From 46238e1d52e1ab0b15b567ed7254e70bd9d39c39 Mon Sep 17 00:00:00 2001 From: Richard Walters Date: Tue, 13 Oct 2020 12:40:24 -0700 Subject: Move authority and codec functions to their own modules --- src/authority.rs | 104 ++++++++++++++++++++++++++++++++++++ src/codec.rs | 56 ++++++++++++++++++++ src/lib.rs | 159 ++----------------------------------------------------- 3 files changed, 165 insertions(+), 154 deletions(-) create mode 100644 src/authority.rs create mode 100644 src/codec.rs diff --git a/src/authority.rs b/src/authority.rs new file mode 100644 index 0000000..b5a377c --- /dev/null +++ b/src/authority.rs @@ -0,0 +1,104 @@ +#![warn(clippy::pedantic)] + +use super::character_classes::{ + REG_NAME_NOT_PCT_ENCODED, + USER_INFO_NOT_PCT_ENCODED, +}; +use super::codec::{decode_element, encode_element}; +use super::context::Context; +use super::error::Error; +use super::parse_host_port::parse_host_port; +use super::validate_ipv6_address::validate_ipv6_address; + +#[derive(Clone, Debug, Default, PartialEq)] +pub struct Authority { + userinfo: Option>, + host: Vec, + port: Option, +} + +impl Authority { + #[must_use = "why u no use host return value?"] + pub fn host(&self) -> &[u8] { + &self.host + } + + #[must_use = "why did you get the port number and then throw it away?"] + pub fn port(&self) -> Option { + self.port + } + + pub fn set_userinfo(&mut self, userinfo: T) + where T: Into>> + { + self.userinfo = userinfo.into(); + } + + pub fn set_host(&mut self, host: T) + where T: Into> + { + self.host = host.into(); + } + + pub fn set_port(&mut self, port: Option) { + self.port = port; + } + + #[must_use = "security breach... security breach... userinfo not used"] + pub fn userinfo(&self) -> Option<&[u8]> { + self.userinfo.as_deref() + } + + #[must_use = "you parsed it; don't you want the results?"] + pub fn parse(authority_string: T) -> Result + where T: AsRef + { + let (userinfo, host_port_string) = Self::parse_userinfo(authority_string.as_ref())?; + let (host, port) = parse_host_port(host_port_string)?; + Ok(Self{ + userinfo, + host, + port, + }) + } + + fn parse_userinfo(authority: &str) -> Result<(Option>, &str), Error> { + Ok(match authority.find('@') { + Some(delimiter) => ( + Some( + decode_element( + &authority[0..delimiter], + &USER_INFO_NOT_PCT_ENCODED, + Context::Userinfo + )? + ), + &authority[delimiter+1..] + ), + None => ( + None, + authority + ) + }) + } +} + +impl std::fmt::Display for Authority { + fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { + if let Some(userinfo) = &self.userinfo { + write!(f, "{}@", encode_element(&userinfo, &USER_INFO_NOT_PCT_ENCODED))?; + } + let host_as_string = String::from_utf8(self.host.clone()); + match host_as_string { + Ok(host_as_string) if validate_ipv6_address(&host_as_string).is_ok() => { + write!(f, "[{}]", host_as_string.to_ascii_lowercase())?; + }, + _ => { + write!(f, "{}", encode_element(&self.host, ®_NAME_NOT_PCT_ENCODED))?; + } + } + if let Some(port) = self.port { + write!(f, ":{}", port)?; + } + Ok(()) + } +} diff --git a/src/codec.rs b/src/codec.rs new file mode 100644 index 0000000..3bba1c6 --- /dev/null +++ b/src/codec.rs @@ -0,0 +1,56 @@ +#![warn(clippy::pedantic)] + +use std::collections::HashSet; +use std::convert::TryFrom; + +use super::context::Context; +use super::error::Error; +use super::percent_encoded_character_decoder::PercentEncodedCharacterDecoder; + +pub fn decode_element( + element: T, + allowed_characters: &'static HashSet, + context: Context +) -> Result, Error> + where T: AsRef +{ + let mut decoding_pec = false; + let mut pec_decoder = PercentEncodedCharacterDecoder::new(); + element + .as_ref() + .chars() + .filter_map(|c| { + if decoding_pec { + pec_decoder + .next(c) + .map_err(Into::into) + .transpose() + .map(|c| { + decoding_pec = false; + c + }) + } else if c == '%' { + decoding_pec = true; + None + } else if allowed_characters.contains(&c) { + Some(Ok(c as u8)) + } else { + Some(Err(Error::IllegalCharacter(context))) + } + }) + .collect() +} + +pub fn encode_element( + element: &[u8], + allowed_characters: &HashSet +) -> String { + element.iter() + .map(|ci| { + match char::try_from(*ci) { + Ok(c) if allowed_characters.contains(&c) => c.to_string(), + _ => format!