From 6350610ef5f7d73680853d39898094f2bf15febb Mon Sep 17 00:00:00 2001
From: Santo Cariotti <santo@dcariotti.me>
Date: Tue, 16 Mar 2021 11:19:53 +0100
Subject: feat: make regex of url to check if it is valid

Currently it works only with GitHub
---
 Cargo.lock               |  1 +
 Cargo.toml               |  1 +
 src/helpers.rs           | 17 +++++++++++++++++
 src/repository/models.rs | 16 ++++++++++++++--
 4 files changed, 33 insertions(+), 2 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index e579fb2..ad4fcfa 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -913,6 +913,7 @@ dependencies = [
  "config",
  "deadpool-postgres",
  "dotenv",
+ "regex",
  "serde 1.0.124",
  "slog",
  "slog-async",
diff --git a/Cargo.toml b/Cargo.toml
index fa12bb6..ae7576f 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -24,3 +24,4 @@ config = "0.10.1"
 serde = { version = "1.0.104", features = ["derive"] }
 chrono = { version = "0.4.19", features = ["serde"] }
 uuid = { version = "0.8.2", features = ["serde", "v4"] }
+regex = "1"
diff --git a/src/helpers.rs b/src/helpers.rs
index d915a50..2dbacfd 100644
--- a/src/helpers.rs
+++ b/src/helpers.rs
@@ -1,3 +1,4 @@
+use regex::Regex;
 use uuid::Uuid;
 
 /// Returns a valid Uuid if `id` is not a valid Uuid
@@ -7,3 +8,19 @@ pub fn uuid_from_string(id: &String) -> Uuid {
         Err(_) => Uuid::parse_str("00000000000000000000000000000000").unwrap(),
     };
 }
+
+/// Check if a path is into the "valid git repositories" and returns the name
+pub fn name_of_git_repository(url: &String) -> Option<String> {
+    const GITHUB_RE: &str = r"^(http(s)?://)?(www.)?github.com/(?P<username>[a-zA-Z0-9-]+)/(?P<repository>[a-zA-Z0-9-]+)";
+    let re = Regex::new(GITHUB_RE).unwrap();
+
+    if !re.is_match(&url) {
+        return None;
+    }
+
+    let captures = re.captures(&url).unwrap();
+    let name = captures.name("username").unwrap().as_str();
+    let repo = captures.name("repository").unwrap().as_str();
+
+    Some(format!("{}/{}", name, repo))
+}
diff --git a/src/repository/models.rs b/src/repository/models.rs
index 1fd1649..1cbf3bb 100644
--- a/src/repository/models.rs
+++ b/src/repository/models.rs
@@ -1,5 +1,6 @@
 use crate::db::get_client;
 use crate::errors::{AppError, AppErrorType};
+use crate::helpers::name_of_git_repository;
 
 use chrono::NaiveDateTime;
 use deadpool_postgres::{Client, Pool};
@@ -133,9 +134,20 @@ impl Repository {
     ) -> Result<Repository, AppError> {
         let client = get_client(pool.clone()).await.unwrap();
 
+        let repo_name: String = match name_of_git_repository(&data.url) {
+            Some(path) => path,
+            None => {
+                return Err(AppError {
+                    message: Some("Repository not found".to_string()),
+                    cause: Some("".to_string()),
+                    error_type: AppErrorType::NotFoundError,
+                });
+            }
+        };
+
         // Search a repository that matches with that url, because if it's
         // exists, the server do not create a clone
-        let repo_search = Repository::search(&client, data.url.clone()).await;
+        let repo_search = Repository::search(&client, repo_name.clone()).await;
         match repo_search {
             Ok(_) => {
                 return Err(AppError {
@@ -168,7 +180,7 @@ impl Repository {
         };
 
         let repo = client
-            .query(&statement, &[&uuid, &data.url, &user_ip])
+            .query(&statement, &[&uuid, &repo_name, &user_ip])
             .await?
             .iter()
             .map(|row| Repository::from_row_ref(row).unwrap())
-- 
cgit v1.2.3-18-g5258