Add superuser endpoint to get user emails with org info (#2211)

Fixes #2203

---------

Co-authored-by: Ilya Kreymer <ikreymer@gmail.com>

backend/btrixcloud/models.py

@@ -208,6 +208,14 @@ class UserOut(BaseModel):
     orgs: List[UserOrgInfoOut]
 
 
+# ============================================================================
+class UserEmailWithOrgInfo(BaseModel):
+    """Output model for getting user email list with org info for each"""
+
+    email: EmailStr
+    orgs: List[UserOrgInfoOut]
+
+
 # ============================================================================
 
 ### CRAWL STATES
@@ -2453,3 +2461,10 @@ class PaginatedCrawlErrorResponse(PaginatedResponse):
     """Response model for crawl errors"""
 
     items: List[CrawlError]
+
+
+# ============================================================================
+class PaginatedUserEmailsResponse(PaginatedResponse):
+    """Response model for user emails with org info"""
+
+    items: List[UserEmailWithOrgInfo]

backend/btrixcloud/users.py

@@ -6,7 +6,7 @@ import os
 from uuid import UUID, uuid4
 import asyncio
 
-from typing import Optional, List, TYPE_CHECKING, cast, Callable
+from typing import Optional, List, TYPE_CHECKING, cast, Callable, Tuple
 
 from fastapi import (
     Request,
@@ -34,6 +34,8 @@ from .models import (
     FailedLogin,
     UpdatedResponse,
     SuccessResponse,
+    UserEmailWithOrgInfo,
+    PaginatedUserEmailsResponse,
 )
 from .pagination import DEFAULT_PAGE_SIZE, paginated_format
 from .utils import is_bool, dt_now
@@ -546,6 +548,30 @@ class UserManager:
             return 0
         return failed_login.get("count", 0)
 
+    async def get_user_emails(
+        self,
+        page_size: int = DEFAULT_PAGE_SIZE,
+        page: int = 1,
+    ) -> Tuple[List[UserEmailWithOrgInfo], int]:
+        """Get user emails with org info for each for paginated endpoint"""
+        # Zero-index page for query
+        page = page - 1
+        skip = page_size * page
+
+        emails: List[UserEmailWithOrgInfo] = []
+
+        total = await self.users.count_documents({"is_superuser": False})
+        async for res in self.users.find(
+            {"is_superuser": False}, skip=skip, limit=page_size
+        ):
+            user = User(**res)
+            user_out = await self.get_user_info_with_orgs(user)
+            emails.append(
+                UserEmailWithOrgInfo(email=user_out.email, orgs=user_out.orgs)
+            )
+
+        return emails, total
+
 
 # ============================================================================
 def init_user_manager(mdb, emailsender, invites):
@@ -706,4 +732,21 @@ def init_users_router(
         )
         return paginated_format(pending_invites, total, page, pageSize)
 
+    @users_router.get(
+        "/emails", tags=["users"], response_model=PaginatedUserEmailsResponse
+    )
+    async def get_user_emails(
+        user: User = Depends(current_active_user),
+        pageSize: int = DEFAULT_PAGE_SIZE,
+        page: int = 1,
+    ):
+        """Get emails of registered users with org information (superuser only)"""
+        if not user.is_superuser:
+            raise HTTPException(status_code=403, detail="not_allowed")
+
+        emails, total = await user_manager.get_user_emails(
+            page_size=pageSize, page=page
+        )
+        return paginated_format(emails, total, page, pageSize)
+
     return users_router

backend/test/test_users.py

@@ -774,3 +774,42 @@ def test_patch_me_invalid_email_in_use(admin_auth_headers, default_org_id):
     )
     assert r.status_code == 400
     assert r.json()["detail"] == "user_already_exists"
+
+
+def test_user_emails_endpoint_non_superuser(crawler_auth_headers, default_org_id):
+    r = requests.get(
+        f"{API_PREFIX}/users/emails",
+        headers=crawler_auth_headers,
+    )
+    assert r.status_code == 403
+    assert r.json()["detail"] == "not_allowed"
+
+
+def test_user_emails_endpoint_superuser(admin_auth_headers, default_org_id):
+    r = requests.get(
+        f"{API_PREFIX}/users/emails",
+        headers=admin_auth_headers,
+    )
+    assert r.status_code == 200
+    data = r.json()
+
+    total = data["total"]
+    user_emails = data["items"]
+
+    assert total > 0
+    assert total == len(user_emails)
+
+    for user in user_emails:
+        assert user["email"]
+        orgs = user.get("orgs")
+        if orgs == []:
+            continue
+
+        for org in orgs:
+            assert org["id"]
+            assert org["name"]
+            assert org["slug"]
+            assert org["default"] in (True, False)
+            role = org["role"]
+            assert role
+            assert isinstance(role, int)

backend/test_nightly/test_storage_quota.py

@@ -15,6 +15,7 @@ config_id = None
 
 storage_quota = None
 
+
 def run_crawl(org_id, headers):
     crawl_data = {
         "runNow": True,