Selecting & copying

The selection plugin lets people select text with the mouse or a finger, and lets your code select the same text. Register interactionPlugin() and selectionPlugin(), and put <SelectionLayer> above the page:

Loading live preview…

<script lang="ts" module>
  import { DocumentGate, Viewer, type OpenInput } from '@embedpdf/svelte/runtime';
  import { RenderLayer, renderPlugin } from '@embedpdf/svelte/render';
  import { Stage, stagePlugin } from '@embedpdf/svelte/stage';
  import { interactionPlugin } from '@embedpdf/svelte/interaction';
  import { SelectionLayer, selectionPlugin } from '@embedpdf/svelte/selection';
  import { localEngine } from '@embedpdf/engine';
  import SelectTitle from './SelectTitle.svelte';

  import '../basic.css';

  const engine = localEngine();
  const plugins = [stagePlugin(), renderPlugin(), interactionPlugin(), selectionPlugin()];

  const ebook = async (): Promise<OpenInput> => {
    const response = await fetch('https://snippet.embedpdf.com/ebook.pdf');
    return { kind: 'bytes', id: 'ebook', bytes: new Uint8Array(await response.arrayBuffer()) };
  };
</script>

<Viewer {engine} {plugins} initialDocuments={[{ source: ebook }]}>
  <DocumentGate>
    {#snippet fallback()}
      <p class="loading">Loading…</p>
    {/snippet}
    <SelectTitle />
    <Stage class="stage">
      <RenderLayer />
      <SelectionLayer />
    </Stage>
  </DocumentGate>
</Viewer>

Try it: the cover’s title starts out selected from code. Drag over text, double-click a word, or triple-click a line. A selection can run from one page onto the next. On a touch screen, long-press a word and drag the handles.

interactionPlugin() is the input system every tool shares: the Stage passes the pointer to it, the selection plugin handles it, and <SelectionLayer> draws the result. So you can restyle or replace the highlight without touching how selecting works.

Selecting uses where the characters are on the page, not the text itself. Nothing reads the page’s text until your code asks for it, and a document can allow selecting while it forbids copying (see Permissions).

Select text from code#

useSelection() does from code what the pointer does, and the highlight follows:

Loading live preview…

<script lang="ts" module>
  import { DocumentGate, Viewer, type OpenInput } from '@embedpdf/svelte/runtime';
  import { RenderLayer, renderPlugin } from '@embedpdf/svelte/render';
  import { Stage, stagePlugin } from '@embedpdf/svelte/stage';
  import { interactionPlugin } from '@embedpdf/svelte/interaction';
  import { SelectionLayer, selectionPlugin } from '@embedpdf/svelte/selection';
  import { localEngine } from '@embedpdf/engine';
  import SelectionToolbar from './SelectionToolbar.svelte';

  import '../programmatic.css';

  const engine = localEngine();
  const plugins = [stagePlugin(), renderPlugin(), interactionPlugin(), selectionPlugin()];

  const ebook = async (): Promise<OpenInput> => {
    const response = await fetch('https://snippet.embedpdf.com/ebook.pdf');
    return { kind: 'bytes', id: 'ebook', bytes: new Uint8Array(await response.arrayBuffer()) };
  };
</script>

<Viewer {engine} {plugins} initialDocuments={[{ source: ebook }]}>
  <DocumentGate>
    {#snippet fallback()}
      <p class="loading">Loading…</p>
    {/snippet}
    <SelectionToolbar />
    <Stage class="stage">
      <RenderLayer />
      <SelectionLayer />
    </Stage>
  </DocumentGate>
</Viewer>

selection.select({ page: page.ref, start: 20, count: 40 }); // 40 characters on one page
selection.selectWordAt(page.ref, { x: 120, y: 80 }); // the word at a point
selection.selectLineAt(page.ref, { x: 120, y: 80 }); // the line at a point
selection.selectPage(page.ref);
selection.selectAll();
selection.clear();

A search result is already a range, so selecting it takes no conversion:

selection.select(hit); // { page, start, count }

For a selection across pages, give where it starts and where it ends:

selection.select({
  start: { page: firstPage.ref, index: 120 },
  end: { page: lastPage.ref, index: 35 },
});

Ranges include start and stop before end, and they count characters as the page stores them, not JavaScript string positions. selection.extendTo(page.ref, point) grows the current selection to a point on a page, as dragging a handle does. Every method that takes a page takes its ref or its index: selection.selectPage(0) selects the first page.

