vuluu2k/builderx_api-mongodb

BuilderX API MongoDB Skill

The builderx_api project integrates MongoDB via the mongodb_driver alongside its primary Postgres (Citus) database. This is used extensively for the Dynamic Database Collections feature in BuilderxApi.DBCollections.DBCollections (lib/builderx_api/db_collections/db_collections.ex).

In this pattern, metadata about the data models (schema) is stored in Postgres (DBCollection), but the actual records are physically stored in MongoDB (MongoRepo) using a single records table separated by table_name and site_id.

1. Interacting with MongoDB Collections

You should generally not interact with MongoRepo directly unless you are inside the builderx_api/db_collections/... scope.

Instead, use DBCollections:

Checking if a record exists

filters = %{"slug" => "my-record"}

# conn must have assigns for customer, account, or is_check_record_creator as required
DBCollections.exists_record(table_name, filters, db_collection_struct, conn)
# => {:ok, true | false}

Querying records

Retrieves customized results based on dynamic schemas.

select = %{"id" => 1, "name" => 1}
filters = %{"status" => "active"}
limit = 10
skip = 0
sort = %{"inserted_at" => -1} # Use 1 for ASC, -1 for DESC
populate = [] # Populate relations if any references are configured
params = %{"site_id" => "site_uuid"}

DBCollections.query_record(
  table_name,
  select,
  filters,
  sort,
  limit,
  skip,
  populate,
  params,
  conn
)
# => List of normalized maps

Inserting records

# attrs is a list of map: [%{"field_name" => "name", "field_value" => "Record 1"}]
# Note that we use a custom key format for dynamic mapping.

{:ok, inserted_record} = DBCollections.insert_record(table_name, attrs, params, conn)

2. Using MongoRepo directly

The BuilderxApi.MongoRepo is an abstraction over :mongo (the mongodb_driver pool). For some administrative actions, it is called directly:

alias BuilderxApi.MongoRepo

table = "records"

# Find
records = MongoRepo.find(table, %{"site_id" => site_id, "table_name" => "users"})

# Find one
record = MongoRepo.find_one(table, %{"_id" => id})

# Update Many
MongoRepo.update_many(
  table,
  %{"site_id" => site_id, "table_name" => "users"},
  %{"$unset" => %{"webcmscol_removed_field" => ""}}
)

# Insert Many
MongoRepo.insert_many(table, list_of_maps)

# Delete Many
MongoRepo.delete_many(table, %{"site_id" => site_id, "table_name" => "users"})

Important Patterns

• webcmscol_: The system prepends webcmscol_ to column names stored in MongoDB to prevent clashes with system variables like _id, site_id, table_name. You will see operations map/unmap this prefix (DBUtils.sanitize_column_name/1).
• Caching Counts: Because counting documents in Mongo can slowly become a bottleneck, the total document count per site collection is cached in Redis: db_collection_records::{site_id}.
• All MongoDB records share the records collection but are differentiated by standard root fields: "site_id" and "table_name".