API reference¶

The blocking service classes, their async twins in water_ogcapi.aio, and the types both return.

Services¶

Bases: OGCService

Blocking client for the USGS Water Data OGC API, as water_ogcapi.aio.NWIS.

Parameters:

Name	Type	Description	Default
`api_key`	`str`	A USGS API key, as for `water_ogcapi.aio.NWIS`.	`None`
`transport`	`Transport`	Sends every request, as for `OGCService`.	`None`

Raises:

Type	Description
`ValidationError`	If the key holds a character other than printable ASCII.

Bases: OGCService

Blocking client for the USGS FabricData service, as water_ogcapi.aio.FabricData.

Parameters:

Name	Type	Description	Default
`transport`	`Transport`	Sends every request, as for `OGCService`.	`None`

Bases: OGCService

Blocking client for the GeoConnex reference features, as water_ogcapi.aio.GeoConnex.

Parameters:

Name	Type	Description	Default
`transport`	`Transport`	Sends every request, as for `OGCService`.	`None`

Blocking client for one OGC API Features service.

Runs a water_ogcapi.aio.OGCService of its own on the background loop. Its methods collections, queryables, schema, iter_pages, collect, and item block on the async methods of the same name. A Transport passed here binds to that loop, so share it only among sync services (D-10); close() closes it, and a service sharing it reconnects on its next request.

Parameters:

Name	Type	Description	Default
`base_url`	`str`	The service's landing page, as for `water_ogcapi.aio.OGCService`.	required
`transport`	`Transport`	Sends every request, as for `water_ogcapi.aio.OGCService`.	`None`

`base_url` `property` ¶

The landing page URL, without a trailing slash.

`collections()` ¶

Return the collection index from /collections.

The index is cached for an hour, then revalidated (D-07); refresh drops it.

Raises:

Type	Description
`ServiceError`	If the request fails or the body is not a JSON object with a `collections` list. Nothing is cached, so the next call asks again.
`ValueError`	If an entry of `collections` is not an object with a string `id`.

`queryables(collection)` ¶

Return the properties a query on collection can filter on.

The document is cached for an hour, then revalidated (D-07); refresh drops it.

Parameters:

Name	Type	Description	Default
`collection`	`str`	The collection id.	required

Raises:

Type	Description
`ServiceError`	If the request fails, with status 404 for an unknown collection, or the body is not a JSON object with a `properties` object.
`ValidationError`	Before any request, if `collection` is empty, `.`, or `..`.

`schema(collection)` ¶

Return the properties each feature of collection carries.

The document is cached for an hour, then revalidated (D-07); refresh drops it.

Parameters:

Name	Type	Description	Default
`collection`	`str`	The collection id.	required

Raises:

Type	Description
`ServiceError`	If the request fails, with status 404 for an unknown collection, or the body is not a JSON object with a `properties` object.
`ValidationError`	Before any request, if `collection` is empty, `.`, or `..`.

`iter_pages(collection, *, params=None, bbox=None, geometry=None, filter=None, ids=None, id_property=None, split_datetime=None, sites_per_piece=None, validate=True, resume=None)` ¶

Yield each page of a collection's items, following next links (D-34).

Only a page without a next link ends the traversal. A page shorter than requested, or one without numberMatched, ends nothing. A split query runs its pieces concurrently and yields pages as they arrive (D-56).

Parameters:

Name	Type	Description	Default
`collection`	`str`	The collection id.	required
`params`	`mapping of str to str`	Query parameters of the first request, such as `limit` or `datetime`, with `f=json` and the profile's `default_limit` added unless given. Later requests take their query from the `next` link.	`None`
`bbox`	`sequence of float`	`(west, south, east, north)` in CRS84, or six values with the minimum and maximum heights, sent as the `bbox` parameter with every digit kept. West above east crosses the antimeridian.	`None`
`geometry`	`mapping or GeoInterface`	A GeoJSON geometry, or an object with `__geo_interface__` such as a shapely polygon, in CRS84. Its bbox is sent in place of the geometry (D-11), so the result holds every feature intersecting that bbox; exact containment is the caller's post-filter.	`None`
`filter`	`str or mapping`	A CQL2 filter, spatial predicates such as `s_intersects` included. A `str` is CQL2-Text, sent as the `filter` parameter of a `GET`. A mapping is CQL2-JSON, sent as the body of a `POST` with `Content-Type: application/query-cql-json`; every `next` link gets the same body, since a `GET` there drops the filter (D-38).	`None`
`ids`	`sequence of str or int`	Values of `id_property` to match, sent as the CQL2-JSON filter `{"op": "in", "args": [{"property": id_property}, ids]}`, as `filter` sends a mapping. Each value goes out with its type, so `"01646500"` keeps its leading zero.	`None`
`id_property`	`str`	The property `ids` matches, such as `id` for feature ids or `monitoring_location_id` for the sites of an NWIS time series.	`None`
`split_datetime`	`('year', 'month')`	NWIS only (D-35, D-56): split the `datetime` interval in `params` at calendar boundaries in UTC into pieces that run concurrently, up to the transport's `max_concurrent` at once, each following its own `next` links. The interval needs both ends, as dates or timestamps with an offset, and each piece ends one microsecond before the next begins. Every piece costs at least one request, so size pieces near one page.	`'year'`
`sites_per_piece`	`int`	NWIS only (D-35, D-56): split the site list, `ids` or a comma-separated `monitoring_location_id` in `params`, into lists of this many sites that run concurrently as `split_datetime` pieces do. A site listed twice is sent once. Not with `split_datetime`.	`None`
`validate`	`bool`	Before the first page request, check `collection` against the collection index, and each name in `params` outside `QUERY_PARAMS`, and `id_property`, against the collection's queryables, from metadata cached for an hour (D-07). pygeoapi drops a name it does not know, so an unchecked typo returns the whole collection. Defaults to `True`; `False` skips the check, for a server that accepts a name its queryables leave out.	`True`
`resume`	`Checkpoint`	Continue a query from a `Checkpoint`: a page's `checkpoint` or a `PaginationError`'s (D-52). Its pending requests are sent as stored, so no query keyword may come with it, `validate` is ignored, and no request it holds is sent again.	`None`

Yields:

Type	Description
`Page`	One per response, the decoded document in `body` and its bytes in `content`. Its `checkpoint` is where the query stands once the caller holds it.

