mime.multipart

mime.multipart splits an in-memory MIME multipart body into individual parts. It preserves their wire order, repeated form field names, and raw body bytes, including binary data and trailing whitespace.

import mime.multipart

body := '--example\r\nContent-Disposition: form-data; name="message"\r\n\r\nhello\r\n--example--\r\n'
parts := multipart.parse(body, 'example')
assert parts.len == 1
assert parts[0].field_name == 'message'
assert parts[0].data == 'hello'

Pass the unquoted boundary parameter value from the outer Content-Type header. Boundaries follow the MIME character restrictions and have a maximum length of 70 bytes. Delimiters must occur at the start of a line and use CRLF line endings. Optional spaces and tabs after delimiters are accepted. Preamble and epilogue are ignored. The closing delimiter can end at EOF without a final CRLF.

Each Part contains field_name and file_name from Content-Disposition, content_type from Content-Type, and data without the delimiter's preceding CRLF. Missing metadata is an empty string. Header names are case insensitive; header order, folded headers, and quoted disposition parameters are supported. Parts without Content-Disposition are also returned, with empty names.

The parse API returns an array rather than an error. An invalid boundary or a body without a delimiter returns an empty array. A malformed header block or incomplete part stops parsing; complete preceding parts remain in the result. Do not use the returned part count alone to validate an entire message.

This module does not decode transfer encodings, charsets, encoded filename parameters, or nested multipart bodies. Filenames are metadata; callers choose how to validate them before using them as filesystem paths.

The framing follows [RFC 2046 section 5.1.1][mime]. Ordered form fields and duplicate names follow [RFC 7578 section 5.2][forms].

[mime]: https://www.rfc-editor.org/rfc/rfc2046.html#section-5.1.1 [forms]: https://www.rfc-editor.org/rfc/rfc7578.html#section-5.2

fn parse(body string, boundary string) []Part

parse returns complete multipart parts in wire order, preserving duplicate field names and body bytes. The boundary is the unquoted MIME parameter value. Invalid boundaries return no parts. Parsing stops at the first malformed or incomplete part and returns the preceding complete parts. Preamble and epilogue are ignored; transfer encodings and nested multipart bodies are not decoded.

struct Part {
pub:
	field_name   string
	file_name    string
	content_type string
	data         string
}

Part represents a single part in a multipart message.