client-api: Add a type for the Content-Disposition HTTP header

This commit is contained in:
Kévin Commaille 2024-06-27 18:51:01 +02:00 committed by Kévin Commaille
parent f73ba5556c
commit 463f89b0eb
14 changed files with 1190 additions and 50 deletions

View File

@ -7,6 +7,9 @@ Breaking changes:
- Change type of `client_secret` field in `ThirdpartyIdCredentials`
from `Box<ClientSecret>` to `OwnedClientSecret`
- Make `id_server` and `id_access_token` in `ThirdpartyIdCredentials` optional
- The `content_disposition` fields of `media::get_content::v3::Response` and
`media::get_content_as_filename::v3::Response` use now the strongly typed
`ContentDisposition` instead of strings.
Improvements:

View File

@ -12,6 +12,7 @@ pub mod v1 {
use http::header::{CONTENT_DISPOSITION, CONTENT_TYPE};
use ruma_common::{
api::{request, response, Metadata},
http_headers::ContentDisposition,
metadata, IdParseError, MxcUri, OwnedServerName,
};
@ -62,12 +63,8 @@ pub mod v1 {
/// The value of the `Content-Disposition` HTTP header, possibly containing the name of the
/// file that was previously uploaded.
///
/// See [MDN] for the syntax.
///
/// [MDN]: https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Content-Disposition#Syntax
#[ruma_api(header = CONTENT_DISPOSITION)]
pub content_disposition: Option<String>,
pub content_disposition: Option<ContentDisposition>,
}
impl Request {

View File

@ -12,6 +12,7 @@ pub mod v1 {
use http::header::{CONTENT_DISPOSITION, CONTENT_TYPE};
use ruma_common::{
api::{request, response, Metadata},
http_headers::ContentDisposition,
metadata, IdParseError, MxcUri, OwnedServerName,
};
@ -66,12 +67,8 @@ pub mod v1 {
/// The value of the `Content-Disposition` HTTP header, possibly containing the name of the
/// file that was previously uploaded.
///
/// See [MDN] for the syntax.
///
/// [MDN]: https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Content-Disposition#Syntax
#[ruma_api(header = CONTENT_DISPOSITION)]
pub content_disposition: Option<String>,
pub content_disposition: Option<ContentDisposition>,
}
impl Request {

View File

@ -3,6 +3,10 @@
use http::{header::HeaderName, HeaderValue};
use ruma_common::api::error::{HeaderDeserializationError, HeaderSerializationError};
pub use ruma_common::http_headers::{
ContentDisposition, ContentDispositionParseError, ContentDispositionType, TokenString,
TokenStringParseError,
};
use web_time::{Duration, SystemTime, UNIX_EPOCH};
/// The [`Cross-Origin-Resource-Policy`] HTTP response header.

View File

@ -12,6 +12,7 @@ pub mod v3 {
use http::header::{CONTENT_DISPOSITION, CONTENT_TYPE};
use ruma_common::{
api::{request, response, Metadata},
http_headers::ContentDisposition,
metadata, IdParseError, MxcUri, OwnedServerName,
};
@ -87,12 +88,8 @@ pub mod v3 {
/// The value of the `Content-Disposition` HTTP header, possibly containing the name of the
/// file that was previously uploaded.
///
/// See [MDN] for the syntax.
///
/// [MDN]: https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Content-Disposition#Syntax
#[ruma_api(header = CONTENT_DISPOSITION)]
pub content_disposition: Option<String>,
pub content_disposition: Option<ContentDisposition>,
/// The value of the `Cross-Origin-Resource-Policy` HTTP header.
///

View File

@ -12,6 +12,7 @@ pub mod v3 {
use http::header::{CONTENT_DISPOSITION, CONTENT_TYPE};
use ruma_common::{
api::{request, response, Metadata},
http_headers::ContentDisposition,
metadata, IdParseError, MxcUri, OwnedServerName,
};
@ -91,12 +92,8 @@ pub mod v3 {
/// The value of the `Content-Disposition` HTTP header, possibly containing the name of the
/// file that was previously uploaded.
///
/// See [MDN] for the syntax.
///
/// [MDN]: https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Content-Disposition#Syntax
#[ruma_api(header = CONTENT_DISPOSITION)]
pub content_disposition: Option<String>,
pub content_disposition: Option<ContentDisposition>,
/// The value of the `Cross-Origin-Resource-Policy` HTTP header.
///

View File

@ -0,0 +1,104 @@
//! Helpers for HTTP headers.
use std::borrow::Cow;
mod content_disposition;
mod rfc8187;
pub use self::content_disposition::{
ContentDisposition, ContentDispositionParseError, ContentDispositionType, TokenString,
TokenStringParseError,
};
/// Whether the given byte is a [`token` char].
///
/// [`token` char]: https://datatracker.ietf.org/doc/html/rfc9110#section-5.6.2
pub const fn is_tchar(b: u8) -> bool {
b.is_ascii_alphanumeric()
|| matches!(
b,
b'!' | b'#'
| b'$'
| b'%'
| b'&'
| b'\''
| b'*'
| b'+'
| b'-'
| b'.'
| b'^'
| b'_'
| b'`'
| b'|'
| b'~'
)
}
/// Whether the given bytes slice is a [`token`].
///
/// [`token`]: https://datatracker.ietf.org/doc/html/rfc9110#section-5.6.2
pub fn is_token(bytes: &[u8]) -> bool {
bytes.iter().all(|b| is_tchar(*b))
}
/// Whether the given string is a [`token`].
///
/// [`token`]: https://datatracker.ietf.org/doc/html/rfc9110#section-5.6.2
pub fn is_token_string(s: &str) -> bool {
is_token(s.as_bytes())
}
/// Whether the given char is a [visible US-ASCII char].
///
/// [visible US-ASCII char]: https://datatracker.ietf.org/doc/html/rfc5234#appendix-B.1
pub const fn is_vchar(c: char) -> bool {
matches!(c, '\x21'..='\x7E')
}
/// Whether the given char is in the US-ASCII character set and allowed inside a [quoted string].
///
/// Contrary to the definition of quoted strings, this doesn't allow `obs-text` characters, i.e.
/// non-US-ASCII characters, as we usually deal with UTF-8 strings rather than ISO-8859-1 strings.
///
/// [quoted string]: https://datatracker.ietf.org/doc/html/rfc9110#section-5.6.4
pub const fn is_ascii_string_quotable(c: char) -> bool {
is_vchar(c) || matches!(c, '\x09' | '\x20')
}
/// Remove characters that do not pass [`is_ascii_string_quotable()`] from the given string.
///
/// [quoted string]: https://datatracker.ietf.org/doc/html/rfc9110#section-5.6.4
pub fn sanitize_for_ascii_quoted_string(value: &str) -> Cow<'_, str> {
if value.chars().all(is_ascii_string_quotable) {
return Cow::Borrowed(value);
}
Cow::Owned(value.chars().filter(|c| is_ascii_string_quotable(*c)).collect())
}
/// If the US-ASCII field value does not contain only token chars, convert it to a [quoted string].
///
/// The string should be sanitized with [`sanitize_for_ascii_quoted_string()`] or should only
/// contain characters that pass [`is_ascii_string_quotable()`].
///
/// [quoted string]: https://datatracker.ietf.org/doc/html/rfc9110#section-5.6.4
pub fn quote_ascii_string_if_required(value: &str) -> Cow<'_, str> {
if !value.is_empty() && is_token_string(value) {
return Cow::Borrowed(value);
}
let value = value.replace('\\', r#"\\"#).replace('"', r#"\""#);
Cow::Owned(format!("\"{value}\""))
}
/// Removes the escape backslashes in the given string.
pub fn unescape_string(s: &str) -> String {
let mut is_escaped = false;
s.chars()
.filter(|c| {
is_escaped = *c == '\\' && !is_escaped;
!is_escaped
})
.collect()
}

