Make Result.serialize work more like Graph.serialize

This patch makes the following changes to `Result.serialize`.

* Return str by default instead of bytes.
* Use "txt" as the default tabular serialization format.
* Use "turtle" as the default graph serialization format.
* Support both typing.IO[bytes] and typing.TextIO destinations.

Corresponding changes are made to the specific serializers also.

This patch also changes how text is written to typing.IO[bytes] in
serializers to ensure that the buffer is flushed and
detatched from the TextIOWrapper once the serialization function
completes so it can be used normally afterwards.

This patch further includes a bunch of additional type hints.

Author: aucampia

rdflib/graph.py

@@ -16,6 +16,7 @@
 import sys
 
 from io import BytesIO, TextIOBase, TextIOWrapper, StringIO, BufferedIOBase
+from typing import Optional, Union
 
 from urllib.request import Request
 from urllib.request import url2pathname
@@ -44,7 +45,7 @@ class Parser(object):
     def __init__(self):
         pass
 
-    def parse(self, source, sink):
+    def parse(self, source, sink, **args):
         pass
 
 
@@ -214,7 +215,12 @@ def __repr__(self):
 
 
 def create_input_source(
-    source=None, publicID=None, location=None, file=None, data=None, format=None
+    source=None,
+    publicID=None,
+    location=None,
+    file=None,
+    data: Optional[Union[str, bytes, bytearray]] = None,
+    format=None,
 ):
     """
     Return an appropriate InputSource instance for the given

rdflib/parser.py

@@ -36,7 +36,21 @@
     UpdateProcessor,
 )
 from rdflib.exceptions import Error
-from typing import Type, TypeVar
+from typing import (
+    TYPE_CHECKING,
+    Any,
+    Dict,
+    Generic,
+    Iterator,
+    Optional,
+    Tuple,
+    Type,
+    TypeVar,
+    overload,
+)
+
+if TYPE_CHECKING:
+    from pkg_resources import EntryPoint
 
 __all__ = ["register", "get", "plugins", "PluginException", "Plugin", "PKGPlugin"]
 
@@ -51,42 +65,47 @@
     "rdf.plugins.updateprocessor": UpdateProcessor,
 }
 
-_plugins = {}
+_plugins: Dict[Tuple[str, Type[Any]], "Plugin"] = {}
 
 
 class PluginException(Error):
     pass
 
 
-class Plugin(object):
-    def __init__(self, name, kind, module_path, class_name):
+PluginT = TypeVar("PluginT")
+
+
+class Plugin(Generic[PluginT]):
+    def __init__(
+        self, name: str, kind: Type[PluginT], module_path: str, class_name: str
+    ):
         self.name = name
         self.kind = kind
         self.module_path = module_path
         self.class_name = class_name
-        self._class = None
+        self._class: Optional[Type[PluginT]] = None
 
-    def getClass(self):
+    def getClass(self) -> Type[PluginT]:
         if self._class is None:
             module = __import__(self.module_path, globals(), locals(), [""])
             self._class = getattr(module, self.class_name)
         return self._class
 
 
-class PKGPlugin(Plugin):
-    def __init__(self, name, kind, ep):
+class PKGPlugin(Plugin[PluginT]):
+    def __init__(self, name: str, kind: Type[PluginT], ep: "EntryPoint"):
         self.name = name
         self.kind = kind
         self.ep = ep
-        self._class = None
+        self._class: Optional[Type[PluginT]] = None
 
-    def getClass(self):
+    def getClass(self) -> Type[PluginT]:
         if self._class is None:
             self._class = self.ep.load()
         return self._class
 
 
-def register(name: str, kind, module_path, class_name):
+def register(name: str, kind: Type[Any], module_path, class_name):
     """
     Register the plugin for (name, kind). The module_path and
     class_name should be the path to a plugin class.
@@ -95,16 +114,13 @@ def register(name: str, kind, module_path, class_name):
     _plugins[(name, kind)] = p
 
 
-PluginT = TypeVar("PluginT")
-
-
 def get(name: str, kind: Type[PluginT]) -> Type[PluginT]:
     """
     Return the class for the specified (name, kind). Raises a
     PluginException if unable to do so.
     """
     try:
-        p = _plugins[(name, kind)]
+        p: Plugin[PluginT] = _plugins[(name, kind)]
     except KeyError:
         raise PluginException("No plugin registered for (%s, %s)" % (name, kind))
     return p.getClass()
@@ -121,7 +137,21 @@ def get(name: str, kind: Type[PluginT]) -> Type[PluginT]:
             _plugins[(ep.name, kind)] = PKGPlugin(ep.name, kind, ep)
 
 
-def plugins(name=None, kind=None):
+@overload
+def plugins(
+    name: Optional[str] = ..., kind: Type[PluginT] = ...
+) -> Iterator[Plugin[PluginT]]:
+    ...
+
+
+@overload
+def plugins(name: Optional[str] = ..., kind: None = ...) -> Iterator[Plugin]:
+    ...
+
+
+def plugins(
+    name: Optional[str] = None, kind: Optional[Type[PluginT]] = None
+) -> Iterator[Plugin]:
     """
     A generator of the plugins.
 

