refactor(web): edit and wrap diagrams with the MCP server's code

- edit_diagram runs the MCP server's editDiagram: every new_xml is checked
  first, one cell per operation, and after the edit only the target page
  is checked, rejecting only errors this edit introduced. An unrelated
  problem elsewhere in the document no longer blocks every edit. The
  error lists each failed operation
- The streaming edit preview uses the MCP applyDiagramOperations
- Delete applyDiagramOperations (292 lines) and wrapWithMxFile from
  lib/utils.ts, and the unused hand-copied scripts/test-diagram-operations.mjs
- One blank document (BLANK_MXFILE) for the web app and the MCP preview,
  replacing four copies
- Saving a .drawio wraps a bare model with normalizeToMxfile
- The empty-diagram check uses hasCells, which also counts cells wrapped
  in a UserObject/object
- DiagramOperation is the MCP type
- The wrapped-cell and empty-diagram tests now run against the MCP code
- New e2e test: edit_diagram changes the canvas, and a failing edit
  leaves it as it was
This commit is contained in:
dayuan.jiang
2026-10-04 12:51:20 +09:00
parent e49e9f0c47
commit bf4e95dcfb
16 changed files with 210 additions and 948 deletions
+83 -26
View File
@@ -1,34 +1,45 @@
import { expect, test } from "@playwright/test"
import { expect, type Page, test } from "@playwright/test"
import { getIframe, sendMessage, waitForCompleteCount } from "./lib/fixtures"
/**
* Checks what draw.io actually shows after display_diagram, not only the
* Checks what draw.io actually shows after the diagram tools, not only the
* tool card. The tool input is streamed in chunks like a real model, and
* the browser tool handler (not the server) completes the tool call.
*/
function streamedToolCall(xml: string) {
function streamedToolCall(toolName: string, input: unknown) {
const toolCallId = `call_${Math.random().toString(36).slice(2)}`
const input = JSON.stringify({ xml })
const chunks = input.match(/[\s\S]{1,40}/g) ?? []
const chunks = JSON.stringify(input).match(/[\s\S]{1,40}/g) ?? []
const events = [
{ type: "start", messageId: `msg_${toolCallId}` },
{ type: "tool-input-start", toolCallId, toolName: "display_diagram" },
{ type: "tool-input-start", toolCallId, toolName },
...chunks.map((inputTextDelta) => ({
type: "tool-input-delta",
toolCallId,
inputTextDelta,
})),
{
type: "tool-input-available",
toolCallId,
toolName: "display_diagram",
input: { xml },
},
{ type: "tool-input-available", toolCallId, toolName, input },
{ type: "finish" },
]
return `${events.map((e) => `data: ${JSON.stringify(e)}\n\n`).join("")}data: [DONE]\n\n`
}
const END_TURN =
'data: {"type":"start"}\n\ndata: {"type":"finish"}\n\ndata: [DONE]\n\n'
/** Answer each chat request with the next reply, then end the turn */
async function mockReplies(p: Page, replies: string[]) {
await p.route("**/api/chat", async (route) => {
await route.fulfill({
status: 200,
contentType: "text/event-stream",
body: replies.shift() ?? END_TURN,
})
})
await p.goto("/", { waitUntil: "networkidle" })
await getIframe(p).waitFor({ state: "visible", timeout: 30000 })
return p.frameLocator("iframe")
}
const cell = (id: string, label: string, x: number) =>
`<mxCell id="${id}" value="${label}" style="rounded=1;" vertex="1" parent="1"><mxGeometry x="${x}" y="40" width="120" height="60" as="geometry"/></mxCell>`
const page = (id: string, cells: string) =>
@@ -46,20 +57,10 @@ const NEW_CELLS =
test("display_diagram replaces the document with the fixed diagram", async ({
page: p,
}) => {
const replies = [TWO_PAGES, NEW_CELLS]
await p.route("**/api/chat", async (route) => {
const xml = replies.shift()
await route.fulfill({
status: 200,
contentType: "text/event-stream",
body: xml
? streamedToolCall(xml)
: 'data: {"type":"start"}\n\ndata: {"type":"finish"}\n\ndata: [DONE]\n\n',
})
})
await p.goto("/", { waitUntil: "networkidle" })
await getIframe(p).waitFor({ state: "visible", timeout: 30000 })
const canvas = p.frameLocator("iframe")
const canvas = await mockReplies(p, [
streamedToolCall("display_diagram", { xml: TWO_PAGES }),
streamedToolCall("display_diagram", { xml: NEW_CELLS }),
])
await sendMessage(p, "Draw two pages")
await waitForCompleteCount(p, 1)
@@ -79,3 +80,59 @@ test("display_diagram replaces the document with the fixed diagram", async ({
await expect(canvas.getByText("Old A")).toHaveCount(0)
await expect(canvas.getByText("Second", { exact: true })).toHaveCount(0)
})
test("edit_diagram applies all operations or none", async ({ page: p }) => {
const canvas = await mockReplies(p, [
streamedToolCall("display_diagram", {
xml: cell("a", "Alpha", 40) + cell("b", "Beta", 220),
}),
streamedToolCall("edit_diagram", {
operations: [
{ operation: "delete", cell_id: "a" },
{
operation: "update",
cell_id: "b",
new_xml: cell("b", "Beta two", 220),
},
{
operation: "add",
cell_id: "c",
new_xml: cell("c", "Gamma", 400),
},
],
}),
// The first operation is fine, the second fails: nothing is kept
streamedToolCall("edit_diagram", {
operations: [
{
operation: "update",
cell_id: "c",
new_xml: cell("c", "Broken", 400),
},
{ operation: "delete", cell_id: "missing" },
],
}),
])
await sendMessage(p, "Draw two boxes")
await waitForCompleteCount(p, 1)
await expect(canvas.getByText("Alpha", { exact: true })).toBeVisible({
timeout: 15000,
})
await sendMessage(p, "Change them")
await waitForCompleteCount(p, 2)
await expect(canvas.getByText("Gamma", { exact: true })).toBeVisible({
timeout: 15000,
})
await expect(canvas.getByText("Beta two", { exact: true })).toBeVisible()
await expect(canvas.getByText("Alpha", { exact: true })).toHaveCount(0)
await sendMessage(p, "Change again")
await expect(p.getByText(/No changes were made/).first()).toBeAttached({
timeout: 15000,
})
await p.waitForTimeout(1000)
await expect(canvas.getByText("Gamma", { exact: true })).toBeVisible()
await expect(canvas.getByText("Broken", { exact: true })).toHaveCount(0)
})