Show the selection in your UI#

useSelectionState() gives you what to show, and your component updates when it changes:

highlight-button.svelte

<script lang="ts">
  import { useSelectionState } from '@embedpdf/svelte/selection';

  const state = useSelectionState(); // state.hasSelection, state.isSelecting, state.pages
</script>

<button disabled={!state.hasSelection}>Highlight</button>

Loading live preview…

<script lang="ts" module>
  import { DocumentGate, Viewer, type OpenInput } from '@embedpdf/svelte/runtime';
  import { RenderLayer, renderPlugin } from '@embedpdf/svelte/render';
  import { Stage, stagePlugin } from '@embedpdf/svelte/stage';
  import { interactionPlugin } from '@embedpdf/svelte/interaction';
  import { SelectionLayer, selectionPlugin } from '@embedpdf/svelte/selection';
  import { localEngine } from '@embedpdf/engine';
  import SelectionStatus from './SelectionStatus.svelte';
  import SelectTitle from './SelectTitle.svelte';

  import '../state.css';

  const engine = localEngine();
  const plugins = [stagePlugin(), renderPlugin(), interactionPlugin(), selectionPlugin()];

  const ebook = async (): Promise<OpenInput> => {
    const response = await fetch('https://snippet.embedpdf.com/ebook.pdf');
    return { kind: 'bytes', id: 'ebook', bytes: new Uint8Array(await response.arrayBuffer()) };
  };
</script>

<Viewer {engine} {plugins} initialDocuments={[{ source: ebook }]}>
  <DocumentGate>
    {#snippet fallback()}
      <p class="loading">Loading…</p>
    {/snippet}
    <SelectTitle />
    <SelectionStatus />
    <Stage class="stage">
      <RenderLayer />
      <SelectionLayer />
    </Stage>
  </DocumentGate>
</Viewer>

Try it: drag over text. isSelecting is true until you let go, and pages counts the pages the selection reaches.

Every field is listed under State. For drawing your own highlight, selection.listSegments(page.ref) gives you each selected line as four corners, turned with the text, and selection.listRects(page.ref) a plain box per line.

Copy the selected text#

Reading the text and using the clipboard are two steps:

const text = await selection.readText();

readText() reads only the selected characters and gives them to you; it doesn’t touch the clipboard. Lines on different pages are joined with a new line, and with nothing selected it’s an empty string. readTextInRange(range) reads any range without selecting it.

For the clipboard, the Svelte package has two helpers. copySelection() copies from your own button:

copy-button.svelte

<script lang="ts">
  import { copySelection, useSelection, useSelectionState } from '@embedpdf/svelte/selection';

  const selection = useSelection();
  const state = useSelectionState();
</script>

<button
  disabled={!state.hasSelection || !selection.canCopy()}
  onclick={() => copySelection(selection)}
>
  Copy
</button>

<SelectionClipboard> makes Ctrl+C and the browser’s Copy menu work. Mount it once per document view:

clipboard.svelte

<script lang="ts">
  import { SelectionClipboard } from '@embedpdf/svelte/selection';
</script>

<!-- Once per document view: Ctrl+C / Cmd+C and the browser's Copy menu -->
<SelectionClipboard />

It reads the text as soon as a selection settles, so the browser’s copy can answer at once. Set prefetch to false to read text only when someone copies. A selection the user may not copy isn’t copied by the keys or by the browser’s Copy menu, and onRefused is called, so you can say so.

With more than one viewer on a page, pass target, the element around one as a function (target={() => area}): Ctrl+C and the Copy menu then copy its selection only while focus is inside it, and the page keeps its own copying everywhere else.

Loading live preview…

<script lang="ts" module>
  import { DocumentGate, Viewer, type OpenInput } from '@embedpdf/svelte/runtime';
  import { RenderLayer, renderPlugin } from '@embedpdf/svelte/render';
  import { Stage, stagePlugin } from '@embedpdf/svelte/stage';
  import { interactionPlugin } from '@embedpdf/svelte/interaction';
  import { SelectionClipboard, SelectionLayer, selectionPlugin } from '@embedpdf/svelte/selection';
  import { localEngine } from '@embedpdf/engine';
  import CopyButton from './CopyButton.svelte';
  import SelectedText from './SelectedText.svelte';
  import SelectTitle from './SelectTitle.svelte';

  import '../copy.css';

  const engine = localEngine();
  const plugins = [stagePlugin(), renderPlugin(), interactionPlugin(), selectionPlugin()];

  const ebook = async (): Promise<OpenInput> => {
    const response = await fetch('https://snippet.embedpdf.com/ebook.pdf');
    return { kind: 'bytes', id: 'ebook', bytes: new Uint8Array(await response.arrayBuffer()) };
  };
</script>

