Merge, split, compress

curl https://api.reflowpdf.com/v1/pdf/merge \
  -H "Authorization: Bearer $REFLOWPDF_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "filename": "Invoice with terms",
    "parts": [
      { "template_id": "invoice" },
      { "pdf_url": "https://example.com/terms.pdf", "pages": "1-2" }
    ]
  }' -o invoice-with-terms.pdf

const pdf = await reflow.pdf.merge({
  filename: 'Invoice with terms',
  parts: [
    { template_id: 'invoice' },   // the template with its own sample data
    { pdf: await readFile('terms.pdf'), pages: '1-2' },   // bytes are sent in base64 for you
  ],
})

pdf = client.pdf.merge(
    filename="Invoice with terms",
    parts=[
        {"template_id": "invoice"},
        {"pdf": open("terms.pdf", "rb").read(), "pages": "1-2"},
    ],
)

A PDF, from anywhere#

With pages, a part gives some of its pages in the order written — "3,1", "2-" (to the end), "1-3,7". An encrypted PDF opens with its password.

Merge#

POST /v1/pdf/merge with parts (1 to 50): their pages one after another in a new PDF, filename naming it. The PDF itself by default; with Accept: application/json, its link (24 hours) and facts, like a print's.

• Each page keeps its content, its links and its form fields.
• The merged file is a new document: the parts' outlines, tags (their structure tree) and standard claims (PDF/A, PDF/UA) do not carry over, so it claims none it no longer keeps. To deliver PDF/UA, print the whole document as one: one HTML, its parts as sections.
• Up to 100 MB a part, 200 MB together, 10,000 pages.

Split#

POST /v1/pdf/split with a PDF and ranges (each a new file), or every (a new file every so many pages). The response is the files' links and, when there are several, a ZIP of them, all kept 24 hours:

{
  "id": "spl_N3wq8sHkT2xV9mRp4LcB1a",
  "object": "split",
  "files": [
    {
      "id": "Hq8vR2cT9xKmL4pW",
      "pages": "1",
      "page_count": 1,
      "url": "https://api.reflowpdf.com/v1/files/Hq8vR2cT9xKmL4pW/report-pages-1.pdf",
      "filename": "report-pages-1.pdf",
      "bytes": 48213
    },
    {
      "id": "Tx2mB7cL1pQ9wE4r",
      "pages": "2-",
      "page_count": 11,
      "url": "https://api.reflowpdf.com/v1/files/Tx2mB7cL1pQ9wE4r/report-pages-2-.pdf",
      "filename": "report-pages-2-.pdf",
      "bytes": 402117
    }
  ],
  "archive": {
    "url": "https://api.reflowpdf.com/v1/files/Zk3nP8vR2cT9xKmL/report-split.zip",
    "filename": "report-split.zip",
    "bytes": 450566
  },
  "expires_at": "2026-11-03T10:22:31.000Z"
}

Compress#

POST /v1/pdf/compress with a PDF: written again losslessly, its streams compressed anew at the highest level, its objects packed in object streams, resources no page uses removed. It is never larger than it was (ReflowPDF-Bytes-Before says its size before; the JSON response, bytes_before), unless linearize: true asks for fast web view, its first page shown before the rest has arrived: that adds its hint tables, a few KB. Its images stay as they are, pixel for pixel; an encrypted PDF keeps its encryption.

Errors#