tests/test_resolve.py

"""Paste-anything target resolution (dee/core/resolve.py).

NO NETWORK — Ensembl/NCBI fetchers are monkeypatched. Locks the input
classifier (sequence vs symbol vs accession) and the resolve dispatch,
including the privacy-preserving rule that raw sequences make no calls.
"""
import pytest

from dee.core import resolve as R
from dee.core import exon as E
from dee.core.exon import Exon, GeneStructure


# ─────────────────────────── classifier ───────────────────────────

@pytest.mark.parametrize("text,kind", [
    ("",                         "empty"),
    ("ACGT" * 10,                "sequence"),          # 40 nt one-liner
    (">hdr\nACGTACGTACGTACGTACGTACGT", "sequence"),    # FASTA
    ("TP53",                     "symbol"),
    ("BRCA1",                    "symbol"),
    ("ENST00000269305",          "ensembl_tx"),
    ("ENSMUST00000000001",       "ensembl_tx"),
    ("ENSG00000141510",          "ensembl_gene"),
    ("NM_000546",                "refseq"),
    ("NM_000546.6",              "refseq"),
    ("XM_011535573",             "refseq"),
    ("!!!",                      "unknown"),
])
def test_classify(text, kind):
    assert R.classify(text)[0] == kind


def test_classify_sequence_strips_to_clean_dna():
    kind, val = R.classify("acgt ACGT\nnnNN gtca gtca gtca")
    assert kind == "sequence"
    assert val == "ACGTACGTNNNNGTCAGTCAGTCA"


# ─────────────────────────── raw sequence (offline) ───────────────────────────

def test_resolve_raw_sequence_makes_no_network_call(monkeypatch):
    # Any network helper firing would be a privacy bug for raw input.
    def _boom(*a, **k):
        raise AssertionError("network call on raw-sequence path!")
    monkeypatch.setattr(E, "_http_get_text", _boom)
    monkeypatch.setattr(E, "_http_get_json", _boom)
    monkeypatch.setattr(E, "fetch_gene_structure", _boom)
    out = R.resolve_target("ACGT" * 20)
    assert out["ok"] and out["kind"] == "sequence"
    assert out["sequence"] == "ACGT" * 20
    assert "pasted sequence" in out["label"]


def test_resolve_short_sequence_errors():
    out = R.resolve_target("ACGTACGT")          # < 23 nt → symbol attempt → fails
    assert out["ok"] is False


# ─────────────────────────── gene symbol ───────────────────────────

def _fake_gene():
    return GeneStructure(
        organism="human", gene_symbol="TP53", transcript_id="ENST00000269305",
        strand=1, cds_sequence="ATG" + "GCT" * 50 + "TAA",
        exons=[Exon(1, 1, 99, 0, 156)], last_junction_cds_pos=0,
    )


def test_resolve_symbol_requires_organism():
    out = R.resolve_target("TP53", organism="")
    assert out["ok"] is False and "Human or" in out["error"]


def test_resolve_symbol_with_organism(monkeypatch):
    monkeypatch.setattr(E, "fetch_gene_structure", lambda org, sym: _fake_gene())
    out = R.resolve_target("TP53", organism="human")
    assert out["ok"] and out["kind"] == "gene"
    assert out["gene_symbol"] == "TP53"
    assert out["sequence"].startswith("ATG")
    assert "ENST00000269305" in out["label"]


def test_resolve_symbol_not_found(monkeypatch):
    monkeypatch.setattr(E, "fetch_gene_structure", lambda org, sym: None)
    out = R.resolve_target("ZZZ9", organism="mouse")
    assert out["ok"] is False and "Couldn't find" in out["error"]


# ─────────────────────────── accessions ───────────────────────────

def test_resolve_ensembl_transcript(monkeypatch):
    monkeypatch.setattr(E, "_http_get_text",
                        lambda url, **k: ">ENST\nATGAAACCCGGGTTTACGTACGTACGT")
    out = R.resolve_target("ENST00000269305")
    assert out["ok"] and out["kind"] == "ensembl"
    assert out["sequence"] == "ATGAAACCCGGGTTTACGTACGTACGT"


def test_resolve_ensembl_gene_resolves_canonical(monkeypatch):
    monkeypatch.setattr(E, "_http_get_json", lambda url, **k: {
        "Transcript": [
            {"id": "ENST_other", "is_canonical": 0},
            {"id": "ENST_canon", "is_canonical": 1},
        ]})
    captured = {}
    def _seq(url, **k):
        captured["url"] = url
        return ">x\nATGCGTACGTACGTACGTACGTACG"
    monkeypatch.setattr(E, "_http_get_text", _seq)
    out = R.resolve_target("ENSG00000141510")
    assert out["ok"] and out["kind"] == "ensembl"
    assert "ENST_canon" in captured["url"]          # used the canonical transcript


def test_resolve_refseq(monkeypatch):
    monkeypatch.setattr(E, "_http_get_text",
                        lambda url, **k: ">NM_000546\nATGGAGGAGCCGCAGTCAGAT")
    out = R.resolve_target("NM_000546")
    assert out["ok"] and out["kind"] == "refseq"
    assert out["sequence"] == "ATGGAGGAGCCGCAGTCAGAT"


def test_resolve_ensembl_fetch_failure(monkeypatch):
    monkeypatch.setattr(E, "_http_get_text", lambda url, **k: None)
    out = R.resolve_target("ENST00000269305")
    assert out["ok"] is False and "Ensembl" in out["error"]


# ─────────────────────────── UniProt / structure (M5) ───────────────────────────

def test_resolve_uniprot_ok(monkeypatch):
    captured = {}
    def _json(url, **k):
        captured["url"] = url
        return {"results": [{"primaryAccession": "P04637"}]}
    monkeypatch.setattr(E, "_http_get_json", _json)
    out = R.resolve_uniprot("human", "TP53")
    assert out["ok"] and out["uniprot"] == "P04637"
    assert out["alphafold_url"] == "https://alphafold.ebi.ac.uk/files/AF-P04637-F1-model_v6.pdb"
    assert out["alphafold_page"] == "https://alphafold.ebi.ac.uk/entry/P04637"
    assert "organism_id%3A9606" in captured["url"] and "reviewed%3Atrue" in captured["url"]


def test_resolve_uniprot_no_hit(monkeypatch):
    monkeypatch.setattr(E, "_http_get_json", lambda url, **k: {"results": []})
    out = R.resolve_uniprot("human", "ZZZ9")
    assert out["ok"] is False


def test_resolve_uniprot_bad_organism():
    assert R.resolve_uniprot("ecoli", "lacZ")["ok"] is False