"""Curatarea textului OCR si alegerea interogarii de cautare.

Fara tesseract si fara retea: `ocr.run` e singurul care atinge procese externe si
nu e testat aici — restul sunt functii pure peste text.
"""

from __future__ import annotations

import ocr


def test_clean_scoate_liniile_goale_si_ramele():
    raw = "Eroare la salvare\n\n   \n|||||\n----\nORA-01722: invalid number\n"
    assert ocr.clean(raw) == "Eroare la salvare\nORA-01722: invalid number"


def test_clean_normalizeaza_spatiile_si_taie_duplicatele():
    raw = "ORA-00001:   unique   constraint\nORA-00001: unique constraint\n"
    # a doua linie e identica dupa normalizare — tesseract repeta uneori randuri
    assert ocr.clean(raw) == "ORA-00001: unique constraint"


def test_clean_plafoneaza_si_marcheaza_trunchierea():
    out = ocr.clean("\n".join(f"linia numarul {i}" for i in range(500)), max_chars=100)
    assert len(out) < 200
    assert out.endswith("[…text trunchiat]")


def test_error_lines_gaseste_codurile_oracle_si_mesajele_romanesti():
    text = (
        "Fisier Editare Ajutor\n"
        "Nr. crt. Denumire Cantitate\n"
        "ORA-12154: TNS:could not resolve\n"
        "Documentul nu exista in baza de date\n"
        "Total general 1.234,00\n"
    )
    assert ocr.error_lines(text) == [
        "ORA-12154: TNS:could not resolve",
        "Documentul nu exista in baza de date",
    ]


def test_error_lines_limiteaza_numarul():
    text = "\n".join(f"eroare {i}" for i in range(20))
    assert len(ocr.error_lines(text, limit=3)) == 3


def test_retrieval_query_pune_legenda_prima_apoi_erorile():
    q = ocr.retrieval_query("nu pot factura", "Meniu principal\nORA-00942: table or view does not exist")
    assert q == "nu pot factura\nORA-00942: table or view does not exist"


def test_retrieval_query_fara_linii_de_eroare_cade_pe_primele_linii():
    # O captura fara niciun cuvant-cheie tot trebuie cautata cumva; mai bine
    # aproximativ decat deloc.
    text = "\n".join(f"linia {i}" for i in range(20))
    q = ocr.retrieval_query("", text)
    assert q.splitlines() == [f"linia {i}" for i in range(8)]


def test_retrieval_query_doar_legenda_cand_ocr_e_gol():
    assert ocr.retrieval_query("cum sterg o factura?", "") == "cum sterg o factura?"