Raises:

Type	Description
`ValidationError`	Before any request, when `collection` is empty, `.`, or `..`, both `bbox` and `geometry`, or `ids` and `filter`, are given, `ids` is a `str` or `bytes`, empty, or lacks `id_property`, `bbox` or `geometry` is malformed, a CQL2-JSON filter holds a NaN or infinite number, a filter is empty, `params` also holds a parameter a keyword sets (`bbox`, `filter`, `filter-lang`) or a credential such as `api_key`, or `limit` is not an integer from 1 to the profile's `max_limit`, or a split is given to a service other than NWIS, both splits are given, or the query has no closed `datetime` interval or site list to split. With `validate`, before the first page request, when the collection, a name in `params`, or `id_property` is unknown; the message lists the valid ones. With `resume`, when a query keyword comes with it or it is a checkpoint of another service or collection.
`TypeError`	Before any request, when a CQL2-JSON filter holds a value JSON cannot encode, such as a set.
`ServiceError`	With `validate`, when a metadata request fails.
`PaginationError`	When a request fails, a body is not a FeatureCollection with well-formed links, or a `next` link repeats a request already answered, leaves the scheme, host, and port of `base_url`, or a page has a link that carries a credential (D-53). It is raised once every page that arrived has been yielded; its failures name each such request or link by URL, and its `checkpoint` is where the query stands.

`collect(collection, *, params=None, bbox=None, geometry=None, filter=None, ids=None, id_property=None, split_datetime=None, sites_per_piece=None, validate=True, keep_pages=False, resume=None)` ¶

Fetch every page of a collection's items into one QueryResult (D-22).

Parameters:

Name	Type	Description	Default
`collection`	`str`	The collection id.	required
`params`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`bbox`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`geometry`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`filter`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`ids`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`id_property`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`split_datetime`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`sites_per_piece`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`validate`	`bool`	Check the query against cached metadata first, as for `iter_pages`.	`True`
`keep_pages`	`bool`	Keep each page's response bytes in `Page.content` (D-23). Page bodies are always dropped. With `resume`, it applies to the new pages; held pages keep what the earlier call kept.	`False`
`resume`	`QueryResult or Checkpoint`	Continue a query that stopped on a failure (D-52): the `result` of a `PartialResultsError`, or the `checkpoint` of a `PaginationError` raised before any page arrived. The result's pages and features come first in the new result, and no request for them is sent. No query keyword may come with it.	`None`

Returns:

Type	Description
`QueryResult`	`geojson` holds the features of all pages in the order they arrived, those of `resume` first.

Raises:

Type	Description
`ValidationError`	On a malformed query or, with `validate`, an unknown name, as for `iter_pages`. With `resume`, before any request, when a query keyword comes with it, it is from another service or collection, a `QueryResult` has no `checkpoint`, or a `Checkpoint` holds pages, which only the `QueryResult` that holds them can carry.
`TypeError`	Before any request, when a CQL2-JSON filter holds a value JSON cannot encode, as for `iter_pages`.
`ServiceError`	With `validate`, when a metadata request fails.
`PartialResultsError`	When a page fails after at least one succeeded, `resume`'s included. Its `result` holds the pages that succeeded and the checkpoint to resume.
`PaginationError`	When no page arrives, as when the first page of an unsplit query fails.

`item(collection, item_id, *, params=None, validate=True)` ¶

Fetch one feature of a collection by its id (D-22).

Parameters:

Name	Type	Description	Default
`collection`	`str`	The collection id.	required
`item_id`	`str or int`	The feature id, percent-encoded into the URL path.	required
`params`	`mapping of str to str`	Query parameters, such as `crs` or `skipGeometry`, with `f=json` added unless given.	`None`
`validate`	`bool`	Before the request, check `collection` against the cached collection index, as `iter_pages` does. Defaults to `True`.	`True`

Returns:

Type	Description
`Page`	`body` holds the decoded Feature and `content` its bytes. `feature_count` is 1, and `number_matched` and `number_returned` are `None`.

Raises:

Type	Description
`ServiceError`	If the request fails, with status 404 for an unknown id, or the body is not a Feature whose `links` member, when present, is a list. With `validate`, also when the metadata request fails.
`ValidationError`	Before any request, if `collection` or `item_id` is empty, `.`, or `..`, which would address another path, or `params` carries a credential such as `api_key`. With `validate`, if the collection is unknown.

`refresh()` ¶

Drop every cached metadata document, so the next call fetches it again.

`close()` ¶

Close the transport, whether this service created it or it was passed.

What one service accepts and sends, read by its service class (D-13, D-41).

Parameters:

Name	Type	Description	Default
`base_url`	`str`	The landing page.	required
`default_limit`	`int`	The `limit` an items query sends when its `params` hold none.	required
`max_limit`	`int`	The largest `limit` the service accepts.	required
`key_env`	`str or None`	The environment variable that holds the service's API key, sent in the `X-Api-Key` header (D-18). `None`, the default, for a service sent no key.	`None`
`rate_limit_window`	`float or None`	Seconds over which the service's `X-RateLimit-Limit` counts, read by its `Governor` (D-49). `None`, the default, for a service that sends no `X-RateLimit-*` headers.	`None`

Bounded, retrying async HTTP client.

The client and its semaphore are created on the first request, so construction performs no I/O. Both then belong to that event loop for the life of the instance: a later request from a different loop raises rather than failing later under contention. Use one transport per loop, or aclose() between loops. Retries cover transport errors and the status codes in RETRY_STATUS.

Parameters:

Name	Type	Description	Default
`timeout`	`float`	Seconds each attempt may take, from sending the request to reading the last body byte, defaults to 30. httpx2 also applies it to each connect, read, write, and pool wait. A timed-out attempt is retried. Waiting for a concurrency permit is not bounded.	`30.0`
`max_concurrent`	`int`	Requests in flight at once, defaults to 4. An explicit semaphore, not a connection-pool limit: over HTTP/2 a single connection multiplexes many streams, so pool size does not bound concurrency. Held only while a request is in flight, never across a backoff wait.	`4`
`max_retries`	`int`	Retries after the first attempt, defaults to 3.	`3`
`backoff`	`float`	Base of the exponential backoff in seconds, defaults to 0.5.	`0.5`
`http2`	`bool`	Negotiate HTTP/2, defaults to `True`.	`True`
`transport`	`AsyncBaseTransport`	Injected transport, for tests.	`None`

Raises:

Type	Description
`ValidationError`	If `max_concurrent` is below 1, or `max_retries` or `backoff` is negative. A zero-permit semaphore would block every request forever.

