Version

main branch as of 2026-06-09

Steps to reproduce

Send a request with binary bytes that are not valid UTF-8, then read request.post_data:

async with page.expect_request("**/upload") as request_info:
    await page.evaluate("""
        () => fetch('/upload', {
            method: 'POST',
            body: new Uint8Array([255, 254])
        })
    """)
request = await request_info.value
print(request.post_data)

Expected behavior

request.post_data should not crash. It can return a best-effort UTF-8 string, while request.post_data_buffer remains the exact binary-safe API.

This would match JavaScript behavior, where Buffer.toString("utf-8") replaces invalid bytes instead of throwing.

Actual behavior

Python decodes request body bytes with strict UTF-8 decoding, so request.post_data can raise:

UnicodeDecodeError: 'utf-8' codec can't decode byte 0xff ...

Potential fix direction

Decode text post data with replacement, for example decode("utf-8", errors="replace"), and add a regression test that verifies:

• request.post_data does not raise for invalid UTF-8 bytes
• request.post_data_buffer still returns the exact original bytes