View File

@ -0,0 +1,736 @@
//! Types to (de)serialize the `Content-Disposition` HTTP header.
use std::{fmt, ops::Deref, str::FromStr};
use ruma_macros::{
AsRefStr, AsStrAsRefStr, DebugAsRefStr, DisplayAsRefStr, OrdAsRefStr, PartialOrdAsRefStr,
};
use super::{
is_tchar, is_token, quote_ascii_string_if_required, rfc8187, sanitize_for_ascii_quoted_string,
unescape_string,
};
/// The value of a `Content-Disposition` HTTP header.
///
/// This implementation supports the `Content-Disposition` header format as defined for HTTP in [RFC
/// 6266].
///
/// The only supported parameter is `filename`. It is encoded or decoded as needed, using a quoted
/// string or the `ext-token = ext-value` format, with the encoding defined in [RFC 8187].
///
/// This implementation does not support serializing to the format defined for the
/// `multipart/form-data` content type in [RFC 7578]. It should however manage to parse the
/// disposition type and filename parameter of the body parts.
///
/// [RFC 6266]: https://datatracker.ietf.org/doc/html/rfc6266
/// [RFC 8187]: https://datatracker.ietf.org/doc/html/rfc8187
/// [RFC 7578]: https://datatracker.ietf.org/doc/html/rfc7578
#[derive(Debug, Clone, PartialEq, Eq, Default)]
#[cfg_attr(not(feature = "unstable-exhaustive-types"), non_exhaustive)]
pub struct ContentDisposition {
/// The disposition type.
pub disposition_type: ContentDispositionType,
/// The filename of the content.
pub filename: Option<String>,
}
impl ContentDisposition {
/// Creates a new `ContentDisposition` with the given disposition type.
pub fn new(disposition_type: ContentDispositionType) -> Self {
Self { disposition_type, filename: None }
}
/// Add the given filename to this `ContentDisposition`.
pub fn with_filename(mut self, filename: Option<String>) -> Self {
self.filename = filename;
self
}
}
impl fmt::Display for ContentDisposition {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
write!(f, "{}", self.disposition_type)?;
if let Some(filename) = &self.filename {
if filename.is_ascii() {
// First, remove all non-quotable characters, that is control characters.
let filename = sanitize_for_ascii_quoted_string(filename);
// We can use the filename parameter.
write!(f, "; filename={}", quote_ascii_string_if_required(&filename))?;
} else {
// We need to use RFC 8187 encoding.
write!(f, "; filename*={}", rfc8187::encode(filename))?;
}
}
Ok(())
}
}
impl TryFrom<&[u8]> for ContentDisposition {
type Error = ContentDispositionParseError;
fn try_from(value: &[u8]) -> Result<Self, Self::Error> {
let mut pos = 0;
skip_ascii_whitespaces(value, &mut pos);
if pos == value.len() {
return Err(ContentDispositionParseError::MissingDispositionType);
}
let disposition_type_start = pos;
// Find the next whitespace or `;`.
while let Some(byte) = value.get(pos) {
if byte.is_ascii_whitespace() || *byte == b';' {
break;
}
pos += 1;
}
let disposition_type =
ContentDispositionType::try_from(&value[disposition_type_start..pos])?;
// The `filename*` parameter (`filename_ext` here) using UTF-8 encoding should be used, but
// it is likely to be after the `filename` parameter containing only ASCII
// characters if both are present.
let mut filename_ext = None;
let mut filename = None;
// Parse the parameters. We ignore parameters that fail to parse for maximum compatibility.
while pos != value.len() {
if let Some(param) = RawParam::parse_next(value, &mut pos) {
if param.name.eq_ignore_ascii_case(b"filename*") {
if let Some(value) = param.decode_value() {
filename_ext = Some(value);
// We can stop parsing, this is the only parameter that we need.
break;
}
} else if param.name.eq_ignore_ascii_case(b"filename") {
if let Some(value) = param.decode_value() {
filename = Some(value);
}
}
}
}
Ok(Self { disposition_type, filename: filename_ext.or(filename) })
}
}
impl FromStr for ContentDisposition {
type Err = ContentDispositionParseError;
fn from_str(s: &str) -> Result<Self, Self::Err> {
s.as_bytes().try_into()
}
}
/// A raw parameter in a `Content-Disposition` HTTP header.
struct RawParam<'a> {
name: &'a [u8],
value: &'a [u8],
is_quoted_string: bool,
}
impl<'a> RawParam<'a> {
/// Parse the next `RawParam` in the given bytes, starting at the given position.
///
/// The position is updated during the parsing.
///
/// Returns `None` if no parameter was found or if an error occurred when parsing the
/// parameter.
fn parse_next(bytes: &'a [u8], pos: &mut usize) -> Option<Self> {
let name = parse_param_name(bytes, pos)?;
skip_ascii_whitespaces(bytes, pos);
if *pos == bytes.len() {
// We are at the end of the bytes and only have the parameter name.
return None;
}
if bytes[*pos] != b'=' {
// We should have an equal sign, there is a problem with the bytes and we can't recover
// from it.
// Skip to the end to stop the parsing.
*pos = bytes.len();
return None;
}
// Skip the equal sign.
*pos += 1;
skip_ascii_whitespaces(bytes, pos);
let (value, is_quoted_string) = parse_param_value(bytes, pos)?;
Some(Self { name, value, is_quoted_string })
}
/// Decode the value of this `RawParam`.
///
/// Returns `None` if decoding the param failed.
fn decode_value(&self) -> Option<String> {
if self.name.ends_with(b"*") {
rfc8187::decode(self.value).ok().map(|s| s.into_owned())
} else {
let s = String::from_utf8_lossy(self.value);
if self.is_quoted_string {
Some(unescape_string(&s))
} else {
Some(s.into_owned())
}
}
}
}
/// Skip ASCII whitespaces in the given bytes, starting at the given position.
///
/// The position is updated to after the whitespaces.
fn skip_ascii_whitespaces(bytes: &[u8], pos: &mut usize) {
while let Some(byte) = bytes.get(*pos) {
if !byte.is_ascii_whitespace() {
break;
}
*pos += 1;
}
}
/// Parse a parameter name in the given bytes, starting at the given position.
///
/// The position is updated while parsing.
///
/// Returns `None` if the end of the bytes was reached, or if an error was encountered.
fn parse_param_name<'a>(bytes: &'a [u8], pos: &mut usize) -> Option<&'a [u8]> {
skip_ascii_whitespaces(bytes, pos);
if *pos == bytes.len() {
// We are at the end of the bytes and didn't find anything.
return None;
}
let name_start = *pos;
// Find the end of the parameter name. The name can only contain token chars.
while let Some(byte) = bytes.get(*pos) {
if !is_tchar(*byte) {
break;
}
*pos += 1;
}
if *pos == bytes.len() {
// We are at the end of the bytes and only have the parameter name.
return None;
}
if bytes[*pos] == b';' {
// We are at the end of the parameter and only have the parameter name, skip the `;` and
// parse the next parameter.
*pos += 1;
return None;
}
let name = &bytes[name_start..*pos];
if name.is_empty() {
// It's probably a syntax error, we cannot recover from it.
*pos = bytes.len();
return None;
}
Some(name)
}
/// Parse a parameter value in the given bytes, starting at the given position.
///
/// The position is updated while parsing.
///
/// Returns a `(value, is_quoted_string)` tuple if parsing succeeded.
/// Returns `None` if the end of the bytes was reached, or if an error was encountered.
fn parse_param_value<'a>(bytes: &'a [u8], pos: &mut usize) -> Option<(&'a [u8], bool)> {
skip_ascii_whitespaces(bytes, pos);
if *pos == bytes.len() {
// We are at the end of the bytes and didn't find anything.
return None;
}
let is_quoted_string = bytes[*pos] == b'"';
if is_quoted_string {
// Skip the start double quote.
*pos += 1;
}
let value_start = *pos;
// Keep track of whether the next byte is escaped with a backslash.
let mut escape_next = false;