`timeout` `property` ¶

Seconds each attempt may take.

`max_concurrent` `property` ¶

Requests in flight at once.

`max_retries` `property` ¶

Retries after the first attempt.

`backoff` `property` ¶

Base of the exponential backoff in seconds.

`http2` `property` ¶

Whether HTTP/2 is negotiated.

`get(url, *, params=None, headers=None, governor=None)` `async` ¶

Send a GET through request, which documents the parameters and errors.

`request(method, url, *, params=None, headers=None, content=None, governor=None)` `async` ¶

Send a request, retrying transient failures.

A POST is retried like a GET: this package posts only queries, which change nothing on the server (docs/design.md D-38).

Parameters:

Name	Type	Description	Default
`method`	`str`	HTTP method, such as `GET` or `POST`.	required
`url`	`str`	Absolute URL to fetch.	required
`params`	`mapping of str to str`	Query parameters, merged into any query already in `url`. A name in both takes its value from `params`. Credentials are not allowed here.	`None`
`headers`	`mapping of str to str`	Request headers, credentials included. Each value must be printable ASCII.	`None`
`content`	`bytes`	Request body, sent unchanged on every attempt.	`None`
`governor`	`Governor`	Paces every attempt, retries included, and reads the rate-limit headers of every response (D-06). Each service class passes its own.	`None`

Returns:

Type	Description
`Response`	The first 2xx response. A 304 is returned, not raised, so conditional revalidation works.

Raises:

Type	Description
`RateLimitError`	On a final 429, or on a 429 whose `Retry-After` exceeds `MAX_SLEEP`. Carries the parsed wait, the remaining count, and the attempt count. Every error carries the method and body of the request (D-38).
`ServiceError`	On any other status of 400 or above, on a redirect, on a malformed URL or undecodable body, when retries are exhausted, or when `aclose()` ran while the call waited for a permit or a backoff.
`RuntimeError`	If called from an event loop other than the one this transport bound to.
`ValidationError`	If `params` carries a credential such as `api_key`, or a header value is not printable ASCII. The message names the header, never its value.

`aclose()` `async` ¶

Close the underlying client and release the loop binding.

Call this from the loop that used the transport, before that loop exits. Closing pooled connections from a different loop is what fails, and the binding is released either way so a failed close cannot strand the instance.

Spaces the starts of one service's requests so its quota lasts (D-06).

Transport.request calls wait before each attempt and observe on each response, retries included. Each X-RateLimit-Limit value keeps its own reading, since a default NWIS key counts /items and metadata separately, and the widest interval of the readings from the last window seconds applies; with none, the static interval does. With remaining of limit left, the interval is window / remaining - 8 * window / limit seconds, from 0 at an eighth of the quota to at most 60.

Parameters:

Name	Type	Description	Default
`window`	`float or None`	Seconds over which the service's `X-RateLimit-Limit` counts. `None` ignores rate-limit headers, for a service that sends none.	required
`interval`	`float`	Seconds between request starts while no reading from the last `window` seconds applies, defaults to 1. The first request never waits.	`STATIC_INTERVAL`

Raises:

Type	Description
`ValidationError`	If `window` is not positive or `interval` is negative.

`window` `property` ¶

Seconds over which the service's X-RateLimit-Limit counts.

`interval` `property` ¶

Seconds between request starts while no recent reading applies.

`wait()` `async` ¶

Wait until the request may start: one interval after the previous start.

A reading or a 429 that arrives during the wait applies to it, and a cancelled wait books nothing.

`observe(status, headers)` ¶

Read the counts on one response, and a 429's Retry-After.

A response a CDN served, or one without both counts, leaves the readings as they were. A 429's Retry-After holds back every later start for that long, up to 60 seconds.

Parameters:

Name	Type	Description	Default
`status`	`int`	HTTP status code.	required
`headers`	`mapping of str to str`	Response headers with lowercase keys.	required

Async API¶

The async service classes, for code that runs its own event loop (D-48).

Each class has the name and constructor of its blocking twin in water_ogcapi, and each method is a coroutine, apart from iter_pages, which returns an async iterator. A Transport binds to the event loop of its first request (D-10).

import asyncio

from water_ogcapi.aio import NWIS


async def main() -> None:
    async with NWIS() as nwis:
        result = await nwis.collect("daily", params={"monitoring_location_id": "USGS-01646500"})


asyncio.run(main())

`OGCService` ¶

Async client for one OGC API Features service.

Construction performs no HTTP (docs/design.md D-10). The first request binds the transport to the running event loop.

Parameters:

Name	Type	Description	Default
`base_url`	`str`	The service's landing page, such as `https://api.waterdata.usgs.gov/ogcapi/v1`. A trailing slash is ignored. Defaults to the profile's `base_url`; required where `profile` is `None`.	`None`
`transport`	`Transport`	Sends every request. The service closes a transport it created and leaves a passed one open, so several services can share one.	`None`

Attributes:

Name	Type	Description
`profile`	`ServiceProfile or None`	The service's page sizes, key convention, and rate-limit window, set by each service class. `None` here: a query sends the server's default `limit`, any `limit` goes unchecked, and requests start 1 s apart (D-49).

`base_url` `property` ¶

The landing page URL, without a trailing slash.

`collections()` `async` ¶

Return the collection index from /collections.

The index is cached for an hour, then revalidated (D-07); refresh drops it.

Raises:

Type	Description
`ServiceError`	If the request fails or the body is not a JSON object with a `collections` list. Nothing is cached, so the next call asks again.
`ValueError`	If an entry of `collections` is not an object with a string `id`.

`queryables(collection)` `async` ¶

Return the properties a query on collection can filter on.

The document is cached for an hour, then revalidated (D-07); refresh drops it.

Parameters:

Name	Type	Description	Default
`collection`	`str`	The collection id.	required

Raises:

Type	Description
`ServiceError`	If the request fails, with status 404 for an unknown collection, or the body is not a JSON object with a `properties` object.
`ValidationError`	Before any request, if `collection` is empty, `.`, or `..`.