<Viewer {engine} {plugins} initialDocuments={[{ source: ebook }]}>
  <DocumentGate>
    {#snippet fallback()}
      <p class="loading">Loading…</p>
    {/snippet}
    <SelectTitle />
    <SelectionClipboard />
    <CopyButton />
    <SelectedText />
    <Stage class="stage">
      <RenderLayer />
      <SelectionLayer />
    </Stage>
  </DocumentGate>
</Viewer>

Try it: select other text: the line above the pages shows what readText() reads once the selection settles. Then press Copy, or Ctrl+C.

<SelectionMenu> places your own UI next to the selection. It hides while the user drags and appears when the selection settles:

Loading live preview…

<script lang="ts" module>
  import { DocumentGate, Viewer, type OpenInput } from '@embedpdf/svelte/runtime';
  import { RenderLayer, renderPlugin } from '@embedpdf/svelte/render';
  import { Stage, stagePlugin } from '@embedpdf/svelte/stage';
  import { interactionPlugin } from '@embedpdf/svelte/interaction';
  import {
    SelectionClipboard,
    SelectionLayer,
    SelectionMenu,
    selectionPlugin,
  } from '@embedpdf/svelte/selection';
  import { localEngine } from '@embedpdf/engine';
  import SelectionActions from './SelectionActions.svelte';
  import SelectTitle from './SelectTitle.svelte';

  import '../menu.css';

  const engine = localEngine();
  const plugins = [stagePlugin(), renderPlugin(), interactionPlugin(), selectionPlugin()];

  const PLACEMENTS = ['top', 'bottom', 'left', 'right'] as const;
  type Placement = (typeof PLACEMENTS)[number];

  const ebook = async (): Promise<OpenInput> => {
    const response = await fetch('https://snippet.embedpdf.com/ebook.pdf');
    return { kind: 'bytes', id: 'ebook', bytes: new Uint8Array(await response.arrayBuffer()) };
  };
</script>

<script lang="ts">
  let placement = $state<Placement>('top');
</script>

<Viewer {engine} {plugins} initialDocuments={[{ source: ebook }]}>
  <DocumentGate>
    {#snippet fallback()}
      <p class="loading">Loading…</p>
    {/snippet}
    <SelectTitle />
    <SelectionClipboard />
    <div class="toolbar" role="group" aria-label="Where the menu goes">
      {#each PLACEMENTS as name (name)}
        <button
          type="button"
          class="segment"
          aria-pressed={placement === name}
          onclick={() => (placement = name)}
        >
          {name}
        </button>
      {/each}
    </div>
    <Stage class="stage">
      <RenderLayer />
      <SelectionLayer />
      {#snippet overlay()}
        <SelectionMenu {placement} gap={8}>
          <SelectionActions />
        </SelectionMenu>
      {/snippet}
    </Stage>
  </DocumentGate>
</Viewer>

Try it: select other text, and the menu follows it. Pick where it goes above the pages.

Mount it in the Stage’s overlay snippet, and it stays attached to the text while the user scrolls and zooms. The menu only places things; what’s in it is yours: copy, comment, highlight, redact. placement is 'top', 'bottom', 'left' or 'right', and gap the space in pixels.

On touch screens#

A finger has no text cursor, so selection works as it does elsewhere on a phone: long-press a word, then drag the handles. Add <SelectionHandles> next to your menu:

selection-handles.svelte

<script lang="ts">
  import { RenderLayer } from '@embedpdf/svelte/render';
  import { SelectionHandles, SelectionLayer, SelectionMenu } from '@embedpdf/svelte/selection';
  import { Stage } from '@embedpdf/svelte/stage';

  import CopyButton from './copy-button.svelte';
</script>

<Stage>
  <RenderLayer />
  <SelectionLayer />

  {#snippet overlay()}
    <SelectionMenu>
      <CopyButton />
    </SelectionMenu>
    <SelectionHandles />
  {/snippet}
</Stage>

