"""Preprocess materialize stage — counts (raw flat_skills), inline experience_band, clean signals, text."""

from __future__ import annotations

import copy
from pathlib import Path

import orjson

from preprocess import DATASET_AS_OF, load_candidates, materialize


def _record() -> dict:
    """A populated candidate (keeps every nested column struct-typed)."""
    return {
        "candidate_id": "CAND_0000001",
        "profile": {"years_of_experience": 7.0, "current_title": "AI Engineer",
                    "current_company": "Acme", "current_industry": "Tech"},
        "career_history": [
            {"title": "AI Engineer", "company": "Acme", "industry": "Tech", "company_size": "51-200",
             "duration_months": 40, "is_current": True, "start_date": "2021-01-01",
             "end_date": None, "description": "ranking systems"}
        ],
        "education": [{"institution": "IIT", "degree": "BTech", "field_of_study": "CS",
                       "start_year": 2014, "end_year": 2018, "tier": "tier_1"}],
        "skills": [
            {"name": "Python", "proficiency": "expert", "endorsements": 9, "duration_months": 60},
            {"name": "SQL", "proficiency": "beginner", "endorsements": 1, "duration_months": 12},
        ],
        "certifications": [{"name": "AWS SAA", "issuer": "AWS", "year": 2022}],
        "redrob_signals": {
            "github_activity_score": 70, "offer_acceptance_rate": -1, "notice_period_days": 15,
            "expected_salary_range_inr_lpa": {"min": 20, "max": 40},
            "skill_assessment_scores": {"Python": 88.0},
        },
    }


def _materialize(tmp_path: Path, records: list[dict]):
    path = tmp_path / "c.jsonl"
    path.write_bytes(b"\n".join(orjson.dumps(r) for r in records))
    return materialize(load_candidates(path), DATASET_AS_OF)


def test_experience_band_is_computed_inline(tmp_path: Path) -> None:
    bands = []
    for yoe in (2.0, 5.0, 8.0, 12.0, 20.0):
        record = _record()
        record["profile"]["years_of_experience"] = yoe
        bands.append(_materialize(tmp_path, [record]).row(0, named=True)["experience_band"])
    assert bands == ["junior", "mid", "senior", "lead", "principal"]


def test_counts_use_raw_flat_skills(tmp_path: Path) -> None:
    row = _materialize(tmp_path, [_record()]).row(0, named=True)
    assert row["num_skills"] == 2  # from flat_skills, no canonicalization
    assert row["num_advanced_skills"] == 1 and row["num_expert_skills"] == 1
    assert row["num_roles"] == 1 and row["num_companies"] == 1
    assert row["num_assessments_taken"] == 1 and row["max_skill_assessment"] == 88.0


def test_sentinel_signals_are_cleaned(tmp_path: Path) -> None:
    row = _materialize(tmp_path, [_record()]).row(0, named=True)
    assert row["has_github"] is True and row["github_score_clean"] == 70.0
    assert row["has_offer_history"] is False and row["offer_acceptance_rate_clean"] is None
    assert row["salary_mid"] == 30.0 and row["is_available_soon"] is True


def test_embedded_text_is_built(tmp_path: Path) -> None:
    row = _materialize(tmp_path, [_record()]).row(0, named=True)
    assert row["text"] == row["summary_text"]
    assert "AI Engineer" in row["text"] and "senior-band" in row["text"]
    assert len(row["embed_source_hash"]) == 16


def test_materialize_never_drops(tmp_path: Path) -> None:
    second = copy.deepcopy(_record())
    second["candidate_id"] = "CAND_0000002"
    assert _materialize(tmp_path, [_record(), second]).height == 2