app/services/mock_data.py

"""Mock data for PoC UI demonstration.

Provides realistic sample PatientProfile, TrialCandidates, and EligibilityLedgers
so the UI is fully demonstrable without real MedGemma/Gemini/MCP calls.

NOTE: NCT IDs use a MOCK-NCT prefix to clearly indicate synthetic data.
Real trial names are used for demo realism, but the IDs are not real
ClinicalTrials.gov identifiers.
"""

from __future__ import annotations

from trialpath.models import (
    Biomarker,
    CriterionAssessment,
    CriterionDecision,
    Demographics,
    Diagnosis,
    EligibilityLedger,
    EvidencePointer,
    GapItem,
    OverallAssessment,
    PatientProfile,
    PerformanceStatus,
    SourceDocument,
    Treatment,
    TrialCandidate,
    TrialEvidencePointer,
    UnknownField,
)

# ---------------------------------------------------------------------------
# Source documents (synthetic)
# ---------------------------------------------------------------------------
_CLINIC_LETTER = SourceDocument(
    doc_id="clinic_letter_2026-01.pdf",
    type="clinic_letter",
    meta={"author": "Dr. Mueller", "date": "2026-01-15"},
)
_PATHOLOGY_REPORT = SourceDocument(
    doc_id="pathology_2025-11.pdf",
    type="pathology",
    meta={"lab": "Charite Berlin", "date": "2025-11-20"},
)
_LAB_RESULTS = SourceDocument(
    doc_id="labs_2026-01.pdf",
    type="lab",
    meta={"date": "2026-01-28"},
)

# ---------------------------------------------------------------------------
# Patient profile
# ---------------------------------------------------------------------------
MOCK_PATIENT_PROFILE = PatientProfile(
    patient_id="MOCK-P001",
    source_docs=[_CLINIC_LETTER, _PATHOLOGY_REPORT, _LAB_RESULTS],
    demographics=Demographics(age=62, sex="Female"),
    diagnosis=Diagnosis(
        primary_condition="Non-Small Cell Lung Cancer",
        histology="adenocarcinoma",
        stage="IIIB",
    ),
    performance_status=PerformanceStatus(
        scale="ECOG",
        value=1,
        evidence=[EvidencePointer(doc_id="clinic_letter_2026-01.pdf", page=1)],
    ),
    biomarkers=[
        Biomarker(
            name="EGFR",
            result="Exon 19 deletion",
            evidence=[
                EvidencePointer(doc_id="pathology_2025-11.pdf", page=2, span_id="egfr_exon19")
            ],
        ),
        Biomarker(
            name="ALK",
            result="Negative",
            evidence=[EvidencePointer(doc_id="pathology_2025-11.pdf", page=2, span_id="alk_neg")],
        ),
        Biomarker(
            name="PD-L1",
            result="45%",
            evidence=[EvidencePointer(doc_id="pathology_2025-11.pdf", page=3, span_id="pdl1_45")],
        ),
    ],
    treatments=[
        Treatment(drug_name="Carboplatin", line=1),
        Treatment(drug_name="Pemetrexed", line=1),
    ],
    unknowns=[
        UnknownField(field="KRAS", reason="Not in records", importance="high"),
        UnknownField(field="Brain MRI", reason="Not available", importance="medium"),
    ],
)

# ---------------------------------------------------------------------------
# Trial candidates
# ---------------------------------------------------------------------------
MOCK_TRIAL_CANDIDATES = [
    TrialCandidate(
        nct_id="MOCK-NCT-KEYNOTE999",
        title="KEYNOTE-999: Pembrolizumab + Chemo for NSCLC",
        conditions=["NSCLC"],
        phase="Phase 3",
        status="Recruiting",
        fingerprint_text="Pembrolizumab combination therapy advanced NSCLC",
    ),
    TrialCandidate(
        nct_id="MOCK-NCT-FLAURA2",
        title="FLAURA-2: Osimertinib + Chemo in EGFR-mutated NSCLC",
        conditions=["NSCLC", "EGFR mutation"],
        phase="Phase 3",
        status="Recruiting",
        fingerprint_text="Osimertinib first-line EGFR mutated advanced NSCLC",
    ),
    TrialCandidate(
        nct_id="MOCK-NCT-CM817",
        title="CheckMate-817: Nivolumab + Ipilimumab in Advanced NSCLC",
        conditions=["NSCLC"],
        phase="Phase 3",
        status="Recruiting",
        fingerprint_text="Nivolumab ipilimumab combination immunotherapy NSCLC",
    ),
]

# ---------------------------------------------------------------------------
# Eligibility ledgers -- evidence-linked per PRD requirement
# ---------------------------------------------------------------------------

# --- KEYNOTE-999: UNCERTAIN (PD-L1 not met at 45% vs >=50%, brain mets unknown) ---
_KEYNOTE_CRITERIA = [
    CriterionAssessment(
        criterion_id="inc_1",
        type="inclusion",
        text="Confirmed NSCLC diagnosis",
        decision=CriterionDecision.MET,
        patient_evidence=[
            EvidencePointer(doc_id="pathology_2025-11.pdf", page=1, span_id="dx_nsclc"),
        ],
        trial_evidence=[
            TrialEvidencePointer(field="eligibility_text.inclusion", offset_start=0, offset_end=42),
        ],
    ),
    CriterionAssessment(
        criterion_id="inc_2",
        type="inclusion",
        text="ECOG performance status 0-1",
        decision=CriterionDecision.MET,
        patient_evidence=[
            EvidencePointer(doc_id="clinic_letter_2026-01.pdf", page=1, span_id="ecog_1"),
        ],
        trial_evidence=[
            TrialEvidencePointer(
                field="eligibility_text.inclusion", offset_start=43, offset_end=89
            ),
        ],
    ),
    CriterionAssessment(
        criterion_id="inc_3",
        type="inclusion",
        text="PD-L1 TPS >= 50%",
        decision=CriterionDecision.NOT_MET,
        patient_evidence=[
            EvidencePointer(doc_id="pathology_2025-11.pdf", page=3, span_id="pdl1_45"),
        ],
        trial_evidence=[
            TrialEvidencePointer(
                field="eligibility_text.inclusion", offset_start=90, offset_end=130
            ),
        ],
    ),
    CriterionAssessment(
        criterion_id="exc_1",
        type="exclusion",
        text="No active brain metastases",
        decision=CriterionDecision.UNKNOWN,
    ),
]

# --- FLAURA-2: LIKELY_ELIGIBLE (EGFR mutation confirmed, stage/ECOG met) ---
_FLAURA_CRITERIA = [
    CriterionAssessment(
        criterion_id="inc_1",
        type="inclusion",
        text="Confirmed NSCLC with EGFR mutation",
        decision=CriterionDecision.MET,
        patient_evidence=[
            EvidencePointer(doc_id="pathology_2025-11.pdf", page=2, span_id="egfr_exon19"),
        ],
        trial_evidence=[
            TrialEvidencePointer(field="eligibility_text.inclusion", offset_start=0, offset_end=54),
        ],
    ),
    CriterionAssessment(
        criterion_id="inc_2",
        type="inclusion",
        text="ECOG 0-1",
        decision=CriterionDecision.MET,
        patient_evidence=[
            EvidencePointer(doc_id="clinic_letter_2026-01.pdf", page=1, span_id="ecog_1"),
        ],
        trial_evidence=[
            TrialEvidencePointer(
                field="eligibility_text.inclusion", offset_start=55, offset_end=78
            ),
        ],
    ),
    CriterionAssessment(
        criterion_id="inc_3",
        type="inclusion",
        text="Locally advanced or metastatic",
        decision=CriterionDecision.MET,
        patient_evidence=[
            EvidencePointer(doc_id="clinic_letter_2026-01.pdf", page=1, span_id="stage_iiib"),
        ],
        trial_evidence=[
            TrialEvidencePointer(
                field="eligibility_text.inclusion", offset_start=79, offset_end=125
            ),
        ],
    ),
]

# --- CheckMate-817: LIKELY_INELIGIBLE (requires Stage IV, patient is IIIB) ---
_CM817_CRITERIA = [
    CriterionAssessment(
        criterion_id="inc_1",
        type="inclusion",
        text="Stage IV NSCLC",
        decision=CriterionDecision.NOT_MET,
        patient_evidence=[
            EvidencePointer(doc_id="clinic_letter_2026-01.pdf", page=1, span_id="stage_iiib"),
        ],
        trial_evidence=[
            TrialEvidencePointer(field="eligibility_text.inclusion", offset_start=0, offset_end=35),
        ],
    ),
    CriterionAssessment(
        criterion_id="inc_2",
        type="inclusion",
        text="No prior immunotherapy",
        decision=CriterionDecision.UNKNOWN,
    ),
]

MOCK_ELIGIBILITY_LEDGERS = [
    EligibilityLedger(
        patient_id="MOCK-P001",
        nct_id="MOCK-NCT-KEYNOTE999",
        overall_assessment=OverallAssessment.UNCERTAIN,
        criteria=_KEYNOTE_CRITERIA,
        gaps=[
            GapItem(
                description="Brain MRI results needed",
                recommended_action="Upload brain MRI report",
                clinical_importance="high",
            ),
        ],
    ),
    EligibilityLedger(
        patient_id="MOCK-P001",
        nct_id="MOCK-NCT-FLAURA2",
        overall_assessment=OverallAssessment.LIKELY_ELIGIBLE,
        criteria=_FLAURA_CRITERIA,
        gaps=[],
    ),
    EligibilityLedger(
        patient_id="MOCK-P001",
        nct_id="MOCK-NCT-CM817",
        overall_assessment=OverallAssessment.LIKELY_INELIGIBLE,
        criteria=_CM817_CRITERIA,
        gaps=[
            GapItem(
                description="KRAS mutation status unknown",
                recommended_action="Request KRAS test from oncologist",
                clinical_importance="medium",
            ),
        ],
    ),
]