Extract all text from a document

To extract a document's plain text, walk its stories in StoryList order and concatenate the Content of every range.

In short: Getting the plain text out of an IDML document means visiting its stories in reading order and concatenating the text of every Content element. You take the order from StoryList on the Document element, then walk each story tree Story → ParagraphStyleRange → CharacterStyleRange → Content, treating Br and Tab as whitespace. No styling needs to be resolved — extraction only cares about which characters appear and where the boundaries are. This page gives the recipe and the cases to get right.

To get the plain text of a document you walk its stories and concatenate the Content of every range, in document order. You do not need to resolve any styling — text extraction only cares about which characters appear and where the boundaries are.

The recipe

1. Open the design map and read StoryList on the Document element. It is a space-separated list of story ids, and it gives you the stories in document order.
2. For each id, open its story part (Stories/Story_<id>.xml). One file holds one Story element.
3. Walk the story tree in order: Story → ParagraphStyleRange → CharacterStyleRange → Content. Append the text of each Content element as you reach it.
4. Treat the markers as whitespace. A Br is a line break; emit \n. A Tab is a tab; emit \t. Our parser already folds these into the run text this way, so a CharacterStyleRange's text comes out with the breaks in place.
5. End each paragraph with a newline. Each ParagraphStyleRange is one paragraph; put a \n between consecutive paragraphs so the output reads as paragraphs, not one run-on line.

Worked over a two-paragraph story

The story below has two paragraphs and one run each. Walking it in order yields:

The first paragraph stands on its own.
The second paragraph follows it.

The first paragraph's Content, then a paragraph break, then the second paragraph's Content. The SpaceBefore on the second range is a layout attribute — extraction ignores 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>The first paragraph stands on its own.</Content>
      </CharacterStyleRange>
    </ParagraphStyleRange>
    <ParagraphStyleRange AppliedParagraphStyle="ParagraphStyle/$ID/[No paragraph style]" SpaceBefore="6">
      <CharacterStyleRange AppliedCharacterStyle="CharacterStyle/$ID/[No character style]">
        <Content>The second paragraph follows it.</Content>
      </CharacterStyleRange>
    </ParagraphStyleRange>
  </Story>
</idPkg:Story>

Things to get right

• Order comes from StoryList, then document order within each story. Don't sort by story id or file name; the listed order is the reading order.
• Some text is metadata, not body copy. Our parser drops HiddenText, Note, and Index / IndexEntry subtrees from the flowed text. If you walk the raw XML yourself, skip those subtrees too, or your output will contain text the reader never sees.
• A TextVariableInstance contributes its ResultText. Where a story carries a running header, page number, or file-name variable, the visible characters are the frozen ResultText — see text variables.
• Threaded stories are still one story. Text that flows across several frames is a single Story part; you extract it once, regardless of how many frames display it. See threading and overset.
• Tables hold their text in cells. A Table nested in a run carries its own Content inside Cell elements; include them if you want the full copy.

Frequently asked questions

What order should I extract stories in? The order given by StoryList on the Document element — a space-separated list of story ids that is the reading order. Don't sort by story id or file name; within each story, follow document order through the tree.

Do I need to resolve styles to extract text? No. Extraction only cares about which characters appear and where the boundaries are, so you can walk Story → ParagraphStyleRange → CharacterStyleRange → Content and concatenate the text without touching Styles.xml.

How do I handle threaded stories that span several frames? Extract them once. Text that flows across several frames is still a single Story part, so the number of frames that display it doesn't change the extraction — see threading and overset.

Why does my extracted text contain copy the reader never sees? Probably because you walked the raw XML without skipping metadata subtrees. Our parser drops HiddenText, Note, and Index / IndexEntry from the flowed text; if you walk the XML yourself, skip those subtrees too, or they'll end up in your output.