jeet/grateful-journal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

added encryption

Browse Source
This commit is contained in:
Jeet Debnath
2026-03-09 10:54:07 +05:30
parent 6e184dc590
commit 6720e28d08
27 changed files with 2093 additions and 709 deletions
Download Patch File Download Diff File Expand all files Collapse all files

32
backend/scripts/create_indexes.py
View File

@@ -15,18 +15,18 @@ from typing import Dict, List, Tuple
def create_indexes():
"""Create all required MongoDB indexes."""
settings = get_settings()
client = MongoClient(settings.mongodb_uri)
db = client[settings.mongodb_db_name]
print(f"✓ Connected to MongoDB: {settings.mongodb_db_name}\n")
indexes_created = []
# ========== USERS COLLECTION INDEXES ==========
print("Creating indexes for 'users' collection...")
# Unique index on email
try:
db.users.create_index(
@@ -38,7 +38,7 @@ def create_indexes():
print(" ✓ Created unique index on email")
except Exception as e:
print(f" ⚠ Email index: {e}")
# Index on createdAt for sorting
try:
db.users.create_index(
@@ -49,10 +49,10 @@ def create_indexes():
print(" ✓ Created index on createdAt")
except Exception as e:
print(f" ⚠ createdAt index: {e}")
# ========== ENTRIES COLLECTION INDEXES ==========
print("\nCreating indexes for 'entries' collection...")
# Compound index: userId + createdAt (for history pagination)
try:
db.entries.create_index(
@@ -63,7 +63,7 @@ def create_indexes():
print(" ✓ Created compound index on (userId, createdAt)")
except Exception as e:
print(f" ⚠ userId_createdAt index: {e}")
# Compound index: userId + entryDate (for calendar queries)
try:
db.entries.create_index(
@@ -74,7 +74,7 @@ def create_indexes():
print(" ✓ Created compound index on (userId, entryDate)")
except Exception as e:
print(f" ⚠ userId_entryDate index: {e}")
# Index on tags for searching (optional, for future)
try:
db.entries.create_index(
@@ -85,7 +85,7 @@ def create_indexes():
print(" ✓ Created index on tags")
except Exception as e:
print(f" ⚠ tags index: {e}")
# Index on entryDate range queries (for calendar)
try:
db.entries.create_index(
@@ -96,7 +96,7 @@ def create_indexes():
print(" ✓ Created index on entryDate")
except Exception as e:
print(f" ⚠ entryDate index: {e}")
# TTL Index on entries (optional: for auto-deleting old entries if needed)
# Uncomment if you want entries to auto-delete after 2 years
# try:
@@ -108,7 +108,7 @@ def create_indexes():
# print(" ✓ Created TTL index on createdAt (2 years)")
# except Exception as e:
# print(f" ⚠ TTL index: {e}")
# ========== SUMMARY ==========
print(f"\n{'='*60}")
print(f"✓ Index Creation Complete")
@@ -116,18 +116,18 @@ def create_indexes():
print(f"Total indexes created: {len(indexes_created)}")
for collection, index_name in indexes_created:
print(f" • {collection}.{index_name}")
# Optional: Print summary of all indexes
print(f"\n{'='*60}")
print("All Indexes Summary")
print(f"{'='*60}")
for collection_name in ["users", "entries"]:
print(f"\n{collection_name}:")
collection = db[collection_name]
for index_info in collection.list_indexes():
print(f" • {index_info['name']}")
client.close()
print("\n✓ Disconnected from MongoDB")

89
backend/scripts/migrate_data.py
View File

@@ -27,21 +27,21 @@ import sys
def migrate_data():
"""Perform complete data migration."""
settings = get_settings()
client = MongoClient(settings.mongodb_uri)
db = client[settings.mongodb_db_name]
print(f"✓ Connected to MongoDB: {settings.mongodb_db_name}\n")
# ========== STEP 1: DEDUPLICATE USERS ==========
print("=" * 70)
print("STEP 1: Deduplicating Users (keeping oldest)")
print("=" * 70)
duplicate_count = 0
user_mapping = {} # Maps old duplicates to canonical user ID
# Group users by email
email_groups = {}
for user in db.users.find():
@@ -49,7 +49,7 @@ def migrate_data():
if email not in email_groups:
email_groups[email] = []
email_groups[email].append(user)
# Process each email group
for email, users in email_groups.items():
if len(users) > 1:
@@ -57,52 +57,53 @@ def migrate_data():
users.sort(key=lambda u: u["createdAt"])
canonical_user = users[0]
canonical_id = canonical_user["_id"]
print(f"\n📧 Email: {email}")
print(f" Found {len(users)} duplicate users")
print(f" Keeping (earliest): {canonical_id}")
# Map all other users to canonical
for dup_user in users[1:]:
dup_id = dup_user["_id"]
user_mapping[str(dup_id)] = canonical_id
duplicate_count += 1
print(f" Deleting (later): {dup_id}")
# Delete duplicate users
for user in users[1:]:
db.users.delete_one({"_id": user["_id"]})
if duplicate_count == 0:
print("\n✓ No duplicate users found")
else:
print(f"\n✓ Removed {duplicate_count} duplicate users")
# ========== STEP 2: MIGRATE ENTRIES ==========
print("\n" + "=" * 70)
print("STEP 2: Migrating Entries (userId string → ObjectId, add entryDate)")
print("=" * 70)
total_entries = db.entries.count_documents({})
entries_updated = 0
entries_with_issues = []
print(f"\nTotal entries to process: {total_entries}\n")
for entry in db.entries.find():
try:
entry_id = entry["_id"]
old_user_id_str = entry.get("userId", "")
# Convert userId: string → ObjectId
if isinstance(old_user_id_str, str):
# Check if this userId is in the duplicate mapping
if old_user_id_str in user_mapping:
new_user_id = user_mapping[old_user_id_str]
print(f" → Entry {entry_id}: userId mapped {old_user_id_str[:8]}... → {str(new_user_id)[:8]}...")
print(
f" → Entry {entry_id}: userId mapped {old_user_id_str[:8]}... → {str(new_user_id)[:8]}...")
else:
new_user_id = ObjectId(old_user_id_str)
update_data = {
"userId": new_user_id,
}
@@ -110,14 +111,15 @@ def migrate_data():
# Already an ObjectId
new_user_id = old_user_id_str
update_data = {}
# Add entryDate if missing (default to createdAt)
if "entryDate" not in entry:
entry_date = entry.get("createdAt", datetime.utcnow())
# Set to start of day
entry_date = entry_date.replace(hour=0, minute=0, second=0, microsecond=0)
entry_date = entry_date.replace(
hour=0, minute=0, second=0, microsecond=0)
update_data["entryDate"] = entry_date
# Add encryption metadata if missing
if "encryption" not in entry:
update_data["encryption"] = {
@@ -125,7 +127,7 @@ def migrate_data():
"iv": None,
"algorithm": None
}
# Perform update if there are changes
if update_data:
update_data["updatedAt"] = datetime.utcnow()
@@ -134,61 +136,65 @@ def migrate_data():
{"$set": update_data}
)
entries_updated += 1
if entries_updated % 100 == 0:
print(f" ✓ Processed {entries_updated}/{total_entries} entries")
print(
f" ✓ Processed {entries_updated}/{total_entries} entries")
except Exception as e:
entries_with_issues.append({
"entry_id": str(entry_id),
"error": str(e)
})
print(f" ⚠ Error processing entry {entry_id}: {e}")
print(f"\n✓ Updated {entries_updated}/{total_entries} entries")
if entries_with_issues:
print(f"\n⚠ {len(entries_with_issues)} entries had issues:")
for issue in entries_with_issues[:5]: # Show first 5
print(f" - {issue['entry_id']}: {issue['error']}")
# ========== STEP 3: VERIFY DATA INTEGRITY ==========
print("\n" + "=" * 70)
print("STEP 3: Verifying Data Integrity")
print("=" * 70)
# Check for orphaned entries (userId doesn't exist in users)
orphaned_count = 0
users_ids = set(str(u["_id"]) for u in db.users.find({}, {"_id": 1}))
for entry in db.entries.find({}, {"userId": 1}):
user_id = entry.get("userId")
if isinstance(user_id, ObjectId):
user_id = str(user_id)
if user_id not in users_ids:
orphaned_count += 1
print(f"\nUsers collection: {db.users.count_documents({})}")
print(f"Entries collection: {db.entries.count_documents({})}")
if orphaned_count > 0:
print(f"\n⚠ WARNING: Found {orphaned_count} orphaned entries (no corresponding user)")
print(
f"\n⚠ WARNING: Found {orphaned_count} orphaned entries (no corresponding user)")
else:
print(f"✓ All entries have valid user references")
# Sample entry check
sample_entry = db.entries.find_one()
if sample_entry:
print(f"\nSample entry structure:")
print(f" _id (entry): {sample_entry['_id']} (ObjectId: {isinstance(sample_entry['_id'], ObjectId)})")
print(f" userId: {sample_entry.get('userId')} (ObjectId: {isinstance(sample_entry.get('userId'), ObjectId)})")
print(
f" _id (entry): {sample_entry['_id']} (ObjectId: {isinstance(sample_entry['_id'], ObjectId)})")
print(
f" userId: {sample_entry.get('userId')} (ObjectId: {isinstance(sample_entry.get('userId'), ObjectId)})")
print(f" entryDate present: {'entryDate' in sample_entry}")
print(f" encryption present: {'encryption' in sample_entry}")
if "entryDate" in sample_entry:
print(f" → entryDate: {sample_entry['entryDate'].isoformat()}")
if "encryption" in sample_entry:
print(f" → encryption: {sample_entry['encryption']}")
# ========== SUMMARY ==========
print(f"\n{'='*70}")
print("✓ Migration Complete")
@@ -196,12 +202,12 @@ def migrate_data():
print(f"Duplicate users removed: {duplicate_count}")
print(f"Entries migrated: {entries_updated}")
print(f"Orphaned entries found: {orphaned_count}")
if orphaned_count == 0:
print("\n✓ Data integrity verified successfully!")
else:
print(f"\n⚠ Please review {orphaned_count} orphaned entries")
client.close()
print("\n✓ Disconnected from MongoDB")
@@ -234,12 +240,13 @@ This script modifies your MongoDB database. Before running:
if __name__ == "__main__":
rollback_warning()
response = input("\nDo you want to proceed with migration? (yes/no): ").strip().lower()
response = input(
"\nDo you want to proceed with migration? (yes/no): ").strip().lower()
if response != "yes":
print("Migration cancelled.")
sys.exit(0)
try:
migrate_data()
except Exception as e: