFormat,
+    CaptureOcrLanguages,
     ExportPdfFilenameTemplate,
     ExportPdfAllBoardsFilenameTemplate,
     ExportPdfCustomWidth,

diff --git a/configurator/src/models/keybindings/field/config/read.rs b/configurator/src/models/keybindings/field/config/read.rs
@@ -133,6 +133,7 @@ impl KeybindingField {
             Self::ExportBoardPdfFile => &config.capture.export_board_pdf_file,
             Self::ExportAllBoardsPdfFile => &config.capture.export_all_boards_pdf_file,
             Self::OpenCaptureFolder => &config.capture.open_capture_folder,
+            Self::CopyTextFromScreen => &config.capture.copy_text_from_screen,
             Self::ToggleFrozenMode => &config.zoom.toggle_frozen_mode,
             Self::ZoomIn => &config.zoom.zoom_in,
             Self::ZoomOut => &config.zoom.zoom_out,

diff --git a/configurator/src/models/keybindings/field/config/write.rs b/configurator/src/models/keybindings/field/config/write.rs
@@ -136,6 +136,7 @@ impl KeybindingField {
             Self::ExportBoardPdfFile => config.capture.export_board_pdf_file = value,
             Self::ExportAllBoardsPdfFile => config.capture.export_all_boards_pdf_file = value,
             Self::OpenCaptureFolder => config.capture.open_capture_folder = value,
+            Self::CopyTextFromScreen => config.capture.copy_text_from_screen = value,
             Self::ToggleFrozenMode => config.zoom.toggle_frozen_mode = value,
             Self::ZoomIn => config.zoom.zoom_in = value,
             Self::ZoomOut => config.zoom.zoom_out = value,

diff --git a/configurator/src/models/keybindings/field/labels.rs b/configurator/src/models/keybindings/field/labels.rs
@@ -128,6 +128,7 @@ impl KeybindingField {
             Self::ExportBoardPdfFile => "Export board: PDF file",
             Self::ExportAllBoardsPdfFile => "Export all boards: PDF file",
             Self::OpenCaptureFolder => "Open capture folder",
+            Self::CopyTextFromScreen => "Copy text from screen (OCR)",
             Self::ToggleFrozenMode => "Toggle freeze",
             Self::ZoomIn => "Zoom in",
             Self::ZoomOut => "Zoom out",
@@ -281,6 +282,7 @@ impl KeybindingField {
             Self::ExportBoardPdfFile => "export_board_pdf_file",
             Self::ExportAllBoardsPdfFile => "export_all_boards_pdf_file",
             Self::OpenCaptureFolder => "open_capture_folder",
+            Self::CopyTextFromScreen => "copy_text_from_screen",
             Self::ToggleFrozenMode => "toggle_frozen_mode",
             Self::ZoomIn => "zoom_in",
             Self::ZoomOut => "zoom_out",

diff --git a/configurator/src/models/keybindings/field/list.rs b/configurator/src/models/keybindings/field/list.rs
@@ -128,6 +128,7 @@ impl KeybindingField {
             Self::ExportBoardPdfFile,
             Self::ExportAllBoardsPdfFile,
             Self::OpenCaptureFolder,
+            Self::CopyTextFromScreen,
             Self::ToggleFrozenMode,
             Self::ZoomIn,
             Self::ZoomOut,

diff --git a/configurator/src/models/keybindings/field/mod.rs b/configurator/src/models/keybindings/field/mod.rs
@@ -113,6 +113,7 @@ pub enum KeybindingField {
     ExportBoardPdfFile,
     ExportAllBoardsPdfFile,
     OpenCaptureFolder,
+    CopyTextFromScreen,
     ToggleFrozenMode,
     ZoomIn,
     ZoomOut,

diff --git a/configurator/src/models/keybindings/field/tab.rs b/configurator/src/models/keybindings/field/tab.rs
@@ -129,6 +129,7 @@ impl KeybindingField {
             | Self::ExportBoardPdfFile
             | Self::ExportAllBoardsPdfFile
             | Self::OpenCaptureFolder
+            | Self::CopyTextFromScreen
             | Self::ToggleFrozenMode
             | Self::ZoomIn
             | Self::ZoomOut

diff --git a/docs/CONFIG.md b/docs/CONFIG.md
@@ -1380,12 +1380,47 @@ copy_to_clipboard = true
 # When false, clipboard-only captures still auto-exit by default.
 # Use --no-exit-after-capture to keep the overlay open for a run.
 exit_after_capture = false
+
+# Languages for "Copy text from screen" (OCR), in Tesseract's plus-separated
+# form. The matching Tesseract language packages must be installed.
+ocr_languages = "eng"
 ```
 
 **Tips:**
 - Set `copy_to_clipboard = false` if you prefer file-only captures.
 - Clipboard-only shortcuts ignore the save directory automatically.
 - `wl-clipboard`, `grim`, and `slurp` are installed automatically by deb/rpm/AUR packages. For source/tarball installs, add them manually; otherwise wayscriber falls back to `xdg-desktop-portal`.
+
+#### Copy text from screen (OCR)
+
+`Copy text from screen` selects a region of the desktop image wayscriber is
+already showing, recognizes the text in it with a local Tesseract, and copies
+the result to the clipboard. It reads the underlying screen capture only —
+never your annotations, the toolbars, or any other wayscriber chrome — and it
+does not change the active tool, the drawing history, or the board.
+
+- The action is `copy_text_from_screen`. It has **no default shortcut**, because
+  `O` is already the orange quick color; bind one in the configurator or in
+  `[keybindings.capture]`.
+- It is also in the command palette (search for "OCR"), and as an optional top
+  toolbar button (`top.utility.ocr`), hidden by default like Screenshot.
+- `ocr_languages` accepts one language or several joined with `+`
+  (`eng`, `eng+deu`). Only letters, digits, `_` and `-` are accepted; anything
+  else falls back to `eng`.
+- OCR obeys `enabled` above: with capture disabled, the action reports that and
+  does nothing.
+- On a solid whiteboard or blackboard with no visible screen capture, OCR
+  refuses rather than reading the board.
+
+**Requirements:** the `tesseract` command plus the language data for every code
+in `ocr_languages`, and `wl-copy` (from `wl-clipboard`) for the clipboard write.
+
+| Distribution | Packages |
+| --- | --- |
+| Arch / Omarchy | `tesseract`, `tesseract-data-eng` |
+| Debian / Ubuntu | `tesseract-ocr` (depends on the English data) |
+| Fedora | `tesseract`, `tesseract-langpack-eng` |
+| Nix | the `tesseract` package/wrapper with `eng` enabled |
 - Use `--exit-after-capture` / `--no-exit-after-capture` to override exit behavior per run.
 
 ### `[export.pdf]` - PDF Export
@@ -1839,6 +1874,10 @@ export_all_boards_pdf_file = []
 # Open the most recent capture folder
 open_capture_folder = ["Ctrl+Alt+O"]
 
+# Select a screen region and copy the text recognized in it (needs Tesseract).
+# Unbound by default: "O" is already the orange quick color.
+copy_text_from_screen = []
+
 # Toggle frozen mode
 toggle_frozen_mode = ["Ctrl+Shift+F"]
 

diff --git a/docs/codebase-overview.md b/docs/codebase-overview.md
@@ -470,6 +470,7 @@ Notifications are sent via `notification::send_notification_async`, keeping all
 | `src/config/` | Config parsing, defaults, keybinding map. |
 | `src/runtime_ui_state/` | Generated UI preference wire format, seed registry, guarded persistence, reset, and recovery state machines. |
 | `src/session/` | Configured and named session persistence, snapshots, sidecars, locks, and catalog metadata. |
+| `src/ocr/` | Screen text recognition: capacity-one controller, Tesseract adapter, typed outcomes. |
 | `src/notification.rs` | Desktop notifications for capture results and the update notice. |
 | `src/update_check/` | Opt-outable "a newer release exists" check: version ordering, manifest trust rules, `curl`/`wget` transport, cache. Installs nothing. |
 | `src/about_window/` | Standalone About dialog: content/layout/interaction split, update card, diagnostics copy. |

diff --git a/packaging/.SRCINFO b/packaging/.SRCINFO
@@ -19,6 +19,8 @@ pkgbase = wayscriber
 	depends = wl-clipboard
 	depends = grim
 	depends = slurp
+	optdepends = tesseract: copy text from screen (OCR)
+	optdepends = tesseract-data-eng: English language data for OCR
 	source = wayscriber-0.9.23.tar.gz::https://github.com/devmobasa/wayscriber/archive/refs/tags/v0.9.23.tar.gz
 	sha256sums = SKIP
 

diff --git a/packaging/PKGBUILD b/packaging/PKGBUILD
@@ -20,6 +20,10 @@ depends=(
     'grim'
     'slurp'
 )
+optdepends=(
