Split découpe un fichier en plusieurs.

import { CloudRakerClient } from "@cloudraker/api";

const client = new CloudRakerClient({ token: "YOUR_API_KEY" });
const packet = { url: "https://example.com/intake-packet.pdf" };

// 1 - Find where each document starts, page by page
const classified = await client.classify({
  body: {
    file: packet,
    granularity: "page",
    classes: [
      { id: "invoice", description: "A vendor invoice with a total due." },
      { id: "w9", description: "An IRS Form W-9." },
      { id: "agreement", description: "A signed services agreement." },
    ],
  },
});

// 2 - Cut the packet on those boundaries (no model runs here)
const run = await client.split({
  body: { file: packet, classifyRunId: classified.id },
});

for (const piece of run.output.splits) {
  console.log(piece.classId, piece.startPage, piece.endPage, piece.confidence);
}

// 3 - Every child is addressable on its own
const data = await client.extract({
  body: {
    files: run.output.documentIds.map((id) => ({ id })),
    schema: {
      type: "object",
      properties: { total_due: { type: ["number", "null"] } },
    },
    citations: true,
  },
});
console.log(data.output?.value);

from cloudraker.client import CloudRaker

client = CloudRaker(token="YOUR_API_KEY")
packet = {"url": "https://example.com/intake-packet.pdf"}

# 1 - Find where each document starts, page by page
classified = client.classify(
    file=packet,
    granularity="page",
    classes=[
        {"id": "invoice", "description": "A vendor invoice with a total due."},
        {"id": "w9", "description": "An IRS Form W-9."},
        {"id": "agreement", "description": "A signed services agreement."},
    ],
)

# 2 - Cut the packet on those boundaries (no model runs here)
run = client.split(file=packet, classify_run_id=classified.id)

for piece in run.output.splits:
    print(piece.class_id, piece.start_page, piece.end_page, piece.confidence)

# 3 - Every child is addressable on its own
data = client.extract(
    files=[{"id": file_id} for file_id in run.output.document_ids],
    schema={
        "type": "object",
        "properties": {"total_due": {"type": ["number", "null"]}},
    },
    citations=True,
)
print(data.output.value)