// Find the end of the value, it's a whitespace or a semi-colon, or a double quote if the string
// is quoted.
while let Some(byte) = bytes.get(*pos) {
if !is_quoted_string && (byte.is_ascii_whitespace() || *byte == b';') {
break;
}
if is_quoted_string && *byte == b'"' && !escape_next {
break;
}
escape_next = *byte == b'\\' && !escape_next;
*pos += 1;
}
let value = &bytes[value_start..*pos];
if is_quoted_string && *pos != bytes.len() {
// Skip the end double quote.
*pos += 1;
}
skip_ascii_whitespaces(bytes, pos);
// Check for parameters separator if we are not at the end of the string.
if *pos != bytes.len() {
if bytes[*pos] == b';' {
// Skip the `;` at the end of the parameter.
*pos += 1;
} else {
// We should have a `;`, there is a problem with the bytes and we can't recover
// from it.
// Skip to the end to stop the parsing.
*pos = bytes.len();
return None;
}
}
Some((value, is_quoted_string))
}
/// An error encountered when trying to parse an invalid [`ContentDisposition`].
#[derive(Debug, Clone, Copy, PartialEq, Eq, thiserror::Error)]
#[non_exhaustive]
pub enum ContentDispositionParseError {
/// The disposition type is missing.
#[error("disposition type is missing")]
MissingDispositionType,
/// The disposition type is invalid.
#[error("invalid disposition type: {0}")]
InvalidDispositionType(#[from] TokenStringParseError),
}
/// A disposition type in the `Content-Disposition` HTTP header as defined in [Section 4.2 of RFC
/// 6266].
///
/// This type can hold an arbitrary [`TokenString`]. To build this with a custom value, convert it
/// from a `TokenString` with `::from()` / `.into()`. To check for values that are not available as
/// a documented variant here, use its string representation, obtained through
/// [`.as_str()`](Self::as_str()).
///
/// Comparisons with other string types are done case-insensitively.
///
/// [Section 4.2 of RFC 6266]: https://datatracker.ietf.org/doc/html/rfc6266#section-4.2
#[derive(
Clone,
Default,
AsRefStr,
DebugAsRefStr,
AsStrAsRefStr,
DisplayAsRefStr,
PartialOrdAsRefStr,
OrdAsRefStr,
)]
#[ruma_enum(rename_all = "lowercase")]
#[non_exhaustive]
pub enum ContentDispositionType {
/// The content can be displayed.
///
/// This is the default.
#[default]
Inline,
/// The content should be downloaded instead of displayed.
Attachment,
#[doc(hidden)]
_Custom(TokenString),
}
impl ContentDispositionType {
/// Try parsing a `&str` into a `ContentDispositionType`.
pub fn parse(s: &str) -> Result<Self, TokenStringParseError> {
Self::from_str(s)
}
}
impl From<TokenString> for ContentDispositionType {
fn from(value: TokenString) -> Self {
if value.eq_ignore_ascii_case("inline") {
Self::Inline
} else if value.eq_ignore_ascii_case("attachment") {
Self::Attachment
} else {
Self::_Custom(value)
}
}
}
impl<'a> TryFrom<&'a [u8]> for ContentDispositionType {
type Error = TokenStringParseError;
fn try_from(value: &'a [u8]) -> Result<Self, Self::Error> {
if value.eq_ignore_ascii_case(b"inline") {
Ok(Self::Inline)
} else if value.eq_ignore_ascii_case(b"attachment") {
Ok(Self::Attachment)
} else {
TokenString::try_from(value).map(Self::_Custom)
}
}
}
impl FromStr for ContentDispositionType {
type Err = TokenStringParseError;
fn from_str(s: &str) -> Result<Self, Self::Err> {
s.as_bytes().try_into()
}
}
impl PartialEq<ContentDispositionType> for ContentDispositionType {
fn eq(&self, other: &ContentDispositionType) -> bool {
self.as_str().eq_ignore_ascii_case(other.as_str())
}
}
impl Eq for ContentDispositionType {}
impl PartialEq<TokenString> for ContentDispositionType {
fn eq(&self, other: &TokenString) -> bool {
self.as_str().eq_ignore_ascii_case(other.as_str())
}
}
impl<'a> PartialEq<&'a str> for ContentDispositionType {
fn eq(&self, other: &&'a str) -> bool {
self.as_str().eq_ignore_ascii_case(other)
}
}
/// A non-empty string consisting only of `token`s as defined in [RFC 9110 Section 3.2.6].
///
/// This is a string that can only contain a limited character set.
///
/// [RFC 7230 Section 3.2.6]: https://datatracker.ietf.org/doc/html/rfc7230#section-3.2.6
#[derive(
Clone,
PartialEq,
Eq,
DebugAsRefStr,
AsStrAsRefStr,
DisplayAsRefStr,
PartialOrdAsRefStr,
OrdAsRefStr,
)]
pub struct TokenString(Box<str>);
impl TokenString {
/// Try parsing a `&str` into a `TokenString`.
pub fn parse(s: &str) -> Result<Self, TokenStringParseError> {
Self::from_str(s)
}
}
impl Deref for TokenString {
type Target = str;
fn deref(&self) -> &Self::Target {
self.as_ref()
}
}
impl AsRef<str> for TokenString {
fn as_ref(&self) -> &str {
&self.0
}
}
impl<'a> PartialEq<&'a str> for TokenString {
fn eq(&self, other: &&'a str) -> bool {
self.as_str().eq(*other)
}
}
impl<'a> TryFrom<&'a [u8]> for TokenString {
type Error = TokenStringParseError;
fn try_from(value: &'a [u8]) -> Result<Self, Self::Error> {
if value.is_empty() {
Err(TokenStringParseError::Empty)
} else if is_token(value) {
let s = std::str::from_utf8(value).expect("ASCII bytes are valid UTF-8");
Ok(Self(s.into()))
} else {
Err(TokenStringParseError::InvalidCharacter)
}
}
}
impl FromStr for TokenString {
type Err = TokenStringParseError;
fn from_str(s: &str) -> Result<Self, Self::Err> {
s.as_bytes().try_into()
}
}
/// The parsed string contains a character not allowed for a [`TokenString`].
#[derive(Debug, Clone, Copy, PartialEq, Eq, thiserror::Error)]
#[non_exhaustive]
pub enum TokenStringParseError {
/// The string is empty.
#[error("string is empty")]
Empty,
/// The string contains an invalid character for a token string.
#[error("string contains invalid character")]
InvalidCharacter,
}
#[cfg(test)]
mod tests {
use std::str::FromStr;
use super::{ContentDisposition, ContentDispositionType};
#[test]
fn parse_content_disposition_valid() {
// Only disposition type.
let content_disposition = ContentDisposition::from_str("inline").unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Inline);
assert_eq!(content_disposition.filename, None);
// Only disposition type with separator.
let content_disposition = ContentDisposition::from_str("attachment;").unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Attachment);
assert_eq!(content_disposition.filename, None);
// Unknown disposition type and parameters.
let content_disposition =
ContentDisposition::from_str("custom; foo=bar; foo*=utf-8''b%C3%A0r'").unwrap();
assert_eq!(content_disposition.disposition_type.as_str(), "custom");
assert_eq!(content_disposition.filename, None);
// Disposition type and filename.
let content_disposition = ContentDisposition::from_str("inline; filename=my_file").unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Inline);
assert_eq!(content_disposition.filename.unwrap(), "my_file");
// Case insensitive.
let content_disposition = ContentDisposition::from_str("INLINE; FILENAME=my_file").unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Inline);
assert_eq!(content_disposition.filename.unwrap(), "my_file");
// Extra spaces.
let content_disposition =
ContentDisposition::from_str(" INLINE ;FILENAME = my_file ").unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Inline);
assert_eq!(content_disposition.filename.unwrap(), "my_file");
// Unsupported filename* is skipped and falls back to ASCII filename.
let content_disposition = ContentDisposition::from_str(
r#"attachment; filename*=iso-8859-1''foo-%E4.html; filename="foo-a.html"#,
)
.unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Attachment);
assert_eq!(content_disposition.filename.unwrap(), "foo-a.html");
