Merge pull request #232 from stac-utils/collections_validation

Validate /collections endpoint

.pre-commit-config.yaml

@@ -1,20 +1,20 @@
 repos:
 -   repo: https://github.com/PyCQA/flake8
-    rev: 6.1.0
+    rev: 7.0.0
     hooks:
     - id: flake8
 -   repo: https://github.com/timothycrosley/isort
-    rev: 5.12.0
+    rev: 5.13.2
     hooks:
     - id: isort
       args: ["--profile", "black"]
 -   repo: https://github.com/psf/black
-    rev: 23.11.0
+    rev: 24.1.1
     hooks:
     - id: black
       language_version: python3.8
 -   repo: https://github.com/pre-commit/mirrors-mypy
-    rev: v1.7.0
+    rev: v1.8.0
     hooks:
       - id: mypy
         exclude: /tests/

CHANGELOG.md

@@ -8,6 +8,7 @@ The format is (loosely) based on [Keep a Changelog](http://keepachangelog.com/)
 
 ### Added
 
+- Added ability to validate response from a /collections endpoint [#220](https://github.com/stac-utils/stac-validator/issues/220)
 - Added mypy to pre-commit config ([#229](https://github.com/stac-utils/stac-validator/pull/224))
 
 ## [v3.3.2] - 2023-11-17

README.md

@@ -106,6 +106,7 @@ Options:
   -m, --max-depth INTEGER  Maximum depth to traverse when recursing. Omit this
                            argument to get full recursion. Ignored if
                            `recursive == False`.
+  --collections            Validate /collections response.
   --item-collection        Validate item collection response. Can be combined
                            with --pages. Defaults to one page.
   -p, --pages INTEGER      Maximum number of pages to validate via --item-

stac_validator/stac_validator.py

@@ -46,6 +46,25 @@ def item_collection_summary(message: List[Dict[str, Any]]) -> None:
     click.secho(f"valid_items: {valid_count}")
 
 
+def collections_summary(message: List[Dict[str, Any]]) -> None:
+    """Prints a summary of the validation results for an item collection response.
+
+    Args:
+        message (List[Dict[str, Any]]): The validation results for the item collection.
+
+    Returns:
+        None
+    """
+    valid_count = 0
+    for collection in message:
+        if "valid_stac" in collection and collection["valid_stac"] is True:
+            valid_count = valid_count + 1
+    click.secho()
+    click.secho("--collections summary", bold=True)
+    click.secho(f"collections_validated: {len(message)}")
+    click.secho(f"valid_collections: {valid_count}")
+
+
 @click.command()
 @click.argument("stac_file")
 @click.option(
@@ -80,6 +99,11 @@ def item_collection_summary(message: List[Dict[str, Any]]) -> None:
     type=int,
     help="Maximum depth to traverse when recursing. Omit this argument to get full recursion. Ignored if `recursive == False`.",
 )
+@click.option(
+    "--collections",
+    is_flag=True,
+    help="Validate /collections response.",
+)
 @click.option(
     "--item-collection",
     is_flag=True,
@@ -102,6 +126,7 @@ def item_collection_summary(message: List[Dict[str, Any]]) -> None:
 )
 def main(
     stac_file: str,
+    collections: bool,
     item_collection: bool,
     pages: int,
     recursive: bool,
@@ -120,6 +145,7 @@ def main(
 
     Args:
         stac_file (str): Path to the STAC file to be validated.
+        collections (bool): Validate response from /collections endpoint.
         item_collection (bool): Whether to validate item collection responses.
         pages (int): Maximum number of pages to validate via `item_collection`.
         recursive (bool): Whether to recursively validate all related STAC objects.
@@ -143,6 +169,7 @@ def main(
     valid = True
     stac = StacValidate(
         stac_file=stac_file,
+        collections=collections,
         item_collection=item_collection,
         pages=pages,
         recursive=recursive,
@@ -155,8 +182,10 @@ def main(
         verbose=verbose,
         log=log_file,
     )
-    if not item_collection:
+    if not item_collection and not collections:
         valid = stac.run()
+    elif collections:
+        stac.validate_collections()
     else:
         stac.validate_item_collection()
 
@@ -169,6 +198,8 @@ def main(
 
     if item_collection:
         item_collection_summary(message)
+    elif collections:
+        collections_summary(message)
 
     sys.exit(0 if valid else 1)
 

stac_validator/validate.py

@@ -25,6 +25,7 @@ class StacValidate:
 
     Attributes:
         stac_file (str): The path or URL to the STAC object to be validated.
+        collections (bool): Validate response from a /collections endpoint.
         item_collection (bool): Whether the STAC object to be validated is an item collection.
         pages (int): The maximum number of pages to validate if `item_collection` is True.
         recursive (bool): Whether to recursively validate related STAC objects.
@@ -45,6 +46,7 @@ class StacValidate:
     def __init__(
         self,
         stac_file: Optional[str] = None,
+        collections: bool = False,
         item_collection: bool = False,
         pages: Optional[int] = None,
         recursive: bool = False,
@@ -58,6 +60,7 @@ def __init__(
         log: str = "",
     ):
         self.stac_file = stac_file
+        self.collections = collections
         self.item_collection = item_collection
         self.pages = pages
         self.message: List = []
@@ -392,6 +395,27 @@ def validate_item_collection_dict(self, item_collection: Dict) -> None:
             self.schema = ""
             self.validate_dict(item)
 
+    def validate_collections(self) -> None:
+        """ "Validate STAC collections from a /collections endpoint.
+
+        Raises:
+            URLError: If there is an issue with the URL used to fetch the item collection.
+            JSONDecodeError: If the item collection content cannot be parsed as JSON.
+            ValueError: If the item collection does not conform to the STAC specification.
+            TypeError: If the item collection content is not a dictionary or JSON object.
+            FileNotFoundError: If the item collection file cannot be found.
+            ConnectionError: If there is an issue with the internet connection used to fetch the item collection.
+            exceptions.SSLError: If there is an issue with the SSL connection used to fetch the item collection.
+            OSError: If there is an issue with the file system (e.g., read/write permissions) while trying to write to the log file.
+
+        Returns:
+            None
+        """
+        collections = fetch_and_parse_file(str(self.stac_file))
+        for collection in collections["collections"]:
+            self.schema = ""
+            self.validate_dict(collection)
+
     def validate_item_collection(self) -> None:
         """Validate a STAC item collection.
 
@@ -429,9 +453,9 @@ def validate_item_collection(self) -> None:
                                 break
         except Exception as e:
             message = {}
-            message[
-                "pagination_error"
-            ] = f"Validating the item collection failed on page {page}: {str(e)}"
+            message["pagination_error"] = (
+                f"Validating the item collection failed on page {page}: {str(e)}"
+            )
             self.message.append(message)
 
     def run(self) -> bool:
@@ -457,7 +481,11 @@ def run(self) -> bool:
         """
         message = {}
         try:
-            if self.stac_file is not None and not self.item_collection:
+            if (
+                self.stac_file is not None
+                and not self.item_collection
+                and not self.collections
+            ):
                 self.stac_content = fetch_and_parse_file(self.stac_file)
 
             stac_type = get_stac_type(self.stac_content).upper()

tests/test_validate_collections.py

@@ -0,0 +1,55 @@
+"""
+Description: Test stac-validator on --collections (/collections validation).
+
+"""
+
+from stac_validator import stac_validator
+
+
+def test_validate_collections_remote():
+    stac_file = "https://earth-search.aws.element84.com/v0/collections"
+    stac = stac_validator.StacValidate(stac_file, collections=True)
+    stac.validate_collections()
+
+    assert stac.message == [
+        {
+            "version": "1.0.0-beta.2",
+            "path": "https://earth-search.aws.element84.com/v0/collections",
+            "schema": [
+                "https://schemas.stacspec.org/v1.0.0-beta.2/collection-spec/json-schema/collection.json"
+            ],
+            "valid_stac": True,
+            "asset_type": "COLLECTION",
+            "validation_method": "default",
+        },
+        {
+            "version": "1.0.0-beta.2",
+            "path": "https://earth-search.aws.element84.com/v0/collections",
+            "schema": [
+                "https://schemas.stacspec.org/v1.0.0-beta.2/collection-spec/json-schema/collection.json"
+            ],
+            "valid_stac": True,
+            "asset_type": "COLLECTION",
+            "validation_method": "default",
+        },
+        {
+            "version": "1.0.0-beta.2",
+            "path": "https://earth-search.aws.element84.com/v0/collections",
+            "schema": [
+                "https://schemas.stacspec.org/v1.0.0-beta.2/collection-spec/json-schema/collection.json"
+            ],
+            "valid_stac": True,
+            "asset_type": "COLLECTION",
+            "validation_method": "default",
+        },
+        {
+            "version": "1.0.0-beta.2",
+            "path": "https://earth-search.aws.element84.com/v0/collections",
+            "schema": [
+                "https://schemas.stacspec.org/v1.0.0-beta.2/collection-spec/json-schema/collection.json"
+            ],
+            "valid_stac": True,
+            "asset_type": "COLLECTION",
+            "validation_method": "default",
+        },
+    ]

tests/test_validate_item_collection.py

@@ -1,9 +1,8 @@
 """
-Description: Test the validator
+Description: Test stac-validator on item-collection validation.
 
 """
 
-
 from stac_validator import stac_validator