Merge pull request #54 from VectorInstitute/add_delete_participants_cmd

Add delete participants command, fix snapshots

Author: amrit110

scripts/collect_coder_analytics.py

@@ -14,7 +14,7 @@
 import os
 import subprocess
 import sys
-from datetime import UTC, datetime
+from datetime import datetime, timezone
 from typing import Any
 
 import requests
@@ -216,15 +216,19 @@ def fetch_user_activity_insights(
         return {}
 
 
-def get_team_mappings() -> dict[str, str]:
-    """Get team mappings from Firestore.
+def get_participant_mappings() -> dict[str, dict[str, Any]]:
+    """Get participant data from Firestore including team and name info.
 
     Returns
     -------
-    dict[str, str]
-        Mapping of github_handle (lowercase) -> team_name
+    dict[str, dict[str, Any]]
+        Mapping of github_handle (lowercase) -> {
+            'team_name': str,
+            'first_name': str | None,
+            'last_name': str | None
+        }
     """
-    print("Fetching team mappings from Firestore...")
+    print("Fetching participant data from Firestore...")
 
     project_id = "coderd"
     database_id = "onboarding"
@@ -238,22 +242,25 @@ def get_team_mappings() -> dict[str, str]:
         data = doc.to_dict()
         if data:
             github_handle = doc.id.lower()
-            team_name = data.get("team_name", "Unassigned")
-            mappings[github_handle] = team_name
+            mappings[github_handle] = {
+                "team_name": data.get("team_name", "Unassigned"),
+                "first_name": data.get("first_name"),
+                "last_name": data.get("last_name"),
+            }
 
-    print(f"✓ Loaded {len(mappings)} participant team mappings")
+    print(f"✓ Loaded {len(mappings)} participant mappings")
     return mappings
 
 
 def fetch_workspaces(
-    team_mappings: dict[str, str], api_url: str, session_token: str
+    participant_mappings: dict[str, dict[str, Any]], api_url: str, session_token: str
 ) -> list[dict[str, Any]]:
     """Fetch workspaces using Coder CLI and enrich with build data.
 
     Parameters
     ----------
-    team_mappings : dict[str, str]
-        Mapping of github_handle -> team_name
+    participant_mappings : dict[str, dict[str, Any]]
+        Mapping of github_handle -> participant data (team_name, first_name, last_name)
     api_url : str
         Coder API base URL
     session_token : str
@@ -262,7 +269,7 @@ def fetch_workspaces(
     Returns
     -------
     list[dict[str, Any]]
-        List of workspace objects with builds, usage hours, and active hours
+        List of workspace objects with builds, usage hours, active hours, and team data
     """
     print("Fetching workspaces from Coder...")
     workspaces = run_command(["coder", "list", "-a", "-o", "json"])
@@ -276,7 +283,8 @@ def fetch_workspaces(
     filtered_workspaces = []
     for ws in workspaces:
         owner_name = ws.get("owner_name", "").lower()
-        team_name = team_mappings.get(owner_name, "Unassigned")
+        participant_data = participant_mappings.get(owner_name, {})
+        team_name = participant_data.get("team_name", "Unassigned")
 
         if team_name not in excluded_teams:
             filtered_workspaces.append(ws)
@@ -299,14 +307,14 @@ def fetch_workspaces(
             for ws in filtered_workspaces
             if ws.get("created_at")
         ),
-        default=datetime.now(UTC),
+        default=datetime.now(timezone.utc),
     )
     # Normalize to midnight (00:00:00) as required by the API
     start_time = earliest_created.replace(
         hour=0, minute=0, second=0, microsecond=0
     ).strftime("%Y-%m-%dT%H:%M:%SZ")
     # Normalize end time to the start of the current hour (required by API)
-    now = datetime.now(UTC)
+    now = datetime.now(timezone.utc)
     end_time = now.replace(minute=0, second=0, microsecond=0).strftime(
         "%Y-%m-%dT%H:%M:%SZ"
     )
@@ -337,6 +345,13 @@ def fetch_workspaces(
         owner_name = workspace.get("owner_name", "").lower()
         workspace["active_hours"] = activity_map.get(owner_name, 0.0)
 
+        # Enrich with participant data (team and name)
+        # This ensures all data needed for dashboard is in the snapshot
+        participant_data = participant_mappings.get(owner_name, {})
+        workspace["team_name"] = participant_data.get("team_name", "Unassigned")
+        workspace["owner_first_name"] = participant_data.get("first_name")
+        workspace["owner_last_name"] = participant_data.get("last_name")
+
         # Progress indicator
         if i % 10 == 0:
             print(f"  Processed {i}/{len(filtered_workspaces)} workspaces...")
@@ -371,7 +386,7 @@ def create_snapshot(
     workspaces: list[dict[str, Any]], templates: list[dict[str, Any]]
 ) -> dict[str, Any]:
     """Create a snapshot object with timestamp."""
-    timestamp = datetime.now(UTC).isoformat().replace("+00:00", "Z")
+    timestamp = datetime.now(timezone.utc).isoformat().replace("+00:00", "Z")
 
     snapshot = {
         "timestamp": timestamp,
@@ -454,11 +469,11 @@ def main() -> None:
     api_url, session_token = get_coder_api_config()
     print(f"✓ Using Coder API: {api_url}")
 
-    # Fetch team mappings first
-    team_mappings = get_team_mappings()
+    # Fetch participant mappings first
+    participant_mappings = get_participant_mappings()
 
     # Fetch data (with filtering and build enrichment)
-    workspaces = fetch_workspaces(team_mappings, api_url, session_token)
+    workspaces = fetch_workspaces(participant_mappings, api_url, session_token)
     templates = fetch_templates()
 