// filename could be UTF-8 for extra compatibility (with `form-data` for example).
let content_disposition =
ContentDisposition::from_str(r#"form-data; name=upload; filename="文件.webp""#)
.unwrap();
assert_eq!(content_disposition.disposition_type.as_str(), "form-data");
assert_eq!(content_disposition.filename.unwrap(), "文件.webp");
}
#[test]
fn parse_content_disposition_invalid_type() {
// Empty.
ContentDisposition::from_str("").unwrap_err();
// Missing disposition type.
ContentDisposition::from_str("; foo=bar").unwrap_err();
}
#[test]
fn parse_content_disposition_invalid_parameters() {
// Unexpected `:` after parameter name, filename parameter is not reached.
let content_disposition =
ContentDisposition::from_str("inline; foo:bar; filename=my_file").unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Inline);
assert_eq!(content_disposition.filename, None);
// Same error, but after filename, so filename was parser.
let content_disposition =
ContentDisposition::from_str("inline; filename=my_file; foo:bar").unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Inline);
assert_eq!(content_disposition.filename.unwrap(), "my_file");
// Missing `;` between parameters, filename parameter is not parsed successfully.
let content_disposition =
ContentDisposition::from_str("inline; filename=my_file foo=bar").unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Inline);
assert_eq!(content_disposition.filename, None);
}
#[test]
fn content_disposition_serialize() {
// Only disposition type.
let content_disposition = ContentDisposition::new(ContentDispositionType::Inline);
let serialized = content_disposition.to_string();
assert_eq!(serialized, "inline");
// Disposition type and ASCII filename without space.
let content_disposition = ContentDisposition::new(ContentDispositionType::Attachment)
.with_filename(Some("my_file".to_owned()));
let serialized = content_disposition.to_string();
assert_eq!(serialized, "attachment; filename=my_file");
// Disposition type and ASCII filename with space.
let content_disposition = ContentDisposition::new(ContentDispositionType::Attachment)
.with_filename(Some("my file".to_owned()));
let serialized = content_disposition.to_string();
assert_eq!(serialized, r#"attachment; filename="my file""#);
// Disposition type and ASCII filename with double quote and backslash.
let content_disposition = ContentDisposition::new(ContentDispositionType::Attachment)
.with_filename(Some(r#""my"\file"#.to_owned()));
let serialized = content_disposition.to_string();
assert_eq!(serialized, r#"attachment; filename="\"my\"\\file""#);
// Disposition type and UTF-8 filename.
let content_disposition = ContentDisposition::new(ContentDispositionType::Attachment)
.with_filename(Some("Mi Corazón".to_owned()));
let serialized = content_disposition.to_string();
assert_eq!(serialized, "attachment; filename*=utf-8''Mi%20Coraz%C3%B3n");
// Sanitized filename.
let content_disposition = ContentDisposition::new(ContentDispositionType::Attachment)
.with_filename(Some("my\r\nfile".to_owned()));
let serialized = content_disposition.to_string();
assert_eq!(serialized, "attachment; filename=myfile");
}
#[test]
fn rfc6266_examples() {
// Basic syntax with unquoted filename.
let unquoted = "Attachment; filename=example.html";
let content_disposition = ContentDisposition::from_str(unquoted).unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Attachment);
assert_eq!(content_disposition.filename.as_deref().unwrap(), "example.html");
let reserialized = content_disposition.to_string();
assert_eq!(reserialized, "attachment; filename=example.html");
// With quoted filename, case insensitivity and extra whitespaces.
let quoted = r#"INLINE; FILENAME= "an example.html""#;
let content_disposition = ContentDisposition::from_str(quoted).unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Inline);
assert_eq!(content_disposition.filename.as_deref().unwrap(), "an example.html");
let reserialized = content_disposition.to_string();
assert_eq!(reserialized, r#"inline; filename="an example.html""#);
// With RFC 8187-encoded UTF-8 filename.
let rfc8187 = "attachment; filename*= UTF-8''%e2%82%ac%20rates";
let content_disposition = ContentDisposition::from_str(rfc8187).unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Attachment);
assert_eq!(content_disposition.filename.as_deref().unwrap(), "€ rates");
let reserialized = content_disposition.to_string();
assert_eq!(reserialized, r#"attachment; filename*=utf-8''%E2%82%AC%20rates"#);
// With RFC 8187-encoded UTF-8 filename with fallback ASCII filename.
let rfc8187_with_fallback =
r#"attachment; filename="EURO rates"; filename*=utf-8''%e2%82%ac%20rates"#;
let content_disposition = ContentDisposition::from_str(rfc8187_with_fallback).unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Attachment);
assert_eq!(content_disposition.filename.as_deref().unwrap(), "€ rates");
}
#[test]
fn rfc8187_examples() {
// Those examples originate from RFC 8187, but are changed to fit the expectations here:
//
// - A disposition type is added
// - The title parameter is renamed to filename
// Basic syntax with unquoted filename.
let unquoted = "attachment; foo= bar; filename=Economy";
let content_disposition = ContentDisposition::from_str(unquoted).unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Attachment);
assert_eq!(content_disposition.filename.as_deref().unwrap(), "Economy");
let reserialized = content_disposition.to_string();
assert_eq!(reserialized, "attachment; filename=Economy");
// With quoted filename.
let quoted = r#"attachment; foo=bar; filename="US-$ rates""#;
let content_disposition = ContentDisposition::from_str(quoted).unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Attachment);
assert_eq!(content_disposition.filename.as_deref().unwrap(), "US-$ rates");
let reserialized = content_disposition.to_string();
assert_eq!(reserialized, r#"attachment; filename="US-$ rates""#);
// With RFC 8187-encoded UTF-8 filename.
let rfc8187 = "attachment; foo=bar; filename*=utf-8'en'%C2%A3%20rates";
let content_disposition = ContentDisposition::from_str(rfc8187).unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Attachment);
assert_eq!(content_disposition.filename.as_deref().unwrap(), "£ rates");
let reserialized = content_disposition.to_string();
assert_eq!(reserialized, r#"attachment; filename*=utf-8''%C2%A3%20rates"#);
// With RFC 8187-encoded UTF-8 filename again.
let rfc8187_other =
r#"attachment; foo=bar; filename*=UTF-8''%c2%a3%20and%20%e2%82%ac%20rates"#;
let content_disposition = ContentDisposition::from_str(rfc8187_other).unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Attachment);
assert_eq!(content_disposition.filename.as_deref().unwrap(), "£ and € rates");
let reserialized = content_disposition.to_string();
assert_eq!(
reserialized,
r#"attachment; filename*=utf-8''%C2%A3%20and%20%E2%82%AC%20rates"#
);
// With RFC 8187-encoded UTF-8 filename with fallback ASCII filename.
let rfc8187_with_fallback = r#"attachment; foo=bar; filename="EURO exchange rates"; filename*=utf-8''%e2%82%ac%20exchange%20rates"#;
let content_disposition = ContentDisposition::from_str(rfc8187_with_fallback).unwrap();
assert_eq!(content_disposition.disposition_type, ContentDispositionType::Attachment);
assert_eq!(content_disposition.filename.as_deref().unwrap(), "€ exchange rates");
let reserialized = content_disposition.to_string();
assert_eq!(reserialized, r#"attachment; filename*=utf-8''%E2%82%AC%20exchange%20rates"#);
}
}