("%{:X}", ci), + } + }) + .collect::() +} diff --git a/src/lib.rs b/src/lib.rs index 98df257..deaec56 100644 --- a/src/lib.rs +++ b/src/lib.rs @@ -7,8 +7,10 @@ extern crate named_tuple; use std::collections::HashSet; -use std::convert::TryFrom; +mod authority; +mod character_classes; +mod codec; mod context; mod error; mod parse_host_port; @@ -16,169 +18,18 @@ mod percent_encoded_character_decoder; mod validate_ipv4_address; mod validate_ipv6_address; +use authority::Authority; +use codec::{decode_element, encode_element}; use context::Context; use error::Error; -use parse_host_port::parse_host_port; -use percent_encoded_character_decoder::PercentEncodedCharacterDecoder; -use validate_ipv6_address::validate_ipv6_address; - -mod character_classes; use character_classes::{ ALPHA, SCHEME_NOT_FIRST, PCHAR_NOT_PCT_ENCODED, QUERY_OR_FRAGMENT_NOT_PCT_ENCODED, QUERY_NOT_PCT_ENCODED_WITHOUT_PLUS, - USER_INFO_NOT_PCT_ENCODED, - REG_NAME_NOT_PCT_ENCODED, }; -fn decode_element( - element: T, - allowed_characters: &'static HashSet, - context: Context -) -> Result, Error> - where T: AsRef -{ - let mut decoding_pec = false; - let mut pec_decoder = PercentEncodedCharacterDecoder::new(); - element - .as_ref() - .chars() - .filter_map(|c| { - if decoding_pec { - pec_decoder - .next(c) - .map_err(Into::into) - .transpose() - .map(|c| { - decoding_pec = false; - c - }) - } else if c == '%' { - decoding_pec = true; - None - } else if allowed_characters.contains(&c) { - Some(Ok(c as u8)) - } else { - Some(Err(Error::IllegalCharacter(context))) - } - }) - .collect() -} - -fn encode_element( - element: &[u8], - allowed_characters: &HashSet -) -> String { - element.iter() - .map(|ci| { - match char::try_from(*ci) { - Ok(c) if allowed_characters.contains(&c) => c.to_string(), - _ => format!("%{:X}", ci), - } - }) - .collect::() -} - -#[derive(Clone, Debug, Default, PartialEq)] -pub struct Authority { - userinfo: Option>, - host: Vec, - port: Option, -} - -impl Authority { - #[must_use = "why u no use host return value?"] - pub fn host(&self) -> &[u8] { - &self.host - } - - #[must_use = "why did you get the port number and then throw it away?"] - pub fn port(&self) -> Option { - self.port - } - - pub fn set_userinfo(&mut self, userinfo: T) - where T: Into>> - { - self.userinfo = userinfo.into(); - } - - pub fn set_host(&mut self, host: T) - where T: Into> - { - self.host = host.into(); - } - - pub fn set_port(&mut self, port: Option) { - self.port = port; - } - - #[must_use = "security breach... security breach... userinfo not used"] - pub fn userinfo(&self) -> Option<&[u8]> { - self.userinfo.as_deref() - } - - #[must_use = "you parsed it; don't you want the results?"] - pub fn parse(authority_string: T) -> Result - where T: AsRef - { - // First, check if there is a UserInfo, and if so, extract it. - let (userinfo, host_port_string) = Self::parse_userinfo(authority_string.as_ref())?; - - // Next, parsing host and port from authority and path. - let (host, port) = parse_host_port(host_port_string)?; - - // Assemble authority from its parts. - Ok(Self{ - userinfo, - host, - port, - }) - } - - fn parse_userinfo(authority: &str) -> Result<(Option>, &str), Error> { - Ok(match authority.find('@') { - Some(delimiter) => ( - Some( - decode_element( - &authority[0..delimiter], - &USER_INFO_NOT_PCT_ENCODED, - Context::Userinfo - )? - ), - &authority[delimiter+1..] - ), - None => ( - None, - authority - ) - }) - } -} - -impl std::fmt::Display for Authority { - fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { - if let Some(userinfo) = &self.userinfo { - write!(f, "{}@", encode_element(&userinfo, &USER_INFO_NOT_PCT_ENCODED))?; - } - let host_as_string = String::from_utf8(self.host.clone()); - match host_as_string { - Ok(host_as_string) if validate_ipv6_address(&host_as_string).is_ok() => { - write!(f, "[{}]", host_as_string.to_ascii_lowercase())?; - }, - _ => { - write!(f, "{}", encode_element(&self.host, ®_NAME_NOT_PCT_ENCODED))?; - } - } - if let Some(port) = self.port { - write!(f, ":{}", port)?; - } - Ok(()) - } -} - #[derive(Clone, Debug, Default, PartialEq)] pub struct Uri { scheme: Option, -- cgit v1.2.3