     # Create snapshot

services/analytics/app/api/snapshot/route.ts

@@ -1,6 +1,5 @@
 import { NextResponse } from 'next/server';
 import { getLatestSnapshot } from '@/lib/gcs';
-import { getTeamMappings } from '@/lib/firestore';
 import {
   enrichWorkspaceData,
   aggregateByTeam,
@@ -24,17 +23,13 @@ export async function GET() {
   try {
     console.log('Fetching latest Coder analytics snapshot...');
 
-    // Fetch data in parallel
-    const [snapshot, teamMappings] = await Promise.all([
-      getLatestSnapshot(),
-      getTeamMappings(),
-    ]);
+    // Fetch snapshot (team data is pre-enriched at collection time)
+    const snapshot = await getLatestSnapshot();
 
     console.log(`Snapshot contains ${snapshot.workspaces.length} workspaces and ${snapshot.templates.length} templates`);
-    console.log(`Loaded ${teamMappings.size} team mappings`);
 
-    // Enrich workspace data with team information
-    const workspaceMetrics = enrichWorkspaceData(snapshot.workspaces, teamMappings);
+    // Enrich workspace data with calculated metrics (team data already in snapshot)
+    const workspaceMetrics = enrichWorkspaceData(snapshot.workspaces);
     console.log(`Enriched ${workspaceMetrics.length} workspace metrics`);
 
     // Calculate aggregated metrics

services/analytics/lib/metrics.ts

@@ -5,7 +5,6 @@ import type {
   TeamMetrics,
   PlatformMetrics,
   TemplateMetrics,
-  ParticipantData,
   ActivityStatus,
   WorkspaceStatus,
   HealthStatus,
@@ -240,36 +239,38 @@ export function getHealthStatus(workspace: CoderWorkspace): HealthStatus {
 // ===== Main Enrichment Functions =====
 
 /**
- * Enrich workspace data with team information and calculated metrics
+ * Enrich workspace data with calculated metrics
+ * Team information is now pre-enriched in the snapshot at collection time
  */
 export function enrichWorkspaceData(
-  workspaces: CoderWorkspace[],
-  teamMappings: Map<string, ParticipantData>
+  workspaces: CoderWorkspace[]
 ): WorkspaceMetrics[] {
   const now = new Date();
 
   return workspaces.map((workspace) => {
-    const ownerHandle = workspace.owner_name.toLowerCase();
-    const participant = teamMappings.get(ownerHandle);
-
     const lastActive = getLastActiveTimestamp(workspace);
     const daysSinceActive = daysBetween(new Date(lastActive), now);
     const daysSinceCreated = daysBetween(new Date(workspace.created_at), now);
     const workspaceHours = getWorkspaceUsageHours(workspace);
     const activeHours = getWorkspaceActiveHours(workspace);
 
-    // Determine full name
+    // Use pre-enriched data from snapshot
+    const teamName = workspace.team_name || 'Unassigned';
+    const firstName = workspace.owner_first_name;
+    const lastName = workspace.owner_last_name;
+
+    // Build owner name
     let ownerName = workspace.owner_name;
-    if (participant?.first_name && participant?.last_name) {
-      ownerName = `${participant.first_name} ${participant.last_name}`;
+    if (firstName && lastName) {
+      ownerName = `${firstName} ${lastName}`;
     }
 
     return {
       workspace_id: workspace.id,
       workspace_name: workspace.name || `${workspace.owner_name}/workspace`,
       owner_github_handle: workspace.owner_name,
       owner_name: ownerName,
-      team_name: participant?.team_name || 'Unassigned',
+      team_name: teamName,
       template_id: workspace.template_id,
       template_name: workspace.template_name,
       template_display_name: workspace.template_display_name,

services/analytics/lib/types.ts

@@ -7,6 +7,12 @@ export interface CoderWorkspace {
   owner_id: string;
   owner_name: string;  // This is the GitHub username!
   owner_avatar_url?: string;
+
+  // Pre-enriched participant data from snapshot (added at collection time)
+  team_name: string;  // Team assignment at snapshot time
+  owner_first_name?: string;  // First name at snapshot time
+  owner_last_name?: string;  // Last name at snapshot time
+
   organization_id: string;
   organization_name: string;
   template_id: string;

src/aieng_platform_onboard/admin/cli.py

@@ -3,6 +3,9 @@
 import argparse
 import sys
 
+from aieng_platform_onboard.admin.delete_participants import (
+    delete_participants_from_csv,
+)
 from aieng_platform_onboard.admin.setup_participants import (
     setup_participants_from_csv,
 )
@@ -46,11 +49,39 @@ def main() -> int:
         help="Validate and show what would be done without making changes",
     )
 
+    # delete-participants subcommand
+    delete_participants_parser = subparsers.add_parser(
+        "delete-participants",
+        help="Delete participants from Firestore database",
+        description="Remove participants and optionally empty teams from Firestore",
+    )
+    delete_participants_parser.add_argument(
+        "csv_file",
+        type=str,
+        help="Path to CSV file with column: github_handle",
+    )
+    delete_participants_parser.add_argument(
+        "--dry-run",
+        action="store_true",
+        help="Validate and show what would be done without making changes",
+    )
+    delete_participants_parser.add_argument(
+        "--keep-empty-teams",
+        action="store_true",
+        help="Keep teams even if they become empty after removing participants",
+    )
+
     args = parser.parse_args()
 
     # Route to appropriate command handler
     if args.command == "setup-participants":
         return setup_participants_from_csv(args.csv_file, dry_run=args.dry_run)
+    if args.command == "delete-participants":
+        return delete_participants_from_csv(
+            args.csv_file,
+            delete_empty_teams=not args.keep_empty_teams,
+            dry_run=args.dry_run,
+        )
 
     # Should never reach here due to required=True
     parser.print_help()