fix bug in parsing drepr model twice, and allow dump version 2

Author: Binh Vu

pydrepr/drepr/models/drepr.py

@@ -1,17 +1,17 @@
-from collections import OrderedDict, defaultdict
+from collections import OrderedDict
 from dataclasses import dataclass, asdict
 from enum import Enum
 from io import StringIO
-from typing import List, Dict, Any, NamedTuple, Optional
+from typing import List, Dict, Any, Optional
 
 import ujson
 from ruamel.yaml import YAML
 
-from drepr.models.parse_v2 import ReprV2Parser
 from drepr.utils.validator import Validator, InputError
 from .align import Alignment, RangeAlignment, AlignmentType, ValueAlignment, AlignedStep
 from .attr import Attr
 from .parse_v1 import ReprV1Parser
+from .parse_v2 import ReprV2Parser
 from .preprocessing import Preprocessing, PMap, PFilter, RMap, PSplit
 from .resource import Resource, CSVProp
 from .sm import SemanticModel, DataNode, ClassNode, LiteralNode
@@ -162,15 +162,15 @@ def is_valid(self):
                 if self.sm.is_rel_iri(node.label):
                     prefix = node.label.split(":", 1)[0]
                     assert prefix in self.sm.prefixes, f"Unknown prefix `{prefix}` of the " \
-                                                                            f"ontology class {node.label}"
+                                                       f"ontology class {node.label}"
         for edge in self.sm.edges.values():
             if self.sm.is_rel_iri(edge.label):
                 prefix = edge.label.split(":", 1)[0]
                 assert prefix in self.sm.prefixes, f"Unknown prefix `{prefix}` of the " \
-                                                                        f"ontology predicate {edge.label}"
+                                                   f"ontology predicate {edge.label}"
 
     def to_lang_format(self, simplify: bool = True, use_json_path: bool = False) -> dict:
-        return ReprV1Parser.dump(self, simplify, use_json_path)
+        return ReprV2Parser.dump(self, simplify, use_json_path)
 
     def to_lang_yml(self, simplify: bool = True, use_json_path: bool = False) -> str:
         model = self.to_lang_format(simplify, use_json_path)

pydrepr/drepr/models/parse_v2/__init__.py

@@ -1,13 +1,18 @@
+from collections import defaultdict
+from dataclasses import asdict
+from typing import List
+
 from drepr.models.parse_v2.path_parser import PathParserV2
 from drepr.utils.validator import *
+from ..align import AlignmentType, RangeAlignment
 
 from ..parse_v1.align_parser import AlignParser
 from ..parse_v1.attr_parser import AttrParser
 from ..parse_v1.preprocessing_parser import PreprocessingParser
 from ..parse_v1.resource_parser import ResourceParser
 
 from .sm_parser import SMParser
-from ..sm import SemanticModel
+from ..sm import SemanticModel, ClassNode, DataNode, LiteralNode
 
 
 class ReprV2Parser:
@@ -57,3 +62,83 @@ def parse(cls, raw: dict):
             sm = None
 