`schema(collection)` `async` ¶

Return the properties each feature of collection carries.

The document is cached for an hour, then revalidated (D-07); refresh drops it.

Parameters:

Name	Type	Description	Default
`collection`	`str`	The collection id.	required

Raises:

Type	Description
`ServiceError`	If the request fails, with status 404 for an unknown collection, or the body is not a JSON object with a `properties` object.
`ValidationError`	Before any request, if `collection` is empty, `.`, or `..`.

`refresh()` ¶

Drop every cached metadata document, so the next call fetches it again.

`iter_pages(collection, *, params=None, bbox=None, geometry=None, filter=None, ids=None, id_property=None, split_datetime=None, sites_per_piece=None, validate=True, resume=None)` `async` ¶

Yield each page of a collection's items, following next links (D-34).

Only a page without a next link ends the traversal. A page shorter than requested, or one without numberMatched, ends nothing. A split query runs its pieces concurrently and yields pages as they arrive (D-56).

Parameters:

Name	Type	Description	Default
`collection`	`str`	The collection id.	required
`params`	`mapping of str to str`	Query parameters of the first request, such as `limit` or `datetime`, with `f=json` and the profile's `default_limit` added unless given. Later requests take their query from the `next` link.	`None`
`bbox`	`sequence of float`	`(west, south, east, north)` in CRS84, or six values with the minimum and maximum heights, sent as the `bbox` parameter with every digit kept. West above east crosses the antimeridian.	`None`
`geometry`	`mapping or GeoInterface`	A GeoJSON geometry, or an object with `__geo_interface__` such as a shapely polygon, in CRS84. Its bbox is sent in place of the geometry (D-11), so the result holds every feature intersecting that bbox; exact containment is the caller's post-filter.	`None`
`filter`	`str or mapping`	A CQL2 filter, spatial predicates such as `s_intersects` included. A `str` is CQL2-Text, sent as the `filter` parameter of a `GET`. A mapping is CQL2-JSON, sent as the body of a `POST` with `Content-Type: application/query-cql-json`; every `next` link gets the same body, since a `GET` there drops the filter (D-38).	`None`
`ids`	`sequence of str or int`	Values of `id_property` to match, sent as the CQL2-JSON filter `{"op": "in", "args": [{"property": id_property}, ids]}`, as `filter` sends a mapping. Each value goes out with its type, so `"01646500"` keeps its leading zero.	`None`
`id_property`	`str`	The property `ids` matches, such as `id` for feature ids or `monitoring_location_id` for the sites of an NWIS time series.	`None`
`split_datetime`	`('year', 'month')`	NWIS only (D-35, D-56): split the `datetime` interval in `params` at calendar boundaries in UTC into pieces that run concurrently, up to the transport's `max_concurrent` at once, each following its own `next` links. The interval needs both ends, as dates or timestamps with an offset, and each piece ends one microsecond before the next begins. Every piece costs at least one request, so size pieces near one page.	`'year'`
`sites_per_piece`	`int`	NWIS only (D-35, D-56): split the site list, `ids` or a comma-separated `monitoring_location_id` in `params`, into lists of this many sites that run concurrently as `split_datetime` pieces do. A site listed twice is sent once. Not with `split_datetime`.	`None`
`validate`	`bool`	Before the first page request, check `collection` against the collection index, and each name in `params` outside `QUERY_PARAMS`, and `id_property`, against the collection's queryables, from metadata cached for an hour (D-07). pygeoapi drops a name it does not know, so an unchecked typo returns the whole collection. Defaults to `True`; `False` skips the check, for a server that accepts a name its queryables leave out.	`True`
`resume`	`Checkpoint`	Continue a query from a `Checkpoint`: a page's `checkpoint` or a `PaginationError`'s (D-52). Its pending requests are sent as stored, so no query keyword may come with it, `validate` is ignored, and no request it holds is sent again.	`None`

Yields:

Type	Description
`Page`	One per response, the decoded document in `body` and its bytes in `content`. Its `checkpoint` is where the query stands once the caller holds it.

Raises:

Type	Description
`ValidationError`	Before any request, when `collection` is empty, `.`, or `..`, both `bbox` and `geometry`, or `ids` and `filter`, are given, `ids` is a `str` or `bytes`, empty, or lacks `id_property`, `bbox` or `geometry` is malformed, a CQL2-JSON filter holds a NaN or infinite number, a filter is empty, `params` also holds a parameter a keyword sets (`bbox`, `filter`, `filter-lang`) or a credential such as `api_key`, or `limit` is not an integer from 1 to the profile's `max_limit`, or a split is given to a service other than NWIS, both splits are given, or the query has no closed `datetime` interval or site list to split. With `validate`, before the first page request, when the collection, a name in `params`, or `id_property` is unknown; the message lists the valid ones. With `resume`, when a query keyword comes with it or it is a checkpoint of another service or collection.
`TypeError`	Before any request, when a CQL2-JSON filter holds a value JSON cannot encode, such as a set.
`ServiceError`	With `validate`, when a metadata request fails.
`PaginationError`	When a request fails, a body is not a FeatureCollection with well-formed links, or a `next` link repeats a request already answered, leaves the scheme, host, and port of `base_url`, or a page has a link that carries a credential (D-53). It is raised once every page that arrived has been yielded; its failures name each such request or link by URL, and its `checkpoint` is where the query stands.

`collect(collection, *, params=None, bbox=None, geometry=None, filter=None, ids=None, id_property=None, split_datetime=None, sites_per_piece=None, validate=True, keep_pages=False, resume=None)` `async` ¶

Fetch every page of a collection's items into one QueryResult (D-22).

Parameters:

Name	Type	Description	Default
`collection`	`str`	The collection id.	required
`params`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`bbox`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`geometry`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`filter`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`ids`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`id_property`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`split_datetime`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`sites_per_piece`	`Mapping[str, str] \| None`	The query and its pieces, as for `iter_pages`.	`None`
`validate`	`bool`	Check the query against cached metadata first, as for `iter_pages`.	`True`
`keep_pages`	`bool`	Keep each page's response bytes in `Page.content` (D-23). Page bodies are always dropped. With `resume`, it applies to the new pages; held pages keep what the earlier call kept.	`False`
`resume`	`QueryResult or Checkpoint`	Continue a query that stopped on a failure (D-52): the `result` of a `PartialResultsError`, or the `checkpoint` of a `PaginationError` raised before any page arrived. The result's pages and features come first in the new result, and no request for them is sent. No query keyword may come with it.	`None`

Returns:

Type	Description
`QueryResult`	`geojson` holds the features of all pages in the order they arrived, those of `resume` first.

Raises:

Type	Description
`ValidationError`	On a malformed query or, with `validate`, an unknown name, as for `iter_pages`. With `resume`, before any request, when a query keyword comes with it, it is from another service or collection, a `QueryResult` has no `checkpoint`, or a `Checkpoint` holds pages, which only the `QueryResult` that holds them can carry.
`TypeError`	Before any request, when a CQL2-JSON filter holds a value JSON cannot encode, as for `iter_pages`.
`ServiceError`	With `validate`, when a metadata request fails.
`PartialResultsError`	When a page fails after at least one succeeded, `resume`'s included. Its `result` holds the pages that succeeded and the checkpoint to resume.
`PaginationError`	When no page arrives, as when the first page of an unsplit query fails.

`item(collection, item_id, *, params=None, validate=True)` `async` ¶

Fetch one feature of a collection by its id (D-22).

Parameters:

Name	Type	Description	Default
`collection`	`str`	The collection id.	required
`item_id`	`str or int`	The feature id, percent-encoded into the URL path.	required
`params`	`mapping of str to str`	Query parameters, such as `crs` or `skipGeometry`, with `f=json` added unless given.	`None`
`validate`	`bool`	Before the request, check `collection` against the cached collection index, as `iter_pages` does. Defaults to `True`.	`True`

Returns:

Type	Description
`Page`	`body` holds the decoded Feature and `content` its bytes. `feature_count` is 1, and `number_matched` and `number_returned` are `None`.

Raises:

Type	Description
`ServiceError`	If the request fails, with status 404 for an unknown id, or the body is not a Feature whose `links` member, when present, is a list. With `validate`, also when the metadata request fails.
`ValidationError`	Before any request, if `collection` or `item_id` is empty, `.`, or `..`, which would address another path, or `params` carries a credential such as `api_key`. With `validate`, if the collection is unknown.

`aclose()` `async` ¶

Close the transport if this service created it.

`NWIS` ¶

Bases: OGCService

Async client for the USGS Water Data OGC API, which serves NWIS data (D-42).

Construction performs no HTTP. Pages hold 10,000 features unless params set a limit, and 50,000 at most (D-41).

Parameters:

Name	Type	Description	Default
`api_key`	`str`	A USGS API key, sent in the `X-Api-Key` header of every request (D-18). Defaults to the `USGS_API_KEY` environment variable, read once here. An empty string sends no key.	`None`
`transport`	`Transport`	Sends every request, as for `OGCService`.	`None`

Raises:

Type	Description
`ValidationError`	If the key holds a character other than printable ASCII, such as a newline.

`FabricData` ¶

Bases: OGCService

Async client for the USGS FabricData service, which is sent no key (D-30).

Construction performs no HTTP. Pages hold 1,000 features unless params set a smaller limit (D-41).

Parameters:

Name	Type	Description	Default
`transport`	`Transport`	Sends every request, as for `OGCService`. Pass `Transport(timeout=..., max_retries=...)` to tune the timeout and retry count.	`None`

`GeoConnex` ¶

Bases: OGCService

Async client for the GeoConnex reference features, which never get the USGS key.

GeoConnex is not a USGS host, so the key never goes there (D-13). Construction performs no HTTP. Pages hold 500 features unless params set a limit, and 10,000 at most (D-41).

Parameters:

Name	Type	Description	Default
`transport`	`Transport`	Sends every request, as for `OGCService`. Sharing one with an `NWIS` service is safe, since the key travels with each NWIS request.	`None`

Results¶

An aggregated query result and the pages behind it (D-03).

Parameters:

Name	Type	Description	Default
`geojson`	`dict`	The aggregated FeatureCollection, nothing coerced. Left out of `repr`.	required
`pages`	`tuple of Page`	Every page received, in order. Bodies are `None`; `content` holds the bytes only with `keep_pages=True` (D-23), as set by the call that fetched each page, so a resumed result can mix the two.	required
`pagination`	`Pagination`	Counts and completeness of the traversal.	required
`crs`	`CRS`	The requested, per-page, and effective CRS, and how it was determined.	required
`checkpoint`	`Checkpoint`	Where a query that stopped on a failure stands, for `collect(resume=...)` (D-52). `None` once nothing is left to send, and after a refused request, which a resume would refuse again.	`None`

One response document and the evidence of how it was fetched (D-22).

Parameters:

Name	Type	Description	Default
`url`	`str`	The request as sent, query included. Credential values are masked, and userinfo and fragment dropped.	required
`params`	`mapping of str to str`	The caller's query parameters, with credential values masked.	required
`status`	`int`	HTTP status code.	required
`headers`	`mapping of str to str`	Response headers in `KEPT_HEADERS`, with lowercase names.	required
`links`	`tuple of Link`	Links from the document.	required
`fetched_at`	`float`	When the response arrived, in seconds since the epoch.	required
`number_matched`	`int or None`	`numberMatched` as the server reported it, `None` when absent.	required
`number_returned`	`int or None`	`numberReturned` as the server reported it, `None` when absent.	required
`feature_count`	`int`	Features received in this document.	required
`body`	`dict`	The decoded document, a Feature or a FeatureCollection. The one mutable field; the caller owns it.	`None`
`content`	`bytes`	The response body after HTTP content decoding, before JSON decoding.	`None`
`method`	`str`	The request's HTTP method, defaults to `GET`.	`'GET'`
`request_body`	`bytes`	The request body as sent, `None` for a request without one. With `method` and `url` it identifies the request (D-38).	`None`
`checkpoint`	`Checkpoint`	Where the query stands once the caller holds this page, set on each page `iter_pages` yields with more left to fetch (D-52). Saving it with the page lets a stopped process resume. `None` on the last page, on a page whose `next` link was refused, on `item` pages, and on the pages of a `QueryResult`, which carries its own.	`None`

Notes

params and headers are read-only copies. body and content are the server's document as received, so redaction does not reach them; the key stays out of them because it never travels in a query (D-18).

`json()` ¶

Decode content into a new object on every call (D-23).

Raises:

Type	Description
`ValueError`	If the page kept no content.

What a traversal fetched and whether it reached the end (D-22).

Parameters:

Name	Type	Description	Default
`number_matched`	`int or None`	The `numberMatched` every page reported and agreed on, `None` when a page left it out or pages disagreed. Pages that report it beside pages that do not leave the result `unknown`.	required
`number_returned`	`int`	Features received across all pages.	required
`page_count`	`int`	Pages received.	required
`completeness`	`('complete', 'limited', 'unknown')`	`complete` when traversal reached the end and every check passed, `limited` when it stopped early, `unknown` otherwise.	`"complete"`
`stop_reason`	`('limit', 'failure')`	Why a `limited` traversal stopped: a caller limit or a failed page.	`"limit"`
`sort_by`	`str`	The sort key the pages were requested with.	`None`
`checks`	`tuple of {"end_of_links", "count"}`	The completeness checks that ran and passed: the last page had no `next` link, and the features received equal the agreed `numberMatched`.	`()`

Raises:

Type	Description
`ValueError`	If `stop_reason` is set without `limited`, or missing with it.

A link from a response document, with credentials masked in href.

Parameters:

Name	Type	Description	Default
`href`	`str`	Target URL.	required
`rel`	`str`	Relation type, such as `next` or `self`.	`None`
`type`	`str`	Media type of the target.	`None`
`title`	`str`	Human-readable label.	`None`
`hreflang`	`str`	Language of the target.	`None`
`extra`	`mapping of str to Any`	Every other member of the link object, read-only and left out of the hash.	`empty_extra()`

`from_dict(doc)` `classmethod` ¶

Parse a link object.

Raises:

Type	Description
`ValueError`	If `doc` has no string `href`.

`to_dict()` ¶

Return the link object, href redacted.

The coordinate reference system of a result and how it was determined (D-37).

Parameters:

Name	Type	Description	Default
`requested`	`str or None`	The `crs` query parameter of the first request as sent, `None` when absent.	required
`page_crs`	`tuple of str or None`	Each page's `Content-Crs` header with its angle brackets removed, `None` for a page without one. A repeated header lists its values comma-separated.	required
`effective`	`str or None`	The CRS of the geometries, `None` when no evidence settles it.	required
`source`	`('content_crs', 'default', 'unconfirmed', 'unknown')`	`content_crs` when every page named the same CRS. Without a header on any page: `default` (CRS84, the OGC API Features default) when no request asked for another CRS, `unconfirmed` when every request asked for the same other CRS, which a server applying it must name in `Content-Crs`, and `unknown` when the requests disagreed. `unknown` also when pages named different CRSs or only some named one.	`"content_crs"`

Where an unfinished items query stands: what it holds and what it has left (D-52).

Every yielded Page with more to fetch carries one, as do PaginationError and a QueryResult that stopped early. Pass it to iter_pages(resume=...), or a QueryResult to collect(resume=...). It pickles, holds no feature, and refuses to resume on another service or collection. Unpickle only a checkpoint you saved: loading a pickle runs code, and it skips the checks a Request makes when built.

Parameters:

Name	Type	Description	Default
`base_url`	`str`	The `base_url` of the service that ran the query.	required
`collection`	`str`	The collection the query reads.	required
`pending`	`tuple of Request`	The requests left to send, in order: the requests that failed, those in flight or fetched and not yet yielded, the `next` links not yet started, the pieces not yet started, then the refused ones (D-56).	required
`held`	`collection of tuple`	The identity (method, URL as sent, body) of each request already answered, in the order answered. A resumed query refuses to send any of them again.	`frozenset[RequestId]()`
`malformed`	`bool`	Whether an answered page held a `numberMatched` or `numberReturned` that is not a non-negative integer, so a resumed `QueryResult` still reads `unknown`.	`False`

Raises:

Type	Description
`TypeError`	If `pending` holds anything but `Request` objects.
`ValueError`	If `pending` is empty: a query with nothing left has no checkpoint.

One items request: its identity (D-38) and the headers it needs.

A Checkpoint holds the requests a query has still to send (D-52). Construction refuses a credential, so a stored or pickled request never carries one.

Parameters:

Name	Type	Description	Default
`url`	`str`	The URL before `params` are added.	required
`method`	`str`	`GET`, the default, or `POST`; every `next` link is requested the same way.	`'GET'`
`params`	`mapping of str to str`	Query parameters of this request only; a `next` link carries its own.	`dict[str, str]()`
`body`	`bytes`	The body, sent to every `next` link as well.	`None`
`headers`	`mapping of str to str`	`Accept` or `Content-Type`, sent with every request beside the service's own headers.	`None`

Raises:

Type	Description
`ValidationError`	If `method` is not `GET` or `POST`, `url` holds userinfo, `url` or `params` carry a credential such as `api_key`, or `headers` holds another header. A credential is refused, since masking it would change the request.

Notes

params and headers are read-only copies, so a change to the mappings passed in never reaches a stored request.

`identity()` ¶

Return the method, URL as sent with credentials masked, and body (D-38).

Collection metadata¶

The /collections document.

Parameters:

Name	Type	Description	Default
`collections`	`tuple of Collection`	Every collection, in the server's order.	`None`
`links`	`tuple of Link`	Links of the document itself.	`None`
`extra`	`mapping of str to Any`	Every other member.	`empty_extra()`

`from_dict(doc)` `classmethod` ¶

Parse the document.

Raises:

Type	Description
`ValueError`	If an entry of `collections` is not an object with a string `id`.

`to_dict()` ¶

Return the document as the server sent it, href redacted in parsed links.

One collection, from /collections or /collections/{id}.

A field is None when the server left its member out or sent it with a shape the field cannot hold, such as a null; that member then stays in extra.

Parameters:

Name	Type	Description	Default
`id`	`str`	Collection identifier, the path segment in its URLs.	required
`title`	`str`	Human-readable title.	`None`
`description`	`str`	Long description.	`None`
`keywords`	`tuple of str`	Keywords.	`None`
`links`	`tuple of Link`	Links, `href` redacted.	`None`
`extent`	`Extent`	Spatial and temporal extent.	`None`
`item_type`	`str`	`itemType`, such as `feature` or `record`.	`None`
`crs`	`tuple of str`	Every CRS the server can return features in.	`None`
`storage_crs`	`str`	`storageCrs`, the CRS the server stores the features in.	`None`
`extra`	`mapping of str to Any`	Every other member.	`empty_extra()`

`from_dict(doc)` `classmethod` ¶

Parse a collection document.

