Step Chaining

Steps execute in order. Each step's result is stored and available to all steps that follow it.

How Steps Chain

step_0 (SQL)    → result stored as step_0
      ↓
step_1 (Lark)   → can read step_0, result stored as step_1
      ↓
step_2 (Expr)   → can read step_0, step_1, and input
      ↓
step_3 (SQL)    → can bind values from step_2 via ?:field
      ↓
step_4 (SQL)    → final result returned if omit: false

Each step number maps to its position in the logic list — zero-indexed.

Controlling Response Output

Only steps with omit: false (or omit not set) appear in the response.

logic:
  - sql: SELECT * FROM users     # step_0 — omit: false (default) → returned

  - lark: transform              # step_1 — omit: true → not returned
    omit: true
    code: |
      result = [...]

  - sql: SELECT * FROM users     # step_2 — omit: false (default) → returned

When multiple steps have omit: false, all their results are returned.

Passing Values Between Steps

SQL → Lark

A SQL result is a list of row objects. Access it in Lark as step_N:

- sql: |
    SELECT id, name, email FROM users
  omit: true

- lark: transform
  code: |
    result = []
    for row in step_0:
        result.append({
            "id": row["id"],
            "name": row["name"],
        })

Lark → SQL (parameter binding)

A Lark step sets result to a dict. SQL binds from it via ?:field_name:

- lark: new_row
  omit: true
  code: |
    row = step_0
    result = {
        "id": row["id"],
        "name": row["name"],
        "email": row["email"],
    }

- sql: |
    INSERT INTO users VALUES(?:id, '?:name', '?:email', now('UTC'))
  omit: true

Expr → Response

An Expr step builds the final response shape from earlier step data:

- sql: |
    SELECT * FROM users
  omit: true

- expr: build_response
  code: |
    {
      "total": len(step_0),
      "users": step_0
    }

Complete Example

Fetch users → transform → insert a copy → return final list:

logic:
  # step_0: fetch all users
  - sql: |
      SELECT * FROM users ORDER BY id DESC
    omit: true

  # step_1: transform — mask emails, slugify names
  - lark: enhance_data
    omit: true
    code: |
      load("strings", "slugify")

      def mask_email(email):
          parts = email.split("@")
          return parts + "***@" + parts[1]

      result = []
      for row in step_0:
          result.append({
              "id": row["id"],
              "name": slugify(row["name"]),
              "email": mask_email(row["email"]),
          })

  # step_2: pick first row for insert
  - lark: new_row
    omit: true
    code: |
      row = step_1
      result = {
          "id": row["id"] + 9999,
          "name": row["name"],
          "email": row["email"],
      }

  # step_3: insert the new row
  - sql: |
      INSERT INTO users VALUES(?:id, '?:name', '?:email', now('UTC'))
    omit: true

  # step_4: return final list
  - sql: |
      SELECT * FROM users

Only step_4 appears in the response. All intermediate steps are omit: true.