View File

@ -0,0 +1,76 @@
//! Encoding and decoding functions according to [RFC 8187].
//!
//! [RFC 8187]: https://datatracker.ietf.org/doc/html/rfc8187
use std::borrow::Cow;
use percent_encoding::{AsciiSet, NON_ALPHANUMERIC};
/// The characters to percent-encode according to the `attr-char` set.
const ATTR_CHAR: AsciiSet = NON_ALPHANUMERIC
.remove(b'!')
.remove(b'#')
.remove(b'$')
.remove(b'&')
.remove(b'+')
.remove(b'-')
.remove(b'.')
.remove(b'^')
.remove(b'_')
.remove(b'`')
.remove(b'|')
.remove(b'~');
/// Encode the given string according to [RFC 8187].
///
/// [RFC 8187]: https://datatracker.ietf.org/doc/html/rfc8187
pub(super) fn encode(s: &str) -> String {
let encoded = percent_encoding::utf8_percent_encode(s, &ATTR_CHAR);
format!("utf-8''{encoded}")
}
/// Decode the given bytes according to [RFC 8187].
///
/// Only the UTF-8 character set is supported, all other character sets return an error.
///
/// [RFC 8187]: https://datatracker.ietf.org/doc/html/rfc8187
pub(super) fn decode(bytes: &[u8]) -> Result<Cow<'_, str>, Rfc8187DecodeError> {
if bytes.is_empty() {
return Err(Rfc8187DecodeError::Empty);
}
let mut parts = bytes.split(|b| *b == b'\'');
let charset = parts.next().ok_or(Rfc8187DecodeError::WrongPartsCount)?;
let _lang = parts.next().ok_or(Rfc8187DecodeError::WrongPartsCount)?;
let encoded = parts.next().ok_or(Rfc8187DecodeError::WrongPartsCount)?;
if parts.next().is_some() {
return Err(Rfc8187DecodeError::WrongPartsCount);
}
if !charset.eq_ignore_ascii_case(b"utf-8") {
return Err(Rfc8187DecodeError::NotUtf8);
}
// For maximum compatibility, do a lossy conversion.
Ok(percent_encoding::percent_decode(encoded).decode_utf8_lossy())
}
/// All errors encountered when trying to decode a string according to [RFC 8187].
///
/// [RFC 8187]: https://datatracker.ietf.org/doc/html/rfc8187
#[derive(Debug, Clone, Copy, PartialEq, Eq, thiserror::Error)]
#[non_exhaustive]
pub(super) enum Rfc8187DecodeError {
/// The string is empty.
#[error("string is empty")]
Empty,
/// The string does not contain the right number of parts.
#[error("string does not contain the right number of parts")]
WrongPartsCount,
/// The character set is not UTF-8.
#[error("character set is not UTF-8")]
NotUtf8,
}

View File

@ -24,6 +24,7 @@ pub mod authentication;
pub mod canonical_json;
pub mod directory;
pub mod encryption;
pub mod http_headers;
mod identifiers;
mod percent_encode;
pub mod power_levels;

View File

@ -7,6 +7,7 @@ mod header_override;
mod manual_endpoint_impl;
mod no_fields;
mod optional_headers;
mod required_headers;
mod ruma_api;
mod ruma_api_macros;
mod status_override;

