Carbon Arc Bucket Access API

openapi: 3.2.0
info:
  title: Carbon Arc API Documentation Bucket Access API
  description: '# Introduction


    Carbon Arc''s Insights Exchange provides a comprehensive consumption-based platform for accessing decision-making data.'
  version: v2
  x-logo:
    url: /static/logo_light.png
tags:
- name: Bucket Access
paths:
  /v2/bucket-access/datasets/{dataset_id}/grant:
    post:
      tags:
      - Bucket Access
      summary: Grant Bucket Access
      description: 'Grant a principal access to a dataset''s S3 bucket by SKU.


        Resolves (dataset_id, lag, cut) → dataset_name via

        admin_config.bulk_bucket_details and delegates to

        BucketPolicyManager.add_principal_for_sku, which is idempotent.


        S3-only: ``log_access=False`` keeps PyCAKE off admin_config, so this endpoint

        never writes there and read-only PSQL_ADMIN is correct for it (PLA-6141).

        CAMS owns the access/audit write, keyed on the ``dataset_name`` returned here.


        Compliance gate (runs before any S3 write): the client_name (=clients.key)

        must resolve to a clients.id (422 unknown_client otherwise), and the SKU

        must not be blocked for that client by the live served-lag restriction

        model (403 compliance_blocked otherwise).'
      operationId: grant_bucket_access_v2_bucket_access_datasets__dataset_id__grant_post
      parameters:
      - name: dataset_id
        in: path
        required: true
        schema:
          type: string
          pattern: ^CA\d+[A-Z]*$
          title: Dataset Id
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/GrantBucketAccessRequest'
      responses:
        '200':
          description: Successful Response
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/GrantBucketAccessResponse'
        '404':
          description: 'sku_not_found / bucket_not_registered: the SKU has no bucket in admin_config.bulk_bucket_details. Terminal — retrying cannot help.'
        '400':
          description: Malformed principal ARN, or missing required identifier.
        '403':
          description: 'compliance_blocked: the SKU is restricted for this client.'
        '409':
          description: 'policy_apply_failed / policy_write_race: the bucket-policy write did not land. Transient — the caller should retry.'
        '413':
          description: 'policy_size_exceeded: the bucket policy hit AWS''s 20KB limit.'
        '422':
          description: 'unknown_client: client_name is not in admin_config.clients.'
  /v2/bucket-access/datasets/{dataset_id}/revoke:
    post:
      tags:
      - Bucket Access
      summary: Revoke Bucket Access
      description: 'Revoke a principal''s S3 bucket access by SKU.


        Either principal_arn (revoke a specific ARN) or client_name (revoke all

        ARNs registered for that client+dataset) must be provided. Idempotent —

        revoking a principal that isn''t currently granted returns revoked=True

        with no policy change (mirrors the CLI behavior).


        S3-only, like the grant path: ``log_access=False``, and CAMS removes the

        admin_config access row and writes the ACCESS_REVOKED audit row (PLA-6141).'
      operationId: revoke_bucket_access_v2_bucket_access_datasets__dataset_id__revoke_post
      parameters:
      - name: dataset_id
        in: path
        required: true
        schema:
          type: string
          pattern: ^CA\d+[A-Z]*$
          title: Dataset Id
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/RevokeBucketAccessRequest'
      responses:
        '200':
          description: Successful Response
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/RevokeBucketAccessResponse'
        '404':
          description: 'sku_not_found / bucket_not_registered: the SKU has no bucket in admin_config.bulk_bucket_details. Terminal — retrying cannot help.'
        '400':
          description: Malformed principal ARN, or missing required identifier.
        '403':
          description: 'compliance_blocked: the SKU is restricted for this client.'
        '409':
          description: 'policy_apply_failed / policy_write_race: the bucket-policy write did not land. Transient — the caller should retry.'
        '413':
          description: 'policy_size_exceeded: the bucket policy hit AWS''s 20KB limit.'
        '422':
          description: 'unknown_client: client_name is not in admin_config.clients.'
  /v2/bucket-access/datasets/{dataset_id}/verify:
    post:
      tags:
      - Bucket Access
      summary: Verify Bucket Access
      description: 'Read which principals are CURRENTLY on a SKU''s bucket policy.