+    'tesseract: copy text from screen (OCR)'
+    'tesseract-data-eng: English language data for OCR'
+)
 makedepends=(
     'cargo'
 )

diff --git a/packaging/package.wayscriber.yaml b/packaging/package.wayscriber.yaml
@@ -122,6 +122,10 @@ overrides:
       - wl-clipboard
       - grim
       - slurp
+    # Screen text recognition is opt-in (no default shortcut, hidden toolbar
+    # button), so the engine and its language data are not a hard dependency.
+    recommends:
+      - tesseract-ocr
   rpm:
     depends:
       # Keep the glibc floor aligned with MAX_GLIBC_VERSION in tools/package.sh.
@@ -135,5 +139,8 @@ overrides:
       - wl-clipboard
       - grim
       - slurp
+    recommends:
+      - tesseract
+      - tesseract-langpack-eng
     rpm:
       arch: x86_64
diff --git a/src/AGENTS.md b/src/AGENTS.md
@@ -8,7 +8,7 @@
 - `src/main.rs` is a thin wrapper that returns `wayscriber::run_from_env()`.
 - `src/lib.rs` owns logging/error wiring and the canonical module graph; its public entry facade routes CLI outcomes through `src/app/` into `src/daemon/` or the active Wayland backend.
 - Reusable modules remain public for integration tests and the configurator, while runtime modules are private library implementation details.
-- Major domains are `domain`, `backend`, `input`, `draw`, `ui`, `capture`, `config`, `session`, `daemon`, `canvas_export`, `paths`, `toolbar_icons`, `render_profiles`, and `runtime_capabilities`.
+- Major domains are `domain`, `backend`, `input`, `draw`, `ui`, `capture`, `ocr`, `config`, `session`, `daemon`, `canvas_export`, `paths`, `toolbar_icons`, `render_profiles`, and `runtime_capabilities`.
 
 ## Invariants
 - Keep Wayland/runtime side effects in backend/about-window code; keep reusable data, validation, rendering helpers, and path logic in library modules.

diff --git a/src/backend/wayland/backend/event_loop/capture.rs b/src/backend/wayland/backend/event_loop/capture.rs
@@ -92,9 +92,11 @@ pub(super) fn handle_pending_actions(
     state.poll_hex_copy_completion();
     state.poll_text_copy_completion();
     state.poll_text_paste_completion();
+    state.poll_ocr_completion();
     state.poll_session_file_dialog_completion(qh);
     state.drain_clipboard_requests();
     state.handle_pending_eyedropper_toggle();
+    state.handle_pending_ocr_request();
     // Copy/paste-hex requests from the color picker popup's pointer release are
     // drained here: unlike the toolbar/key paths, that release has no other
     // drain site. An accepted quick-color recolor is queued from the same

diff --git a/src/backend/wayland/handlers/compositor.rs b/src/backend/wayland/handlers/compositor.rs
@@ -60,6 +60,7 @@ impl CompositorHandler for WaylandState {
         self.zoom
             .handle_resize(phys_w, phys_h, &mut self.input_state);
         self.cancel_eyedropper_if_source_missing();
+        self.cancel_ocr_if_source_missing();
         self.toolbar
             .maybe_update_scale(self.surface.current_output().as_ref(), scale);
         self.toolbar.mark_dirty();
@@ -170,6 +171,7 @@ impl CompositorHandler for WaylandState {
         self.zoom
             .handle_resize(phys_w, phys_h, &mut self.input_state);
         self.cancel_eyedropper_if_source_missing();
+        self.cancel_ocr_if_source_missing();
 
         // If freeze-on-start was requested, trigger it once the surface is configured and active.
         if self.pending_freeze_on_start() {

diff --git a/src/backend/wayland/handlers/keyboard/mod.rs b/src/backend/wayland/handlers/keyboard/mod.rs
@@ -117,6 +117,16 @@ impl KeyboardHandler for WaylandState {
         // Any fresh key press ends the previous auto-repeat; a repeatable one
         // re-arms it at the end of this handler.
         self.clear_key_repeat();
+        if self.input_state.ocr_is_engaged() {
+            // Every other shortcut is swallowed while the selector is up so a
+            // key cannot change the active tool mid-drag.
+            if matches!(key, Key::Escape)
+                || self.input_state.action_for_key(key) == Some(Action::CopyTextFromScreen)
+            {
+                self.cancel_ocr();
+            }
+            return;
+        }
         if self.input_state.eyedropper_is_engaged() {
             if matches!(key, Key::Escape)
                 || self.input_state.action_for_key(key) == Some(Action::PickScreenColor)