rdflib/plugin.py

@@ -41,6 +41,7 @@
 from rdflib.graph import Graph
 from rdflib.term import URIRef, Literal, BNode
 from rdflib.namespace import RDF, XSD
+from typing import IO, Optional
 
 from ..shared.jsonld.context import Context, UNDEF
 from ..shared.jsonld.util import json
@@ -53,10 +54,16 @@
 
 
 class JsonLDSerializer(Serializer):
-    def __init__(self, store):
+    def __init__(self, store: Graph):
         super(JsonLDSerializer, self).__init__(store)
 
-    def serialize(self, stream, base=None, encoding=None, **kwargs):
+    def serialize(
+        self,
+        stream: IO[bytes],
+        base: Optional[str] = None,
+        encoding: Optional[str] = None,
+        **kwargs
+    ):
         # TODO: docstring w. args and return value
         encoding = encoding or "utf-8"
         if encoding not in ("utf-8", "utf-16"):

rdflib/plugins/serializers/jsonld.py

@@ -14,7 +14,7 @@ class N3Serializer(TurtleSerializer):
 
     short_name = "n3"
 
-    def __init__(self, store, parent=None):
+    def __init__(self, store: Graph, parent=None):
         super(N3Serializer, self).__init__(store)
         self.keywords.update({OWL.sameAs: "=", SWAP_LOG.implies: "=>"})
         self.parent = parent
@@ -109,7 +109,7 @@ def p_clause(self, node, position):
             self.write("{")
             self.depth += 1
             serializer = N3Serializer(node, parent=self)
-            serializer.serialize(self.stream)
+            serializer.serialize(self.stream.buffer)
             self.depth -= 1
             self.write(self.indent() + "}")
             return True

rdflib/plugins/serializers/n3.py

@@ -1,5 +1,7 @@
+from typing import IO, Optional
 import warnings
 
+from rdflib.graph import ConjunctiveGraph, Graph
 from rdflib.term import Literal
 from rdflib.serializer import Serializer
 
@@ -9,15 +11,22 @@
 
 
 class NQuadsSerializer(Serializer):
-    def __init__(self, store):
+    def __init__(self, store: Graph):
         if not store.context_aware:
             raise Exception(
                 "NQuads serialization only makes " "sense for context-aware stores!"
             )
 
         super(NQuadsSerializer, self).__init__(store)
-
-    def serialize(self, stream, base=None, encoding=None, **args):
+        self.store: ConjunctiveGraph
+
+    def serialize(
+        self,
+        stream: IO[bytes],
+        base: Optional[str] = None,
+        encoding: Optional[str] = None,
+        **args
+    ):
         if base is not None:
             warnings.warn("NQuadsSerializer does not support base.")
         if encoding is not None and encoding.lower() != self.encoding.lower():

rdflib/plugins/serializers/nquads.py

@@ -3,12 +3,17 @@
 See <http://www.w3.org/TR/rdf-testcases/#ntriples> for details about the
 format.
 """
+from typing import IO, Optional
+
+from rdflib.graph import Graph
 from rdflib.term import Literal
 from rdflib.serializer import Serializer
 
 import warnings
 import codecs
 
+from rdflib.util import as_textio
+
 __all__ = ["NTSerializer"]
 
 
@@ -17,19 +22,32 @@ class NTSerializer(Serializer):
     Serializes RDF graphs to NTriples format.
     """
 
-    def __init__(self, store):
+    def __init__(self, store: Graph):
         Serializer.__init__(self, store)
         self.encoding = "ascii"  # n-triples are ascii encoded
 
-    def serialize(self, stream, base=None, encoding=None, **args):
+    def serialize(
+        self,
+        stream: IO[bytes],
+        base: Optional[str] = None,
+        encoding: Optional[str] = None,
+        **args
+    ):
         if base is not None:
             warnings.warn("NTSerializer does not support base.")
         if encoding is not None and encoding.lower() != self.encoding.lower():
             warnings.warn("NTSerializer does not use custom encoding.")
         encoding = self.encoding
-        for triple in self.store:
-            stream.write(_nt_row(triple).encode(self.encoding, "_rdflib_nt_escape"))
-        stream.write("\n".encode("latin-1"))
+
+        with as_textio(
+            stream,
+            encoding=self.encoding,
+            errors="_rdflib_nt_escape",
+            write_through=True,
+        ) as text_stream:
+            for triple in self.store:
+                text_stream.write(_nt_row(triple))
+            text_stream.write("\n")
 
 
 class NT11Serializer(NTSerializer):
@@ -39,7 +57,7 @@ class NT11Serializer(NTSerializer):
     Exactly like nt - only utf8 encoded.
     """
 
-    def __init__(self, store):
+    def __init__(self, store: Graph):
         Serializer.__init__(self, store)  # default to utf-8