Files
next-ai-draw-io/app/api/validate-diagram/route.ts
T
dayuan.jiang c0fa997186 fix: older defects (batch C) and the second batch's review
Chats:
- New Chat right after an answer saves that chat once. Saves run one at
  a time and read the chat on screen when their turn comes; a save
  scheduled for a chat that is no longer on screen is dropped. A chat
  whose id was still on its way to the URL no longer comes back after
  New Chat (the next answer went into it).
- Crossing the 768 px breakpoint keeps the chat panel: a streaming answer,
  unsaved messages and attachments stay. The panel gets the sizes of each
  side, and a panel collapsed on desktop opens on mobile.
- The chat's export waits for its own reply: an edit's history export
  still on its way no longer answers it with the older diagram, and two
  file saves at once no longer swap results.
- A second edit in one answer is previewed on the first edit's result.
- Stop also ends a running screenshot check; a chat that cannot be saved
  (storage full) can be left with "Continue without saving".
- Small diagrams with shapes count as diagrams; the tool card no longer
  crashes on malformed operations.

Quota and providers:
- Requests that reach the server's own endpoints count toward the quota:
  EdgeOne (always its own endpoint now), a private base URL whatever key
  header is sent, keyless Ollama without a URL. With the quota on, a
  redirect is followed only to a public address. The output cap applies
  to these requests too.
- Stop records the tokens of the steps that finished; the screenshot check
  counts its tokens without counting a request.
- EdgeOne configured only by AI_PROVIDER works, also in the admin Test,
  which forwards the access code. Azure set up only in the admin panel
  works in chat. The Test sends a Bedrock session token.
- The admin panel's Test of an entry without a URL uses the server's URL
  as the server does (no private address check for it); the admin panel
  no longer writes an Ollama URL.

MCP server:
- Write tools and start_session run one at a time, so two at once never
  drop each other's change; a cancelled call waiting its turn is skipped.
  get_diagram and export_diagram keep the session they started with.
- Export to .drawio first gets the user's latest edits from the browser.
- History thumbnails: one that arrives after the next AI write is
  dropped; a sync reply keeps the image; a version that changed only page
  settings is its own entry.
- A diagram over the 10 MB limit is saved without its image, or the user
  is told to download it (the server now answers 413 instead of cutting
  the connection).
- Labels holding text like id='1' or parent='1' are no longer read as
  attributes (a layer or a parent was deleted). A broken bare
  <mxGraphModel> file is refused.
- After a sync reply the tab no longer sends its autosave copy again.

Desktop and files:
- A newer switch of the same preset is not rolled back by an older one
  that failed. .env values with escaped quotes are read whole.
- MCP saved files: a file that could not be read stays protected while a
  folder without permission hides it, and is saved again once deleted.
- The desktop app reports "no chats" only when the count was read and no
  model settings are stored.
2026-10-05 18:57:02 +09:00

