Address review comments

Author: john-dupuy

src/stac_api_validator/__main__.py

@@ -65,8 +65,10 @@
     help="Query parameter key and value to pass for authorization, e.g., 'key=xyz'.",
 )
 @click.option(
-    "--auth-headers",
-    help="Auth headers to attach to the request, e.g., {'Authorization': 'api-key <api-key>'}",
+    "-H",
+    "--headers",
+    multiple=True,
+    help="Headers to attach to the main request and dependent pystac requests, curl syntax",
 )
 def main(
     log_level: str,
@@ -76,20 +78,29 @@ def main(
     geometry: Optional[str],
     auth_bearer_token: Optional[str] = None,
     auth_query_parameter: Optional[str] = None,
-    auth_headers: Optional[str] = None,
+    headers: Optional[List[str]] = None,
 ) -> int:
     """STAC API Validator."""
     logging.basicConfig(stream=sys.stdout, level=log_level)
 
     try:
+        processed_headers = {}
+        if headers:
+            processed_headers.update(
+                {
+                    key.strip(): value.strip()
+                    for key, value in (header.split(":") for header in headers)
+                }
+            )
+
         (warnings, errors) = validate_api(
             root_url=root_url,
             ccs_to_validate=conformance_classes,
             collection=collection,
             geometry=geometry,
             auth_bearer_token=auth_bearer_token,
             auth_query_parameter=auth_query_parameter,
-            auth_headers=auth_headers,
+            headers=processed_headers,
         )
     except Exception as e:
         click.secho(

src/stac_api_validator/validations.py

@@ -1,5 +1,4 @@
 """Validations module."""
-import ast
 import itertools
 import json
 import logging
@@ -279,7 +278,8 @@ def is_geojson_type(maybe_type: Optional[str]) -> bool:
 def get_catalog(data_dict: Dict[str, Any], r_session: Session) -> Catalog:
     stac_io = StacIO.default()
     if r_session.headers and r_session.headers.get("Authorization"):
-        stac_io.headers = {"Authorization": str(r_session.headers["Authorization"])}
+        stac_io.headers = r_session.headers
+        stac_io.headers["Accept-Encoding"] = "*"
     catalog = Catalog.from_dict(data_dict)
     catalog._stac_io = stac_io
     return catalog
@@ -513,7 +513,7 @@ def validate_api(
     geometry: Optional[str],
     auth_bearer_token: Optional[str],
     auth_query_parameter: Optional[str],
-    auth_headers: Optional[str],
+    headers: Optional[Dict[str, str]],
 ) -> Tuple[Warnings, Errors]:
     warnings = Warnings()
     errors = Errors()
@@ -525,8 +525,8 @@ def validate_api(
     if auth_query_parameter and (xs := auth_query_parameter.split("=", 1)):
         r_session.params = {xs[0]: xs[1]}
 
-    if auth_headers:
-        r_session.headers.update(ast.literal_eval(auth_headers))
+    if headers:
+        r_session.headers.update(headers)
 
     _, landing_page_body, landing_page_headers = retrieve(
         Method.GET, root_url, errors, Context.CORE, r_session
@@ -596,9 +596,6 @@ def validate_api(
 
     if not errors:
         try:
-            headers = {}
-            if r_session.headers and r_session.headers.get("Authorization"):
-                headers["Authorization"] = str(r_session.headers["Authorization"])
             catalog = Client.open(root_url, headers=headers)
             catalog.validate()
             for child in catalog.get_children():
@@ -743,11 +740,7 @@ def validate_browseable(
                     link["href"],
                     errors,
                     Context.BROWSEABLE,
-                    params={
-                        "ids": item.id,
-                        "collections": getattr(item, "collection", None)
-                        or getattr(item, "collection_id", None),
-                    },
+                    params={"ids": item.id, "collections": item.collection_id},
                     r_session=r_session,
                 )
                 if body and len(body.get("features", [])) != 1:

tests/resources/sample-item.json

@@ -0,0 +1,53 @@
+{
+  "type": "Feature",
+  "stac_version": "1.0.0",
+  "id": "CS3-20160503_132131_05",
+  "properties": {
+    "datetime": "2016-05-03T13:22:30.040000Z",
+    "title": "A CS3 item",
+    "license": "PDDL-1.0",
+    "providers": [
+      {
+        "name": "CoolSat",
+        "roles": ["producer", "licensor"],
+        "url": "https://cool-sat.com/"
+      }
+    ]
+  },
+  "geometry": {
+    "type": "Polygon",
+    "coordinates": [
+      [
+        [-122.308150179, 37.488035566],
+        [-122.597502109, 37.538869539],
+        [-122.576687533, 37.613537207],
+        [-122.2880486, 37.562818007],
+        [-122.308150179, 37.488035566]
+      ]
+    ]
+  },
+  "links": [
+    {
+      "rel": "collection",
+      "href": "https://raw.githubusercontent.com/radiantearth/stac-spec/v0.8.1/collection-spec/examples/sentinel2.json"
+    }
+  ],
+  "assets": {
+    "analytic": {
+      "href": "http://cool-sat.com/catalog/CS3-20160503_132130_04/analytic.tif",
+      "title": "4-Band Analytic",
+      "product": "http://cool-sat.com/catalog/products/analytic.json",
+      "type": "image/tiff; application=geotiff; profile=cloud-optimized",
+      "roles": ["data", "analytic"]
+    },
+    "thumbnail": {
+      "href": "http://cool-sat.com/catalog/CS3-20160503_132130_04/thumbnail.png",
+      "title": "Thumbnail",
+      "type": "image/png",
+      "roles": ["thumbnail"]
+    }
+  },
+  "bbox": [-122.59750209, 37.48803556, -122.2880486, 37.613537207],
+  "stac_extensions": [],
+  "collection": "CS3"
+}

tests/test_main.py

@@ -26,7 +26,14 @@ def test_retrieve_called_with_auth_headers(
             "The import hook that typeguard uses seems to break the mock below."
         )
 
