Edit PDF Tags

About Edit PDF Tags

The structure tree (/StructTreeRoot) of a tagged PDF decides what a screen reader reads and in what order. When auto-tagging turned the cover title into a paragraph or hung a table caption in the wrong section, this tool lets you fix the tree itself.

Select a node to change its tag name or alternate text (/Alt), move it up or down or under another parent, create new nodes, or delete a node with its subtree. The text and images drawn on the page don't change; only renaming a tag to Artifact changes how its content is marked, to decoration.

The tree and the page are linked only by MCIDs

A structure element points at real text through MCID numbers attached to marked-content sequences (BDC … EMC) in the content stream. The count shown beside each node is how many such links it carries. Moving a node takes its links with it, so a title retagged as H1 still points at the same glyphs.

The tool does not check whether a move still makes sense. Drag a page-3 paragraph under the page-1 heading and a screen reader, following tree order, will read the page-3 sentence right after that heading. Only move linked nodes to positions consistent with where their content sits. Deleting a node also clears the ParentTree slots that pointed at it, but its content is left untagged, which checkers report as "content not tagged".

"Artifact" is not a tag name

In the PDF specification, decorative content is marked as /Artifact in the content stream; it is not an element of the structure tree. So when you rename a node to Artifact, this tool does not create an element called Artifact. It takes the node (and its subtree) out of the structure tree and rewrites the marked content the node pointed to by MCID (such as /P <</MCID 3>> BDC) as /Artifact BMC, clearing the node's ParentTree slots as well. Decorative images, running headers and page numbers are then skipped by screen readers, and checkers count them as artifacts rather than untagged content. Content inside Form XObjects (MCR /Stm) can't be rewritten, so check such nodes with the accessibility checker afterwards.

Structures you will fix most often

Avoid skipping heading levels: an H3 straight after an H1 looks like a broken hierarchy in a screen reader's headings list. Tables need TR inside Table and TH or TD inside TR for row and column headers to be announced. If TD cells sit directly under Table, create a TR node and move them into it.

Lists work the same way: LI inside L lets a screen reader say "list, 3 items, 1 of 3". Put a Caption as a child of, or directly next to, the Table or Figure it describes.

Worked examples

How to use it

1. Open the PDF; the tag tree appears as an indented outline.
2. Select nodes to rename, edit alt text, move, add or delete.
3. Apply, save, and re-run the PDF accessibility checker.

Frequently asked questions