From 82028ee32047fec2a600ed61e7133f18babec916 Mon Sep 17 00:00:00 2001
From: "Shahin M. Shahin" <peregrineshahin@gmail.com>
Date: Sat, 17 Aug 2024 14:36:11 +0300
Subject: [PATCH] Introduce GitHub OAuth Authentication

This implements the long waited GitHub Auth
Closes #2132
---
 server/fishtest/__init__.py          |   4 +
 server/fishtest/schemas.py           |  31 +++--
 server/fishtest/templates/login.mak  |  10 ++
 server/fishtest/templates/signup.mak |   9 ++
 server/fishtest/templates/user.mak   |  18 ++-
 server/fishtest/userdb.py            |  53 ++++++--
 server/fishtest/views.py             | 183 ++++++++++++++++++++++++++-
 7 files changed, 281 insertions(+), 27 deletions(-)

diff --git a/server/fishtest/__init__.py b/server/fishtest/__init__.py
index 3fcf99ec6..7adae958f 100644
--- a/server/fishtest/__init__.py
+++ b/server/fishtest/__init__.py
@@ -177,5 +177,9 @@ def group_finder(username, request):
     config.add_route("api_actions", "/api/actions")
     config.add_route("api_calc_elo", "/api/calc_elo")
 
+    # GitHub OAuth Routes
+    config.add_route("github_oauth", "/github/oauth")
+    config.add_route("github_callback", "/github/callback")
+
     config.scan()
     return config.make_wsgi_app()
diff --git a/server/fishtest/schemas.py b/server/fishtest/schemas.py
index 45ed5db4c..9a7c5ab3b 100644
--- a/server/fishtest/schemas.py
+++ b/server/fishtest/schemas.py
@@ -87,18 +87,25 @@ def size_is_length(x):
     size_is_length,
 )
 
-user_schema = {
-    "_id?": ObjectId,
-    "username": username,
-    "password": str,
-    "registration_time": datetime_utc,
-    "pending": bool,
-    "blocked": bool,
-    "email": email,
-    "groups": [str, ...],
-    "tests_repo": union("", url),
-    "machine_limit": uint,
-}
+user_schema = intersect(
+    {
+        "_id?": ObjectId,
+        "username": username,
+        "password?": str,
+        "registration_time": datetime_utc,
+        "pending": bool,
+        "blocked": bool,
+        "email?": email,
+        "github_id?": str,
+        "linked_github_username?": str,
+        "github_access_token?": str,
+        "groups": [str, ...],
+        "tests_repo?": union("", url),
+        "machine_limit": uint,
+    },
+    at_least_one_of("email", "github_id"),
+    at_least_one_of("password", "github_access_token"),
+)
 
 
 worker_schema = {
diff --git a/server/fishtest/templates/login.mak b/server/fishtest/templates/login.mak
index 7160b2922..1bcfcef1c 100644
--- a/server/fishtest/templates/login.mak
+++ b/server/fishtest/templates/login.mak
@@ -58,6 +58,16 @@
 
     <button type="submit" class="btn btn-primary w-100">Login</button>
   </form>
+
+  <div class="text-md-center my-4">
+    <p>Or</p>
+    <form action="${request.route_url('github_oauth')}" method="post">
+      <input type="hidden" name="action" value="login">
+      <button type="submit" class="btn btn-secondary w-100">
+        <i class="fa-brands fa-github"></i> Log in with GitHub
+      </button>
+    </form>
+  </div>
 </div>
 
 <script
diff --git a/server/fishtest/templates/signup.mak b/server/fishtest/templates/signup.mak
index ac5d68e8d..e80e253f0 100644
--- a/server/fishtest/templates/signup.mak
+++ b/server/fishtest/templates/signup.mak
@@ -117,6 +117,15 @@
 
     <button type="submit" class="btn btn-primary w-100">Register</button>
   </form>
+  <div class="text-md-center my-4">
+    <p>Or</p>
+    <form action="${request.route_url('github_oauth')}" method="post">
+      <input type="hidden" name="action" value="signup">
+      <button type="submit" class="btn btn-secondary w-100">
+        <i class="fa-brands fa-github"></i> Sign up with GitHub
+      </button>
+    </form>
+  </div>
 </div>
 
