Declarative source manifest (next)

A source for an HTTP API, as data. Expressions are CEL. Templates are text with {expression} holes; {{ and }} are literal braces. A template that is one hole keeps the hole's value: null leaves a parameter or header out, and a list repeats a parameter.

A manifest is a whole source for an HTTP API, which OpenETL runs itself, with no executable. Apply one with openetl connector apply -f FILE, or name one as a role's manifest in openetl-connector.yaml. openetl connector schema manifest prints its JSON Schema.

This page is generated from internal/declarative/manifest.schema.json. A field that isn't listed is refused, unless its section says other fields are allowed.

Top-level fields

Field	Type	Required	Description
`apiVersion`	string	Yes	Is always `openetl.io/v1alpha1`.
`kind`	string	Yes	Is always `DeclarativeSource`.
`metadata`	object	Yes	Which connector and version this is, and what the console shows about it.
`spec`	object	Yes	The source: its config, how it calls the API, and its streams.

`metadata`

Which connector and version this is, and what the console shows about it.

Field	Type	Required	Description
`name`	string	Yes	The connector's package: owner/name, or name for the official owner, openetl. Each part is lowercase letters, digits and hyphens. Matches `^([a-z0-9]([-a-z0-9]{0,61}[a-z0-9])?/)?[a-z0-9]([-a-z0-9]{0,61}[a-z0-9])?$`.
`version`	string	Yes	Semantic version, without build metadata (`+…`): versions that differ only by it compare as equal. Matches `^(0\|[1-9][0-9])\.(0\|[1-9][0-9])\.(0\|[1-9][0-9]*)(-[0-9A-Za-z.-]+)?$`.
`description`	string	No	One line. Has at most 500 characters.
`license`	string	No	An SPDX expression, such as Apache-2.0. Has at most 100 characters.
`repository`	string	No	An https URL of the source repository. Matches `^https://`.
`docs`	string	No	An https URL. Matches `^https://`.
`tags`	array of string	No	Words the catalog shows with the connector, such as crm. Has at most 20 items. Each item matches `^[a-z0-9][a-z0-9-]{0,31}$`.
`author`	string	No	Who made the connector. Has at most 200 characters.
`logo`	string	No	An https URL of the connector's logo, which the UI shows. Matches `^https://`.
`release`	object	No	The git release this version comes from.

`metadata.release`

The git release this version comes from.

Field	Type	Required	Description
`tag`	string	No	The git tag. Has at most 255 characters.
`url`	string	No	An https URL, such as the release's page. Matches `^https://`.

`spec`

The source: its config, how it calls the API, and its streams.

Field	Type	Required	Description
`resources`	object	No	What the source asks of its jobs.
`config`	object	Yes	The JSON Schema of the connector's config: an object. Top-level string properties with x-secret: true are secrets, which only auth and sign can reference, by name. Expressions see the other properties as config.NAME.
`http`	object	Yes	How the source calls the API: its base URL, headers, auth, signing, rate limit and error rules.
`check`	object	No	What Check does: fetch the first page of a stream. Default: the first stream without a parent.
`streams`	array of object	Yes	The streams the source reads. Has at least one item.

`spec.resources`

What the source asks of its jobs.

Field	Type	Required	Description
`hint`	string	No	The job size a pipeline of this source starts at when it's sized automatically. Is one of `xs`, `s`, `m`, `l` or `xl`.

`spec.config`

The JSON Schema of the connector's config: an object. Top-level string properties with x-secret: true are secrets, which only auth and sign can reference, by name. Expressions see the other properties as config.NAME.

Field	Type	Required	Description
`type`	string	Yes	Is always `object`.
`properties`	object	No	The config's properties, by name, each a JSON Schema. Each name matches `^[A-Za-z_][A-Za-z0-9_]*$`.

Other fields are allowed.

`spec.http`

How the source calls the API: its base URL, headers, auth, signing, rate limit and error rules.

Field	Type	Required	Description
`base_url`	`template`	Yes	A template over config. Request paths are appended to it, and next-page URLs must have its scheme, host and port.
`headers`	`templates`	No	Headers for every request: templates, as in a stream's request.
`auth`	object	No	How requests authenticate. The host adds the credential when a request leaves, and only to the base URL's scheme, host and port.
`sign`	object	No	Signs every request with a key the host holds. hmac puts a MAC of message in a header or a query parameter; jwt sends a JWT as the Authorization header.
`rate_limit`	object	No	How fast the source sends requests.
`errors`	`errors`	No	Error rules for every stream, after the stream's own.

`spec.http.auth`

How requests authenticate. The host adds the credential when a request leaves, and only to the base URL's scheme, host and port.

