GEWU/tests/domain/proposal_fixtures.py

"""Golden pilot proposal fixtures for deterministic evidence and policy checks."""

from datetime import UTC, datetime

from gewu.domain.contracts import (
    AutoAcceptPolicy,
    Claim,
    Confidence,
    EvidenceReference,
    ExtractionBlock,
    ExtractionRecord,
    KnowledgeNode,
    LensFit,
    Lifecycle,
    Locator,
    NodeType,
    ProjectRecord,
    ProjectStatus,
    ProposalOperation,
    ProposalRevision,
    ReliabilityGrade,
    ReviewState,
    Scope,
    SourceRevisionRecord,
    SupportAssessment,
    SupportStatus,
)
from gewu.domain.validation import (
    assertion_digest,
    evidence_digest,
    extraction_digest,
    proposal_payload_digest,
)

BASE_COMMIT = "1" * 40
NOW = datetime(2026, 10, 13, tzinfo=UTC)
SOURCE_REVISION = SourceRevisionRecord(
    source_id="shared:pilot-report",
    revision_id="shared:pilot-report:r1",
    artifact_sha256="2" * 64,
    media_type="text/markdown",
    byte_length=100,
    captured_at=NOW,
    original_url=None,
    title="Pilot dome installation report",
    authors=("Pilot team",),
    publisher="GEWU fixtures",
    license="internal",
    source_type="internal",
    reliability_grade=ReliabilityGrade.MEDIUM,
    ingested_for=("dome-performance",),
    archive_status="archived",
)
BLOCK_TEXT = "The installation hosted 400 visitors during the 2026 season."
_EXTRACTION_UNHASHED = ExtractionRecord(
    extraction_id="0" * 64,
    source_revision_id=SOURCE_REVISION.revision_id,
    parser_name="gewu-text",
    parser_version="0.1.0",
    parser_config_sha256="4" * 64,
    normalization_profile="unicode-nfc-lf-v1",
    status="complete",
    blocks=(
        ExtractionBlock(
            block_id="block-1",
            ordinal=0,
            original_text=BLOCK_TEXT,
            normalized_text=BLOCK_TEXT,
            normalization_map=tuple((index, index + 1) for index in range(len(BLOCK_TEXT))),
            page=None,
        ),
    ),
)
EXTRACTION = _EXTRACTION_UNHASHED.model_copy(
    update={"extraction_id": extraction_digest(_EXTRACTION_UNHASHED)}
)
CLAIM_TEXT = "The installation hosted 400 visitors during the 2026 season."
CLAIM_DIGEST = assertion_digest(CLAIM_TEXT)
START = 0
END = len(BLOCK_TEXT)
EVIDENCE = EvidenceReference(
    source_revision_id=SOURCE_REVISION.revision_id,
    extraction_id=EXTRACTION.extraction_id,
    assertion_path="claims[0].text",
    assertion_digest=CLAIM_DIGEST,
    quote=BLOCK_TEXT,
    locator=Locator(kind="text_span", block_id="block-1", start=START, end=END),
    normalization_profile="unicode-nfc-lf-v1",
)
CLAIM = Claim(
    claim_id="dome:claim-visitors",
    text=CLAIM_TEXT,
    evidence_type="reported",
    evidence=(EVIDENCE,),
    assessment=SupportAssessment(
        assessment_id="assessment-1",
        assertion_digest=CLAIM_DIGEST,
        evidence_digests=(evidence_digest(EVIDENCE),),
        assessor_id="agent-pilot",
        run_id="run-pilot-1",
        method="paired source claim check",
        status=SupportStatus.SUPPORTED,
        rationale="The quoted sentence states the count and period.",
    ),
)
NODE = KnowledgeNode(
    node_id="dome:season-visitors",
    project_id="dome-performance",
    node_type=NodeType.DATAPOINT,
    title="Dome season visitors",
    summary="The installation hosted 400 visitors in the 2026 season.",
    created=NOW,
    updated=NOW,
    review_state=ReviewState.DRAFT,
    lifecycle=Lifecycle.ACTIVE,
    confidence=Confidence.MEDIUM,
    claim_support="supported",
    tags=("attendance",),
    audience=("product",),
    domain=("production",),
    use_for=("comparison",),
    lens_fit={
        "dome-performance": LensFit(
            score=0.9,
            horizon="H1",
            key_question="How do audiences engage?",
            reason="Reports a representative audience measure.",
        ),
    },
    run_id="run-pilot-1",
    agent_id="agent-pilot",
    model_id="fixture-model",
    claims=(CLAIM,),
    as_of=NOW,
    stale_after=datetime(2027, 1, 1, tzinfo=UTC),
    value=400,
    unit="visitors",
    sample_size=None,
    reporting_year=2026,
)
PROJECTS = {
    "dome-performance": ProjectRecord(
        project_id="dome-performance",
        title="Dome interactive content",
        status=ProjectStatus.ACTIVE,
        north_star="Create compelling interactive experiences for dome audiences.",
        key_questions=("How do audiences engage?",),
        autonomy_policy_id="dome-policy",
    ),
    "interactive-video": ProjectRecord(
        project_id="interactive-video",
        title="Interactive video",
        status=ProjectStatus.ACTIVE,
        north_star="Understand useful interactive video experiences.",
        key_questions=("How do teams control branching?",),
        autonomy_policy_id="video-policy",
    ),
}
POLICIES = {
    "dome-policy": AutoAcceptPolicy(
        policy_id="dome-policy",
        policy_version=1,
        project_id="dome-performance",
        automatic_node_types=("datapoint", "problem", "technique", "case"),
        automatic_source_registration=True,
        require_supported_assessment=True,
    ),
    "video-policy": AutoAcceptPolicy(
        policy_id="video-policy",
        policy_version=1,
        project_id="interactive-video",
        automatic_node_types=("datapoint", "problem", "technique", "case"),
        automatic_source_registration=True,
        require_supported_assessment=True,
    ),
    "shared": AutoAcceptPolicy(
        policy_id="shared",
        policy_version=1,
        project_id="shared",
        automatic_source_registration=True,
        require_supported_assessment=True,
    ),
}


def proposal(
    *,
    node: KnowledgeNode | None = NODE,
    source: SourceRevisionRecord | None = SOURCE_REVISION,
    scope: Scope | None = None,
    base_commit: str = BASE_COMMIT,
) -> ProposalRevision:
    operations = tuple(
        operation
        for operation in (
            ProposalOperation(operation="register_source", source=source) if source else None,
            ProposalOperation(operation="upsert_node", node=node) if node else None,
        )
        if operation is not None
    )
    candidate = ProposalRevision(
        proposal_id="proposal-pilot-1",
        revision=1,
        initiating_project_id="dome-performance",
        scope=scope or Scope(projects=("dome-performance",), include_shared=True),
        base_commit=base_commit,
        payload_digest="0" * 64,
        proposer_id="agent-pilot",
        run_id="run-pilot-1",
        operations=operations,
    )
    return candidate.model_copy(update={"payload_digest": proposal_payload_digest(candidate)})