| """Stres testinde (pilot/) bulunan hatalar için regresyon testleri.""" |
| import sqlite3 |
| import sys |
| from pathlib import Path |
|
|
| ROOT = Path(__file__).resolve().parent.parent |
| sys.path.insert(0, str(ROOT)) |
|
|
| from t2s.extract import ResolvedTable, extract_html, resolve |
| from t2s.pipeline import merge_continuations |
| from t2s.schema import TOTAL_RE, build_table, column_name, connect, write_table |
| from t2s.trnum import fold |
|
|
|
|
| def _rt(caption, page, body, src="a.pdf", hdr=(["Ad"], ["Değer"])): |
| return ResolvedTable(caption, [list(h) for h in hdr], body, src, page, "test") |
|
|
|
|
| def test_parantezli_yil_korunur(): |
| assert column_name(["Nüfus (2025)"], 0) == "nufus_2025" |
| assert column_name(["Ücret (TL)"], 0) == "ucret" |
|
|
|
|
| def test_il_toplami_toplam_satiri(): |
| assert TOTAL_RE.search(fold("İl Toplamı")) |
| assert TOTAL_RE.search(fold("Genel Toplam")) |
| assert not TOTAL_RE.search(fold("Toplamı bilinmeyen kalem a")) |
|
|
|
|
| def test_sayfalara_bolunmus_tablo_birlesir_uc_sayfa(): |
| parts = [_rt("Tablo 6: Mahalleler", 1, [["A", "1"]]), _rt("", 2, [["B", "2"]]), _rt("", 3, [["C", "3"]]), |
| _rt("Tablo 7: Başka", 4, [["D", "4"]])] |
| out = merge_continuations(parts) |
| assert len(out) == 2 and [r[0] for r in out[0].body] == ["A", "B", "C"] |
|
|
|
|
| def test_farkli_baslikli_ardisik_tablo_birlesmez(): |
| out = merge_continuations([_rt("Bilanço", 1, [["A", "1"]]), _rt("Kar veya Zarar Tablosu", 2, [["B", "2"]])]) |
| assert len(out) == 2 |
|
|
|
|
| def test_adsiz_tablolar_birbirinin_uzerine_yazilmaz(tmp_path): |
| con = connect(tmp_path / "t.db") |
| n1 = write_table(con, build_table(_rt("", 1, [["A", "1"]], src="x.pdf"))) |
| n2 = write_table(con, build_table(_rt("", 3, [["B", "2"]], src="y.pdf"))) |
| assert n1 != n2 and "sutun" not in (n1, n2) |
| assert {t for (t,) in con.execute("SELECT tablo FROM _tablolar")} == {n1, n2} |
|
|
|
|
| def test_ayni_adli_iki_tablo_ayni_sayfada_korunur(tmp_path): |
| con = connect(tmp_path / "t.db") |
| a = write_table(con, build_table(_rt("Tablo 1: Satışlar", 1, [["A", "1"]]))) |
| b = write_table(con, build_table(_rt("Tablo 2: Satışlar", 1, [["B", "2"]]))) |
| assert a != b |
| |
| b2 = write_table(con, build_table(_rt("Tablo 2: Satışlar", 1, [["B", "3"]]))) |
| assert b2 == b and con.execute("SELECT COUNT(*) FROM _tablolar").fetchone()[0] == 2 |
|
|
|
|
| def test_tfoot_veri_sayilmaz(): |
| html = ("<table><caption>Tablo 1: X</caption><thead><tr><th>İl</th><th>Değer</th></tr></thead>" |
| "<tbody><tr><td>A</td><td>1</td></tr></tbody><tfoot><tr><td colspan='2'>Kaynak: Y</td></tr></tfoot></table>") |
| rt = resolve(extract_html(html)[0]) |
| assert [r[0] for r in rt.body] == ["A"] |
|
|
|
|
| def test_donem_satiri_baslik_bolum_satiri_govde(): |
| from t2s.extract import Cell, RawTable |
| g = [["", "Dipnot", "Cari Dönem", "Önceki Dönem"], ["", "", "31.12.2025", "31.12.2024"], |
| ["VARLIKLAR", "", "", ""], ["Nakit", "5", "1.000", "900"], ["Stoklar", "9", "2.000", "1.500"]] |
| cells = [Cell(r, c, v) for r, row in enumerate(g) for c, v in enumerate(row)] |
| rt = resolve(RawTable(cells, len(g), 4, caption="Bilanço", source="f.pdf", page=1, extractor="test")) |
| assert rt.n_header_rows == 2 and rt.body[0][0] == "VARLIKLAR" |
| bt = build_table(rt) |
| names = [c.name for c in bt.columns] |
| assert any("2025" in n for n in names) |
| assert next(c for c in bt.columns if c.original.startswith("Dipnot")).semantic == "kod" |
|
|
|
|
| def test_replace_islevi_serbest_yazma_komutu_reddedilir(): |
| from t2s.errors import static_checks |
| con = sqlite3.connect(":memory:") |
| assert not static_checks("SELECT CAST(REPLACE(REPLACE('x', '.', ''), ',', '.') AS REAL)", con) |
| for bad in ("REPLACE INTO t VALUES (1)", "WITH x AS (SELECT 1) DELETE FROM t", "SELECT 1; DROP TABLE t"): |
| assert [e.tip for e in static_checks(bad, con)] == ["SOZDIZIMI"] |
|
|
|
|
| def test_toplam_uyarisi_yalnizca_toplam_satiri_sonuca_girebiliyorsa(tmp_path): |
| from t2s.errors import static_checks |
| con = connect(tmp_path / "t.db") |
| rt = _rt("Tablo 1: İller", 1, [["Ankara", "Marmara", "10"], ["İzmir", "Ege", "20"], ["Toplam", "", "30"]], |
| hdr=(["İl"], ["Bölge"], ["Değer"])) |
| name = write_table(con, build_table(rt)) |
| assert [e.tip for e in static_checks(f'SELECT SUM(deger) FROM "{name}"', con)] == ["SAYISAL_AKIL"] |
| assert static_checks(f"SELECT SUM(deger) FROM \"{name}\" WHERE bolge = 'Ege'", con) == [] |
|
|