 <script
diff --git a/server/fishtest/templates/user.mak b/server/fishtest/templates/user.mak
index 1172e4a0e..899c18e77 100644
--- a/server/fishtest/templates/user.mak
+++ b/server/fishtest/templates/user.mak
@@ -40,7 +40,7 @@
         <li class="list-group-item bg-transparent text-break">Registered: ${format_date(user['registration_time'] if 'registration_time' in user else 'Unknown')}</li>
         % if not profile:
           <li class="list-group-item bg-transparent text-break">Tests Repository: 
-            % if user['tests_repo']:
+            % if 'tests_repo' in user and user['tests_repo']:
               <a class="alert-link" href="${user['tests_repo']}">${extract_repo_from_link(user['tests_repo'])}</a>
             % else:
               <span>-</span>
@@ -55,6 +55,14 @@
         <li class="list-group-item bg-transparent text-break">
           Groups: ${format_group(user['groups'])}
         </li>
+        <li class="list-group-item bg-transparent text-break">
+        Linked GitHub user: 
+        % if 'linked_github_username' in user and user['linked_github_username']:
+          <a class="alert-link" href="/${user["linked_github_username"]}">GitHub/${user["linked_github_username"]}</a>
+        % else:
+          <span>No accounts linked</span>
+        % endif
+        </li>
         <li class="list-group-item bg-transparent text-break">Machine Limit: ${limit}</li>
         <li class="list-group-item bg-transparent text-break">CPU-Hours: ${hours}</li>
       </ul>
@@ -273,6 +281,14 @@
       % endif
     % endif
   </form>
+  % if profile and not 'linked_github_username' in user:
+    <form action="${request.route_url('github_oauth')}" method="post">
+      <input type="hidden" name="action" value="link">
+      <button type="submit" class="btn btn-secondary w-100">
+        <i class="fa-brands fa-github"></i> Link GitHub Account
+      </button>
+    </form>
+  % endif
 </div>
 
 <script
diff --git a/server/fishtest/userdb.py b/server/fishtest/userdb.py
index 6c2957b7f..bc52b766f 100644
--- a/server/fishtest/userdb.py
+++ b/server/fishtest/userdb.py
@@ -92,6 +92,9 @@ def get_blocked(self):
     def get_user(self, username):
         return self.find_by_username(username)
 
+    def get_user_by_github_id(self, github_id):
+        return self.users.find_one({"github_id": github_id})
+
     def get_user_groups(self, username):
         user = self.get_user(username)
         if user is not None:
@@ -105,27 +108,51 @@ def add_user_group(self, username, group):
         self.users.replace_one({"_id": user["_id"]}, user)
         self.clear_cache()
 
-    def create_user(self, username, password, email, tests_repo):
+    def create_user(
+        self,
+        username,
+        password=None,
+        email="",
+        tests_repo="",
+        linked_github_username=None,
+        github_id=None,
+        github_access_token=None,
+    ):
         try:
             if self.find_by_username(username) or self.find_by_email(email):
                 return False
             # insert the new user in the db
-            user = {
-                "username": username,
-                "password": password,
-                "registration_time": datetime.now(timezone.utc),
-                "pending": True,
-                "blocked": False,
-                "email": email,
-                "groups": [],
-                "tests_repo": tests_repo,
-                "machine_limit": DEFAULT_MACHINE_LIMIT,
-            }
+            if github_id is not None and github_access_token is not None:
+                user = {
+                    "username": username,
+                    "registration_time": datetime.now(timezone.utc),
+                    "pending": True,
+                    "blocked": False,
+                    "github_id": github_id,
+                    "github_access_token": github_access_token,
+                    "linked_github_username": linked_github_username,
+                    "email": email,
+                    "groups": [],
+                    "tests_repo": tests_repo,
+                    "machine_limit": DEFAULT_MACHINE_LIMIT,
+                }
+            else:
+                user = {
+                    "username": username,
+                    "password": password,
+                    "registration_time": datetime.now(timezone.utc),
+                    "pending": True,
+                    "blocked": False,
+                    "email": email,
+                    "groups": [],
+                    "tests_repo": tests_repo,
+                    "machine_limit": DEFAULT_MACHINE_LIMIT,
+                }
+
             validate_user(user)
             self.users.insert_one(user)
             self.last_pending_time = 0
             self.last_blocked_time = 0
-
             return True
         except:
             return None
diff --git a/server/fishtest/views.py b/server/fishtest/views.py
index e560520c2..4d36071ea 100644
--- a/server/fishtest/views.py
+++ b/server/fishtest/views.py
@@ -4,6 +4,7 @@
 import json
 import os
 import re
+import subprocess
 import threading
 import time
 from datetime import datetime, timezone
@@ -25,7 +26,7 @@
     password_strength,
     update_residuals,
 )
-from pyramid.httpexceptions import HTTPFound, HTTPNotFound
+from pyramid.httpexceptions import HTTPBadRequest, HTTPFound, HTTPNotFound
 from pyramid.security import forget, remember
 from pyramid.view import forbidden_view_config, notfound_view_config, view_config
 from requests.exceptions import ConnectionError, HTTPError
@@ -156,6 +157,186 @@ def login(request):
     return {}
 
