Remove incorrect test, throw error when base IRI is invalid and improve parsing code

Author: mielvds

lib/pyld/iri_resolver.py

@@ -222,10 +222,14 @@ def unresolve(absolute_iri: str, base_iri: str = ""):
     """
     # TODO: better sync with jsonld.js version
     # skip IRI processing
-    if base_iri is None:
+    if not base_iri:
         return absolute_iri
 
     base = parse_url(base_iri)
+
+    if not base.scheme:
+        raise ValueError(f"Found invalid baseIRI '{base_iri}' for value '{absolute_iri}'")
+    
     rel = parse_url(absolute_iri)
 
     # schemes and network locations (authorities) don't match, don't alter IRI
@@ -264,7 +268,8 @@ def parse_url(url):
     m = re.match(p, url)
     # remove default http and https ports
     g = list(m.groups())
-    if ((g[0] == 'https' and g[1].endswith(':443')) or
+    
+    if g[1] is not None and ((g[0] == 'https' and g[1].endswith(':443')) or
             (g[0] == 'http' and g[1].endswith(':80'))):
         g[1] = g[1][:g[1].rfind(':')]
     return ParsedUrl(*g)

tests/test_iri_resolver.py

@@ -302,16 +302,13 @@ def test_hash_absolute(self):
     def test_colon_in_value_ignores_base(self):
         assert unresolve('http:abc', 'http://base.org/') == 'http:abc'
 
-    def test_colon_in_value_removes_dots(self):
-        assert unresolve('http://abc/../../', 'http://base.org/') == 'http://abc/'
-
-    # def test_non_absolute_base_error(self):
-    #     with pytest.raises(ValueError, match=r"Found invalid baseIRI 'def' for value 'abc'"):
-    #         unresolve('abc', 'def')
+    def test_non_absolute_base_error(self):
+        with pytest.raises(ValueError, match=r"Found invalid baseIRI 'def' for value 'http://base.org/abc'"):
+            unresolve('http://base.org/abc', 'def')
 
-    # def test_non_absolute_base_empty_value_error(self):
-    #     with pytest.raises(ValueError, match=r"Found invalid baseIRI 'def' for value ''"):
-    #         unresolve('', 'def')
+    def test_non_absolute_base_empty_value_error(self):
+        with pytest.raises(ValueError, match=r"Found invalid baseIRI 'def' for value ''"):
+            unresolve('', 'def')
 
     def test_base_without_path_slash(self):
         assert unresolve('http://base.org/abc', 'http://base.org') == 'abc'