View File

@ -1,6 +1,11 @@
use http::header::LOCATION;
use assert_matches2::assert_matches;
use http::header::{CONTENT_DISPOSITION, LOCATION};
use ruma_common::{
api::{request, response, Metadata},
api::{
request, response, IncomingRequest, IncomingResponse, MatrixVersion, Metadata,
OutgoingRequest, OutgoingResponse, SendAccessToken,
},
http_headers::{ContentDisposition, ContentDispositionType},
metadata,
};
@ -13,16 +18,128 @@ const METADATA: Metadata = metadata! {
}
};
/// Request type for the `no_fields` endpoint.
/// Request type for the `optional_headers` endpoint.
#[request]
pub struct Request {
#[ruma_api(header = LOCATION)]
pub location: Option<String>,
#[ruma_api(header = CONTENT_DISPOSITION)]
pub content_disposition: Option<ContentDisposition>,
}
/// Response type for the `no_fields` endpoint.
/// Response type for the `optional_headers` endpoint.
#[response]
pub struct Response {
#[ruma_api(header = LOCATION)]
pub stuff: Option<String>,
#[ruma_api(header = CONTENT_DISPOSITION)]
pub content_disposition: Option<ContentDisposition>,
}
#[test]
fn request_serde_no_header() {
let req = Request { location: None, content_disposition: None };
let http_req = req
.clone()
.try_into_http_request::<Vec<u8>>(
"https://homeserver.tld",
SendAccessToken::None,
&[MatrixVersion::V1_1],
)
.unwrap();
assert_matches!(http_req.headers().get(LOCATION), None);
assert_matches!(http_req.headers().get(CONTENT_DISPOSITION), None);
let req2 = Request::try_from_http_request::<_, &str>(http_req, &[]).unwrap();
assert_eq!(req2.location, None);
assert_eq!(req2.content_disposition, None);
}
#[test]
fn request_serde_with_header() {
let location = "https://other.tld/page/";
let content_disposition = ContentDisposition::new(ContentDispositionType::Attachment)
.with_filename(Some("my_file".to_owned()));
let req = Request {
location: Some(location.to_owned()),
content_disposition: Some(content_disposition.clone()),
};
let mut http_req = req
.clone()
.try_into_http_request::<Vec<u8>>(
"https://homeserver.tld",
SendAccessToken::None,
&[MatrixVersion::V1_1],
)
.unwrap();
assert_matches!(http_req.headers().get(LOCATION), Some(_));
assert_matches!(http_req.headers().get(CONTENT_DISPOSITION), Some(_));
let req2 = Request::try_from_http_request::<_, &str>(http_req.clone(), &[]).unwrap();
assert_eq!(req2.location.unwrap(), location);
assert_eq!(req2.content_disposition.unwrap(), content_disposition);
// Try removing the headers.
http_req.headers_mut().remove(LOCATION).unwrap();
http_req.headers_mut().remove(CONTENT_DISPOSITION).unwrap();
let req3 = Request::try_from_http_request::<_, &str>(http_req.clone(), &[]).unwrap();
assert_eq!(req3.location, None);
assert_eq!(req3.content_disposition, None);
// Try setting invalid header.
http_req.headers_mut().insert(CONTENT_DISPOSITION, ";".try_into().unwrap());
let req4 = Request::try_from_http_request::<_, &str>(http_req, &[]).unwrap();
assert_eq!(req4.location, None);
assert_eq!(req4.content_disposition, None);
}
#[test]
fn response_serde_no_header() {
let res = Response { stuff: None, content_disposition: None };
let http_res = res.clone().try_into_http_response::<Vec<u8>>().unwrap();
assert_matches!(http_res.headers().get(LOCATION), None);
assert_matches!(http_res.headers().get(CONTENT_DISPOSITION), None);
let res2 = Response::try_from_http_response(http_res).unwrap();
assert_eq!(res2.stuff, None);
assert_eq!(res2.content_disposition, None);
}
#[test]
fn response_serde_with_header() {
let location = "https://other.tld/page/";
let content_disposition = ContentDisposition::new(ContentDispositionType::Attachment)
.with_filename(Some("my_file".to_owned()));
let res = Response {
stuff: Some(location.to_owned()),
content_disposition: Some(content_disposition.clone()),
};
let mut http_res = res.clone().try_into_http_response::<Vec<u8>>().unwrap();
assert_matches!(http_res.headers().get(LOCATION), Some(_));
assert_matches!(http_res.headers().get(CONTENT_DISPOSITION), Some(_));
let res2 = Response::try_from_http_response(http_res.clone()).unwrap();
assert_eq!(res2.stuff.unwrap(), location);
assert_eq!(res2.content_disposition.unwrap(), content_disposition);
// Try removing the headers.
http_res.headers_mut().remove(LOCATION).unwrap();
http_res.headers_mut().remove(CONTENT_DISPOSITION).unwrap();
let res3 = Response::try_from_http_response(http_res.clone()).unwrap();
assert_eq!(res3.stuff, None);
assert_eq!(res3.content_disposition, None);
// Try setting invalid header.
http_res.headers_mut().insert(CONTENT_DISPOSITION, ";".try_into().unwrap());
let res4 = Response::try_from_http_response(http_res).unwrap();
assert_eq!(res4.stuff, None);
assert_eq!(res4.content_disposition, None);
}

