mirror of
https://github.com/trustgraph-ai/trustgraph.git
synced 2026-04-25 08:26:21 +02:00
29b4300808
* Provenance tests * Embeddings tests * Test librarian * Test triples stream * Test concurrency * Entity centric graph writes * Agent tool service tests * Structured data tests * RDF tests * Addition LLM tests * Reliability tests
144 lines
4.7 KiB
Python
144 lines
4.7 KiB
Python
"""
|
|
Tests for pipeline metadata preservation: DocumentMetadata and
|
|
ProcessingMetadata round-trip through translators, field preservation,
|
|
and default handling.
|
|
"""
|
|
|
|
import pytest
|
|
|
|
from trustgraph.schema import DocumentMetadata, ProcessingMetadata, Triple, Term, IRI
|
|
from trustgraph.messaging.translators.metadata import (
|
|
DocumentMetadataTranslator,
|
|
ProcessingMetadataTranslator,
|
|
)
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
# DocumentMetadata translator
|
|
# ---------------------------------------------------------------------------
|
|
|
|
class TestDocumentMetadataTranslator:
|
|
|
|
def setup_method(self):
|
|
self.tx = DocumentMetadataTranslator()
|
|
|
|
def test_full_round_trip(self):
|
|
data = {
|
|
"id": "doc-123",
|
|
"time": 1710000000,
|
|
"kind": "application/pdf",
|
|
"title": "Test Document",
|
|
"comments": "No comments",
|
|
"metadata": [],
|
|
"user": "alice",
|
|
"tags": ["finance", "q4"],
|
|
"parent-id": "doc-100",
|
|
"document-type": "page",
|
|
}
|
|
obj = self.tx.to_pulsar(data)
|
|
assert obj.id == "doc-123"
|
|
assert obj.time == 1710000000
|
|
assert obj.kind == "application/pdf"
|
|
assert obj.title == "Test Document"
|
|
assert obj.user == "alice"
|
|
assert obj.tags == ["finance", "q4"]
|
|
assert obj.parent_id == "doc-100"
|
|
assert obj.document_type == "page"
|
|
|
|
wire = self.tx.from_pulsar(obj)
|
|
assert wire["id"] == "doc-123"
|
|
assert wire["user"] == "alice"
|
|
assert wire["parent-id"] == "doc-100"
|
|
assert wire["document-type"] == "page"
|
|
|
|
def test_defaults_for_missing_fields(self):
|
|
obj = self.tx.to_pulsar({})
|
|
assert obj.parent_id == ""
|
|
assert obj.document_type == "source"
|
|
|
|
def test_metadata_triples_preserved(self):
|
|
triple_wire = [{
|
|
"s": {"t": "i", "i": "http://example.org/s"},
|
|
"p": {"t": "i", "i": "http://example.org/p"},
|
|
"o": {"t": "i", "i": "http://example.org/o"},
|
|
}]
|
|
data = {"metadata": triple_wire}
|
|
obj = self.tx.to_pulsar(data)
|
|
assert len(obj.metadata) == 1
|
|
assert obj.metadata[0].s.iri == "http://example.org/s"
|
|
|
|
def test_none_metadata_handled(self):
|
|
data = {"metadata": None}
|
|
obj = self.tx.to_pulsar(data)
|
|
assert obj.metadata == []
|
|
|
|
def test_empty_tags_preserved(self):
|
|
data = {"tags": []}
|
|
obj = self.tx.to_pulsar(data)
|
|
wire = self.tx.from_pulsar(obj)
|
|
assert wire["tags"] == []
|
|
|
|
def test_falsy_fields_omitted_from_wire(self):
|
|
"""Empty string fields should be omitted from wire format."""
|
|
obj = DocumentMetadata(id="", time=0, user="")
|
|
wire = self.tx.from_pulsar(obj)
|
|
assert "id" not in wire
|
|
assert "user" not in wire
|
|
|
|
|
|
# ---------------------------------------------------------------------------
|
|
# ProcessingMetadata translator
|
|
# ---------------------------------------------------------------------------
|
|
|
|
class TestProcessingMetadataTranslator:
|
|
|
|
def setup_method(self):
|
|
self.tx = ProcessingMetadataTranslator()
|
|
|
|
def test_full_round_trip(self):
|
|
data = {
|
|
"id": "proc-1",
|
|
"document-id": "doc-123",
|
|
"time": 1710000000,
|
|
"flow": "default",
|
|
"user": "alice",
|
|
"collection": "my-collection",
|
|
"tags": ["tag1"],
|
|
}
|
|
obj = self.tx.to_pulsar(data)
|
|
assert obj.id == "proc-1"
|
|
assert obj.document_id == "doc-123"
|
|
assert obj.flow == "default"
|
|
assert obj.user == "alice"
|
|
assert obj.collection == "my-collection"
|
|
assert obj.tags == ["tag1"]
|
|
|
|
wire = self.tx.from_pulsar(obj)
|
|
assert wire["id"] == "proc-1"
|
|
assert wire["document-id"] == "doc-123"
|
|
assert wire["user"] == "alice"
|
|
assert wire["collection"] == "my-collection"
|
|
|
|
def test_missing_fields_use_defaults(self):
|
|
obj = self.tx.to_pulsar({})
|
|
assert obj.id is None
|
|
assert obj.user is None
|
|
assert obj.collection is None
|
|
|
|
def test_tags_none_omitted(self):
|
|
obj = ProcessingMetadata(tags=None)
|
|
wire = self.tx.from_pulsar(obj)
|
|
assert "tags" not in wire
|
|
|
|
def test_tags_empty_list_preserved(self):
|
|
obj = ProcessingMetadata(tags=[])
|
|
wire = self.tx.from_pulsar(obj)
|
|
assert wire["tags"] == []
|
|
|
|
def test_user_and_collection_preserved(self):
|
|
"""Core pipeline routing fields must survive round-trip."""
|
|
data = {"user": "bob", "collection": "research"}
|
|
obj = self.tx.to_pulsar(data)
|
|
wire = self.tx.from_pulsar(obj)
|
|
assert wire["user"] == "bob"
|
|
assert wire["collection"] == "research"
|