Update user names in crawls and workflows after username update (#1299)

Fixes #1275

backend/btrixcloud/basecrawls.py

@@ -206,6 +206,12 @@ class BaseCrawlOps:
             {"$set": data},
         )
 
+    async def update_usernames(self, userid: uuid.UUID, updated_name: str) -> None:
+        """Update username references matching userid"""
+        await self.crawls.update_many(
+            {"userid": userid}, {"$set": {"userName": updated_name}}
+        )
+
     async def shutdown_crawl(self, crawl_id: str, org: Organization, graceful: bool):
         """stop or cancel specified crawl"""
         crawl = await self.get_crawl_raw(crawl_id, org)
@@ -764,3 +770,5 @@ def init_base_crawls_api(
         org: Organization = Depends(org_crawl_dep),
     ):
         return await ops.delete_crawls_all_types(delete_list, org, user)
+
+    return ops

backend/btrixcloud/crawlconfigs.py

@@ -340,6 +340,14 @@ class CrawlConfigOps:
             ret["started"] = crawl_id
         return ret
 
+    async def update_usernames(self, userid: uuid.UUID, updated_name: str) -> None:
+        """Update username references matching userid"""
+        for workflow_field in ["createdBy", "modifiedBy", "lastStartedBy"]:
+            await self.crawl_configs.update_many(
+                {workflow_field: userid},
+                {"$set": {f"{workflow_field}Name": updated_name}},
+            )
+
     async def get_crawl_configs(
         self,
         org: Organization,

backend/btrixcloud/main.py

@@ -76,8 +76,6 @@ def main():
 
     event_webhook_ops = init_event_webhooks_api(mdb, org_ops, app_root)
 
-    user_manager.set_org_ops(org_ops)
-
     # pylint: disable=import-outside-toplevel
     if not os.environ.get("KUBERNETES_SERVICE_HOST"):
         print(
@@ -106,7 +104,7 @@ def main():
 
     coll_ops = init_collections_api(app, mdb, org_ops, storage_ops, event_webhook_ops)
 
-    init_base_crawls_api(
+    base_crawl_ops = init_base_crawls_api(
         app,
         mdb,
         user_manager,
@@ -118,6 +116,8 @@ def main():
         current_active_user,
     )
 
+    user_manager.set_ops(org_ops, crawl_config_ops, base_crawl_ops)
+
     crawls = init_crawls_api(
         app,
         mdb,

backend/btrixcloud/main_op.py

@@ -41,8 +41,6 @@ def main():
 
     event_webhook_ops = EventWebhookOps(mdb, org_ops)
 
-    user_manager.set_org_ops(org_ops)
-
     # pylint: disable=import-outside-toplevel
     if not os.environ.get("KUBERNETES_SERVICE_HOST"):
         print(
@@ -66,6 +64,8 @@ def main():
         profile_ops,
     )
 
+    user_manager.set_ops(org_ops, crawl_config_ops, None)
+
     coll_ops = CollectionOps(mdb, crawl_manager, org_ops, event_webhook_ops)
 
     crawl_ops = CrawlOps(

backend/btrixcloud/users.py

@@ -51,21 +51,26 @@ from .auth import (
 
 
 # ============================================================================
-# pylint: disable=raise-missing-from, too-many-public-methods
+# pylint: disable=raise-missing-from, too-many-public-methods, too-many-instance-attributes
 class UserManager:
     """Browsertrix UserManager"""
 
     def __init__(self, mdb, email, invites):
         self.users = mdb.get_collection("users")
+        self.crawl_config_ops = None
+        self.base_crawl_ops = None
         self.email = email
         self.invites = invites
         self.org_ops = None
 
         self.registration_enabled = is_bool(os.environ.get("REGISTRATION_ENABLED"))
 