Loading live preview…

<script lang="ts" module>
  import { DocumentGate, Viewer, type OpenInput } from '@embedpdf/svelte/runtime';
  import { RenderLayer, renderPlugin } from '@embedpdf/svelte/render';
  import { Stage, stagePlugin } from '@embedpdf/svelte/stage';
  import { interactionPlugin } from '@embedpdf/svelte/interaction';
  import {
    SelectionHandles,
    SelectionLayer,
    SelectionMenu,
    selectionPlugin,
  } from '@embedpdf/svelte/selection';
  import { localEngine } from '@embedpdf/engine';
  import CopyButton from './CopyButton.svelte';
  import SelectWord from './SelectWord.svelte';

  import '../handles.css';

  const engine = localEngine();
  const plugins = [stagePlugin(), renderPlugin(), interactionPlugin(), selectionPlugin()];

  const ebook = async (): Promise<OpenInput> => {
    const response = await fetch('https://snippet.embedpdf.com/ebook.pdf');
    return { kind: 'bytes', id: 'ebook', bytes: new Uint8Array(await response.arrayBuffer()) };
  };
</script>

<Viewer {engine} {plugins} initialDocuments={[{ source: ebook }]}>
  <DocumentGate>
    {#snippet fallback()}
      <p class="loading">Loading…</p>
    {/snippet}
    <SelectWord />
    <Stage class="stage">
      <RenderLayer />
      <SelectionLayer />
      {#snippet overlay()}
        <!-- Clear of the start handle's grip, which sits above the first line. -->
        <SelectionMenu gap={20}>
          <CopyButton />
        </SelectionMenu>
        <SelectionHandles />
      {/snippet}
    </Stage>
  </DocumentGate>
</Viewer>

Try it: drag a handle’s round grip, with a finger or the mouse: the selection grows or shrinks from the other end.

Each handle is the selection’s own edge, with a round grip. Dragging one grows the selection from the other end, across pages too. Grabbing a handle never scrolls the page, and everywhere else a finger still scrolls and pinch-zooms. For the small tick a native app gives on long-press, see haptic feedback.

Methods#

Everything useSelection() gives you.

Method	What it does
`select(range)`	Selects `{ page, start, count }` on one page, or `{ start, end }` across pages. An empty range clears. Select text from code
`selectWordAt(page, point)`	Selects the word at a point, like a double-click. Returns `false` when there’s no text there.
`selectLineAt(page, point)`	Selects the line at a point, like a triple-click. Returns `false` when there’s no text there.
`selectPage(page)`	Selects every character on a page
`selectAll()`	Selects the whole document
`extendTo(page, point)`	Grows the selection to a point, as dragging a handle does
`clear()`	Clears the selection. Always allowed.
`readText()`	The selected text, or `''` with nothing selected. Needs permission to copy. Copy the selected text
`readTextInRange(range)`	The text of any range, without selecting it
`listSegments(page)`	Each selected line on a page, as four corners turned with the text
`listRects(page)`	A plain box around each selected line on a page
`getAnchor()`	Where a menu attaches: `{ page, bounds }` on the page where the selection ends, or `null`. `<SelectionMenu>` uses it.
`getSnapshot()`	Everything at once: each page’s lines, both ends with their direction, and the range. For drawing your own highlight and handles.
`canSelect()`, `canCopy()`	Whether selecting, and reading the selected text, are allowed. Permissions
`getSettings()`, `updateSettings(changes)`, `resetSettings()`	Reads the settings, changes them while the app runs, or goes back to what you registered. In a component, use `useSelectionSettings()`. Settings

Every method that returns a promise also takes a signal in its options, to cancel it.

State#

Everything useSelectionState() gives you, as one reactive object: read its fields, and your markup updates when they change. The method in the last column reads the same value once.

Field	What it is	Read once
`hasSelection`	Whether anything is selected	`hasSelection()`
`isSelecting`	Whether the user is still dragging	`isSelecting()`
`range`	The selection as `{ start, end }`, to store and pass back to `select()`, or `null`	`getRange()`
`pages`	The pages the selection is on, in document order	`listSelectedPages()`

Settings#

Set them when you register the plugin, and change them while the app runs with selection.updateSettings({ … }); selection.resetSettings() goes back to what you registered, and useSelectionSettings() gives you the current settings.

