{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "$id": "https://quarka.org/schema/assistant/definitions/transform-step.schema.json",
  "title": "Transform step",
  "description": "One operation in a data_source transform pipeline. Exactly one operation per step.",
  "type": "object",
  "oneOf": [
    {
      "additionalProperties": false,
      "required": ["sort"],
      "properties": {
        "sort": { "type": "object", "minProperties": 1, "additionalProperties": { "enum": ["asc", "desc"] } }
      }
    },
    {
      "additionalProperties": false,
      "required": ["limit"],
      "properties": { "limit": { "type": "integer", "minimum": 1 } }
    },
    {
      "additionalProperties": false,
      "required": ["filter"],
      "properties": {
        "filter": {
          "type": "object",
          "minProperties": 1,
          "additionalProperties": { "$ref": "#/definitions/filterCondition" }
        }
      }
    },
    {
      "additionalProperties": false,
      "required": ["group_by", "agg"],
      "properties": {
        "group_by": { "type": "string", "minLength": 1 },
        "agg": { "type": "object", "minProperties": 1, "additionalProperties": { "enum": ["sum", "avg", "count", "count_distinct", "min", "max", "first"] } },
        "keep": { "type": "array", "items": { "type": "string" } }
      }
    },
    {
      "additionalProperties": false,
      "required": ["tally"],
      "properties": {
        "tally": {
          "type": "object",
          "additionalProperties": false,
          "required": ["buckets"],
          "properties": {
            "buckets": {
              "type": "array",
              "minItems": 1,
              "items": {
                "type": "object",
                "additionalProperties": false,
                "required": ["into"],
                "properties": {
                  "into": { "type": "string", "pattern": "^[a-z][a-z0-9_]*$" },
                  "count": { "type": "boolean", "enum": [true] },
                  "count_distinct": { "type": "string", "minLength": 1 },
                  "when": {
                    "type": "object",
                    "minProperties": 1,
                    "additionalProperties": { "$ref": "#/definitions/filterCondition" }
                  }
                },
                "oneOf": [
                  { "required": ["count"], "not": { "required": ["count_distinct"] } },
                  { "required": ["count_distinct"], "not": { "required": ["count"] } }
                ]
              }
            }
          }
        }
      }
    },
    {
      "additionalProperties": false,
      "required": ["calc", "expr"],
      "properties": {
        "calc": { "type": "string", "minLength": 1 },
        "expr": { "type": "string", "minLength": 1 },
        "scope": { "enum": ["row", "global"] }
      }
    },
    {
      "additionalProperties": false,
      "required": ["lookup", "key", "into"],
      "properties": {
        "lookup": { "type": "string", "minLength": 1 },
        "key": { "type": "string", "minLength": 1 },
        "into": { "type": "string", "minLength": 1 }
      }
    },
    {
      "additionalProperties": false,
      "required": ["set_var", "expr"],
      "properties": {
        "set_var": { "type": "string", "minLength": 1 },
        "expr": { "type": "string", "minLength": 1 }
      }
    },
    {
      "$comment": "Issue #1307 (declaration/implementation sync, step 1): extract_host has been implemented since the 2026-06-12 qa-labo sync but was never declared, so a manifest using it was rejected by both validators. Adopted as first-class vocabulary per the 2026-07-22 roadmap ruling (small, unambiguous, useful on its own). The multi-query layer (queries[] / for_each / combine / sources_template / loading_text / native_qal / params) is explicitly FROZEN and must NOT be added here — keeping it out of the schema is what stops new manifests from being written against it.",
      "additionalProperties": false,
      "required": ["extract_host"],
      "properties": {
        "extract_host": {
          "type": "object",
          "additionalProperties": false,
          "required": ["from", "into"],
          "properties": {
            "from": { "type": "string", "minLength": 1 },
            "into": { "type": "string", "minLength": 1 }
          }
        }
      }
    }
  ],
  "definitions": {
    "filterCondition": {
      "description": "transform.filter supports exactly 8 operators (Issue #1235). prefix / between are NOT supported here (QAL make.filter only). Issue #1308: comparison semantics are shared with tally when / step if — all operands go through $var resolution, eq/neq/in/contains compare string forms, gt/gte/lt/lte coerce both sides with Number().",
      "type": "object",
      "minProperties": 1,
      "additionalProperties": false,
      "properties": {
        "eq": {},
        "neq": {},
        "gt": {},
        "gte": {},
        "lt": {},
        "lte": {},
        "in": {
          "$comment": "Issue #1308: the \"$var\" string form has been implemented since #1235 (resolveForEachIn) and documented in spec 2.10.0, but the schema only accepted a literal array — a manifest written exactly as the spec teaches was rejected. Aligned here because this PR owns the in operator's resolution rule. The wider declaration/implementation sweep is Issue #1307.",
          "oneOf": [
            { "type": "array" },
            { "type": "string", "pattern": "^\\$" }
          ]
        },
        "contains": { "type": "string" }
      }
    }
  }
}