JSONLDParser now uses resolver directly

We pass an explicit resolver around, defaulting to
`get_default_resolver()` if not specified.

`source_to_json()` has been removed as it uses `create_input_source()`,
which trusts all locations passed to it. Instead, we parse json
explicitly in the two places where it was called.

Resolves #1369.

rdflib/plugins/parsers/jsonld.py

@@ -32,17 +32,16 @@
 
 # NOTE: This code reads the entire JSON object into memory before parsing, but
 # we should consider streaming the input to deal with arbitrarily large graphs.
-
+import contextlib
 import warnings
 from rdflib.graph import ConjunctiveGraph
-from rdflib.parser import Parser, URLInputSource
+from rdflib.parser import Parser, URLInputSource, create_input_source
 from rdflib.namespace import RDF, XSD
 from rdflib.term import URIRef, BNode, Literal
 
 from ..shared.jsonld.context import Context, Term, UNDEF
 from ..shared.jsonld.util import (
     json,
-    source_to_json,
     VOCAB_DELIMS,
     context_from_urlinputsource,
 )
@@ -69,6 +68,8 @@
 
 
 # Add jsonld suffix so RDFLib can guess format from file name
+from ...resolver import get_default_resolver
+
 try:
     from rdflib.util import SUFFIX_FORMAT_MAP
 
@@ -108,7 +109,9 @@ def parse(self, source, sink, **kwargs):
 
         generalized_rdf = kwargs.get("generalized_rdf", False)
 
-        data = source_to_json(source)
+        input_source = create_input_source(source)
+        with contextlib.closing(input_source.getByteStream()) as stream:
+            data = json.load(stream)
 
         # NOTE: A ConjunctiveGraph parses into a Graph sink, so no sink will be
         # context_aware. Keeping this check in case RDFLib is changed, or
@@ -118,7 +121,15 @@ def parse(self, source, sink, **kwargs):
         else:
             conj_sink = sink
 
-        to_rdf(data, conj_sink, base, context_data, version, generalized_rdf)
+        to_rdf(
+            data,
+            conj_sink,
+            base,
+            context_data,
+            version,
+            generalized_rdf,
+            resolver=self.resolver,
+        )
 
 
 def to_rdf(
@@ -129,15 +140,18 @@ def to_rdf(
     version=None,
     generalized_rdf=False,
     allow_lists_of_lists=None,
+    resolver=None,
 ):
+    resolver = resolver or get_default_resolver()
+
     # TODO: docstring w. args and return value
-    context = Context(base=base, version=version)
+    context = Context(base=base, version=version, resolver=resolver)
     if context_data:
         context.load(context_data)
     parser = Parser(
         generalized_rdf=generalized_rdf, allow_lists_of_lists=allow_lists_of_lists
     )
-    return parser.parse(data, context, dataset)
+    return parser.parse(data, context, dataset, resolver=resolver)
 
 
 class Parser(object):
@@ -149,7 +163,7 @@ def __init__(self, generalized_rdf=False, allow_lists_of_lists=None):
             else ALLOW_LISTS_OF_LISTS
         )
 
-    def parse(self, data, context, dataset):
+    def parse(self, data, context, dataset, resolver):
         topcontext = False
 
         if isinstance(data, list):

rdflib/plugins/serializers/jsonld.py

@@ -119,7 +119,7 @@ def from_rdf(
         context = context_data
         context_data = context.to_dict()
     else:
-        context = Context(context_data, base=base)
+        context = Context(context_data, base=base, resolver=None)
 
     converter = Converter(context, use_native_types, use_rdf_type)
     result = converter.convert(graph)

rdflib/plugins/shared/jsonld/context.py

@@ -6,7 +6,8 @@
 
 """
 # https://github.com/RDFLib/rdflib-jsonld/blob/feature/json-ld-1.1/rdflib_jsonld/context.py
-
+import contextlib
+import json
 from collections import namedtuple
 from rdflib.namespace import RDF
 