Setting	What it does	Default	From CSS
`dragThreshold`	How far, in pixels, the pointer moves before a drag selects	`4`	—
`color`	The color of selected text	the viewer’s accent at 35%	`--epdf-text-selection`
`handles.color`	The grips at its ends, on touch screens	the viewer’s accent	`--epdf-text-selection-handle`
`handles.shadow`	Their shadow	`'0 1px 4px rgb(0 0 0 / 0.35)'`	`--epdf-text-selection-handle-shadow`

A CSS variable wins over the setting, for dark mode or a different look per viewer: see Colors & theming.

Loading live preview…

<script lang="ts" module>
  import { DocumentGate, Viewer, type OpenInput } from '@embedpdf/svelte/runtime';
  import { RenderLayer, renderPlugin } from '@embedpdf/svelte/render';
  import { Stage, stagePlugin } from '@embedpdf/svelte/stage';
  import { interactionPlugin } from '@embedpdf/svelte/interaction';
  import { SelectionHandles, SelectionLayer, selectionPlugin } from '@embedpdf/svelte/selection';
  import { localEngine } from '@embedpdf/engine';
  import ColorPicker from './ColorPicker.svelte';
  import SelectTitle from './SelectTitle.svelte';

  import '../colors.css';

  const engine = localEngine();
  const plugins = [
    stagePlugin(),
    renderPlugin(),
    interactionPlugin(),
    selectionPlugin({ handles: { shadow: '0 1px 3px rgb(0 0 0 / 0.3)' } }),
  ];

  const ebook = async (): Promise<OpenInput> => {
    const response = await fetch('https://snippet.embedpdf.com/ebook.pdf');
    return { kind: 'bytes', id: 'ebook', bytes: new Uint8Array(await response.arrayBuffer()) };
  };
</script>

<script lang="ts">
  let fromCss = $state(false);
</script>

<Viewer {engine} {plugins} initialDocuments={[{ source: ebook }]}>
  <DocumentGate>
    {#snippet fallback()}
      <p class="loading">Loading…</p>
    {/snippet}
    <SelectTitle />
    <div class="toolbar">
      <ColorPicker />
      <label class="option">
        <input type="checkbox" bind:checked={fromCss} />
        Override with CSS
      </label>
    </div>
    <Stage class={fromCss ? 'stage from-css' : 'stage'}>
      <RenderLayer />
      <SelectionLayer />
      {#snippet overlay()}
        <SelectionHandles />
      {/snippet}
    </Stage>
  </DocumentGate>
</Viewer>

Try it: pick a color, and selection.updateSettings({ color, handles }) changes the selection and its handles at once. Then turn on Override with CSS: the CSS variables win over the color you picked.

Events#

selection-committed.svelte

<script lang="ts">
  import { useSelectionEvent } from '@embedpdf/svelte/selection';

  import { prefetchTranslation } from './translation';

  useSelectionEvent(
    (selection) => selection.onCommitted,
    () => prefetchTranslation(),
  );
</script>

Event	Fires when	Carries
`onChanged`	The selection changed, while dragging too	`range`, `pages`
`onCommitted`	A drag or a tap finished selecting	`range`
`onCleared`	The selection was cleared	—
`onSettingsChanged`	The settings changed	`settings`, `changed`

Use the change event for things that follow the selection as it grows, and the commit event for things that should happen once, such as creating a highlight.

Permissions#

Selecting and copying are two permissions:

Permission	What it allows	Check
`doc.text.select`	Selecting text, and highlighting it	`selection.canSelect()`
`doc.text.copy`	Reading the selected text with `readText()`	`selection.canCopy()`

Without doc.text.select, pointer selection does nothing and select() is refused; clear() always works. Without doc.text.copy, people can still select and highlight, but readText() is refused and no page text is sent. So a document can allow markup without copy.

Good to know#

Characters, not string positions#

A range counts characters as the page stores them. Search results, annotations and selections all use the same counting, so a range from one goes straight into another.

No HTML text layer#

The selection works from where each character is, so there’s no hidden HTML text over the page. That keeps large documents fast, and it’s why a document can allow selecting without giving away its text.

Outside the browser#

The selection plugin itself has no clipboard or HTML in it, so readText() also works in tests and in Node. The clipboard is the Svelte package’s job.

Was this page helpful?

Your feedback goes directly to the documentation team.

Selecting & copying

Select text from code#

Show the selection in your UI#

Copy the selected text#

Add a selection menu#

On touch screens#

Methods#

State#

Settings#

Events#

Permissions#

Good to know#

Characters, not string positions#

No HTML text layer#

Outside the browser#