test_validator.py

"""Smoke tests for the submission validator. Run: ``python test_validator.py``.

No pytest dependency — plain asserts against real sample files in the repo plus a
deliberately corrupted in-memory copy. Exits non-zero on first failure.
"""

from __future__ import annotations

import csv
import os
import tempfile

import openpyxl

import hypotheses
import validator

REPO = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))

GOOD_XLSX = [
    ("260428_coalescing_with_John2/M3H1_decreases_phagocytosis_papers_and_findings_Oishi.xlsx",
     hypotheses.BY_SLUG["3h1-decreased-phagocytosis"]),
    ("260421_New_files/abca1-reduces-ad-risk_submission.xlsx", None),
    ("260421_New_files/apoe4-increases-amyloid_submission.xlsx", None),
]


def _p(rel: str) -> str:
    return os.path.join(REPO, rel)


def check(name: str, cond: bool, detail: str = "") -> None:
    status = "ok  " if cond else "FAIL"
    print(f"[{status}] {name}{(' — ' + detail) if detail else ''}")
    if not cond:
        raise SystemExit(1)


def test_good_xlsx() -> None:
    for rel, hyp in GOOD_XLSX:
        parsed = validator.parse(_p(rel))
        rep = validator.validate(parsed, hyp)
        n = rep.summary
        check(f"good xlsx parses+validates: {os.path.basename(rel)}", rep.ok,
              f"{n['n_papers']}p/{n['n_findings']}f errors={rep.errors}")
        check(f"  has papers: {os.path.basename(rel)}", n["n_papers"] > 0)


def test_csv_roundtrip() -> None:
    """Export a good xlsx to CSV and confirm it validates identically."""
    rel, hyp = GOOD_XLSX[0]
    parsed_x = validator.parse(_p(rel))
    grid, _ = validator._load_grid(_p(rel))  # noqa: SLF001 - intentional in test
    with tempfile.NamedTemporaryFile("w", suffix=".csv", delete=False, newline="") as f:
        csv.writer(f).writerows(
            [["" if c is None else c for c in row] for row in grid])
        csv_path = f.name
    parsed_c = validator.parse(csv_path)
    os.unlink(csv_path)
    check("csv export validates", validator.validate(parsed_c, hyp).ok)
    check("csv paper count matches xlsx",
          len(parsed_c.papers) == len(parsed_x.papers),
          f"{len(parsed_c.papers)} vs {len(parsed_x.papers)}")


def test_corrupted() -> None:
    """Build a broken sheet: bad relevance (5), literal 'None' DOI, a paper with no findings."""
    wb = openpyxl.Workbook(); ws = wb.active
    ws.append(validator.CANONICAL_HEADERS)
    ws.append([""] + [None] * 13)  # empty hypothesis
    ws.append([None, "None", "PMID", "111", None, "P1", None])  # 'None' DOI, no paper_id useful
    ws.append([None, None, None, None, "a finding", "P1.F1", 5])  # relevance out of range
    ws.append([None, "10.1/x", "PMID", "222", None, "P2", None])  # P2 has no findings
    with tempfile.NamedTemporaryFile(suffix=".xlsx", delete=False) as f:
        wb.save(f.name); path = f.name
    parsed = validator.parse(path)
    os.unlink(path)
    rep = validator.validate(parsed, hypotheses.BY_CODE["3H1"])
    check("corrupted file fails validation", not rep.ok, f"errors={len(rep.errors)}")
    joined = " | ".join(rep.errors)
    check("  flags empty hypothesis", "A2" in joined)
    check("  flags out-of-range relevance", "out of the 0–1" in joined)
    check("  flags paper with no findings", "no findings" in joined)

    err_issues = [i for i in rep.issues if i.severity == "error"]
    check("  every error has an actionable fix",
          all(i.fix for i in err_issues), f"{len(err_issues)} error issues")
    check("  every error is anchored to a location",
          all(i.anchor for i in err_issues))
    a2 = next(i for i in rep.issues if "A2" in i.message)
    check("  empty-hypothesis issue anchors to A2", a2.anchor == "A2")


def test_is_valid_url() -> None:
    check("accepts https url", validator.is_valid_url("https://github.com/you/repo"))
    check("accepts http url", validator.is_valid_url("http://example.com/wiki"))
    check("rejects bare text", not validator.is_valid_url("not a url"))
    check("rejects scheme-less host", not validator.is_valid_url("github.com/you"))
    check("rejects empty", not validator.is_valid_url(""))


def test_canonical_emit() -> None:
    parsed = validator.parse(_p(GOOD_XLSX[0][0]))
    xlsx_bytes, js = validator.to_canonical(parsed)
    check("canonical xlsx is non-empty", len(xlsx_bytes) > 0)
    check("json has hypothesis+papers", "hypothesis" in js and len(js["papers"]) > 0)
    check("json findings_text is plain strings",
          all(isinstance(t, str) for p in js["papers"] for t in p["findings_text"]))


if __name__ == "__main__":
    test_good_xlsx()
    test_csv_roundtrip()
    test_corrupted()
    test_is_valid_url()
    test_canonical_emit()
    print("\nAll validator smoke tests passed.")