Hybrid tree formatting + link mapping

[seanmcguire12]

why

• reduce tokens & redundancy in the hybrid tree
• improve formatting of the hybrid tree (by removing & collapsing unnecessary nodes)

what changed

• this PR reduces redundancy in the hybrid tree by combining StaticText nodes into their parent node when possible
• it also excludes StaticText nodes if they are a child of a link
• links are now formatted inline with the following format: [@id]
• the TreeResult type now holds an idToUrl dictionary where node IDs are keys and URLs are values

example

here is what the tree looked like before:

here is what the tree looks like now:

future PRs (fast follow)

• inform the LLM on the change to link formatting
• when an LLM chooses an ID, lookup the URL in the idToUrl mapping, and include that in the extraction result
• remove text_extract, just have one extract function
• reintroduce chunking

test plan

• evals

[changeset-bot]

🦋 Changeset detected

Latest commit: a1075e4

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
@browserbasehq/stagehand	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR