Collection search (#735)

* create BaseSearch class

* add collection search functionality

* update tests

* replace Z with +00:00 in datetime strings for Python 3.10

* move ItemSearch back to search.py

* reject extra args if client-side filter

* fix matched method

* fix error for collection search support check

* moar tests!

* add warning expectation to test

* add collection search functionality to cli

* add collection search examples to quickstart

* quote search tokens with special characters

* add collection search example to intro notebook

* add CollectionSearch entry to api docs

* clean up client docs a bit

* add collection_list_as_dict method

* add CollectionSearch section to usage docs

* fix lint error

* reinstate item_search.py

* clean up warnings, tests

* address review comments

* improve matched logic

* actually clean up matched logic

* update changelog

.gitignore

@@ -620,3 +620,4 @@ $RECYCLE.BIN/
 # Windows shortcuts
 *.lnk
 
+uv.lock

CHANGELOG.md

@@ -7,6 +7,10 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
 
 ## [Unreleased]
 
+### Added
+
+- Support for collection search via `CollectionSearch` class and associated client methods [#735](https://github.com/stac-utils/pystac-client/pull/735)
+
 ### Removed
 
 - Python 3.9 support [#724](https://github.com/stac-utils/pystac-client/pull/724)

docs/api.rst

@@ -29,6 +29,16 @@ endpoint, if supported.
    :members:
    :undoc-members:
 
+Collection Search
+-----------------
+
+The `CollectionSearch` class represents a search of collections in a STAC API.
+
+.. autoclass:: pystac_client.CollectionSearch
+   :members:
+   :undoc-members:
+   :member-order: bysource
+
 Item Search
 -----------
 
@@ -39,6 +49,7 @@ The `ItemSearch` class represents a search of a STAC API.
    :undoc-members:
    :member-order: bysource
 
+
 STAC API IO
 -----------
 

docs/quickstart.rst

@@ -7,7 +7,8 @@ Python library.
 CLI
 ~~~
 
-Use the CLI to quickly make searches and output or save the results.
+Use the CLI to quickly make item- or collection-level searches and
+output or save the results.
 
 The ``--matched`` switch performs a search with limit=1 so does not get
 any Items, but gets the total number of matches which will be output to
@@ -18,6 +19,15 @@ the screen (if supported by the STAC API).
     $ stac-client search https://earth-search.aws.element84.com/v1 -c sentinel-2-l2a --bbox -72.5 40.5 -72 41 --matched
     3141 items matched
 
+The ``--matched`` flag can also be used for collection search to get
+the total number of collections that match your search terms.
+
+
+.. code-block:: console
+
+    $ stac-client collections https://emc.spacebel.be --q sentinel-2 --matched
+    76 collections matched
+
 If the same URL is to be used over and over, define an environment
 variable to be used in the CLI call:
 
@@ -87,6 +97,26 @@ than once to use additional operators.
     $ stac-client search ${STAC_API_URL} -c sentinel-2-l2a --bbox -72.5 40.5 -72 41 --datetime 2020-01-01/2020-01-31 --query "eo:cloud_cover<10" "eo:cloud_cover>5" --matched
     4 items matched
 
+
+Collection searches can also use multiple filters like this example
+search for collections that include the term ``"biomass"`` and have
+a spatial extent that intersects Scandinavia.
+
+.. code-block:: console
+
+    $ stac-client collections https://emc.spacebel.be --q biomass --bbox 0.09 54.72 33.31 71.36  --matched
+    43 items matched
+
+Since most STAC APIs have not yet implemented the `collection search 
+extension <https://github.com/stac-api-extensions/collection-search>`_, 
+``pystac-client`` will perform a limited client-side 
+filter on the full list of collections using only the ``bbox``, 
+``datetime``, and ``q`` (free-text search) parameters.
+In the case that the STAC API does not support collection search, a
+warning will be displayed to inform you that the filter is being
+applied client-side.
+
+
 Python
 ~~~~~~
 
@@ -99,7 +129,7 @@ specific STAC API (use the root URL):
 
     client = Client.open("https://earth-search.aws.element84.com/v1")
 
-Create a search:
+Create an item-level search:
 
 .. code-block:: python
 
@@ -125,3 +155,23 @@ The ``ItemCollection`` can then be saved as a GeoJSON FeatureCollection.
 
