Merge pull request #1046 from dwinston/autodetect-parse-format

Autodetect parse() format

3 files changed, 55 insertions, 9 deletions

diff --git a/rdflib/graph.py b/rdflib/graph.py
index 12d18dce..145224b8 100644
--- a/rdflib/graph.py
+++ b/rdflib/graph.py
@@ -2,6 +2,8 @@ from __future__ import absolute_import
 from __future__ import division
 from __future__ import print_function
 
+from xml.sax import SAXParseException
+
 from rdflib.term import Literal  # required for doctests
 from rdflib.namespace import Namespace  # required for doctests
 
@@ -21,6 +23,7 @@ from rdflib.parser import create_input_source
 from rdflib.namespace import NamespaceManager
 from rdflib.resource import Resource
 from rdflib.collection import Collection
+import rdflib.util  # avoid circular dependency
 
 import os
 import shutil
@@ -31,6 +34,7 @@ from urllib.parse import urlparse
 
 assert Literal  # avoid warning
 assert Namespace  # avoid warning
+
 logger = logging.getLogger(__name__)
 
 
@@ -1066,13 +1070,24 @@ class Graph(Node):
         )
         if format is None:
             format = source.content_type
+        assumed_xml = False
         if format is None:
-            # raise Exception("Could not determine format for %r. You can" + \
-            # "expicitly specify one with the format argument." % source)
-            format = "application/rdf+xml"
+            if (hasattr(source, "file")
+                    and getattr(source.file, "name", None)
+                    and isinstance(source.file.name, str)):
+                format = rdflib.util.guess_format(source.file.name)
+            if format is None:
+                format = "application/rdf+xml"
+                assumed_xml = True
         parser = plugin.get(format, Parser)()
         try:
             parser.parse(source, self, **args)
+        except SAXParseException as saxpe:
+            if assumed_xml:
+                logger.warning(
+                    "Could not guess format for %r, so assumed xml."
+                    " You can explicitly specify format using the format argument." % source)
+            raise saxpe
         finally:
             if source.auto_close:
                 source.close()
diff --git a/rdflib/util.py b/rdflib/util.py
index 57b20915..92996ec7 100644
--- a/rdflib/util.py
+++ b/rdflib/util.py
@@ -47,8 +47,7 @@ from rdflib.exceptions import ContextTypeError
 from rdflib.exceptions import ObjectTypeError
 from rdflib.exceptions import PredicateTypeError
 from rdflib.exceptions import SubjectTypeError
-from rdflib.graph import Graph
-from rdflib.graph import QuotedGraph
+import rdflib.graph  # avoid circular dependency
 from rdflib.namespace import Namespace
 from rdflib.namespace import NamespaceManager
 from rdflib.term import BNode
@@ -161,7 +160,7 @@ def from_n3(s, default=None, backend=None, nsm=None):
         >>> from rdflib import RDFS
         >>> from_n3('rdfs:label') == RDFS['label']
         True
-        >>> nsm = NamespaceManager(Graph())
+        >>> nsm = NamespaceManager(rdflib.graph.Graph())
         >>> nsm.bind('dbpedia', 'http://dbpedia.org/resource/')
         >>> berlin = URIRef('http://dbpedia.org/resource/Berlin')
         >>> from_n3('dbpedia:Berlin', nsm=nsm) == berlin
@@ -207,16 +206,16 @@ def from_n3(s, default=None, backend=None, nsm=None):
         return Literal(int(s))
     elif s.startswith("{"):
         identifier = from_n3(s[1:-1])
-        return QuotedGraph(backend, identifier)
+        return rdflib.graph.QuotedGraph(backend, identifier)
     elif s.startswith("["):
         identifier = from_n3(s[1:-1])
-        return Graph(backend, identifier)
+        return rdflib.graph.Graph(backend, identifier)
     elif s.startswith("_:"):
         return BNode(s[2:])
     elif ":" in s:
         if nsm is None:
             # instantiate default NamespaceManager and rely on its defaults
-            nsm = NamespaceManager(Graph())
+            nsm = NamespaceManager(rdflib.graph.Graph())
         prefix, last_part = s.split(":", 1)
         ns = dict(nsm.namespaces())[prefix]
         return Namespace(ns)[last_part]
diff --git a/test/test_parse_file_guess_format.py b/test/test_parse_file_guess_format.py
new file mode 100644
index 00000000..abb039df
--- /dev/null
+++ b/test/test_parse_file_guess_format.py
@@ -0,0 +1,32 @@
+import unittest
+from pathlib import Path
+from shutil import copyfile
+from tempfile import TemporaryDirectory
+
+from xml.sax import SAXParseException
+
+from rdflib import Graph, logger as graph_logger
+
+
+class FileParserGuessFormatTest(unittest.TestCase):
+    def test_ttl(self):
+        g = Graph()
+        self.assertIsInstance(g.parse("test/w3c/turtle/IRI_subject.ttl"), Graph)
+
+    def test_n3(self):
+        g = Graph()
+        self.assertIsInstance(g.parse("test/n3/example-lots_of_graphs.n3"), Graph)
+
+    def test_warning(self):
+        g = Graph()
+        with TemporaryDirectory() as tmpdirname:
+            newpath = Path(tmpdirname).joinpath("no_file_ext")
+            copyfile("test/w3c/turtle/IRI_subject.ttl", str(newpath))
+            with self.assertLogs(graph_logger, "WARNING") as log_cm:
+                with self.assertRaises(SAXParseException):
+                    g.parse(str(newpath))
+            self.assertTrue(any("Could not guess format" in msg for msg in log_cm.output))
+
+
+if __name__ == '__main__':
+    unittest.main()