 
+# Hide these
+GITHUB_CLIENT_ID = ""
+GITHUB_CLIENT_SECRET = ""
+
+
+def get_github_access_token(code, redirect_uri):
+    token_url = "https://github.com/login/oauth/access_token"
+    curl_command = [
+        "curl",
+        "-X",
+        "POST",
+        token_url,
+        "-H",
+        "Accept: application/json",
+        "-d",
+        f"client_id={GITHUB_CLIENT_ID}",
+        "-d",
+        f"client_secret={GITHUB_CLIENT_SECRET}",
+        "-d",
+        f"code={code}",
+        "-d",
+        f"redirect_uri={redirect_uri}",
+    ]
+    result = subprocess.run(curl_command, stdout=subprocess.PIPE, text=True)
+    return json.loads(result.stdout)["access_token"]
+
+
+def extract_primary_email(emails):
+    if isinstance(emails, list):
+        return next(
+            (
+                email.get("email")
+                for email in emails
+                if isinstance(email, dict) and email.get("primary")
+            ),
+            None,
+        )
+    return None
+
+
+def get_github_user_data(access_token):
+    user_url = "https://api.github.com/user"
+    emails_url = "https://api.github.com/user/emails"
+
+    user_command = [
+        "curl",
+        "-H",
+        f"Authorization: bearer {access_token}",
+        "-H",
+        "Accept: application/json",
+        user_url,
+    ]
+    user_result = subprocess.run(user_command, stdout=subprocess.PIPE, text=True)
+    user_data = json.loads(user_result.stdout)
+
+    emails_command = [
+        "curl",
+        "-H",
+        f"Authorization: bearer {access_token}",
+        "-H",
+        "Accept: application/json",
+        emails_url,
+    ]
+    emails_result = subprocess.run(emails_command, stdout=subprocess.PIPE, text=True)
+
+    try:
+        emails = json.loads(emails_result.stdout)
+    except json.JSONDecodeError:
+        emails = []
+
+    primary_email = extract_primary_email(emails)
+
+    user_data["primary_email"] = primary_email
+    return user_data, emails
+
+
+@view_config(route_name="github_oauth", request_method="POST")
+def github_oauth(request):
+    action = request.POST.get("action")
+    request.session["github_action"] = action  # Store action in session
+
+    github_authorize_url = "https://github.com/login/oauth/authorize"
+    redirect_uri = request.route_url("github_callback")
+
+    params = {
+        "client_id": GITHUB_CLIENT_ID,
+        "redirect_uri": redirect_uri,
+        "scope": "user:email",
+    }
+    url = requests.Request("GET", github_authorize_url, params=params).prepare().url
+    return HTTPFound(location=url)
+
+
+@view_config(route_name="github_callback")
+def github_callback(request):
+    code = request.params.get("code")
+    if not code:
+        return HTTPBadRequest("No code provided")
+
+    action = request.session.get("github_action", "login")  # Default to login
+    github_access_token = get_github_access_token(
+        code, request.route_url("github_callback")
+    )
+    user_data, user_email_data = get_github_user_data(github_access_token)
+    username = user_data.get("login")
+    if not username:
+        request.session.flash("Failed to retrieve GitHub username", "error")
+        return HTTPFound(location=request.route_url("login"))
+
+    github_id = str(user_data.get("id"))
+    email = extract_primary_email(user_email_data)
+
+    if action == "login":
+        return handle_github_login(
+            request, github_id, username, email, github_access_token
+        )
+    elif action == "signup":
+        return handle_github_signup(
+            request, github_id, username, email, github_access_token
+        )
+    elif action == "link":
+        return handle_github_link(request, github_id, username, github_access_token)
+    else:
+        return HTTPBadRequest("Invalid action specified")
+
+
+def handle_github_login(request, github_id, username, email, github_access_token):
+    user = request.userdb.get_user_by_github_id(github_id)
+    if user is None:
+        request.session.flash("User not found. Please register first.", "error")
+        return HTTPFound(location=request.route_url("signup"))
+
+    # Remember the user in the session
+    headers = remember(request, user["username"], max_age=60 * 60 * 24 * 365)
+    return HTTPFound(location=request.route_url("home"), headers=headers)
+
+
+def handle_github_signup(request, github_id, username, email, github_access_token):
+    user = request.userdb.get_user_by_github_id(github_id)
+    if user:
+        request.session.flash("User already exists. Please login instead.", "error")
+        return HTTPFound(location=request.route_url("login"))
+
+    request.userdb.create_user(
+        username=username,
+        email=email,
+        github_id=github_id,
+        github_access_token=github_access_token,
+        linked_github_username=username,
+    )
+
+    headers = remember(request, username, max_age=60 * 60 * 24 * 365)
+    return HTTPFound(location=request.route_url("home"), headers=headers)
+
+
+def handle_github_link(request, github_id, username, github_access_token):
+    existing_user = request.userdb.get_user_by_github_id(github_id)
+    if existing_user:
+        request.session.flash(
+            "This GitHub account is already linked with another user.", "error"
+        )
+        return HTTPFound(location=request.route_url("profile"))
+
+    auth_user_id = request.authenticated_userid
+    if not auth_user_id:
+        request.session.flash(
+            "You need to be logged in to link a GitHub account.", "error"
+        )
+        return HTTPFound(location=request.route_url("login"))
+
+    user = request.userdb.get_user(auth_user_id)
+    user["github_id"] = github_id
+    user["github_access_token"] = github_access_token
+    user["linked_github_username"] = username
+    request.userdb.save_user(user)
+
+    request.session.flash("GitHub account linked successfully.", "success")
+    return HTTPFound(location=request.route_url("profile"))
+
+
 # Note that the allowed length of mailto URLs on Chrome/Windows is severely
 # limited.