Raises:

Type	Description
`ValueError`	If `doc` has no string `id`.

`to_dict()` ¶

Return the document as the server sent it, href redacted in parsed links.

A collection's extent.

Parameters:

Name	Type	Description	Default
`spatial`	`SpatialExtent`	Bounding boxes.	`None`
`temporal`	`TemporalExtent`	Time intervals.	`None`
`extra`	`mapping of str to Any`	Every other member.	`empty_extra()`

`from_dict(doc)` `classmethod` ¶

Parse the member.

`to_dict()` ¶

Return the member as the server sent it.

The spatial member of a collection's extent.

Parameters:

Name	Type	Description	Default
`bbox`	`tuple of tuple of float`	Every bounding box the server sent, in order: 4 or 6 numbers each. The first covers the whole collection.	`None`
`crs`	`str`	CRS of the boxes.	`None`
`extra`	`mapping of str to Any`	Every other member.	`empty_extra()`

`from_dict(doc)` `classmethod` ¶

Parse the member.

`to_dict()` ¶

Return the member as the server sent it.

The temporal member of a collection's extent.

Parameters:

Name	Type	Description	Default
`interval`	`tuple of (str or None, str or None)`	Every interval the server sent, in order. `None` marks an open end.	`None`
`trs`	`str`	Temporal reference system.	`None`
`extra`	`mapping of str to Any`	Every other member.	`empty_extra()`

`from_dict(doc)` `classmethod` ¶

Parse the member.

`to_dict()` ¶

Return the member as the server sent it.

A /queryables or /schema document.

/queryables lists the properties a query can filter on, /schema the properties a feature carries.

Parameters:

Name	Type	Description	Default
`title`	`str`	Human-readable title.	`None`
`properties`	`mapping of str to Property`	Every property by name, in the server's order. Read-only.	`None`
`extra`	`mapping of str to Any`	Every other member, such as `$id` and `$schema`.	`empty_extra()`

`from_dict(doc)` `classmethod` ¶

Parse the document.

`to_dict()` ¶

Return the document as the server sent it.

One property of a /queryables or /schema document, a JSON Schema.

Parameters:

Name	Type	Description	Default
`type`	`str`	JSON type, such as `string` or `integer`.	`None`
`title`	`str`	Human-readable name.	`None`
`description`	`str`	Long description.	`None`
`format`	`str`	Format, such as `date-time` or `geometry-any`.	`None`
`role`	`str`	`x-ogc-role`, such as `id` or `primary-geometry`.	`None`
`extra`	`mapping of str to Any`	Every other member, such as `enum` or `example`.	`empty_extra()`

`from_dict(doc)` `classmethod` ¶

Parse the property.

`to_dict()` ¶

Return the property as the server sent it.

Logging and errors¶

Set where the package logger writes, replacing the previous configuration.

Each call states the whole configuration: an argument left out returns to its default, so configure_logger(level="DEBUG") after configure_logger(file=...) turns file logging off. configure_logger() restores the import-time setup, a console handler on stderr at WARNING. https://docs.hyriver.io/water-ogcapi/logging/ covers applications with their own logging, AWS Lambda, and tests.

Parameters:

Name	Type	Description	Default
`verbose`	`bool`	Shortcut for the console level: `True` is DEBUG, `False` is WARNING. `level` wins when both are given.	`None`
`level`	`str or int`	Console level, such as `"INFO"`. Defaults to WARNING.	`None`
`file`	`str or Path`	Also write to this file, encoded as UTF-8. Parent directories are created.	`None`
`file_level`	`str or int`	File level. Defaults to DEBUG.	`'DEBUG'`
`file_mode`	`('a', 'w')`	Append to the file or overwrite it. Defaults to `'a'`.	`'a'`
`file_only`	`bool`	Write to the file and not the console. Requires `file`.	`False`

Raises:

Type	Description
`ValidationError`	If a level or `file_mode` is invalid, or `file_only` is set without `file`. The previous configuration stays in place.
`OSError`	If the file cannot be opened. The previous configuration stays in place.

Exceptions for water_ogcapi.

`WaterOGCAPIError` ¶

Bases: Exception

Base exception for all water_ogcapi errors.

`ValidationError` ¶

Bases: WaterOGCAPIError, ValueError

Raised before a request is sent when caller input is invalid (docs/design.md D-45).

Its message names the valid options where a finite set exists, such as the collections of a service or the properties a query can filter on.

`ServiceError` ¶

Bases: WaterOGCAPIError

Raised when a request fails: an error status, a redirect, a body of the wrong shape, or a URL refused before sending (attempts 0).

Parameters:

Name	Type	Description	Default
`message`	`str`	Description of the failure.	required
`url`	`str`	The URL that triggered the error, included in the message when given, with credential query values masked (D-04).	`None`
`status`	`int`	HTTP status code from the failed response, when available.	`None`
`attempts`	`int`	Requests made before giving up, defaults to 1; 0 when the request was refused before sending. Anything above 1 means the transport retried, so an outer retry budget can tell a single failure from an exhausted one.	`1`
`retry_after`	`float`	Seconds the server asked the caller to wait, parsed from `Retry-After`.	`None`
`method`	`str`	The request's HTTP method, defaults to `GET`.	`'GET'`
`request_body`	`bytes`	The request body as sent, `None` for a request without one (D-38).	`None`

`RateLimitError` ¶

Bases: ServiceError

Raised when an OGC API service returns HTTP 429 Too Many Requests.

Parameters:

Name	Type	Description	Default
`message`	`str`	Description of the failure.	required
`url`	`str`	The URL that triggered the error.	`None`
`retry_after`	`float`	Seconds to wait before retrying, parsed from the `Retry-After` header. Both integer-second and HTTP-date forms are supported.	`None`
`remaining`	`int`	Value of the `x-ratelimit-remaining` header when present.	`None`
`attempts`	`int`	Requests made before giving up, defaults to 1.	`1`
`method`	`str`	The request's HTTP method, defaults to `GET`.	`'GET'`
`request_body`	`bytes`	The request body as sent, `None` for a request without one (D-38).	`None`

`PaginationError` ¶

Bases: WaterOGCAPIError

Raised when page requests of a traversal fail (docs/design.md D-24).

