change scheme strategy

snakebids/tests/strategies.py

@@ -37,22 +37,24 @@
 valid_entities: tuple[str, ...] = tuple(BidsConfig.load("bids").entities.keys())
 
 path_characters = st.characters(blacklist_characters=["/", "\x00"], codec="UTF-8")
-# see StackOverflow
-# https://stackoverflow.com/questions/1547899/which-characters-make-a-url-invalid#1547940
-scheme_characters_str = (
-    "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-._~?#[]@!$&'()*+,;="
-)
-scheme_characters = st.sampled_from(scheme_characters_str)
 
 
 def nothing() -> Any:
     return st.nothing()  # type: ignore
 
 
-def schemes() -> st.SearchStrategy[str | None]:  # generate the prefix part of a url
+def schemes() -> st.SearchStrategy[str | None]:
+    # Generate the prefix part of a url.
+    #
+    # we only sample from a fixed set of schemes. If schemes are randomly generated
+    # then something like file:// could be a problem because it may be either an
+    # abs or relative path and will fail the os.path.isabs() code
+    #
+    # Note for its current status we don't guarantee the following schemes to work
+    # but only requiring them to be retained correctly i.e. not from 'gs://' to
+    # 'gs:/' when fed into generate_inputs() and path expansion
     fixed = st.sampled_from(["C://", "c:\\", "gs://", "s3://"])
-    random = st.text(scheme_characters, min_size=1, max_size=5)
-    return st.one_of(st.none(), random, fixed)
+    return st.one_of(st.none(), fixed)
 
 
 def paths(

snakebids/tests/test_generate_inputs.py

@@ -1915,35 +1915,48 @@ def scheme_bids_component_strategy(draw, **kwargs):
 
 class TestParseBidsPath:
     @given(
-        component=scheme_bids_component_strategy(max_values=1, restrict_patterns=True)
+        component=sb_st.bids_components(max_values=1, restrict_patterns=True),
+        scheme=sb_st.schemes(),
     )
-    def test_splits_wildcards_from_path(self, component: BidsComponent):
+    def test_splits_wildcards_from_path(
+        self, component: BidsComponent, scheme: str | None
+    ):
         path = component.expand()[0]
+        if scheme is not None:
+            path = f"{scheme}{path}"
         entities = [BidsEntity.normalize(e).entity for e in component.zip_lists]
         tpl_path, matches = _parse_bids_path(path, entities)
         assert tpl_path.format(**matches) == path
 
     @given(
-        component=scheme_bids_component_strategy(max_values=1, restrict_patterns=True)
+        component=sb_st.bids_components(max_values=1, restrict_patterns=True),
+        scheme=sb_st.schemes(),
     )
-    def test_one_match_found_for_each_entity(self, component: BidsComponent):
+    def test_one_match_found_for_each_entity(
+        self, component: BidsComponent, scheme: str | None
+    ):
         path = component.expand()[0]
+        if scheme is not None:
+            path = f"{scheme}{path}"
         entities = [BidsEntity.normalize(e).entity for e in component.zip_lists]
         _, matches = _parse_bids_path(path, entities)
         assert set(matches.items()) == {
             (key, val[0]) for key, val in component.zip_lists.items()
         }
 
     @given(
-        component=scheme_bids_component_strategy(
+        component=sb_st.bids_components(
             max_values=1, restrict_patterns=True, extra_entities=False
         ),
+        scheme=sb_st.schemes(),
         entity=sb_st.bids_entity(),
     )
     def test_missing_match_leads_to_error(
-        self, component: BidsComponent, entity: BidsEntity
+        self, component: BidsComponent, scheme: str | None, entity: BidsEntity
     ):
         path = component.expand()[0]
+        if scheme is not None:
+            path = f"{scheme}{path}"
         entities = [BidsEntity.normalize(e).entity for e in component.zip_lists]
         assume(entity.entity not in entities)
         with pytest.raises(BidsParseError) as err: