BREAKING CHANGE: eliminate rdflib.term.Genid and `rdflib.term.RDFLi…

…bGenid` TBD ... - Fixes <RDFLib#1828>
aucampia · Jun 19, 2023 · 07b43db · 07b43db
1 parent f278b86
commit 07b43db
Show file tree

Hide file tree

Showing 2 changed files with 51 additions and 58 deletions.
diff --git a/rdflib/graph.py b/rdflib/graph.py
@@ -1773,6 +1773,10 @@ def do_skolemize2(t: _TripleType) -> _TripleType:
     def de_skolemize(
         self, new_graph: Optional[Graph] = None, uriref: Optional[URIRef] = None
     ) -> Graph:
+        """
+        .. warning::
+            De-skolemization is not a well defined process.
+        """
         def do_de_skolemize(uriref: URIRef, t: _TripleType) -> _TripleType:
             (s, p, o) = t
             if s == uriref:
@@ -1788,19 +1792,19 @@ def do_de_skolemize(uriref: URIRef, t: _TripleType) -> _TripleType:
         def do_de_skolemize2(t: _TripleType) -> _TripleType:
             (s, p, o) = t
 
-            if RDFLibGenid._is_rdflib_skolem(s):
-                # type error: Argument 1 to "RDFLibGenid" has incompatible type "Node"; expected "str"
-                s = RDFLibGenid(s).de_skolemize()  # type: ignore[arg-type]
-            elif Genid._is_external_skolem(s):
-                # type error: Argument 1 to "Genid" has incompatible type "Node"; expected "str"
-                s = Genid(s).de_skolemize()  # type: ignore[arg-type]
-
-            if RDFLibGenid._is_rdflib_skolem(o):
-                # type error: Argument 1 to "RDFLibGenid" has incompatible type "Node"; expected "str"
-                o = RDFLibGenid(o).de_skolemize()  # type: ignore[arg-type]
-            elif Genid._is_external_skolem(o):
-                # type error: Argument 1 to "Genid" has incompatible type "Node"; expected "str"
-                o = Genid(o).de_skolemize()  # type: ignore[arg-type]
+            # if RDFLibGenid._is_rdflib_skolem(s):
+            #     # type error: Argument 1 to "RDFLibGenid" has incompatible type "Node"; expected "str"
+            #     s = RDFLibGenid(s).de_skolemize()  # type: ignore[arg-type]
+            # elif Genid._is_external_skolem(s):
+            #     # type error: Argument 1 to "Genid" has incompatible type "Node"; expected "str"
+            #     s = Genid(s).de_skolemize()  # type: ignore[arg-type]
+
+            # if RDFLibGenid._is_rdflib_skolem(o):
+            #     # type error: Argument 1 to "RDFLibGenid" has incompatible type "Node"; expected "str"
+            #     o = RDFLibGenid(o).de_skolemize()  # type: ignore[arg-type]
+            # elif Genid._is_external_skolem(o):
+            #     # type error: Argument 1 to "Genid" has incompatible type "Node"; expected "str"
+            #     o = Genid(o).de_skolemize()  # type: ignore[arg-type]
 
             return s, p, o
 

diff --git a/rdflib/term.py b/rdflib/term.py
@@ -20,6 +20,7 @@
 * Numerical Ranges
 
 """
+from __future__ import annotations
 import re
 from fractions import Fraction
 
@@ -57,7 +58,7 @@
     TypeVar,
     Union,
 )
-from urllib.parse import urldefrag, urljoin, urlparse
+from urllib.parse import urldefrag, urljoin, urlparse, urlsplit
 
 from isodate import (
     Duration,
@@ -79,9 +80,10 @@
 _SKOLEM_DEFAULT_AUTHORITY = "https://rdflib.github.io"
 
 logger = logging.getLogger(__name__)
-skolem_genid = "/.well-known/genid/"
-rdflib_skolem_genid = "/.well-known/genid/rdflib/"
-skolems: Dict[str, "BNode"] = {}
+_WELL_KNOWN_GENID = "/.well-known/genid/"
+_RDFLIB_GENID_SUFFIX = "rdflib/"
+_RDFLIB_GENID_PATH = "/.well-known/genid/" + _RDFLIB_GENID_SUFFIX
+# _skolems: Dict[str, "BNode"] = {}
 
 
 _invalid_uri_chars = '<>" {}|\\^`'
@@ -347,16 +349,19 @@ def __radd__(self, other) -> "URIRef":
     def __mod__(self, other) -> "URIRef":
         return self.__class__(str(self) % other)
 
-    def de_skolemize(self) -> "BNode":
-        """Create a Blank Node from a skolem URI, in accordance
-        with http://www.w3.org/TR/rdf11-concepts/#section-skolemization.
-        This function accepts only rdflib type skolemization, to provide
-        a round-tripping within the system.
-
-        .. versionadded:: 4.0
-        """
-        if isinstance(self, RDFLibGenid):
-            parsed_uri = urlparse("%s" % self)
+class _Deskolemizer:
+    def __init__(self) -> None:
+        self._skolems: Dict[str, BNode] = {}
+
+    def __call__(self, uri: str) -> Union[BNode, str]:
+        parsed_uri = urlsplit(uri)
+        if parsed_uri.query != "" or parsed_uri.fragment != "":
+            # Behaviour is undefined for skolem URIs with query or fragment, so just return the URI
+            return uri
+        if parsed_uri.path.startswith(_WELL_KNOWN_GENID):
+            genid_suffix = parsed_uri.path[len(_WELL_KNOWN_GENID) :]
+            if genid_suffix.startswith(_RDFLIB_GENID_SUFFIX):
+                genid_suffix = genid_suffix[len(_RDFLIB_GENID_SUFFIX) :]
             return BNode(value=parsed_uri.path[len(rdflib_skolem_genid) :])
         elif isinstance(self, Genid):
             bnode_id = "%s" % self
@@ -370,38 +375,22 @@ def de_skolemize(self) -> "BNode":
             raise Exception("<%s> is not a skolem URI" % self)
 
 
-class Genid(URIRef):
-    __slots__ = ()
-
-    @staticmethod
-    def _is_external_skolem(uri: Any) -> bool:
-        if not isinstance(uri, str):
-            uri = str(uri)
-        parsed_uri = urlparse(uri)
-        gen_id = parsed_uri.path.rfind(skolem_genid)
-        if gen_id != 0:
-            return False
-        return True
-
+# def _is_external_skolem(uri: str) -> bool:
+#     parsed_uri = urlparse(uri)
+#     gen_id = parsed_uri.path.rfind(skolem_genid)
+#     if gen_id != 0:
+#         return False
+#     return True
 
-class RDFLibGenid(Genid):
-    __slots__ = ()
 
-    @staticmethod
-    def _is_rdflib_skolem(uri: Any) -> bool:
-        if not isinstance(uri, str):
-            uri = str(uri)
-        parsed_uri = urlparse(uri)
-        if (
-            parsed_uri.params != ""
-            or parsed_uri.query != ""
-            or parsed_uri.fragment != ""
-        ):
-            return False
-        gen_id = parsed_uri.path.rfind(rdflib_skolem_genid)
-        if gen_id != 0:
-            return False
-        return True
+# def _is_rdflib_skolem(uri: str) -> bool:
+#     parsed_uri = urlparse(uri)
+#     if parsed_uri.params != "" or parsed_uri.query != "" or parsed_uri.fragment != "":
+#         return False
+#     gen_id = parsed_uri.path.rfind(rdflib_skolem_genid)
+#     if gen_id != 0:
+#         return False
+#     return True
 
 
 def _unique_id() -> str:
@@ -499,7 +488,7 @@ def skolemize(
         if authority is None:
             authority = _SKOLEM_DEFAULT_AUTHORITY
         if basepath is None:
-            basepath = rdflib_skolem_genid
+            basepath = _RDFLIB_SKOLEM_GENID
         skolem = "%s%s" % (basepath, str(self))
         return URIRef(urljoin(authority, skolem))