View File

@ -0,0 +1,130 @@
use assert_matches2::assert_matches;
use http::header::{CONTENT_DISPOSITION, LOCATION};
use ruma_common::{
api::{
error::{
DeserializationError, FromHttpRequestError, FromHttpResponseError,
HeaderDeserializationError,
},
request, response, IncomingRequest, IncomingResponse, MatrixVersion, Metadata,
OutgoingRequest, OutgoingResponse, SendAccessToken,
},
http_headers::{ContentDisposition, ContentDispositionType},
metadata,
};
const METADATA: Metadata = metadata! {
method: GET,
rate_limited: false,
authentication: None,
history: {
unstable => "/_matrix/my/endpoint",
}
};
/// Request type for the `required_headers` endpoint.
#[request]
pub struct Request {
#[ruma_api(header = LOCATION)]
pub location: String,
#[ruma_api(header = CONTENT_DISPOSITION)]
pub content_disposition: ContentDisposition,
}
/// Response type for the `required_headers` endpoint.
#[response]
pub struct Response {
#[ruma_api(header = LOCATION)]
pub stuff: String,
#[ruma_api(header = CONTENT_DISPOSITION)]
pub content_disposition: ContentDisposition,
}
#[test]
fn request_serde() {
let location = "https://other.tld/page/";
let content_disposition = ContentDisposition::new(ContentDispositionType::Attachment)
.with_filename(Some("my_file".to_owned()));
let req =
Request { location: location.to_owned(), content_disposition: content_disposition.clone() };
let mut http_req = req
.clone()
.try_into_http_request::<Vec<u8>>(
"https://homeserver.tld",
SendAccessToken::None,
&[MatrixVersion::V1_1],
)
.unwrap();
assert_matches!(http_req.headers().get(LOCATION), Some(_));
assert_matches!(http_req.headers().get(CONTENT_DISPOSITION), Some(_));
let req2 = Request::try_from_http_request::<_, &str>(http_req.clone(), &[]).unwrap();
assert_eq!(req2.location, location);
assert_eq!(req2.content_disposition, content_disposition);
// Try removing the headers.
http_req.headers_mut().remove(LOCATION).unwrap();
http_req.headers_mut().remove(CONTENT_DISPOSITION).unwrap();
let err = Request::try_from_http_request::<_, &str>(http_req.clone(), &[]).unwrap_err();
assert_matches!(
err,
FromHttpRequestError::Deserialization(DeserializationError::Header(
HeaderDeserializationError::MissingHeader(_)
))
);
// Try setting invalid header.
http_req.headers_mut().insert(LOCATION, location.try_into().unwrap());
http_req.headers_mut().insert(CONTENT_DISPOSITION, ";".try_into().unwrap());
let err = Request::try_from_http_request::<_, &str>(http_req, &[]).unwrap_err();
assert_matches!(
err,
FromHttpRequestError::Deserialization(DeserializationError::Header(
HeaderDeserializationError::InvalidHeader(_)
))
);
}
#[test]
fn response_serde() {
let location = "https://other.tld/page/";
let content_disposition = ContentDisposition::new(ContentDispositionType::Attachment)
.with_filename(Some("my_file".to_owned()));
let res =
Response { stuff: location.to_owned(), content_disposition: content_disposition.clone() };
let mut http_res = res.clone().try_into_http_response::<Vec<u8>>().unwrap();
assert_matches!(http_res.headers().get(LOCATION), Some(_));
assert_matches!(http_res.headers().get(CONTENT_DISPOSITION), Some(_));
let res2 = Response::try_from_http_response(http_res.clone()).unwrap();
assert_eq!(res2.stuff, location);
assert_eq!(res2.content_disposition, content_disposition);
// Try removing the headers.
http_res.headers_mut().remove(LOCATION).unwrap();
http_res.headers_mut().remove(CONTENT_DISPOSITION).unwrap();
let err = Response::try_from_http_response(http_res.clone()).unwrap_err();
assert_matches!(
err,
FromHttpResponseError::Deserialization(DeserializationError::Header(
HeaderDeserializationError::MissingHeader(_)
))
);
// Try setting invalid header.
http_res.headers_mut().insert(LOCATION, location.try_into().unwrap());
http_res.headers_mut().insert(CONTENT_DISPOSITION, ";".try_into().unwrap());
let err = Response::try_from_http_response(http_res).unwrap_err();
assert_matches!(
err,
FromHttpResponseError::Deserialization(DeserializationError::Header(
HeaderDeserializationError::InvalidHeader(_)
))
);
}

View File

@ -1,11 +1,12 @@
//! Common types for implementing federation authorization.
use std::{borrow::Cow, fmt, str::FromStr};
use std::{fmt, str::FromStr};
use headers::authorization::Credentials;
use http::HeaderValue;
use http_auth::ChallengeParser;
use ruma_common::{
http_headers::quote_ascii_string_if_required,
serde::{Base64, Base64DecodeError},
IdParseError, OwnedServerName, OwnedServerSigningKeyId,
};
@ -119,40 +120,19 @@ impl fmt::Debug for XMatrix {
}
}
/// Whether the given char is a [token char].
///
/// [token char]: https://www.rfc-editor.org/rfc/rfc9110#section-5.6.2
fn is_tchar(c: char) -> bool {
const TOKEN_CHARS: [char; 15] =
['!', '#', '$', '%', '&', '\'', '*', '+', '-', '.', '^', '_', '`', '|', '~'];
c.is_ascii_alphanumeric() || TOKEN_CHARS.contains(&c)
}
/// If the field value does not contain only token chars, convert it to a [quoted string].
///
/// [quoted string]: https://www.rfc-editor.org/rfc/rfc9110#section-5.6.4
fn escape_field_value(value: &str) -> Cow<'_, str> {
if !value.is_empty() && value.chars().all(is_tchar) {
return Cow::Borrowed(value);
}
let value = value.replace('\\', r#"\\"#).replace('"', r#"\""#);
Cow::Owned(format!("\"{value}\""))
}
impl fmt::Display for XMatrix {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
let Self { origin, destination, key, sig } = self;
let origin = escape_field_value(origin.as_str());
let key = escape_field_value(key.as_str());
let origin = quote_ascii_string_if_required(origin.as_str());
let key = quote_ascii_string_if_required(key.as_str());
let sig = sig.encode();
let sig = escape_field_value(&sig);
let sig = quote_ascii_string_if_required(&sig);
write!(f, r#"{} "#, Self::SCHEME)?;
if let Some(destination) = destination {
let destination = escape_field_value(destination.as_str());
let destination = quote_ascii_string_if_required(destination.as_str());
write!(f, r#"destination={destination},"#)?;
}