     item_collection = search.item_collection()
     item_collection.save_object('my_itemcollection.json')
+
+
+Create a collection-level search:
+
+.. code-block:: python
+
+    collection_search = client.collection_search(
+        q='"sentinel-2" OR "sentinel-1"',
+    )
+    print(f"{collection_search.matched()} collections found")
+
+
+The ``collections()`` iterator method can be used to iterate through all
+resulting collections.
+
+.. code-block:: python
+
+    for collection in collection_search.collections():
+        print(collection.id)
+

docs/tutorials/pystac-client-introduction.ipynb

@@ -40,7 +40,9 @@
    "cell_type": "code",
    "execution_count": null,
    "id": "98942e75",
-   "metadata": {},
+   "metadata": {
+    "scrolled": true
+   },
    "outputs": [],
    "source": [
     "# STAC API root URL\n",
@@ -74,6 +76,48 @@
     "    print(collection)"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "ebab2724-cab3-4fba-b25b-fdfb4e537014",
+   "metadata": {},
+   "source": [
+    "# Collection Search\n",
+    "\n",
+    "Sometimes, it can be challenging to identify which collection you want to work with. The `collection_search` method allows you to discover collections by applying search filters that will help you find the specific collection(s) you need. Since many STAC APIs have not implemented the collection search extension, `pystac-client` will perform a limited client-side filter if the API does not conform to the collection search spec."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "a23a53ec-5b5f-421d-9f0e-01dbde8c3697",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "collection_search = cat.collection_search(\n",
+    "    q=\"ASTER\",\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "90b3d014-9c8f-4c5b-a94e-bfb7f17380ad",
+   "metadata": {},
+   "source": [
+    "The `collections` method lets you iterate through the results of the search so you can inspect the details of matching collections."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "006f13fd-5e58-4f3f-bd5a-707cd830caa1",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "for result in collection_search.collections():\n",
+    "    print(result.id, f\"{collection.description}\", sep=\"\\n\")\n",
+    "    print(\"\\n\")"
+   ]
+  },
   {
    "cell_type": "code",
    "execution_count": null,
@@ -233,7 +277,7 @@
    "hash": "6b6313dbab648ff537330b996f33bf845c0da10ea77ae70864d6ca8e2699c7ea"
   },
   "kernelspec": {
-   "display_name": "Python 3.9.11 ('.venv': venv)",
+   "display_name": "Python 3 (ipykernel)",
    "language": "python",
    "name": "python3"
   },
@@ -247,7 +291,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.9.11"
+   "version": "3.12.3"
   }
  },
  "nbformat": 4,

docs/usage.rst

@@ -229,10 +229,10 @@ creating your :class:`Client<pystac_client.Client>`.
 CollectionClient
 ++++++++++++++++
 
-STAC APIs may optionally implement a ``/collections`` endpoint as describe in the
+STAC APIs may optionally implement a ``/collections`` endpoint as described in the
 `STAC API - Collections spec
-<https://github.com/radiantearth/stac-api-spec/tree/master/collections>`__. This endpoint
-allows clients to search or inspect items within a particular collection.
+<https://github.com/radiantearth/stac-api-spec/tree/release/v1.0.0/ogcapi-features#stac-api---collections>`__. 
+This endpoint allows clients to search or inspect items within a particular collection.
 
 .. code-block:: python
 
@@ -245,7 +245,7 @@ allows clients to search or inspect items within a particular collection.
 PySTAC will get items by iterating through all children until it gets to an ``item`` link.
 PySTAC client will use the API endpoint instead: `/collections/<collection_id>/items`
 (as long as `STAC API - Item Search spec
-<https://github.com/radiantearth/stac-api-spec/tree/master/item-search>`__ is supported).
+<https://github.com/radiantearth/stac-api-spec/tree/release/v1.0.0/item-search>`__ is supported).
 
 .. code-block:: python
 
@@ -254,15 +254,113 @@ PySTAC client will use the API endpoint instead: `/collections/<collection_id>/i
 Note that calling list on this iterator will take a really long time since it will be retrieving
 every itme for the whole ``"sentinel-2-l2a"`` collection.
 