185 lines
6.2 KiB
TypeScript

/**
* API endpoint for VLM-based diagram validation.
* Accepts a PNG image and streams validation results using useObject-compatible format.
*/
import { Output, streamText } from "ai"
import { checkAccessCode, rejectCrossSite } from "@/lib/access-code"
import { getValidationModel } from "@/lib/ai-providers"
import {
checkAndIncrementRequest,
isQuotaEnabled,
recordTokenUsage,
} from "@/lib/dynamo-quota-manager"
import { getUserIdFromRequest } from "@/lib/user-id"
import { VALIDATION_SYSTEM_PROMPT } from "@/lib/validation-prompts"
import {
type ValidationResult,
ValidationResultSchema,
} from "@/lib/validation-schema"
export const maxDuration = 30
// Data URL length cap (~3.75 MB of PNG), well above a normal diagram capture
const MAX_IMAGE_DATA_LENGTH = 5 * 1024 * 1024
interface ValidateDiagramRequest {
imageData: string // Base64 PNG data URL
sessionId?: string
}
// Default valid result for disabled/error cases
const DEFAULT_VALID_RESULT: ValidationResult = {
valid: true,
issues: [],
suggestions: [],
}
/** A fixed result in the text format useObject reads */
function createStreamingResponse(result: ValidationResult): Response {
return new Response(JSON.stringify(result), {
headers: { "Content-Type": "text/plain; charset=utf-8" },
})
}
export async function POST(req: Request): Promise<Response> {
const crossSite = rejectCrossSite(req)
if (crossSite) return crossSite
// Uses the server's model credentials, so require the access code
const accessError = checkAccessCode(req)
if (accessError) return accessError
try {
// Check if VLM validation is enabled (default: true)
const enableValidation = process.env.ENABLE_VLM_VALIDATION !== "false"
if (!enableValidation) {
return createStreamingResponse(DEFAULT_VALID_RESULT)
}
const body: ValidateDiagramRequest = await req.json()
const { imageData, sessionId } = body
if (!imageData) {
return Response.json(
{ error: "Missing imageData" },
{ status: 400 },
)
}
// Validate image data format
if (
!imageData.startsWith("data:image/png;base64,") &&
!imageData.startsWith("data:image/")
) {
return Response.json(
{ error: "Invalid image data format" },
{ status: 400 },
)
}
if (imageData.length > MAX_IMAGE_DATA_LENGTH) {
return Response.json(
{ error: "Image data too large" },
{ status: 413 },
)
}
// It runs the server's vision model: with the quota on, the daily
// and per-minute token limits apply, and its tokens are counted. Not
// the request limit, which is for chats: the day's last chat still
// gets its check, and a check does not count as a chat.
const userId = getUserIdFromRequest(req)
const countsQuota = isQuotaEnabled() && userId !== "anonymous"
if (countsQuota) {
const quotaCheck = await checkAndIncrementRequest(
userId,
{
requests: 0,
tokens: Number(process.env.DAILY_TOKEN_LIMIT) || 200000,
tpm: Number(process.env.TPM_LIMIT) || 20000,
},
0,
)
if (!quotaCheck.allowed) {
return Response.json(
{
error: quotaCheck.error,
type: quotaCheck.type,
used: quotaCheck.used,
limit: quotaCheck.limit,
},
{ status: 429 },
)
}
}
// Get the validation model
let model
try {
model = getValidationModel()
} catch (error) {
console.warn(
"[validate-diagram] Validation model not available:",
error,
)
// Return valid if no vision model is configured
return createStreamingResponse(DEFAULT_VALID_RESULT)
}
// Parse timeout with validation (minimum 1000ms, default 10000ms)
const timeout =
Math.max(
1000,
parseInt(process.env.VALIDATION_TIMEOUT || "10000", 10),
) || 10000
// Stream the VLM response for useObject consumption
const result = streamText({
model,
output: Output.object({ schema: ValidationResultSchema }),
system: VALIDATION_SYSTEM_PROMPT,
messages: [
{
role: "user",
content: [
{
type: "image",
image: imageData,
},
{
type: "text",
text: "Please analyze this diagram for visual quality issues.",
},
],
},
],
maxOutputTokens: 1024,
abortSignal: AbortSignal.timeout(timeout),
onFinish: ({ output, totalUsage }) => {
if (countsQuota && totalUsage) {
recordTokenUsage(
userId,
(totalUsage.inputTokens || 0) +
(totalUsage.outputTokens || 0),
)
}
if (sessionId && output) {
console.log(
`[validate-diagram] Session ${sessionId}: valid=${output.valid}, issues=${output.issues?.length ?? 0}`,
)
}
},
})
return result.toTextStreamResponse()
} catch (error) {
// Log with session context if available
const errorMessage =
error instanceof Error ? error.message : String(error)
console.error("[validate-diagram] Error:", errorMessage)
// On error, return valid to not block the user
return createStreamingResponse(DEFAULT_VALID_RESULT)
}
}