Fix for Issue #724

rdflib/graph.py

@@ -389,6 +389,7 @@ def add(self, triple):
         assert isinstance(o, Node), "Object %s must be an rdflib term" % (o,)
         self.__store.add((s, p, o), self, quoted=False)
 
+
     def addN(self, quads):
         """Add a sequence of triple with context"""
 
@@ -989,6 +990,7 @@ def parse(
         location=None,
         file=None,
         data=None,
+        ignore_errors=False,
         **args
     ):
         """
@@ -1055,7 +1057,7 @@ def parse(
         >>> os.remove(file_name)
 
         """
-
+        errorFlag = ignore_errors
         source = create_input_source(
             source=source,
             publicID=publicID,
@@ -1072,14 +1074,15 @@ def parse(
             format = "application/rdf+xml"
         parser = plugin.get(format, Parser)()
         try:
-            parser.parse(source, self, **args)
+            parser.parse(source, self, ignore_errors=errorFlag,**args  )
         finally:
             if source.auto_close:
                 source.close()
         return self
 
-    def load(self, source, publicID=None, format="xml"):
-        self.parse(source, publicID, format)
+    def load(self, source, publicID=None, format="xml" , ignore_errors=False):
+        errorFlag = ignore_errors
+        self.parse(source, publicID, format , ignore_errors =errorFlag)
 
     def query(
         self,
@@ -1516,6 +1519,7 @@ def parse(
         location=None,
         file=None,
         data=None,
+        ignore_errors = False,
         **args
     ):
         """
@@ -1529,7 +1533,7 @@ def parse(
         The graph into which the source was parsed. In the case of n3
         it returns the root context.
         """
-
+        errorFlag = ignore_errors
         source = create_input_source(
             source=source,
             publicID=publicID,
@@ -1545,7 +1549,7 @@ def parse(
 
         context = Graph(store=self.store, identifier=g_id)
         context.remove((None, None, None))  # hmm ?
-        context.parse(source, publicID=publicID, format=format, **args)
+        context.parse(source, publicID=publicID, format=format, ignore_errors=errorFlag,**args)
         return context
 
     def __reduce__(self):
@@ -1700,10 +1704,12 @@ def parse(
         location=None,
         file=None,
         data=None,
+        ignore_errors=False,
         **args
     ):
+        errorFlag = ignore_errors
         c = ConjunctiveGraph.parse(
-            self, source, publicID, format, location, file, data, **args
+            self, source, publicID, format, location, file, data, ignore_errors=errorFlag,**args
         )
         self.graph(c)
         return c
@@ -1762,7 +1768,7 @@ def add(self, triple):
 
     def addN(self, quads):
         """Add a sequence of triple with context"""
-
+        
         self.store.addN(
             (s, p, o, c)
             for s, p, o, c in quads
@@ -1998,7 +2004,7 @@ def namespaces(self):
     def absolutize(self, uri, defrag=1):
         raise UnSupportedAggregateOperation()
 
-    def parse(self, source, publicID=None, format="xml", **args):
+    def parse(self, source, publicID=None, format="xml", ignore_errors=False,**args):
         raise ModificationException()
 
     def n3(self):

rdflib/parser.py

@@ -44,7 +44,7 @@ class Parser(object):
     def __init__(self):
         pass
 
-    def parse(self, source, sink):
+    def parse(self, source, sink , ignore_errors=False):
         pass
 
 

rdflib/plugins/parsers/notation3.py

@@ -42,6 +42,10 @@
 
 from uuid import uuid4
 
+import logging
+log = logging.getLogger(__name__) #For logging the statements that could not be parsed
+
+
 from rdflib.term import URIRef, BNode, Literal, Variable, _XSD_PFX, _unique_id
 from rdflib.graph import QuotedGraph, ConjunctiveGraph, Graph
 from rdflib.compat import long_type
@@ -351,6 +355,7 @@ def __init__(
         genPrefix="",
         why=None,
         turtle=False,
+        ignore_errors=False,
     ):
         """ note: namespace names should *not* end in  # ;
         the  # will get added during qname processing """
@@ -363,7 +368,7 @@ def __init__(
         self._store = store
         if genPrefix:
             store.setGenPrefix(genPrefix)  # pass it on
-
+        self.errorFlag = ignore_errors
         self._thisDoc = thisDoc
         self.lines = 0  # for error handling
         self.startOfLine = 0  # For calculating character number
@@ -1637,7 +1642,11 @@ def UEscape(self, argstr, i, startline):
         return self._unicodeEscape(argstr, i, startline, unicodeEscape8, 8, "U")
 
     def BadSyntax(self, argstr, i, msg):
-        raise BadSyntax(self._thisDoc, self.lines, argstr, i, msg)
+        if self.errorFlag:
+            log.info("Bad Synatx at line %i: %s" %(self.lines+1 , msg))
+        else: 
+            raise BadSyntax(self._thisDoc, self.lines, argstr, i, msg)
+
 
 
 # If we are going to do operators then they should generate
@@ -1884,8 +1893,8 @@ class TurtleParser(Parser):
     def __init__(self):
         pass
 
-    def parse(self, source, graph, encoding="utf-8", turtle=True):
-
+    def parse(self, source, graph, encoding="utf-8", turtle=True , ignore_errors=False):
+        errorFlag = ignore_errors
         if encoding not in [None, "utf-8"]:
             raise Exception(
                 ("N3/Turtle files are always utf-8 encoded, ", "I was passed: %s")
@@ -1895,7 +1904,7 @@ def parse(self, source, graph, encoding="utf-8", turtle=True):
         sink = RDFSink(graph)
 
         baseURI = graph.absolutize(source.getPublicId() or source.getSystemId() or "")
-        p = SinkParser(sink, baseURI=baseURI, turtle=turtle)
+        p = SinkParser(sink, baseURI=baseURI, turtle=turtle , ignore_errors=errorFlag)
 
         p.loadStream(source.getByteStream())
 
@@ -1915,7 +1924,8 @@ class N3Parser(TurtleParser):
     def __init__(self):
         pass
 
-    def parse(self, source, graph, encoding="utf-8"):
+    def parse(self, source, graph, encoding="utf-8" , ignore_errors = False):
+        errorFlag=ignore_errors
         # we're currently being handed a Graph, not a ConjunctiveGraph
         assert graph.store.context_aware  # is this implied by formula_aware
         assert graph.store.formula_aware
@@ -1926,7 +1936,7 @@ def parse(self, source, graph, encoding="utf-8"):
         # TODO: update N3Processor so that it can use conj_graph as the sink
         conj_graph.namespace_manager = graph.namespace_manager
 
-        TurtleParser.parse(self, source, conj_graph, encoding, turtle=False)
+        TurtleParser.parse(self, source, conj_graph, encoding, turtle=False , ignore_errors=errorFlag)
 
 
 def _test():  # pragma: no cover
@@ -1945,7 +1955,7 @@ def main():  # pragma: no cover
     sink = RDFSink(g)
     base_uri = "file://" + os.path.join(os.getcwd(), sys.argv[1])
 
-    p = SinkParser(sink, baseURI=base_uri)
+    p = SinkParser(sink, baseURI=base_uri , ignore_errors=False)
     p._bindings[""] = p._baseURI + "#"
     p.startDoc()
 

rdflib/plugins/parsers/nquads.py

@@ -38,10 +38,14 @@
 
 __all__ = ["NQuadsParser"]
 
+import logging
+log = logging.getLogger(__name__) #For logging the statements that could not be parsed
+
 
 class NQuadsParser(NTriplesParser):
-    def parse(self, inputsource, sink, **kwargs):
+    def parse(self, inputsource, sink, ignore_errors=False , **kwargs ):
         """Parse f as an N-Triples file."""
+        errorFlag=ignore_errors
         assert sink.store.context_aware, (
             "NQuadsParser must be given" " a context aware store."
         )
@@ -63,7 +67,10 @@ def parse(self, inputsource, sink, **kwargs):
             try:
                 self.parseline()
             except ParseError as msg:
-                raise ParseError("Invalid line (%s):\n%r" % (msg, __line))
+                if not errorFlag:
+                    raise ParseError("Invalid line (%s):\n%r" % (msg, __line))
+                else:
+                    log.info("Invalid line (%s):\n%r" % (msg, __line))
 
         return self.sink
 

rdflib/plugins/parsers/nt.py

@@ -20,8 +20,9 @@ class NTParser(Parser):
     def __init__(self):
         super(NTParser, self).__init__()
 
-    def parse(self, source, sink, baseURI=None):
+    def parse(self, source, sink, baseURI=None , ignore_errors=False):
+        errorFlag = ignore_errors
         f = source.getByteStream()  # TODO getCharacterStream?
         parser = NTriplesParser(NTSink(sink))
-        parser.parse(f)
+        parser.parse(f , ignore_errors=errorFlag)
         f.close()

rdflib/plugins/parsers/ntriples.py

@@ -12,6 +12,10 @@
 import re
 import codecs
 
+import logging
+log = logging.getLogger(__name__) #For logging the statements that could not be parsed
+
+
 from rdflib.term import URIRef as URI
 from rdflib.term import BNode as bNode
 from rdflib.term import Literal
@@ -130,8 +134,9 @@ def __init__(self, sink=None):
         else:
             self.sink = Sink()
 
-    def parse(self, f):
+    def parse(self, f , ignore_errors=False):
         """Parse f as an N-Triples file."""
+        errorFlag = ignore_errors
         if not hasattr(f, "read"):
             raise ParseError("Item to parse must be a file-like object.")
 
