dkg 8.0.12__py3-none-any.whl → 8.0.14__py3-none-any.whl

dkg/utils/knowledge_collection_tools.py

@@ -5,7 +5,8 @@ from dkg.exceptions import DatasetInputFormatNotSupported, InvalidDataset
 from dkg.types import JSONLD, NQuads
 from pyld import jsonld
 from dkg.constants import DEFAULT_RDF_FORMAT, DEFAULT_CANON_ALGORITHM, ESCAPE_MAP
-from rdflib import Graph, BNode, URIRef, Literal as RDFLiteral
+from rdflib import Graph, BNode, URIRef, Dataset
+from rdflib.exceptions import ParserError as RDFParserError
 from uuid import uuid4
 from web3 import Web3
 import math
@@ -154,29 +155,26 @@ def generate_missing_ids_for_blank_nodes(nquads_list: list[str] | None) -> list[
 
         return term  # Return IRIs or Literals unchanged
 
-    # Create a temporary graph for parsing individual quads
-    result = []
+    all_nquads = "\n".join(nquad for nquad in nquads_list if nquad.strip())
 
-    # Process each N-Quad string individually to maintain order
-    for nquad in nquads_list:
-        if not nquad.strip():
-            continue
+    # Create a single Dataset
+    g = Graph()
+    try:
+        g.parse(data=all_nquads, format="nt")
+    except RDFParserError:
+        raise UnsupportedJSONLD(nquads_list)
 
-        # Parse single N-Quad
-        g = Graph()
-        g.parse(data=nquad, format="nquads")
-
-        # Get the triple and replace blank nodes
-        for s, p, o in g:
-            updated_quad = (
-                replace_blank_node(s),
-                replace_blank_node(p),
-                replace_blank_node(o),
-            )
-            # Format as N-Quad string
-            result.append(
-                f"{updated_quad[0].n3()} {updated_quad[1].n3()} {updated_quad[2].n3()} ."
-            )
+    # Process all quads
+    result = []
+    for s, p, o in g:
+        updated_quad = (
+            replace_blank_node(s),
+            replace_blank_node(p),
+            replace_blank_node(o),
+        )
+        result.append(
+            f"{updated_quad[0].n3()} {updated_quad[1].n3()} {updated_quad[2].n3()} ."
+        )
 
     return result
 
@@ -184,16 +182,13 @@ def generate_missing_ids_for_blank_nodes(nquads_list: list[str] | None) -> list[
 def group_nquads_by_subject(nquads_list: list[str], sort: bool = False):
     grouped = {}
 
-    # Process each quad in original order
-    for nquad in nquads_list:
-        if not nquad.strip():  # Skip empty lines
-            continue
+    all_nquads = "\n".join(nquad for nquad in nquads_list if nquad.strip())
 
-        # Parse single quad
-        g = Graph()
-        g.parse(data=nquad, format="nquads")
-        quad = next(iter(g))
-        subject, predicate, obj = quad
+    d = Dataset()
+    d.parse(data=all_nquads, format="nquads")
+
+    for quad in d:
+        subject, predicate, obj, graph = quad
 
         # Get subject key
         subject_key = (
@@ -206,11 +201,8 @@ def group_nquads_by_subject(nquads_list: list[str], sort: bool = False):
         if subject_key not in grouped:
             grouped[subject_key] = []
 
-        # Format object
-        object_value = f'"{obj}"' if isinstance(obj, RDFLiteral) else f"<{obj}>"
-
         # Add quad to group
-        quad_string = f"{subject_key} <{predicate}> {object_value} ."
+        quad_string = f"{subject.n3()} {predicate.n3()} {obj.n3()} ."
         grouped[subject_key].append(quad_string)
 
     # Return grouped quads (sorted if requested)
@@ -266,3 +258,44 @@ def escape_literal_dict(obj):
         return escape_literal_string(s=obj)
     else:
         return obj
+
+
+# Used when JSON-LD parsing fails due to quads being passed instead of triples
+class UnsupportedJSONLD(Exception):
+    def __init__(self, nquads_list):
+        self.nquads_list = nquads_list
+        self.message = f"""
+Unsupported JSON-LD input detected
+
+After parsing the JSON-LD input, the parser detected creation of new named graphs.
+The DKG does not support custom named graphs.
+
+Problematic Quads:
+
+{self.find_problematic_quads()}
+
+Full Parsed N-Quads Array:
+
+{self.format_nquads_list()}
+
+"""
+        super().__init__(self.message)
+
+    def __str__(self):
+        return f"{self.__class__.__name__}: {self.message}"
+
+    def format_nquads_list(self):
+        return "\n".join(nquad.strip() for nquad in self.nquads_list)
+
+    def find_problematic_quads(self):
+        problematic = []
+        g = Graph()
+        for quad in self.nquads_list:
+            if not quad.strip():
+                continue
+            try:
+                g.parse(data=quad, format="nt")
+            except RDFParserError:
+                problematic.append(quad)
+
+        return "\n".join(f"{i + 1}. {quad}" for i, quad in enumerate(problematic))

{dkg-8.0.12.dist-info → dkg-8.0.14.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: dkg
-Version: 8.0.12
+Version: 8.0.14
 Summary: Python library for interacting with the OriginTrail Decentralized Knowledge Graph
 License: Apache-2.0
 Author: Uladzislau Hubar

{dkg-8.0.12.dist-info → dkg-8.0.14.dist-info}/RECORD

@@ -68,15 +68,15 @@ dkg/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 dkg/utils/blockchain_request.py,sha256=CMz3sNRhfmYirm0-VUK2HTpbRfgc1gtG2SoPIuRYKb0,12446
 dkg/utils/decorators.py,sha256=uUh0xI9wv1gE6JF44eYEfUPGSFEBLR6Ua5D9Dhc3s10,1566
 dkg/utils/knowledge_asset_tools.py,sha256=uehOshzaYg03hDU8RSIV_k2mNaASVgqyyErzj6GwBxw,81
-dkg/utils/knowledge_collection_tools.py,sha256=hPCFcIdyWDCgfbo5uPQizvfin0hHvEElAYpvKYVDIBs,7951
+dkg/utils/knowledge_collection_tools.py,sha256=qGEZqsAo25J4P0HCwXA6EI37fVsAf7Nne1zEXR0Ot0Y,8776
 dkg/utils/merkle.py,sha256=924kloBAnCXydteVtWMj_QLP5CRJf2GbHMZ-lbIK0KE,5141
 dkg/utils/metadata.py,sha256=483OroYwGNfZ_cCXfH3-xUrZgiR4mjjo9iU_Ie5RYRs,1658
 dkg/utils/node_request.py,sha256=wppF8Xf0RkuAAcURcYgyjGAdMsXm0L5YEem8wFGE2g8,6517
 dkg/utils/rdf.py,sha256=AvlcxZEeP58UbaGGvPX_ss69O-tgTXOJ9y9COZqVgkw,2973
 dkg/utils/string_transformations.py,sha256=eR51fVwTF9QKxEqXo9_1Bfw_k8iQajdXD6rKuTvhs70,972
 dkg/utils/ual.py,sha256=g7PFyS4Sbwjmwkq-eB20uRULEC2wlPGZr31BVQjs5OQ,1569
-dkg-8.0.12.dist-info/LICENSE,sha256=Dr70w2zcW8-jrPGlpTTTlJPL8lR4j2zpDD32tdEFgjY,11375
-dkg-8.0.12.dist-info/METADATA,sha256=qoW7Du0ik1jwQC-qkG3nBQgZQLMoRtUrh4FMJXLJzcQ,10821
-dkg-8.0.12.dist-info/NOTICE,sha256=Rk5toFR2ZqPwVZ3P_P4wE6U1xCnWR9KD3rNBqfPY7h8,368
-dkg-8.0.12.dist-info/WHEEL,sha256=fGIA9gx4Qxk2KDKeNJCbOEwSrmLtjWCwzBz351GyrPQ,88
-dkg-8.0.12.dist-info/RECORD,,
+dkg-8.0.14.dist-info/LICENSE,sha256=Dr70w2zcW8-jrPGlpTTTlJPL8lR4j2zpDD32tdEFgjY,11375
+dkg-8.0.14.dist-info/METADATA,sha256=KnPQYF84DahhRYTXoH3vd478uHS7KeF829qLuSZDiV0,10821
+dkg-8.0.14.dist-info/NOTICE,sha256=Rk5toFR2ZqPwVZ3P_P4wE6U1xCnWR9KD3rNBqfPY7h8,368
+dkg-8.0.14.dist-info/WHEEL,sha256=fGIA9gx4Qxk2KDKeNJCbOEwSrmLtjWCwzBz351GyrPQ,88
+dkg-8.0.14.dist-info/RECORD,,