Schema Definition

The schema file defines how to map your database tables to Elasticsearch/OpenSearch documents.

Basic Example

[
  {
    "database": "mydb",
    "index": "books",
    "nodes": {
      "table": "book",
      "columns": ["isbn", "title"],
      "children": [
        {
          "table": "author",
          "columns": ["name"]
        }
      ]
    }
  }
]

Full Schema Reference

[
  {
    "database": "<database name>",
    "index": "<index name>",
    "setting": { },
    "plugins": ["<Plugin A>", "<Plugin B>"],
    "pipeline": "<pipeline>",
    "routing": "<routing>",
    "nodes": {
      "table": "<table name>",
      "schema": "<schema name>",
      "columns": ["<column 1>", "<column 2>"],
      "label": "<label>",
      "children": [
        {
          "table": "<child table>",
          "columns": ["<column>"],
          "label": "<label>",
          "relationship": {
            "variant": "object | scalar",
            "type": "one_to_one | one_to_many",
            "through_tables": ["<junction table>"]
          },
          "transform": {
            "rename": { },
            "mapping": { },
            "concat": { }
          }
        }
      ]
    }
  }
]

---

Root Properties

Property	Required	Description
database	Yes	Database name
index	No	Elasticsearch/OpenSearch index name (defaults to database name)
nodes	Yes	Root node defining the document structure
plugins	No	List of plugins to apply
pipeline	No	Elasticsearch ingest pipeline
routing	No	Elasticsearch routing field
setting	No	Elasticsearch index settings

Index Settings

{
  "setting": {
    "analysis": {
      "analyzer": {
        "ngram_analyzer": {
          "filter": ["lowercase"],
          "type": "custom",
          "tokenizer": "ngram_tokenizer"
        }
      },
      "tokenizer": {
        "ngram_tokenizer": {
          "token_chars": ["letter", "digit"],
          "min_gram": "2",
          "max_gram": "10",
          "type": "ngram"
        }
      }
    }
  }
}

---

Node Properties

Property	Required	Description
table	Yes	Database table name
schema	No	Database schema (defaults to public)
columns	No	Columns to include (defaults to all)
label	No	Field name in output document (defaults to table name)
children	No	Child nodes (nested documents)
transform	No	Transform operations
relationship	No	Relationship configuration (for child nodes)

---

Relationship

Defines how parent and child nodes are related.

Property	Values	Description
variant	object, scalar	Output format
type	one_to_one, one_to_many	Relationship cardinality
through_tables	["table_name"]	Junction tables for many-to-many

Variant Examples

objectscalar

Returns the full object:

{
  "author": {
    "id": 1,
    "name": "George Orwell"
  }
}

Returns only the value(s):

{
  "author": ["George Orwell", "Aldous Huxley"]
}

---

Transform

Transform operations modify the output document.

rename

Rename columns in the output:

"transform": {
  "rename": {
    "isbn": "book_isbn",
    "title": "book_title"
  }
}

mapping

Specify Elasticsearch field mappings:

"transform": {
  "mapping": {
    "isbn": { "type": "keyword" },
    "title": { "type": "text" },
    "price": { "type": "float" }
  }
}

concat

Concatenate columns into a new field:

"transform": {
  "concat": {
    "columns": ["first_name", "last_name"],
    "destination": "full_name",
    "delimiter": " "
  }
}

---

Re-indexing Required

Changing the schema changes the document structure. You must re-index after schema changes.