adds --dump screen.pdf and CDP's Page.printToPDF (along with IO.read and
IO.read). Builds ontop of https://github.com/lightpanda-io/browser/pull/3231
This generates a text-based PDF, not just our PNG placed into a PDF. The biggest
change to the existing (isolated) code was adding href information so that links
are actually clickable.
Like 3231 the hard parts are all Claude but isolated. The one place it diverged
from the plan was NOT using the subsetter rust crate for trimming the size of
the embedded font. It felt that the dependencies were too much (e.g. a second
version of skrifa, ...). One subsetter was removed, the pdf rendering, while
still a blackbox, was moved from Rust to Zig (pdf.zig), which is kind of nice.
Zig 0.16 forwards the result type through the unary float builtins, so
@trunc produces the integer directly and @intFromFloat is deprecated as
redundant with it. Every site was a truncating cast, or already held an
integral value, so the behaviour is unchanged.
Track delivery sessions that repeatedly hit the recursion cap without draining. Disconnect the frame's observers after 100 such sessions so timer-rearmed mutation cascades cannot consume memory indefinitely.
Assisted-By: devx/1a7af220-8162-4dc8-b16c-088baab97324
Allocate mutable CSS property values from the factory slab instead of the frame-lifetime arena. Replacing or removing a property now returns its owned value storage for reuse without attempting to free interned strings.
Assisted-By: devx/1a7af220-8162-4dc8-b16c-088baab97324
The transport already drops its staging buffer past 256 KB, but the
connection buffer that receives the copy lives for the whole keep-alive
connection and was only ever cleared, so one screenshot pinned its
capacity per open connection. Apply the same threshold there.
Claude-Session: https://claude.ai/code/session_01S9t1TX3vTKunaXdnatbcjB
The renderer knows that layout reflows to the width and that height 0
means the whole content, so the bounding rule moves next to measure()
and only measures when the strip isn't already fixed: the default
viewport shot no longer pays a second layout pass. The limits are
constants spliced into the tool description, not a caller-tunable that
nothing tuned. expireImages runs inside prune so there is one end-of-
turn hook, and re-homes a stripped result instead of casting away const.
An inline screenshot is re-sent on every request for as long as it sits
in history, and a full-page render could reach 1920x16384. Inline images
are now rendered at most 1280 wide and 4096 tall (measured after the
reflow, so short pages aren't padded), files keep full size; the
conversation keeps only the newest two images, older tool results keep
their text with a note; the MCP transport releases its buffer after a
large response.
The slash path's result has two consumers: the terminal, which can't
show an image, and the conversation, which can. Opt in when a model is
attached and forward the image through the same adapter the model-driven
path uses, and stop mapping a failed adapter to a text-only success.
MCP's ImageContent and CallToolResult take their payload type like
TextContent does; resolveScope reuses resolveTarget; needsLocator folds
into replayRequires.
zenai tool results now carry image parts (lightpanda-io/zenai#12, #13),
so the model-driven tool path opts into inline images: a screenshot
without `path` reaches the model as text plus the PNG, on every backend.
The slash-command path still needs `path`, its result goes to the
terminal.
Callers pass CallOpts.inline_image; execScreenshot rejects a path-less
call before navigating or rendering, which removes the per-consumer
guards and covers the model-driven tool path that had none. MCP image
content is a protocol type, the screenshot recording rule joins the
recorder's replayRequires predicate, and the viewport-to-Opts mapping,
node-scope ladder and save-path helpers are shared instead of copied.
The PNG renderer was reachable from CDP and fetch --dump png only. The
tool renders the page or one node; with `path` it writes the file and
returns its location (agent, PandaScript, MCP), without it MCP returns
the image inline as base64 content. ToolResult carries the prepared
image so the transport streams it; the agent and script runtime reject
the inline form since their tool results are text. An inline screenshot
is not recorded, as it has no replayable form.
Agent held the same Browser/Session/Notification/registry fields and
teardown order as mcp/Server.zig, plus its own "enableConsoleCapture
after every newSession" for /reset. ToolSession.restartSession() owns
that now; init is written in terms of it.
Claude-Session: https://claude.ai/code/session_01M6WGk8wZSE28efFQkYT9SK
Two sessions on one thread, each entered only around its own use, with
JS state proving the isolates are separate and a balanced park/enter/
deinit teardown. Pins what the C API relies on without Server in the loop.
Claude-Session: https://claude.ai/code/session_01M6WGk8wZSE28efFQkYT9SK
stdio kept the isolate permanently entered while HTTP parked it between
requests, so a handler touching page JS without enterIsolate passed the
stdio-mode tests and only failed under --port. Bracket unconditionally:
stdio is the one-isolate case of the same discipline.
The flag survives only as `multi_session`, which is what it always gated:
the session tools need a transport that routes by id.
Claude-Session: https://claude.ai/code/session_01M6WGk8wZSE28efFQkYT9SK
Extract the Browser+Session+Notification+node Registry quartet, with its
ordering-sensitive init/teardown and isolate enter/exit, into
src/ToolSession.zig. mcp/Server.zig's sessions map now holds
*lp.ToolSession directly; the session id lives only as the map key.
Claude-Session: https://claude.ai/code/session_01M6WGk8wZSE28efFQkYT9SK
findElement and interactiveElements had their own role table and name
algorithm, which knew nothing of <label for>, wrapping labels or
aria-labelledby, so a control the tree showed as "Email address" was
unfindable by that name. Both now come from AXNode, with the element's
text as the name of last resort for elements that have no role (a div
interactive only by its listener or tabindex). getName takes an
optional label index so a walk resolves <label for> with one document
scan; the semantic tree walk uses it too.
The renderer knows that layout reflows to the width and that height 0
means the whole content, so the bounding rule moves next to measure()
and only measures when the strip isn't already fixed: the default
viewport shot no longer pays a second layout pass. The limits are
constants spliced into the tool description, not a caller-tunable that
nothing tuned. expireImages runs inside prune so there is one end-of-
turn hook, and re-homes a stripped result instead of casting away const.
An inline screenshot is re-sent on every request for as long as it sits
in history, and a full-page render could reach 1920x16384. Inline images
are now rendered at most 1280 wide and 4096 tall (measured after the
reflow, so short pages aren't padded), files keep full size; the
conversation keeps only the newest two images, older tool results keep
their text with a note; the MCP transport releases its buffer after a
large response.
The slash path's result has two consumers: the terminal, which can't
show an image, and the conversation, which can. Opt in when a model is
attached and forward the image through the same adapter the model-driven
path uses, and stop mapping a failed adapter to a text-only success.
MCP's ImageContent and CallToolResult take their payload type like
TextContent does; resolveScope reuses resolveTarget; needsLocator folds
into replayRequires.
zenai tool results now carry image parts (lightpanda-io/zenai#12, #13),
so the model-driven tool path opts into inline images: a screenshot
without `path` reaches the model as text plus the PNG, on every backend.
The slash-command path still needs `path`, its result goes to the
terminal.
Callers pass CallOpts.inline_image; execScreenshot rejects a path-less
call before navigating or rendering, which removes the per-consumer
guards and covers the model-driven tool path that had none. MCP image
content is a protocol type, the screenshot recording rule joins the
recorder's replayRequires predicate, and the viewport-to-Opts mapping,
node-scope ladder and save-path helpers are shared instead of copied.
The PNG renderer was reachable from CDP and fetch --dump png only. The
tool renders the page or one node; with `path` it writes the file and
returns its location (agent, PandaScript, MCP), without it MCP returns
the image inline as base64 content. ToolResult carries the prepared
image so the transport streams it; the agent and script runtime reject
the inline form since their tool results are text. An inline screenshot
is not recorded, as it has no replayable form.
Remove singleton and give each Browser a screenshot.Renderer.
Also, add tests to make sure the Rust and Zig structures/constants match (this
has proven useful with zig-v8-fork).
Karl Seguin