* M12: run XFA forms' JavaScript: calculations, validation, rows, visibility (#60) Laid-out XFA forms had their fields but none of their behaviour: totals didn't compute, "Add row" buttons did nothing, and subforms a check box should reveal stayed hidden. The js crate gains the XFA scripting object model as native boa objects over a snapshot of the form tree: xfa.form navigation by name, SOM resolution (absolute, relative with the containing-subform search, descendants, indexes and wildcards), rawValue/presence/access, instance managers (_name, addInstance, removeInstance, setInstances, count), xfa.host (messageBox, resetData, print, setFocus, gotoURL), xfa.layout page numbers, xfa.event, app, console and util, plus unqualified sibling names as XFA resolves them. Scripts run with `this` bound to the node and yield their completion value; everything they change or ask for comes back as effects. The xfa crate builds the live form tree (instances from the data, values, presence and access with overrides) and the event table from the template, keeps presence and access overrides on the AcroForm (/PCXfaOverrides) so undo and reopening preserve them, adds and removes row instances in the datasets packet through the existing merge pipeline, and lays the form out again on demand. The engine runs initialize and calculate scripts on open (an appended revision, nothing dirty), exit, validate and calculate after a field changes (a failing validate shows its message and keeps the value, as Acrobat does), and click scripts for buttons through run_javascript, so the app's buttons and the js_run tool work unchanged. Values a script sets go to the fields and the data at once; rows and visibility changes re-lay the form out. All of it is one undo step. JavaScript off turns it off. FormCalc scripts are recognised and reported, not run. Bounds: script length and nesting as before, loop and recursion limits, at most 1000 instances added per run, 2000 scripts and 50 re-layouts per event; a calculate script that adds rows, or a click script looping on addInstance/setInstances, ends quickly with a bounded form (tested). Checked on IRCC's IMM 5669: opens with its page-number scripts, the Print button's xfa.host.print runs through js_run, fill and save keep the original bytes. * M12: Save and SaveAs named actions run the file commands (#60) XFA forms' Save buttons are app.execMenuItem("SaveAs"), which the layout turns into a SaveAs named action; the shell answered it with "the SaveAs action isn't supported yet". Named Save and SaveAs now run file.save and file.save_as, like Named Print opens the print dialog. The buttons test covers Save against a file on disk (SaveAs opens the native dialog, which a headless test must not). * M12: bound XFA form scripts against hostile forms (#60) Review fixes for the XFA scripting PR (#230): - Effects: values, presence and access set again on the same object merge (last wins, behind order-dependent effects); 10k effects, 100 message boxes and 1000 console lines per script and per event, reported when hit. An event's data writes are kept in memory and written as one rewrite of the datasets packet; later writes in the same edit replace that stream in place. - form_tree stops at 200k objects and says so; forms without scripts build no live form and run nothing. - Only click scripts may print, save, open links or move the focus; others are noted in the console. Open-time messages and errors show right after open. - Lower loop limits for initialize/calculate/validate, a time budget per event after which remaining scripts are skipped and reported, and a per-script time limit after which a runaway script is abandoned and the document's scripts turned off. - What open-time scripts change is noted in the XFA warnings. - A document's Named /Save no longer writes the file on click (SaveAs opens the Save As dialog, as a script's execMenuItem does). - XFA click scripts are no longer written as /S /JavaScript actions: the button carries /PCXfaClick and runs by SOM path. Private keys documented. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Brandon Thomas <echelon@gmail.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
pdfcraft-automation
Agent control for PdfCraft: a headless tool table over the engine, and an opt-in MCP server.
- Layer: L7 (architecture §13), but headless. It depends on
engine,renderandorganize, never on a UI toolkit. - Status: over a hundred tools;
pdfcraft-cli toolslists the current set with their schemas (the count isn't repeated here, so it can't go stale). They cover inspection, rendering, text, page edits and organizing, page labels, bookmarks, comments and their review, form filling and authoring (fields, properties, scripts, data exchange, detection), redaction and sanitizing, password protection and signatures, calibrated 2D measurements and CSV export, adding text, images, links and stamps, optimizing, OCR, printing, export, metadata, undo/redo, save, combine, extract and split. Comment tools take geometry in the same top-left-origin points as everything else, andcomment_addcan mark text by searching for it (find). The MCP server also serves the open documents as resources, page images included. The running app's UI is driven separately, through theui.*control channel inpdfcraft-ui-egui.
API
let mut a = Automation::new().with_root("/work")?; // optional: confine all paths
let doc = a.call("doc_open", &json!({ "path": "in.pdf" }))?; // Vec<Content>
a.write_output("page1.png", &png)?; // save a result yourself, under the same root rules
tools() -> Vec<ToolDef> // name, title, description, input_schema, read_only, destructive, command
mcp::McpServer::new(a).serve(stdin, stdout)? // newline-delimited JSON-RPC 2.0
Content is Json(Value) or Png { data, width, height }. Errors are UnknownTool, InvalidArgs (the call didn't match the schema) or Failed (a readable message for the agent).
Conventions
- Tool names are
snake_case(page_rotate): MCP clients reject dots.ToolDef::commandlinks a tool to the registry id it automates (page.rotate), andcommand_listreports the link. - Pages and positions are 1-based. Rectangles are PDF points with the origin at the top-left of the displayed page.
- Unknown arguments are rejected, so typos fail loudly.
- Tools that change a document return its summary (
doc,pages,dirty,undo,redo, …). doc_closerefuses to drop unsaved changes unlessdiscard_changes: true.doc_savewrites atomically, incrementally in place, and in full for a new path.- With a root set, every read and write path must resolve inside it (symlinks and
..included). Relative paths resolve inside the root, and..is resolved by name before the check. Every path outside the root gets the same refusal (<path> is outside the allowed directory <root>), whether or not it exists, so a confined agent can't probe the rest of the disk; on Windows another network share or device path (\\host\share,\\?\UNC\…,\\.\…) is refused without being contacted. "Not found" and other filesystem errors are reported only for paths inside the root. Tools that write several files into a folder name them after the document or input file, with separators, colons and control characters replaced by_, so a name can't lead them out of the folder. The root itself must be a folder.
MCP server
Opt-in only. Nothing starts it automatically, and it opens no port. It runs while pdfcraft-cli mcp [--root DIR] runs, normally launched by an agent from its MCP configuration. It stops when stdin closes. The CLI's mcp Cargo feature (on by default) compiles it out entirely when disabled.
Implemented: initialize (protocol 2025-06-18, 2025-03-26, 2024-11-05), ping, tools/list (with readOnlyHint/destructiveHint annotations), tools/call (JSON results also returned as structuredContent; images as image/png), and resources/list, resources/templates/list and resources/read. The resources expose the open documents read-only: pdfcraft://doc/{doc}/info (JSON), …/text, …/page/{page}/text and …/page/{page}/image{?dpi} (PNG, 1–600 dpi). Tool failures come back as isError: true results, so the agent can read and recover from them.
Adding a tool
- Add the engine capability first, with its tests (the tool is a thin adapter).
- Add a
ToolDefinsrc/tools.rs: a precise description, the schema,ro()/destructive(), andcmd()if a registry command exists. - Handle it in
Automation::call, validating pages and positions with the existing helpers. - Add an end-to-end test in
tests/automation.rs.tool_table_is_well_formedchecks names, schemas and command links.