         return DRepr(resources, preprocessing, attrs, aligns, sm)
+
+    @classmethod
+    def dump(cls, drepr: 'DRepr', simplify: bool = True, use_json_path: bool = False):
+        version = '2'
+        sm = OrderedDict()
+
+        class_counter = defaultdict(int)
+        class_ids: Dict[str, str] = {}
+        for node in drepr.sm.nodes.values():
+            if isinstance(node, ClassNode):
+                class_counter[node.label] += 1
+                class_ids[node.node_id] = f"{node.label}:{class_counter[node.label]}"
+                sm[class_ids[node.node_id]] = OrderedDict([
+                    ("properties", []),
+                    ("static_properties", []),
+                    ("links", [])
+                ])
+
+        for node in drepr.sm.nodes.values():
+            if isinstance(node, DataNode):
+                edge = [e for e in drepr.sm.edges.values() if e.target_id == node.node_id][0]
+                if node.data_type is not None:
+                    prop = (edge.label, node.attr_id, node.data_type.value)
+                else:
+                    prop = (edge.label, node.attr_id)
+                sm[class_ids[edge.source_id]]['properties'].append(prop)
+
+            if isinstance(node, LiteralNode):
+                edge = [e for e in drepr.sm.edges if e.target_id == node.node_id][0]
+                if node.data_type is not None:
+                    prop = (edge.label, node.value, node.data_type.value)
+                else:
+                    prop = (edge.label, node.value)
+                sm[class_ids[edge.source_id]]['static_properties'].append(prop)
+
+        for edge in drepr.sm.edges.values():
+            if isinstance(drepr.sm.nodes[edge.source_id], ClassNode) and isinstance(
+                    drepr.sm.nodes[edge.target_id], ClassNode):
+                sm[class_ids[edge.source_id]]['links'].append((edge.label, class_ids[edge.target_id]))
+            if edge.is_subject:
+                sm[class_ids[edge.source_id]]['subject'] = drepr.sm.nodes[edge.target_id].attr_id
+
+        sm['prefixes'] = drepr.sm.prefixes
+
+        preprocessing: List[dict] = []
+        for prepro in drepr.preprocessing:
+            preprocessing.append(OrderedDict([("type", prepro.type.value)]))
+            for k, v in asdict(prepro.value).items():
+                preprocessing[-1][k] = v
+            preprocessing[-1]["path"] = prepro.value.path.to_lang_format(use_json_path)
+
+        return OrderedDict(
+            [("version", version),
+             ("resources",
+              OrderedDict(
+                  [(res.id,
+                    OrderedDict([("type", res.type.value)] + (
+                        [(k, v)
+                         for k, v in asdict(res.prop).items()] if res.prop is not None else [])))
+                   for res in drepr.resources])), ("preprocessing", preprocessing),
+             ("attributes",
+              OrderedDict([(attr.id,
+                            OrderedDict([("resource_id", attr.resource_id),
+                                         ("path", attr.path.to_lang_format(use_json_path)),
+                                         ("unique", attr.unique), ("sorted", attr.sorted.value),
+                                         ("value_type", attr.value_type.value),
+                                         ("missing_values", attr.missing_values)]))
+                           for attr in drepr.attrs])),
+             ("alignments", [
+                 OrderedDict([("type", AlignmentType.range.value), ("source", align.source),
+                              ("target", align.target),
+                              ("aligned_dims", [
+                                  OrderedDict([
+                                      ("source", step.source_idx),
+                                      ("target", step.target_idx),
+                                  ]) for step in align.aligned_steps
+                              ])]) if isinstance(align, RangeAlignment) else
+                 OrderedDict([("type", AlignmentType.value.value), ("source", align.source),
+                              ("target", align.target)]) for align in drepr.aligns
+             ]), ("semantic_model", sm)])

pydrepr/drepr/models/parse_v2/sm_parser.py

@@ -1,3 +1,4 @@
+import copy
 import re
 
 from drepr.utils.validator import Validator, InputError
@@ -34,8 +35,10 @@ class SMParser:
     def parse(cls, sm: dict) -> SemanticModel:
         nodes = {}
         edges = {}
-
+        # shallow copy
+        sm = copy.copy(sm)
         prefixes = sm.pop('prefixes', {})
+
         trace0 = f"Parsing `prefixes` of the semantic model"
         Validator.must_be_dict(prefixes, trace0)
         for prefix, uri in prefixes.items():

pydrepr/drepr/version.py

@@ -1,3 +1,3 @@
-__version__ = "2.9.2"   # ___PKG_VERSION___: DO NOT MODIFY the version here. Update it via version_manager.py!
+__version__ = "2.9.3"   # ___PKG_VERSION___: DO NOT MODIFY the version here. Update it via version_manager.py!
 __engine_version__ = "1.0.8"   # ___PKG_VERSION___: DO NOT MODIFY the version here. Update it via version_manager.py!
 __engine_release_tag__ = "2.7"   # ___PKG_VERSION___: DO NOT MODIFY the version here. Update it via version_manager.py!

version_manager.py

@@ -4,7 +4,7 @@
 
 # ================================================================================
 # Note: update the version by changing variables before the `=...=` line
-DREPR_PYLIB_VESRION = "2.9.2"
+DREPR_PYLIB_VESRION = "2.9.3"
 DREPR_ENGINE_VERSION = "1.0.8"
 # this tag marks the release which contains the pre-built engine in it.
 DREPR_ENGINE_RELEASE_TAG = "2.7"