@@ -35,7 +36,7 @@
     VOCAB,
 )
 from .errors import INVALID_REMOTE_CONTEXT, RECURSIVE_CONTEXT_INCLUSION
-from .util import source_to_json, urljoin, urlsplit, split_iri, norm_url
+from .util import urljoin, urlsplit, split_iri, norm_url
 
 
 NODE_KEYS = {GRAPH, ID, INCLUDED, JSON, LIST, NEST, NONE, REV, SET, TYPE, VALUE, LANG}
@@ -52,7 +53,7 @@ class Defined(int):
 
 
 class Context(object):
-    def __init__(self, source=None, base=None, version=None):
+    def __init__(self, source=None, base=None, version=None, *, resolver=None):
         self.version = version or 1.0
         self.language = None
         self.vocab = None
@@ -67,6 +68,7 @@ def __init__(self, source=None, base=None, version=None):
         self.parent = None
         self.propagate = True
         self._context_cache = {}
+        self.resolver = resolver
         if source:
             self.load(source)
 
@@ -93,7 +95,7 @@ def subcontext(self, source, propagate=True):
         return parent._subcontext(source, propagate)
 
     def _subcontext(self, source, propagate):
-        ctx = Context(version=self.version)
+        ctx = Context(version=self.version, resolver=self.resolver)
         ctx.propagate = propagate
         ctx.parent = self
         ctx.language = self.language
@@ -406,7 +408,9 @@ def _fetch_context(self, source, base, referenced_contexts):
         if source_url in self._context_cache:
             return self._context_cache[source_url]
 
-        source = source_to_json(source_url)
+        input_source = self.resolver.resolve(source_url, format="json-ld")
+        with contextlib.closing(input_source.getByteStream()) as stream:
+            source = json.load(stream)
         if source and CONTEXT not in source:
             raise INVALID_REMOTE_CONTEXT
         self._context_cache[source_url] = source

rdflib/plugins/shared/jsonld/util.py

@@ -18,17 +18,6 @@
 from io import StringIO
 
 
-def source_to_json(source):
-    # TODO: conneg for JSON (fix support in rdflib's URLInputSource!)
-    source = create_input_source(source, format="json-ld")
-
-    stream = source.getByteStream()
-    try:
-        return json.load(StringIO(stream.read().decode("utf-8")))
-    finally:
-        stream.close()
-
-
 VOCAB_DELIMS = ("#", "/", ":")
 
 

test/jsonld/runner.py

@@ -8,6 +8,7 @@
 
 # monkey-patch N-Quads parser via it's underlying W3CNTriplesParser to keep source bnode id:s ..
 from rdflib.plugins.parsers.ntriples import W3CNTriplesParser, r_nodeid, bNode
+from rdflib.resolver import PermissiveResolver
 
 
 def _preserving_nodeid(self, bnode_context=None):
@@ -29,6 +30,7 @@ def do_test_json(suite_base, cat, num, inputpath, expectedpath, context, options
         base=input_uri,
         context_data=context,
         generalized_rdf=True,
+        resolver=PermissiveResolver(),
     )
     expected_json = _load_json(expectedpath)
     use_native_types = True  # CONTEXT in input_obj
@@ -79,6 +81,7 @@ def do_test_parser(suite_base, cat, num, inputpath, expectedpath, context, optio
         base=options.get("base", input_uri),
         version=version,
         generalized_rdf=options.get("produceGeneralizedRdf", False),
+        resolver=PermissiveResolver(),
     )
     assert isomorphic(result_graph, expected_graph), "Expected:\n%s\nGot:\n%s" % (
         expected_graph.serialize(),

test/jsonld/test_context.py

@@ -1,14 +1,18 @@
 """
 JSON-LD Context Spec
 """
-
+import json
 from functools import wraps
+
+from rdflib.parser import StringInputSource
 from rdflib.plugins.shared.jsonld.context import Context, Term
 from rdflib.plugins.shared.jsonld import context
 from rdflib.plugins.shared.jsonld import errors
+from rdflib.resolver import Resolver, url_resolver
 
 
 # exception utility (see also nose.tools.raises)
+
 def _expect_exception(expected_error):
     def _try_wrapper(f):
         @wraps(f)
@@ -128,85 +132,79 @@ def test_prefix_like_vocab():
     assert term.id == "ex:term"
 
