Anchored objects

An anchored object is a <Rectangle>, <TextFrame>, or <Group> written as a child of a CharacterStyleRange, carrying one <AnchoredObjectSetting>.

In short: An anchored object is built from two parts in IDML. The first is an ordinary frame element — a <Rectangle>, a <TextFrame>, or a <Group>, the same elements you would find on a spread — but written inside a run of text rather than on the spread. The second is a single extra child, <AnchoredObjectSetting>, that records where the frame should land relative to the anchor character. Our parser reads the frame's geometry and styling onto the host paragraph and reads the setting alongside it. This page is the reference for both: the frame element's attributes and every attribute of <AnchoredObjectSetting>, each with its honest support status.

An anchored object is a frame written as a child of a CharacterStyleRange. The frame is one of three familiar elements — a <Rectangle>, a <TextFrame>, or a <Group> — and it carries a single extra child, <AnchoredObjectSetting>, that describes where it should land relative to the anchor character. The parser reads the frame's geometry and styling onto the host paragraph and reads the setting alongside it.

<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
<idPkg:Story xmlns:idPkg="http://ns.adobe.com/AdobeInDesign/idml/1.0/packaging" DOMVersion="20.0">
  <Story Self="ustory">
    <ParagraphStyleRange AppliedParagraphStyle="ParagraphStyle/$ID/[No paragraph style]">
      <CharacterStyleRange AppliedCharacterStyle="CharacterStyle/$ID/[No character style]">
        <Content>A paragraph that carries an anchored object.</Content>
        <Rectangle Self="uanchor" AppliedObjectStyle="ObjectStyle/$ID/[None]" ItemTransform="1 0 0 1 0 0" GeometricBounds="0 0 36 72" FillColor="Color/Black" StrokeColor="Swatch/None" StrokeWeight="0">
          <AnchoredObjectSetting AnchoredPosition="InlinePosition" SpineRelative="false" LockPosition="false" AnchorPoint="BottomLeftAnchor" AnchorXoffset="0" AnchorYoffset="0"/>
        </Rectangle>
      </CharacterStyleRange>
    </ParagraphStyleRange>
  </Story>
</idPkg:Story>

The frame element

Any of three elements may appear directly inside a CharacterStyleRange to become an anchored object. Which one it is, the parser records as a frame kind; the rest of the attributes it reads mirror how the same element is read at spread level.

A few things the parser deliberately does not recurse into on an anchored frame: nested transparency settings, and — beyond a nested <Image>'s link URI and transform on an anchored <Rectangle> — image styling. A <Group> records its child frames in z-order; a <TextFrame> records its ParentStory so the renderer can chase the story content once placement lands. Parsed, not yet renderedanchored frames read but not flowed — paged-parse/src/story.rs:264

AnchoredObjectSetting

The placement block. It sits inside the frame element and records intent — where the object should land relative to the anchor — without itself moving anything. The parser stores each attribute's value verbatim; an absent attribute falls back to the cascaded default.

Why the count doesn't change

The example above adds a whole <Rectangle> to the story, yet the renderer still reports frames: 1, paragraphs: 1, runs: 1. That is the point of an anchored object being inside a run: it is recorded on the host paragraph, not promoted to a spread-level frame, so it does not inflate the frame count — and the run that hosts it is still one run. The frame on the spread (frames: 1) is the page's text frame; the anchored rectangle is data hanging off the paragraph within it.

Next

What AnchoredPosition, AnchorPoint, and the reference/alignment attributes actually do — and how much of it the renderer carries today — is the subject of inline vs anchored positioning.

Frequently asked questions

Which elements can be an anchored object? Exactly three: a <Rectangle>, a <TextFrame>, or a <Group>, written directly inside a CharacterStyleRange. These are the same elements you would find on a spread; the parser records which one it is as the frame's kind and reads the rest of its attributes the way it reads them at spread level.

What does <AnchoredObjectSetting> do? It is the placement block nested inside the frame element. It records intent — where the object should land relative to the anchor, via attributes like AnchoredPosition, AnchorPoint, and the offset and reference/alignment pairs — without itself moving anything. The parser stores each attribute verbatim, and an absent attribute falls back to its cascaded default.

Why doesn't adding an anchored object change the frame, paragraph, or run count? Because the object is recorded on its host paragraph rather than promoted to a spread-level frame. The frame count reflects the page's text frame, the paragraph still has one paragraph, and the run that hosts the object is still one run — so the renderer reports frames: 1, paragraphs: 1, runs: 1 even though a whole <Rectangle> was added to the story.

Does the parser read everything inside an anchored frame? No. It deliberately skips nested transparency settings and, beyond a nested <Image>'s link URI and transform on an anchored <Rectangle>, image styling. A <Group> records its child frames in z-order, and a <TextFrame> records its ParentStory so the renderer can chase the story content once placement lands.