-    def set_org_ops(self, ops):
+    # pylint: disable=attribute-defined-outside-init
+    def set_ops(self, org_ops, crawl_config_ops, base_crawl_ops):
         """set org ops"""
-        self.org_ops = ops
+        self.org_ops = org_ops
+        self.crawl_config_ops = crawl_config_ops
+        self.base_crawl_ops = base_crawl_ops
 
     async def init_index(self):
         """init lookup index"""
@@ -476,6 +481,10 @@ class UserManager:
 
         await self.update_email_name(user, user_update.email, user_update.name)
 
+        if user_update.name:
+            await self.base_crawl_ops.update_usernames(user.id, user_update.name)
+            await self.crawl_config_ops.update_usernames(user.id, user_update.name)
+
     async def update_verified(self, user: User) -> None:
         """Update verified status for user"""
         await self.users.find_one_and_update(

backend/test/test_users.py

@@ -1,7 +1,13 @@
 import requests
 import time
 
-from .conftest import API_PREFIX, CRAWLER_USERNAME, ADMIN_PW, ADMIN_USERNAME
+from .conftest import (
+    API_PREFIX,
+    CRAWLER_USERNAME,
+    ADMIN_PW,
+    ADMIN_USERNAME,
+    FINISHED_STATES,
+)
 
 VALID_USER_EMAIL = "validpassword@example.com"
 VALID_USER_PW = "validpassw0rd!"
@@ -230,14 +236,67 @@ def test_reset_valid_password(admin_auth_headers, default_org_id):
     assert r.json()["updated"] == True
 
 
-def test_patch_me_endpoint(admin_auth_headers, default_org_id):
+def test_patch_me_endpoint(admin_auth_headers, default_org_id, admin_userid):
+    # Start a new crawl
+    crawl_data = {
+        "runNow": True,
+        "name": "name change test crawl",
+        "config": {
+            "seeds": [{"url": "https://specs.webrecorder.net/", "depth": 1}],
+        },
+    }
+    r = requests.post(
+        f"{API_PREFIX}/orgs/{default_org_id}/crawlconfigs/",
+        headers=admin_auth_headers,
+        json=crawl_data,
+    )
+    data = r.json()
+    crawl_id = data["run_now_job"]
+
+    # Wait for it to complete
+    while True:
+        r = requests.get(
+            f"{API_PREFIX}/orgs/{default_org_id}/crawls/{crawl_id}/replay.json",
+            headers=admin_auth_headers,
+        )
+        data = r.json()
+        if data["state"] in FINISHED_STATES:
+            break
+        time.sleep(5)
+
+    # Change user name and email
+    new_name = "New Admin"
     r = requests.patch(
         f"{API_PREFIX}/users/me",
         headers=admin_auth_headers,
-        json={"email": "admin2@example.com", "name": "New Admin"},
+        json={"email": "admin2@example.com", "name": new_name},
     )
     assert r.status_code == 200
 
+    # Verify that name was updated in workflows and crawls
+    for workflow_field in ["createdBy", "modifiedBy", "lastStartedBy"]:
+        r = requests.get(
+            f"{API_PREFIX}/orgs/{default_org_id}/crawlconfigs?{workflow_field}={admin_userid}",
+            headers=admin_auth_headers,
+        )
+        assert r.status_code == 200
+        data = r.json()
+        assert data["total"] > 0
+        for workflow in data["items"]:
+            if workflow[workflow_field] == admin_userid:
+                assert workflow[f"{workflow_field}Name"] == new_name
+
+    r = requests.get(
+        f"{API_PREFIX}/orgs/{default_org_id}/crawls?userid={admin_userid}",
+        headers=admin_auth_headers,
+    )
+    assert r.status_code == 200
+    data = r.json()
+    assert data["total"] > 0
+    for item in data["items"]:
+        if item["userid"] == admin_userid:
+            assert item["userName"] == new_name
+
 
 def test_patch_me_invalid_email_in_use(admin_auth_headers, default_org_id):
     r = requests.patch(