 
-# Mock external sources loading
-SOURCES = {}
-_source_to_json = context.source_to_json
-
+class MockContextResolver(Resolver):
+    def __init__(self, contexts):
+        self.contexts = contexts
 
-def _mock_source_loader(f):
-    @wraps(f)
-    def _wrapper():
-        context.source_to_json = SOURCES.get
-        f()
-        context.source_to_json = _source_to_json
+    def is_resolution_allowed(self, scheme: str, location: str) -> bool:
+        return str(location) in self.contexts
 
-    return _wrapper
+    @url_resolver(schemes={'http', 'https'})
+    def resolve_http(self, url, format, scheme, trust=False):
+        return StringInputSource(
+            json.dumps(self.contexts[str(url)]),
+            system_id=str(url),
+        )
 
 
-@_mock_source_loader
 def test_loading_contexts():
     # Given context data:
     source1 = "http://example.org/base.jsonld"
     source2 = "http://example.org/context.jsonld"
-    SOURCES[source1] = {"@context": {"@vocab": "http://example.org/vocab/"}}
-    SOURCES[source2] = {"@context": [source1, {"n": "name"}]}
+    SOURCES = {
+        source1: {"@context": {"@vocab": "http://example.org/vocab/"}},
+        source2: {"@context": [source1, {"n": "name"}]},
+    }
 
     # Create a context:
-    ctx = Context(source2)
+    ctx = Context(source2, resolver=MockContextResolver(SOURCES))
     assert ctx.expand("n") == "http://example.org/vocab/name"
 
     # Context can be a list:
-    ctx = Context([source2])
+    ctx = Context([source2], resolver=MockContextResolver(SOURCES))
     assert ctx.expand("n") == "http://example.org/vocab/name"
 
 
-@_mock_source_loader
 def test_use_base_in_local_context():
     ctx = Context({"@base": "/local"})
     assert ctx.base == "/local"
 
 
-@_mock_source_loader
 def test_override_base():
     ctx = Context(
         base="http://example.org/app/data/item", source={"@base": "http://example.org/"}
     )
     assert ctx.base == "http://example.org/"
 
 
-@_mock_source_loader
 def test_resolve_relative_base():
     ctx = Context(base="http://example.org/app/data/item", source={"@base": "../"})
     assert ctx.base == "http://example.org/app/"
     assert ctx.resolve_iri("../other") == "http://example.org/other"
 
 
-@_mock_source_loader
 def test_set_null_base():
     ctx = Context(base="http://example.org/app/data/item", source={"@base": None})
     assert ctx.base is None
     assert ctx.resolve_iri("../other") == "../other"
 
 
-@_mock_source_loader
 def test_ignore_base_remote_context():
     ctx_url = "http://example.org/remote-base.jsonld"
-    SOURCES[ctx_url] = {"@context": {"@base": "/remote"}}
-    ctx = Context(ctx_url)
+    SOURCES = {ctx_url: {"@context": {"@base": "/remote"}}}
+    ctx = Context(ctx_url, resolver=MockContextResolver(SOURCES))
     assert ctx.base == None
 
 
 @_expect_exception(errors.RECURSIVE_CONTEXT_INCLUSION)
-@_mock_source_loader
 def test_recursive_context_inclusion_error():
     ctx_url = "http://example.org/recursive.jsonld"
-    SOURCES[ctx_url] = {"@context": ctx_url}
-    ctx = Context(ctx_url)
+    SOURCES = {ctx_url: {"@context": ctx_url}}
+    ctx = Context(ctx_url, resolver=MockContextResolver(SOURCES))
 
 
 @_expect_exception(errors.INVALID_REMOTE_CONTEXT)
-@_mock_source_loader
 def test_invalid_remote_context():
     ctx_url = "http://example.org/recursive.jsonld"
-    SOURCES[ctx_url] = {"key": "value"}
-    ctx = Context(ctx_url)
+    SOURCES = {ctx_url: {"key": "value"}}
+    ctx = Context(ctx_url, resolver=MockContextResolver(SOURCES))