Field	Type	Required	Description
`type`	string	Yes	How requests authenticate. Is one of `bearer`, `basic`, `api_key`, `oauth2` or `aws_sigv4`.

`spec.http.auth` when `type` is `bearer`

Field	Type	Required	Description
`token`	`secret`	Yes	The bearer token.

`spec.http.auth` when `type` is `basic`

Field	Type	Required	Description
`username`	`template`	Yes	A template over config.
`password`	`secret`	Yes	The password.

`spec.http.auth` when `type` is `api_key`

Field	Type	Required	Description
`key`	`secret`	Yes	The API key.
`header`	string	No	The header the key goes in, such as X-API-Key. Isn't empty.
`query`	string	No	The query parameter the key goes in. Isn't empty.
`prefix`	string	No	Put before the key, such as "Token ".

Has exactly one of header or query.

`spec.http.auth` when `type` is `oauth2`

Field	Type	Required	Description
`token_url`	`template`	Yes	A template over config.
`client_id`	`templateOrSecret`	Yes	The client's ID.
`client_secret`	`secret`	Yes	The client's secret.
`refresh_token`	`secret`	No	A refresh token, for the refresh token grant. Without one, the host uses the client credentials grant.
`scopes`	array of string	No	The scopes to ask for.
`auth_style`	string	No	Where the client credentials go in token requests. Default: the host tries both. Is one of `header` or `params`.

`spec.http.auth` when `type` is `aws_sigv4`

Field	Type	Required	Description
`access_key_id`	`templateOrSecret`	Yes	The AWS access key ID.
`secret_access_key`	`secret`	Yes	The AWS secret access key.
`session_token`	`secret`	No	An AWS session token, for temporary credentials.
`region`	`template`	Yes	A template over config.
`service`	string	Yes	The AWS service's signing name, such as execute-api. Isn't empty.

`spec.http.sign`

Signs every request with a key the host holds. hmac puts a MAC of message in a header or a query parameter; jwt sends a JWT as the Authorization header.

Field	Type	Required	Description
`type`	string	Yes	How requests are signed. Is one of `hmac` or `jwt`.
`key`	`secret`	Yes	The signing key.

`spec.http.sign` when `type` is `hmac`

Field	Type	Required	Description
`hash`	string	Yes	The HMAC's hash function. Is one of `sha1`, `sha256` or `sha512`.
`message`	`expression`	Yes	An expression over request (method, url, path, query, headers, body), config and now, returning a string or bytes.
`encoding`	string	No	Default: hex. Is one of `hex` or `base64`.
`header`	string	No	The header the MAC goes in. Isn't empty.
`query`	string	No	The query parameter the MAC goes in. Isn't empty.
`prefix`	string	No	Put before the MAC.

Has exactly one of header or query.

`spec.http.sign` when `type` is `jwt`

Field	Type	Required	Description
`algorithm`	string	Yes	The JWT's signing algorithm. Is one of `HS256`, `RS256` or `ES256`.
`claims`	`expression`	Yes	An expression over config and now, returning a map.
`jose_header`	`expression`	No	An expression returning extra header fields, such as kid. The host sets alg and typ.
`prefix`	string	No	Default: "Bearer ".

`spec.http.rate_limit`

How fast the source sends requests.

Field	Type	Required	Description
`rps`	number	Yes	Requests per second at most. The rate adapts below it when the API throttles. Is more than 0.

`spec.check`

What Check does: fetch the first page of a stream. Default: the first stream without a parent.

Field	Type	Required	Description
`stream`	string	Yes	The stream whose first page Check fetches.

`spec.streams[]`

A stream the source reads: how to request it, page through it, and find its records.

Field	Type	Required	Description
`name`	string	Yes	The stream's name. Isn't empty.
`primary_key`	array of string	No	The top-level properties that make up a record's primary key.
`request`	object	Yes	The request for each page. Templates over config, state, page, parent, partition and now.
`records`	`expression`	No	An expression over response, returning the page's records, each an object. Default: response.body.
`paginator`	object	No	How to get the next page. A resume point follows every page. Without a paginator, a stream is one page.
`incremental`	object	No	Read from a cursor floor onwards. Requests see the floor, less the lookback, as state.cursor, which is absent on the first sync. The API must return every record whose cursor is at or after it.
`parent`	object	No	Read the stream once per record of another stream, which the request sees as parent. The parent must have a primary key and no parent of its own.
`partitions`	object	No	Read the stream in splits, concurrently where the job has room, each with its own place and floor. Requests see each one as partition. Not for a stream with a parent.
`errors`	`errors`	No	Error rules for this stream, tried before the API's.
`schema`	object	No	JSON Schema for some or all of the record's top-level properties, which declares their types. The others are inferred.