Every page that succeeded has already been yielded, so the error holds no page.

Parameters:

Name	Type	Description	Default
`failures`	`sequence of ServiceError`	One error per failed request, each naming the request by its URL as sent.	required
`checkpoint`	`Checkpoint`	Where the query stands, the failed requests first, for `resume` (D-52). `None` when nothing but refused requests remains, since a resume would refuse them again.	`None`

`PartialResultsError` ¶

Bases: PaginationError

Raised by collect() when a page failed after at least one page succeeded.

Parameters:

Name	Type	Description	Default
`failures`	`sequence of ServiceError`	One error per failed request, each naming the request by its URL as sent.	required
`result`	`QueryResult`	Every page that succeeded, with `completeness` `limited` and stop reason `failure` (docs/design.md D-24). Its `checkpoint`, also this error's, lets `collect(resume=result)` continue it (D-52).	required

API reference¶

Services¶

base_url property ¶

collections() ¶

queryables(collection) ¶

schema(collection) ¶

iter_pages(collection, *, params=None, bbox=None, geometry=None, filter=None, ids=None, id_property=None, split_datetime=None, sites_per_piece=None, validate=True, resume=None) ¶

collect(collection, *, params=None, bbox=None, geometry=None, filter=None, ids=None, id_property=None, split_datetime=None, sites_per_piece=None, validate=True, keep_pages=False, resume=None) ¶

item(collection, item_id, *, params=None, validate=True) ¶

refresh() ¶

close() ¶

timeout property ¶

max_concurrent property ¶

max_retries property ¶

backoff property ¶

http2 property ¶

get(url, *, params=None, headers=None, governor=None) async ¶

request(method, url, *, params=None, headers=None, content=None, governor=None) async ¶

aclose() async ¶

window property ¶

interval property ¶

wait() async ¶

observe(status, headers) ¶

Async API¶

OGCService ¶

base_url property ¶

collections() async ¶

queryables(collection) async ¶

schema(collection) async ¶

refresh() ¶

iter_pages(collection, *, params=None, bbox=None, geometry=None, filter=None, ids=None, id_property=None, split_datetime=None, sites_per_piece=None, validate=True, resume=None) async ¶

collect(collection, *, params=None, bbox=None, geometry=None, filter=None, ids=None, id_property=None, split_datetime=None, sites_per_piece=None, validate=True, keep_pages=False, resume=None) async ¶

item(collection, item_id, *, params=None, validate=True) async ¶

aclose() async ¶

NWIS ¶

FabricData ¶

GeoConnex ¶

Results¶

json() ¶

from_dict(doc) classmethod ¶

to_dict() ¶

identity() ¶

Collection metadata¶

from_dict(doc) classmethod ¶

to_dict() ¶

from_dict(doc) classmethod ¶

to_dict() ¶

from_dict(doc) classmethod ¶

to_dict() ¶

from_dict(doc) classmethod ¶

to_dict() ¶

from_dict(doc) classmethod ¶

to_dict() ¶

from_dict(doc) classmethod ¶

to_dict() ¶

from_dict(doc) classmethod ¶

to_dict() ¶

Logging and errors¶

WaterOGCAPIError ¶

ValidationError ¶

ServiceError ¶

RateLimitError ¶

PaginationError ¶

PartialResultsError ¶

`base_url` `property` ¶

`collections()` ¶

`queryables(collection)` ¶

`schema(collection)` ¶

`iter_pages(collection, *, params=None, bbox=None, geometry=None, filter=None, ids=None, id_property=None, split_datetime=None, sites_per_piece=None, validate=True, resume=None)` ¶

`collect(collection, *, params=None, bbox=None, geometry=None, filter=None, ids=None, id_property=None, split_datetime=None, sites_per_piece=None, validate=True, keep_pages=False, resume=None)` ¶

`item(collection, item_id, *, params=None, validate=True)` ¶

`refresh()` ¶

`close()` ¶

`timeout` `property` ¶

`max_concurrent` `property` ¶

`max_retries` `property` ¶

`backoff` `property` ¶

`http2` `property` ¶

`get(url, *, params=None, headers=None, governor=None)` `async` ¶

`request(method, url, *, params=None, headers=None, content=None, governor=None)` `async` ¶

`aclose()` `async` ¶

`window` `property` ¶

`interval` `property` ¶

`wait()` `async` ¶

`observe(status, headers)` ¶

`OGCService` ¶

`base_url` `property` ¶

`collections()` `async` ¶

`queryables(collection)` `async` ¶

`schema(collection)` `async` ¶

`refresh()` ¶

`iter_pages(collection, *, params=None, bbox=None, geometry=None, filter=None, ids=None, id_property=None, split_datetime=None, sites_per_piece=None, validate=True, resume=None)` `async` ¶

`collect(collection, *, params=None, bbox=None, geometry=None, filter=None, ids=None, id_property=None, split_datetime=None, sites_per_piece=None, validate=True, keep_pages=False, resume=None)` `async` ¶

`item(collection, item_id, *, params=None, validate=True)` `async` ¶

`aclose()` `async` ¶

`NWIS` ¶

`FabricData` ¶

`GeoConnex` ¶

`json()` ¶

`from_dict(doc)` `classmethod` ¶

`to_dict()` ¶

`identity()` ¶

`from_dict(doc)` `classmethod` ¶

`to_dict()` ¶

`from_dict(doc)` `classmethod` ¶

`to_dict()` ¶

`from_dict(doc)` `classmethod` ¶

`to_dict()` ¶

`from_dict(doc)` `classmethod` ¶

`to_dict()` ¶

`from_dict(doc)` `classmethod` ¶

`to_dict()` ¶

`from_dict(doc)` `classmethod` ¶

`to_dict()` ¶

`from_dict(doc)` `classmethod` ¶

`to_dict()` ¶

`WaterOGCAPIError` ¶

`ValidationError` ¶

`ServiceError` ¶

`RateLimitError` ¶

`PaginationError` ¶

`PartialResultsError` ¶