Allow instance_uri configuration value to contain URI scheme

CHANGELOG.md

@@ -6,6 +6,10 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 
 ## [Unreleased]
 
+### Changed
+
+- Allow `instance_uri` configuration value to contain URI scheme.
+
 ## [1.13.1] - 2023-02-09
 
 ### Fixed

config.yaml.example

@@ -5,7 +5,8 @@ web_client_dir: null
 http_host: '127.0.0.1'
 http_port: 8380
 
-instance_uri: 127.0.0.1:8380
+instance_uri: http://127.0.0.1:8380
+
 instance_title: Mitra
 instance_short_description: My instance
 instance_description: My instance

contrib/mitra_config.yaml

@@ -12,7 +12,8 @@ http_port: 8383
 #log_level: info
 
 # Domain name
-instance_uri: example.tld
+instance_uri: https://example.tld
+
 instance_title: example
 instance_short_description: my instance
 # Long description can contain markdown syntax

src/config/main.rs

@@ -9,7 +9,7 @@ use serde::{
 };
 use url::Url;
 
-use crate::utils::urls::guess_protocol;
+use crate::utils::urls::normalize_url;
 
 use super::blockchain::BlockchainConfig;
 use super::environment::Environment;
@@ -78,7 +78,9 @@ pub struct Config {
     pub log_level: LogLevel,
 
     // Domain name or <IP address>:<port>
+    // URI scheme is optional
     instance_uri: String,
+
     pub instance_title: String,
     pub instance_short_description: String,
     pub instance_description: String,
@@ -121,14 +123,7 @@ pub struct Config {
 
 impl Config {
     pub(super) fn try_instance_url(&self) -> Result<Url, url::ParseError> {
-        let scheme = match self.environment {
+        normalize_url(&self.instance_uri)
-            Environment::Development => "http",
-            Environment::Production => guess_protocol(&self.instance_uri),
-        };
-        let url_str = format!("{}://{}", scheme, self.instance_uri);
-        let url = Url::parse(&url_str)?;
-        url.host().ok_or(url::ParseError::EmptyHost)?; // validates URL
-        Ok(url)
     }
 
     pub fn instance(&self) -> Instance {

src/utils/urls.rs

@@ -1,4 +1,4 @@
-use std::net::Ipv6Addr;
+use std::net::{Ipv4Addr, Ipv6Addr};
 use url::{Host, ParseError, Url};
 
 pub fn get_hostname(url: &str) -> Result<String, ParseError> {
@@ -14,13 +14,13 @@ pub fn get_hostname(url: &str) -> Result<String, ParseError> {
 }
 
 pub fn guess_protocol(hostname: &str) -> &'static str {
-    let maybe_ipv6_address = hostname.parse::<Ipv6Addr>();
+    let maybe_ipv4_address = hostname.parse::<Ipv4Addr>();
-    if let Ok(ipv6_address) = maybe_ipv6_address {
+    if let Ok(_ipv4_address) = maybe_ipv4_address {
-        let prefix = ipv6_address.segments()[0];
-        if (0x0200..=0x03ff).contains(&prefix) {
-            // Yggdrasil
         return "http";
     };
+    let maybe_ipv6_address = hostname.parse::<Ipv6Addr>();
+    if let Ok(_ipv6_address) = maybe_ipv6_address {
+        return "http";
     };
     if hostname.ends_with(".onion") || hostname.ends_with(".i2p") {
         // Tor / I2P
@@ -31,9 +31,36 @@ pub fn guess_protocol(hostname: &str) -> &'static str {
     }
 }
 
+pub fn normalize_url(url: &str) -> Result<Url, url::ParseError> {
+    let normalized_url = if
+        url.starts_with("http://") ||
+        url.starts_with("https://")
+    {
+        url.to_string()
+    } else {
+        // Add scheme
+        // Doesn't work for IPv6
+        let hostname = if let Some((hostname, _port)) = url.split_once(':') {
+            hostname
+        } else {
+            url
+        };
+        let url_scheme = guess_protocol(hostname);
+        format!(
+            "{}://{}",
+            url_scheme,
+            url,
+        )
+    };
+    let url = Url::parse(&normalized_url)?;
+    url.host().ok_or(ParseError::EmptyHost)?; // validates URL
+    Ok(url)
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
+
     #[test]
     fn test_get_hostname() {
         let url = "https://example.org/objects/1";
@@ -41,6 +68,13 @@ mod tests {
         assert_eq!(hostname, "example.org");
     }
 
+    #[test]
+    fn test_get_hostname_if_port_number() {
+        let url = "http://127.0.0.1:8380/objects/1";
+        let hostname = get_hostname(url).unwrap();
+        assert_eq!(hostname, "127.0.0.1");
+    }
+
     #[test]
     fn test_get_hostname_tor() {
         let url = "http://2gzyxa5ihm7nsggfxnu52rck2vv4rvmdlkiu3zzui5du4xyclen53wid.onion/objects/1";
@@ -76,9 +110,25 @@ mod tests {
             guess_protocol("zzz.i2p"),
             "http",
         );
+        // Yggdrasil
         assert_eq!(
             guess_protocol("319:3cf0:dd1d:47b9:20c:29ff:fe2c:39be"),
             "http",
         );
+        // localhost
+        assert_eq!(
+            guess_protocol("127.0.0.1"),
+            "http",
+        );
+    }
+
+    #[test]
+    fn test_normalize_url() {
+        let result = normalize_url("https://test.net").unwrap();
+        assert_eq!(result.to_string(), "https://test.net/");
+        let result = normalize_url("example.com").unwrap();
+        assert_eq!(result.to_string(), "https://example.com/");
+        let result = normalize_url("127.0.0.1:8380").unwrap();
+        assert_eq!(result.to_string(), "http://127.0.0.1:8380/");
     }
 }