@@ -147,17 +152,22 @@ def parse(self, f):
             try:
                 self.parseline()
             except ParseError:
-                raise ParseError("Invalid line: %r" % self.line)
+                if not errorFlag:
+                    raise ParseError("Invalid line: %r" % self.line)
+                else :
+                    log.info("Invalid line: %r" % self.line)
+                    continue
         return self.sink
 
-    def parsestring(self, s):
+    def parsestring(self, s , ignore_errors=False):
         """Parse s as an N-Triples string."""
+        errorFlag=ignore_errors
         if not isinstance(s, str):
             raise ParseError("Item to parse must be a string instance.")
         f = BytesIO()
         f.write(cast_bytes(s))
         f.seek(0)
-        self.parse(f)
+        self.parse(f , ignore_errors=errorFlag)
 
     def readline(self):
         """Read an N-Triples line from buffered input."""

rdflib/plugins/parsers/rdfxml.py

@@ -5,6 +5,8 @@
 from xml.sax import make_parser
 from xml.sax.handler import ErrorHandler
 from xml.sax.saxutils import handler, quoteattr, escape
+import logging
+log = logging.getLogger(__name__) #For logging the statements that could not be parsed
 
 
 from urllib.parse import urldefrag, urljoin
@@ -115,9 +117,10 @@ def next_li(self):
 
 
 class RDFXMLHandler(handler.ContentHandler):
-    def __init__(self, store):
+    def __init__(self, store , ignore_errors=False):
         self.store = store
         self.preserve_bnode_ids = False
+        self.errorFlag = ignore_errors
         self.reset()
 
     def reset(self):
@@ -208,8 +211,10 @@ def error(self, message):
             locator.getLineNumber(),
             locator.getColumnNumber(),
         )
-        raise ParserError(info + message)
-
+        if not self.errorFlag:
+            raise ParserError(info + message)
+        else: 
+            log.info(info + message)
     def get_current(self):
         return self.stack[-2]
 
@@ -561,7 +566,8 @@ def literal_element_end(self, name, qname):
         self.parent.object += self.current.object + end
 
 
-def create_parser(target, store):
+def create_parser(target, store , ignore_errors=False):
+    errorFlag = ignore_errors
     parser = make_parser()
     try:
         # Workaround for bug in expatreader.py. Needed when
@@ -570,7 +576,7 @@ def create_parser(target, store):
     except AttributeError:
         pass  # Not present in Jython (at least)
     parser.setFeature(handler.feature_namespaces, 1)
-    rdfxml = RDFXMLHandler(store)
+    rdfxml = RDFXMLHandler(store , ignore_errors = errorFlag)
     rdfxml.setDocumentLocator(target)
     # rdfxml.setDocumentLocator(_Locator(self.url, self.parser))
     parser.setContentHandler(rdfxml)
@@ -582,8 +588,9 @@ class RDFXMLParser(Parser):
     def __init__(self):
         pass
 
-    def parse(self, source, sink, **args):
-        self._parser = create_parser(source, sink)
+    def parse(self, source, sink , ignore_errors=False , **args):
+        errorFlag = ignore_errors
+        self._parser = create_parser(source, sink , ignore_errors=errorFlag)
         content_handler = self._parser.getContentHandler()
         preserve_bnode_ids = args.get("preserve_bnode_ids", None)
         if preserve_bnode_ids is not None:

rdflib/plugins/parsers/trig.py

@@ -130,8 +130,8 @@ class TrigParser(Parser):
     def __init__(self):
         pass
 
-    def parse(self, source, graph, encoding="utf-8"):
-
+    def parse(self, source, graph, encoding="utf-8" , ignore_errors=False):
+        errorFlag = ignore_errors
         if encoding not in [None, "utf-8"]:
             raise Exception(
                 ("TriG files are always utf-8 encoded, ", "I was passed: %s") % encoding
@@ -151,7 +151,7 @@ def parse(self, source, graph, encoding="utf-8"):
         baseURI = conj_graph.absolutize(
             source.getPublicId() or source.getSystemId() or ""
         )
-        p = TrigSinkParser(sink, baseURI=baseURI, turtle=True)
+        p = TrigSinkParser(sink, baseURI=baseURI, turtle=True , ignore_errors=errorFlag)
 
         p.loadStream(source.getByteStream())