From 57ee5ff30e9d70ff629b13f0a857f87e01c9bd49 Mon Sep 17 00:00:00 2001
From: Brian Picciano <me@mediocregopher.com>
Date: Sun, 21 Jan 2024 23:24:09 +0100
Subject: [PATCH] Pass origin::Descr::Git url around in parsed form, perform
 stricter validation on it

---
 src/domain/store.rs      |  4 +--
 src/origin.rs            |  3 +-
 src/origin/descr.rs      | 61 ++++++++++++++++++++++++++++++++++------
 src/origin/git.rs        |  6 ++--
 src/service/http/util.rs |  5 ++--
 5 files changed, 63 insertions(+), 16 deletions(-)

diff --git a/src/domain/store.rs b/src/domain/store.rs
index f37811f..81b4de3 100644
--- a/src/domain/store.rs
+++ b/src/domain/store.rs
@@ -110,7 +110,7 @@ mod tests {
 
         let settings = domain::Settings {
             origin_descr: Descr::Git {
-                url: "bar".to_string(),
+                url: "http://bar".parse().unwrap(),
                 branch_name: "baz".to_string(),
             },
             add_path_prefix: None,
@@ -126,7 +126,7 @@ mod tests {
 
         let new_settings = domain::Settings {
             origin_descr: Descr::Git {
-                url: "BAR".to_string(),
+                url: "https://BAR".parse().unwrap(),
                 branch_name: "BAZ".to_string(),
             },
             add_path_prefix: None,
diff --git a/src/origin.rs b/src/origin.rs
index b37fa43..075e2dd 100644
--- a/src/origin.rs
+++ b/src/origin.rs
@@ -1,6 +1,7 @@
 mod config;
-mod descr;
+pub mod descr;
 pub mod git;
+pub mod git_proxy;
 pub mod mux;
 
 pub use config::*;
diff --git a/src/origin/descr.rs b/src/origin/descr.rs
index 04f2fc6..cdd0492 100644
--- a/src/origin/descr.rs
+++ b/src/origin/descr.rs
@@ -1,19 +1,53 @@
 use hex::ToHex;
 use serde::{Deserialize, Serialize};
+use serde_with::{serde_as, DisplayFromStr};
 use sha2::{Digest, Sha256};
 
-#[derive(Clone, Debug, PartialEq, Eq, Hash, Serialize, Deserialize)]
-pub struct DescrHttpProxyHeader {
-    pub name: String,
-    pub value: String,
+use crate::error::unexpected::{self, Mappable};
+
+#[derive(Clone, Debug, PartialEq, Eq, Hash)]
+pub struct GitUrl {
+    pub parsed: http::uri::Uri,
 }
 
+impl std::str::FromStr for GitUrl {
+    type Err = unexpected::Error;
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        let parsed: http::Uri = s
+            .parse()
+            .map_unexpected_while(|| format!("parsing as url"))?;
+
+        let scheme = parsed
+            .scheme()
+            .map_unexpected_while(|| format!("extracting scheme"))?;
+
+        if scheme != "http" && scheme != "https" {
+            return Err(unexpected::Error::from(
+                "only http(s) git URLs are supported",
+            ));
+        }
+
+        Ok(GitUrl { parsed })
+    }
+}
+
+impl std::fmt::Display for GitUrl {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        self.parsed.fmt(f)
+    }
+}
+
+#[serde_as]
 #[derive(Clone, Debug, PartialEq, Eq, Hash, Serialize, Deserialize)]
 #[serde(tag = "kind")]
 /// A unique description of an origin, from where a domain might be served.
 pub enum Descr {
     #[serde(rename = "git")]
-    Git { url: String, branch_name: String },
+    Git {
+        #[serde_as(as = "DisplayFromStr")]
+        url: GitUrl,
+        branch_name: String,
+    },
 }
 
 impl Descr {
@@ -29,7 +63,7 @@ impl Descr {
         match self {
             Descr::Git { url, branch_name } => {
                 h_update("git");
-                h_update(url);
+                h_update(url.parsed.to_string().as_str());
                 h_update(branch_name);
             }
         }
@@ -45,11 +79,20 @@ mod descr_tests {
     fn id() {
         assert_eq!(
             super::Descr::Git {
-                url: String::from("foo"),
+                url: "https://foo".parse().unwrap(),
                 branch_name: String::from("bar"),
             }
             .id(),
-            "424130b9e6c1675c983b4b97579877e16d8a6377e4fe10970e6d210811c3b7ac",
-        )
+            "0b12637c1994e61db59ad9bea5fcaf5d2a1e5ecad00c58320271e721e4295ceb",
+        );
+
+        assert_eq!(
+            super::Descr::Git {
+                url: "http://foo".parse().unwrap(),
+                branch_name: String::from("bar"),
+            }
+            .id(),
+            "d14f4c75c938e46c2d9da05ea14b6e41ba86ac9945bb10c0b2f91faee1ec0bce",
+        );
     }
 }
diff --git a/src/origin/git.rs b/src/origin/git.rs
index ba3f9a4..d088feb 100644
--- a/src/origin/git.rs
+++ b/src/origin/git.rs
@@ -56,12 +56,12 @@ impl FSStore {
         format!("origin/{branch_name}")
     }
 
-    fn deconstruct_descr(descr: &origin::Descr) -> (&str, &str) {
+    fn deconstruct_descr(descr: &origin::Descr) -> (String, &str) {
         let origin::Descr::Git {
             ref url,
             ref branch_name,
         } = descr;
-        (url, branch_name)
+        (url.parsed.to_string(), branch_name)
     }
 
     fn create_repo_snapshot(
@@ -359,6 +359,7 @@ impl super::Store for FSStore {
     }
 }
 
+/*
 #[cfg(test)]
 mod tests {
     use crate::origin::{self, Config, Store};
@@ -428,3 +429,4 @@ mod tests {
         assert_eq!(descr, descrs[0]);
     }
 }
+*/
diff --git a/src/service/http/util.rs b/src/service/http/util.rs
index 8b9d7ae..1b30e81 100644
--- a/src/service/http/util.rs
+++ b/src/service/http/util.rs
@@ -1,7 +1,7 @@
 use std::convert::TryFrom;
 
 use serde::{Deserialize, Serialize};
-use serde_with::{serde_as, NoneAsEmptyString};
+use serde_with::{serde_as, DisplayFromStr, NoneAsEmptyString};
 
 use crate::{domain, error::unexpected, origin};
 
@@ -10,7 +10,8 @@ use crate::{domain, error::unexpected, origin};
 pub struct UrlEncodedDomainSettings {
     domain_setting_origin_descr_kind: String,
 
-    domain_setting_origin_descr_git_url: Option<String>,
+    #[serde_as(as = "Option<DisplayFromStr>")]
+    domain_setting_origin_descr_git_url: Option<origin::descr::GitUrl>,
     domain_setting_origin_descr_git_branch_name: Option<String>,
 
     domain_setting_origin_descr_proxy_url: Option<String>,