`spec.streams[].request`

The request for each page. Templates over config, state, page, parent, partition and now.

Field	Type	Required	Description
`method`	string	No	Default: GET. Is one of `GET` or `POST`.
`path`	`template`	Yes	Appended to the base URL. Holes are escaped as path segments.
`params`	`templates`	No	Query parameters.
`headers`	`templates`	No	Headers, besides the API's.
`body`	`expression`	No	An expression whose value is sent as JSON.
`idempotent`	boolean	No	Lets the host retry a POST, such as a search that changes nothing.

`spec.streams[].paginator`

How to get the next page. A resume point follows every page. Without a paginator, a stream is one page.

Field	Type	Required	Description
`type`	string	Yes	How the next page is found. Is one of `cursor`, `next_url`, `link`, `offset` or `page`.
`stop`	`expression`	No	An expression over response and records; true ends pagination.

`spec.streams[].paginator` when `type` is `cursor`

Field	Type	Required	Description
`next`	`expression`	Yes	An expression over response and records: the next page's token, or null or an empty string at the end. Requests see it as page.token.
`param`	string	No	Also send the token as this query parameter. Isn't empty.

`spec.streams[].paginator` when `type` is `next_url`

Field	Type	Required	Description
`next`	`expression`	Yes	An expression over response and records: the next page's URL, absolute or relative to the base URL, or null or an empty string at the end. It's fetched as it is, with no params added.

`spec.streams[].paginator` when `type` is `link`

The next page is the Link header's rel="next" URL, fetched as it is.

`spec.streams[].paginator` when `type` is `offset`

Field	Type	Required	Description
`param`	string	No	Default: offset. Isn't empty.
`start`	integer	No	Default: 0. Is at least 0.
`size`	integer	No	Records in a full page. A shorter page is the last. Is at least 1.
`size_param`	string	No	Send size as this query parameter. Isn't empty.

`spec.streams[].paginator` when `type` is `page`

Field	Type	Required	Description
`param`	string	No	Default: page. Isn't empty.
`start`	integer	No	Default: 1. Is at least 0.
`size`	integer	No	Records in a full page. A shorter page is the last. Is at least 1.
`size_param`	string	No	Send size as this query parameter. Isn't empty.

`spec.streams[].incremental`

Read from a cursor floor onwards. Requests see the floor, less the lookback, as state.cursor, which is absent on the first sync. The API must return every record whose cursor is at or after it.

Field	Type	Required	Description
`cursor_field`	string	Yes	A top-level property of the records. Isn't empty.
`type`	string	Yes	`datetime`: RFC 3339 text, and `state.cursor` is a timestamp. `unix`: seconds since 1970, and `state.cursor` is an int. `integer` and `string`: compared as they are. Is one of `datetime`, `unix`, `integer` or `string`.
`lookback`	string	No	Read again this far behind the floor, for late updates. datetime and unix only. A duration such as 90s, 5m or 1h30m. Matches `^([0-9]+(\.[0-9]+)?(ns\|us\|µs\|ms\|s\|m\|h))+$`.
`ordered`	boolean	No	The API returns records sorted by the cursor, oldest first, so the floor advances after every page.

`spec.streams[].parent`

Read the stream once per record of another stream, which the request sees as parent. The parent must have a primary key and no parent of its own.

Field	Type	Required	Description
`stream`	string	Yes	The parent stream's name.

`spec.streams[].partitions`

Read the stream in splits, concurrently where the job has room, each with its own place and floor. Requests see each one as partition. Not for a stream with a parent.

Field	Type	Required	Description
`values`	`expression`	No	An expression over config and now, returning a list: a partition per entry.
`range`	object	No	Partitions of {start, end}, in RFC 3339, a step at a time.

`spec.streams[].partitions.range`

Partitions of {start, end}, in RFC 3339, a step at a time.

Field	Type	Required	Description
`start`	`expression`	Yes	An expression over config and now, returning a timestamp.
`end`	`expression`	No	As start. Default: now.
`step`	string	Yes	A duration, such as 24h.

`spec.streams[].schema`

JSON Schema for some or all of the record's top-level properties, which declares their types. The others are inferred.

Field	Type	Required	Description
`properties`	map of names to object	No	The properties' JSON Schemas, by name.

Other fields are allowed.

Field	Type	Required	Description
`secret`	string	Yes	The name of a config property marked x-secret. Isn't empty.