+CollectionSearch
+++++++++++++++++
+
+STAC API services may optionally implement a ``/collections`` endpoint as described in the
+`STAC API - Collections spec
+<https://github.com/radiantearth/stac-api-spec/tree/release/v1.0.0/ogcapi-features#stac-api---collections>`__.
+The ``/collections`` endpoint can be extended with the 
+`STAC API - Collection Search Extension <https://github.com/stac-api-extensions/collection-search>`__ 
+which adds the capability to apply filter parameters to the collection-level metadata. 
+See the `Query Parameters and Fields
+<https://github.com/stac-api-extensions/collection-search?tab=readme-ov-file#query-parameters-and-fields>`__
+from that spec for details on the meaning of each parameter.
+
+The :meth:`pystac_client.Client.collection_search` method provides an interface for making
+requests to a service's "collections" endpoint. This method returns a
+:class:`pystac_client.CollectionSearch` instance.
+
+.. code-block:: python
+
+    >>> from pystac_client import Client
+    >>> catalog = Client.open('https://planetarycomputer.microsoft.com/api/stac/v1')
+    >>> results = catalog.collection_search(
+    ...     q="biomass",
+    ...     datetime="2022/.."
+    ... )
+
+Instances of :class:`~pystac_client.CollectionSearch` have a handful of methods for
+getting matching collections as Python objects. The right method to use depends on
+how many of the matches you want to consume (a single collection at a time, a
+page at a time, or everything) and whether you want plain Python dictionaries
+representing the collections, or :class:`pystac.Collection` objects.
+
+The following table shows the :class:`~pystac_client.CollectionSearch` methods for fetching
+matches, according to which set of matches to return and whether to return them as
+``pystac`` objects or plain dictionaries.
+
+====================== ======================================================= ===============================================================
+Matches to return      PySTAC objects                                          Plain dictionaries
+====================== ======================================================= ===============================================================
+**Single collections** :meth:`~pystac_client.CollectionSearch.collections`     :meth:`~pystac_client.CollectionSearch.collections_as_dicts`
+**Pages**              :meth:`~pystac_client.CollectionSearch.pages`           :meth:`~pystac_client.CollectionSearch.pages_as_dicts`
+**Everything**         :meth:`~pystac_client.CollectionSearch.collection_list` :meth:`~pystac_client.CollectionSearch.collection_list_as_dict`
+====================== ======================================================= ===============================================================
+
+Additionally, the ``matched`` method can be used to access result metadata about
+how many total items matched the query:
+
+* :meth:`CollectionSearch.matched <pystac_client.CollectionSearch.matched>`: returns the number
+  of hits (collections) for this search. If the API supports the STAC API Context Extension this
+  value will be returned directly from a search result with ``limit=1``. Otherwise ``pystac-client``
+  will count the results and return a value with an associated warning.
+
+.. code-block:: python
+
+    >>> for collection in results.collections():
+    ...     print(item.id)
+    fia
+    modis-13Q1-061
+    modis-13A1-061
+    sentinel-3-olci-lfr-l2-netcdf
+
+The :meth:`~pystac_client.CollectionSearch.collections` and related methods handle retrieval of
+successive pages of results
+by finding links with a ``"rel"`` type of ``"next"`` and parsing them to construct the
+next request. The default
+implementation of this ``"next"`` link parsing assumes that the link follows the spec for
+an extended STAC link as
+described in the
+`STAC API - Collections: Collection Paging <https://github.com/radiantearth/stac-api-spec/blob/main/ogcapi-features/README.md#collection-pagination>`__
+section.
+
+Alternatively, the Collections can be returned as a list, where each
+list is one page of results retrieved from search:
+
+.. code-block:: python
+
+    >>> for page in results.pages():
+    ...     for collection in page.collections():
+    ...         print(collection.id)
+    fia
+    modis-13Q1-061
+    modis-13A1-061
+    sentinel-3-olci-lfr-l2-netcdf
+
+If you do not need the :class:`pystac.Collection` instances, you can instead use
+:meth:`CollectionSearch.collections_as_dicts <pystac_client.CollectionSearch.collections_as_dicts>`
+to retrieve dictionary representation of the collections, without incurring the cost of
+creating the Collection objects.
+
+.. code-block:: python
+
+    >>> for collection_dict in results.collections_as_dicts():
+    ...     print(collection_dict["id"])
+    fia
+    modis-13Q1-061
+    modis-13A1-061
+    sentinel-3-olci-lfr-l2-netcdf
+
 ItemSearch
 ++++++++++
 
 STAC API services may optionally implement a ``/search`` endpoint as describe in the
 `STAC API - Item Search spec
-<https://github.com/radiantearth/stac-api-spec/tree/master/item-search>`__. This
+<https://github.com/radiantearth/stac-api-spec/tree/main/item-search`__. This
 endpoint allows clients to query STAC Items across the entire service using a variety
 of filter parameters. See the `Query Parameter Table
-<https://github.com/radiantearth/stac-api-spec/tree/master/item-search#query-parameter-table>`__
+<https://github.com/radiantearth/stac-api-spec/tree/main/item-search#query-parameter-table>`__
 from that spec for details on the meaning of each parameter.
 
 The :meth:`pystac_client.Client.search` method provides an interface for making
@@ -280,10 +378,10 @@ requests to a service's "search" endpoint. This method returns a
     ... )
 
 Instances of :class:`~pystac_client.ItemSearch` have a handful of methods for
-getting matching items into Python objects. The right method to use depends on
+getting matching items as Python objects. The right method to use depends on
 how many of the matches you want to consume (a single item at a time, a
 page at a time, or everything) and whether you want plain Python dictionaries
-representing the items, or proper ``pystac`` objects.
+representing the items, or :class:`pystac.Item` objects.
 
 The following table shows the :class:`~pystac_client.ItemSearch` methods for fetching
 matches, according to which set of matches to return and whether to return them as

pystac_client/__init__.py

@@ -1,6 +1,7 @@
 __all__ = [
     "Client",
     "CollectionClient",
+    "CollectionSearch",
     "ConformanceClasses",
     "ItemSearch",
     "Modifiable",
@@ -10,6 +11,7 @@
 from pystac_client._utils import Modifiable
 from pystac_client.client import Client
 from pystac_client.collection_client import CollectionClient
+from pystac_client.collection_search import CollectionSearch
 from pystac_client.conformance import ConformanceClasses
 from pystac_client.item_search import ItemSearch
 from pystac_client.version import __version__