-    expected_auth_header = {"Authorization": "api-key fake-api-key-value"}
+    expected_headers = {
+        "User-Agent": "python-requests/2.28.2",
+        "Accept-Encoding": "gzip, deflate",
+        "Accept": "*/*",
+        "Connection": "keep-alive",
+        "Authorization": "api-key fake-api-key-value",
+    }
+
     with unittest.mock.patch(
         "stac_api_validator.validations.retrieve"
     ) as retrieve_mock:
@@ -37,12 +44,10 @@ def test_retrieve_called_with_auth_headers(
                 "https://invalid",
                 "--conformance",
                 "core",
-                "--auth-headers",
-                f"{expected_auth_header}",
+                "-H",
+                "Authorization: api-key fake-api-key-value",
             ],
         )
         assert retrieve_mock.call_count == 1
         r_session = retrieve_mock.call_args.args[-1]
-        assert (
-            r_session.headers["Authorization"] == expected_auth_header["Authorization"]
-        )
+        assert r_session.headers == expected_headers

tests/test_validations.py

@@ -0,0 +1,129 @@
+"""
+Test cases for the 'validations' module
+"""
+import json
+import os
+import pathlib
+import unittest.mock
+from copy import copy
+
+import pystac
+import pytest
+import requests
+
+from stac_api_validator import validations
+
+
+@pytest.fixture
+def r_session():
+    yield requests.Session()
+
+
+@pytest.fixture
+def catalog_dict():
+    current_path = pathlib.Path(os.path.dirname(os.path.abspath(__file__)))
+
+    with open(current_path / "resources" / "landing_page.json") as f:
+        # Load the contents of the file into a Python dictionary
+        data = json.load(f)
+
+    yield data
+
+
+@pytest.fixture
+def sample_item():
+    current_path = pathlib.Path(os.path.dirname(os.path.abspath(__file__)))
+
+    with open(current_path / "resources" / "sample-item.json") as f:
+        # Load the contents of the file into a Python dictionary
+        data = json.load(f)
+
+    yield pystac.Item.from_dict(data)
+
+
+@pytest.fixture
+def expected_headers():
+    yield {
+        "User-Agent": "python-requests/2.28.2",
+        "Accept-Encoding": "gzip, deflate",
+        "Accept": "*/*",
+        "Connection": "keep-alive",
+        "Authorization": "api-key fake-api-key-value",
+    }
+
+
+def test_get_catalog(r_session, catalog_dict, expected_headers):
+    r_session.headers = copy(expected_headers)
+    expected_headers.update({"Accept-Encoding": "*"})
+
+    catalog = validations.get_catalog(catalog_dict, r_session)
+    assert catalog._stac_io.headers == expected_headers
+
+
+def test_retrieve(r_session, expected_headers):
+    headers = {"Authorization": "api-key fake-api-key-value"}
+    r_session.send = unittest.mock.MagicMock()
+    r_session.send.status_code = 500
+
+    validations.retrieve(
+        validations.Method.GET,
+        "https://invalid",
+        validations.Errors(),
+        validations.Context.CORE,
+        r_session=r_session,
+        headers=headers,
+    )
+    assert r_session.send.call_count == 1
+    prepared_request_headers = r_session.send.call_args_list[0].args[0].headers
+    assert prepared_request_headers == expected_headers
+
+
+def test_validate_api(request, r_session, expected_headers):
+    if request.config.getoption("typeguard_packages"):
+        pytest.skip(
+            "The import hook that typeguard uses seems to break the mock below."
+        )
+    headers = {"Authorization": "api-key fake-api-key-value"}
+
+    with unittest.mock.patch(
+        "stac_api_validator.validations.retrieve"
+    ) as retrieve_mock:
+        retrieve_mock.return_value = None, None, None
+        validations.validate_api(
+            "https://invalid",
+            ccs_to_validate=["core"],
+            collection=None,
+            geometry=None,
+            auth_bearer_token=None,
+            auth_query_parameter=None,
+            headers=headers,
+        )
+        assert retrieve_mock.call_count == 1
+        r_session = retrieve_mock.call_args.args[-1]
+        assert r_session.headers == expected_headers
+
+
+def test_validate_browseable(
+    request, r_session, catalog_dict, sample_item, expected_headers
+):
+    if request.config.getoption("typeguard_packages"):
+        pytest.skip(
+            "The import hook that typeguard uses seems to break the mock below."
+        )
+
+    r_session.headers = copy(expected_headers)
+
+    with unittest.mock.patch(
+        "stac_api_validator.validations.get_catalog"
+    ) as get_catalog_mock:
+        get_catalog_mock.get_all_items.return_value = [sample_item]
+
+        validations.validate_browseable(
+            catalog_dict,
+            errors=validations.Errors(),
+            warnings=validations.Warnings(),
+            r_session=r_session,
+        )
+        assert get_catalog_mock.call_count == 1
+        session_from_mock = get_catalog_mock.call_args.args[-1]
+        assert session_from_mock.headers == expected_headers