        Read-only: never mutates the policy. One SKU maps to one bucket, so a single

        call verifies every client on that SKU.


        Callers use this to answer "does this client actually have access right now",

        which no CAMS-side record can answer — the grant ledger and the sample

        settings rows record what we *intended*, and a lost concurrent write makes

        the two disagree.


        Two outcomes are deliberately 200s with flags rather than HTTP errors,

        because the caller must render them as "unknown" and a 4xx/5xx invites a

        silent fallback to intent instead:


        - ``bucket_found=false`` — the SKU or its bucket is not registered.

        - ``read_ok=false`` — the live read failed (throttle, access-denied).

        ``principal_count`` is null and ``present``/``missing`` are empty; do NOT

        read this as "nobody has access".


        Defined as sync ``def`` on purpose: BucketPolicyManager is fully blocking

        (boto3 + a sync Postgres handle), so FastAPI runs this in its threadpool

        rather than stalling the event loop.'
      operationId: verify_bucket_access_v2_bucket_access_datasets__dataset_id__verify_post
      parameters:
      - name: dataset_id
        in: path
        required: true
        schema:
          type: string
          pattern: ^CA\d+[A-Z]*$
          title: Dataset Id
      requestBody:
        required: true
        content:
          application/json:
            schema:
              $ref: '#/components/schemas/VerifyBucketAccessRequest'
      responses:
        '200':
          description: Successful Response
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/VerifyBucketAccessResponse'
        '404':
          detail: Not found
          description: Not Found
        '422':
          description: Validation Error
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/HTTPValidationError'
components:
  schemas:
    GrantBucketAccessResponse:
      properties:
        dataset_id:
          type: string
          title: Dataset Id
        dataset_name:
          type: string
          title: Dataset Name
        lag:
          anyOf:
          - type: string
          - type: 'null'
          title: Lag
        cut:
          anyOf:
          - type: string
          - type: 'null'
          title: Cut
        principal_arn:
          type: string
          title: Principal Arn
        client_name:
          type: string
          title: Client Name
        granted:
          type: boolean
          title: Granted
      type: object
      required:
      - dataset_id
      - dataset_name
      - lag
      - cut
      - principal_arn
      - client_name
      - granted
      title: GrantBucketAccessResponse
    RevokeBucketAccessResponse:
      properties:
        dataset_id:
          type: string
          title: Dataset Id
        dataset_name:
          type: string
          title: Dataset Name
        lag:
          anyOf:
          - type: string
          - type: 'null'
          title: Lag
        cut:
          anyOf:
          - type: string
          - type: 'null'
          title: Cut
        principal_arn:
          anyOf:
          - type: string
          - type: 'null'
          title: Principal Arn
        client_name:
          anyOf:
          - type: string
          - type: 'null'
          title: Client Name
        revoked:
          type: boolean
          title: Revoked
      type: object
      required:
      - dataset_id
      - dataset_name
      - lag
      - cut
      - principal_arn
      - client_name
      - revoked
      title: RevokeBucketAccessResponse
    GrantBucketAccessRequest:
      properties:
        principal_arn:
          type: string
          maxLength: 2048
          minLength: 1
          pattern: ^(arn:aws:iam::\d{12}:[\w+=,.@\-/]+|\d{12})$
          title: Principal Arn
          description: IAM principal ARN to grant access to (e.g. 'arn:aws:iam::123456789012:user/alice'), or a 12-digit AWS account id which is converted to '<id>:root'.
        client_name:
          type: string
          maxLength: 128
          minLength: 1
          pattern: ^[a-zA-Z0-9_-]+$
          title: Client Name
          description: Client name; must exist in admin_config.clients.
        lag:
          anyOf:
          - type: string
          - type: 'null'
          title: Lag
          description: Lag of the dataset SKU to grant access to. Omit (or null) for the no-lag variant — bulk_bucket_details stores no-lag rows with lag=null.
        cut:
          anyOf:
          - type: string
          - type: 'null'
          title: Cut
          description: Cut of the dataset SKU to grant access to. Omit (or null) for the no-cut variant — bulk_bucket_details stores no-cut rows with cut=null.
      type: object
      required:
      - principal_arn
      - client_name
      title: GrantBucketAccessRequest
    HTTPValidationError:
      properties:
        detail:
          items:
            $ref: '#/components/schemas/ValidationError'
          type: array
          title: Detail
      type: object
      title: HTTPValidationError
    VerifyBucketAccessResponse:
      properties:
        dataset_id:
          type: string
          title: Dataset Id
        lag:
          anyOf:
          - type: string
          - type: 'null'
          title: Lag
        cut:
          anyOf:
          - type: string
          - type: 'null'
          title: Cut
        bucket_found:
          type: boolean
          title: Bucket Found
        read_ok:
          type: boolean
          title: Read Ok
        principal_count:
          anyOf:
          - type: integer
          - type: 'null'
          title: Principal Count
          description: Total client-access principals on the live policy. Null when the read failed — NOT zero, which would read as 'nobody has access'.
        present:
          items:
            type: string
          type: array
          title: Present
        missing:
          items:
            type: string
          type: array
          title: Missing
        read_error:
          anyOf:
          - type: string
          - type: 'null'
          title: Read Error
          description: 'Why the read failed, when read_ok is false: invalid_credentials | no_credentials | access_denied | throttled | bucket_missing | unknown. A fixed vocabulary, never raw AWS text, since the caller renders it and AWS messages can carry bucket names and ARNs.'
        checked_at:
          type: string
          format: date-time
          title: Checked At
      type: object
      required:
      - dataset_id
      - lag
      - cut
      - bucket_found
      - read_ok
      - present
      - missing
      - checked_at
      title: VerifyBucketAccessResponse
    VerifyBucketAccessRequest:
      properties:
        lag:
          anyOf:
          - type: string
          - type: 'null'
          title: Lag
          description: Lag of the dataset SKU. Null for the no-lag variant.
        cut:
          anyOf:
          - type: string
          - type: 'null'
          title: Cut
          description: Cut of the dataset SKU. Null for the no-cut variant.
        principal_arns:
          items:
            type: string
          type: array
          maxItems: 500
          title: Principal Arns
          description: ARNs to check for. Each is reported in `present` or `missing`. Pass an empty list to get only `principal_count`.
      type: object
      title: VerifyBucketAccessRequest
    ValidationError:
      properties:
        loc:
          items:
            anyOf:
            - type: string
            - type: integer
          type: array
          title: Location
        msg:
          type: string
          title: Message
        type:
          type: string
          title: Error Type
        input:
          title: Input
        ctx:
          type: object
          title: Context
      type: object
      required:
      - loc
      - msg
      - type
      title: ValidationError
    RevokeBucketAccessRequest:
      properties:
        principal_arn:
          anyOf:
          - type: string
            maxLength: 2048
            pattern: ^(arn:aws:iam::\d{12}:[\w+=,.@\-/]+|\d{12})$
          - type: 'null'
          title: Principal Arn
          description: IAM principal ARN to revoke. Either principal_arn or client_name must be provided.
        client_name:
          anyOf:
          - type: string
            maxLength: 128
            minLength: 1
            pattern: ^[a-zA-Z0-9_-]+$
          - type: 'null'
          title: Client Name
          description: Client name to revoke all ARNs for. Either principal_arn or client_name must be provided.
        lag:
          anyOf:
          - type: string
          - type: 'null'
          title: Lag
          description: Lag of the dataset SKU. Null for the no-lag variant.
        cut:
          anyOf:
          - type: string
          - type: 'null'
          title: Cut
          description: Cut of the dataset SKU. Null for the no-cut variant.
      type: object
      title: RevokeBucketAccessRequest