Field	Type	Required	Description
`when`	`expression`	Yes	An expression over response.
`action`	string	Yes	`fail`: fail with `code` and `message`. `ignore`: treat the response as an empty last page. `expired`: the page token a resumed read started from was rejected, so the stream starts over from its floor. Is one of `fail`, `ignore` or `expired`.
`code`	string	No	For fail. Default: how the host classified the response. Is one of `TRANSIENT`, `RATE_LIMITED`, `AUTH_EXPIRED`, `AUTH_INVALID`, `PERMISSION_DENIED`, `CONFIG_INVALID`, `CONTRACT_CHANGED` or `RESOURCE_EXHAUSTED`.
`message`	`template`	No	For fail: a template over response.

Declarative source manifest

Top-level fields

`metadata`

`metadata.release`

`spec`

`spec.resources`

`spec.config`

`spec.http`

`spec.http.auth`

`spec.http.auth` when `type` is `bearer`

`spec.http.auth` when `type` is `basic`

`spec.http.auth` when `type` is `api_key`

`spec.http.auth` when `type` is `oauth2`

`spec.http.auth` when `type` is `aws_sigv4`

`spec.http.sign`

`spec.http.sign` when `type` is `hmac`

`spec.http.sign` when `type` is `jwt`

`spec.http.rate_limit`

`spec.check`

`spec.streams[]`

`spec.streams[].request`

`spec.streams[].paginator`

`spec.streams[].paginator` when `type` is `cursor`

`spec.streams[].paginator` when `type` is `next_url`

`spec.streams[].paginator` when `type` is `link`

`spec.streams[].paginator` when `type` is `offset`

`spec.streams[].paginator` when `type` is `page`

`spec.streams[].incremental`

`spec.streams[].parent`

`spec.streams[].partitions`

`spec.streams[].partitions.range`

`spec.streams[].schema`

Types

`template`

`templates`

`secret`

`templateOrSecret`

`expression`

`errors`

On this page

Declarative source manifest

Top-level fieldsCopy Anchor Link

metadataCopy Anchor Link

metadata.releaseCopy Anchor Link

specCopy Anchor Link

spec.resourcesCopy Anchor Link

spec.configCopy Anchor Link

spec.httpCopy Anchor Link

spec.http.authCopy Anchor Link

spec.http.auth when type is bearerCopy Anchor Link

spec.http.auth when type is basicCopy Anchor Link

spec.http.auth when type is api_keyCopy Anchor Link

spec.http.auth when type is oauth2Copy Anchor Link

spec.http.auth when type is aws_sigv4Copy Anchor Link

spec.http.signCopy Anchor Link

spec.http.sign when type is hmacCopy Anchor Link

spec.http.sign when type is jwtCopy Anchor Link

spec.http.rate_limitCopy Anchor Link

spec.checkCopy Anchor Link

spec.streams[]Copy Anchor Link

spec.streams[].requestCopy Anchor Link

spec.streams[].paginatorCopy Anchor Link

spec.streams[].paginator when type is cursorCopy Anchor Link

spec.streams[].paginator when type is next_urlCopy Anchor Link

spec.streams[].paginator when type is linkCopy Anchor Link

spec.streams[].paginator when type is offsetCopy Anchor Link

spec.streams[].paginator when type is pageCopy Anchor Link

spec.streams[].incrementalCopy Anchor Link

spec.streams[].parentCopy Anchor Link

spec.streams[].partitionsCopy Anchor Link

spec.streams[].partitions.rangeCopy Anchor Link

spec.streams[].schemaCopy Anchor Link

TypesCopy Anchor Link

templateCopy Anchor Link

templatesCopy Anchor Link

secretCopy Anchor Link

templateOrSecretCopy Anchor Link

expressionCopy Anchor Link

errorsCopy Anchor Link

On this page

Top-level fields

`metadata`

`metadata.release`

`spec`

`spec.resources`

`spec.config`

`spec.http`

`spec.http.auth`

`spec.http.auth` when `type` is `bearer`

`spec.http.auth` when `type` is `basic`

`spec.http.auth` when `type` is `api_key`

`spec.http.auth` when `type` is `oauth2`

`spec.http.auth` when `type` is `aws_sigv4`

`spec.http.sign`

`spec.http.sign` when `type` is `hmac`

`spec.http.sign` when `type` is `jwt`

`spec.http.rate_limit`

`spec.check`

`spec.streams[]`

`spec.streams[].request`

`spec.streams[].paginator`

`spec.streams[].paginator` when `type` is `cursor`

`spec.streams[].paginator` when `type` is `next_url`

`spec.streams[].paginator` when `type` is `link`

`spec.streams[].paginator` when `type` is `offset`

`spec.streams[].paginator` when `type` is `page`

`spec.streams[].incremental`

`spec.streams[].parent`

`spec.streams[].partitions`

`spec.streams[].partitions.range`

`spec.streams[].schema`

Types

`template`

`templates`

`secret`

`templateOrSecret`

`expression`

`errors`