Files
LerNota/tests/test_categorization.py
T

421 lines
18 KiB
Python

"""Testes de categorização: palavra-chave, categorize_supplier (cache/limite/IA) e CRUD de categoria."""
import os
import re
import tempfile
import unittest
from pathlib import Path
from unittest.mock import patch
class _DbTestCase(unittest.TestCase):
def setUp(self):
self._tmp = tempfile.TemporaryDirectory()
os.environ["DB_PATH"] = str(Path(self._tmp.name) / "cat.sqlite3")
from app.config import get_settings
get_settings.cache_clear()
from app import categorization
categorization.reset_cache()
def tearDown(self):
os.environ.pop("DB_PATH", None)
os.environ.pop("OPENAI_API_KEY", None)
os.environ.pop("CATEGORIZATION_MAX_AI_CALLS_PER_BATCH", None)
os.environ.pop("ADMIN_USERNAME", None)
os.environ.pop("ADMIN_PASSWORD", None)
from app.config import get_settings
get_settings.cache_clear()
from app import categorization
categorization.reset_cache()
self._tmp.cleanup()
class FindCategoriaByKeywordTests(_DbTestCase):
def test_match_found(self):
from app import database as db
with db.session() as conn:
db.create_categoria(conn, "Alimentação", "MERCADO")
row = db.find_categoria_by_keyword(conn, "Supermercado Bom Preço")
self.assertIsNotNone(row)
self.assertEqual(row["categoria"], "Alimentação")
def test_no_match(self):
from app import database as db
with db.session() as conn:
db.create_categoria(conn, "Alimentação", "MERCADO")
row = db.find_categoria_by_keyword(conn, "Posto de Gasolina XYZ")
self.assertIsNone(row)
def test_case_insensitive(self):
from app import database as db
with db.session() as conn:
db.create_categoria(conn, "Transporte", "posto")
row = db.find_categoria_by_keyword(conn, "POSTO SHELL CENTRO")
self.assertIsNotNone(row)
self.assertEqual(row["categoria"], "Transporte")
def test_multiple_candidates_first_by_id_wins(self):
from app import database as db
with db.session() as conn:
first_id = db.create_categoria(conn, "Geral", "MERCADO")
db.create_categoria(conn, "Específico", "MERCADO CENTRAL")
row = db.find_categoria_by_keyword(conn, "Mercado Central Ltda")
self.assertEqual(row["id"], first_id)
self.assertEqual(row["categoria"], "Geral")
def test_reserved_row_never_returned_even_with_empty_keyword(self):
from app import database as db
with db.session() as conn:
# id=1 "Não Encontrado" já existe com palavra_chave='' (substring de tudo).
row = db.find_categoria_by_keyword(conn, "Qualquer Fornecedor")
self.assertIsNone(row)
self.assertEqual(db.get_categoria(conn, 1)["categoria"], "Não Encontrado")
class CategorizeSupplierTests(_DbTestCase):
def test_cache_hit_skips_ai(self):
from app import database as db
from app import categorization
os.environ["OPENAI_API_KEY"] = "test-key"
from app.config import get_settings
get_settings.cache_clear()
with db.session() as conn:
cat_id = db.create_categoria(conn, "Alimentação", "MERCADO")
batch_id = db.create_batch(conn)
with patch("app.categorization._call_ai", return_value="Alimentação") as mocked:
first = categorization.categorize_supplier("Padaria X", conn, batch_id)
second = categorization.categorize_supplier("padaria x", conn, batch_id)
self.assertEqual(first, cat_id)
self.assertEqual(second, cat_id)
mocked.assert_called_once()
def test_batch_limit_reached_skips_ai_and_uses_keyword_fallback(self):
from app import database as db
from app import categorization
os.environ["OPENAI_API_KEY"] = "test-key"
os.environ["CATEGORIZATION_MAX_AI_CALLS_PER_BATCH"] = "1"
from app.config import get_settings
get_settings.cache_clear()
with db.session() as conn:
cat_id = db.create_categoria(conn, "Alimentação", "PADARIA")
batch_id = db.create_batch(conn)
with patch("app.categorization._call_ai", return_value=None) as mocked:
categorization.categorize_supplier("Fornecedor Um", conn, batch_id)
self.assertEqual(mocked.call_count, 1)
result = categorization.categorize_supplier("Padaria Dois", conn, batch_id)
# 2º fornecedor: limite já atingido -> IA não é chamada, cai no fallback de palavra-chave
self.assertEqual(mocked.call_count, 1)
self.assertEqual(result, cat_id)
def test_ai_returns_valid_category(self):
from app import database as db
from app import categorization
os.environ["OPENAI_API_KEY"] = "test-key"
from app.config import get_settings
get_settings.cache_clear()
with db.session() as conn:
cat_id = db.create_categoria(conn, "Saúde", "FARMACIA")
batch_id = db.create_batch(conn)
with patch("app.categorization._call_ai", return_value="Saúde"):
result = categorization.categorize_supplier("Drogaria Central", conn, batch_id)
self.assertEqual(result, cat_id)
def test_ai_fails_or_invalid_falls_back_to_keyword(self):
from app import database as db
from app import categorization
os.environ["OPENAI_API_KEY"] = "test-key"
from app.config import get_settings
get_settings.cache_clear()
with db.session() as conn:
cat_id = db.create_categoria(conn, "Transporte", "POSTO")
batch_id = db.create_batch(conn)
with patch("app.categorization._call_ai", return_value=None):
result = categorization.categorize_supplier("Posto Ipiranga", conn, batch_id)
self.assertEqual(result, cat_id)
def test_no_method_finds_category_returns_reserved(self):
from app import database as db
from app import categorization
with db.session() as conn:
batch_id = db.create_batch(conn)
result = categorization.categorize_supplier("Fornecedor Desconhecido", conn, batch_id)
self.assertEqual(result, db.RESERVED_CATEGORIA_ID)
def test_unexpected_internal_failure_never_propagates_and_falls_back_to_reserved(self):
"""3.2: qualquer exceção não tratada dentro do módulo -> categoria_id=1,
nunca propaga para o chamador (que criaria fiscal_documents)."""
from app import database as db
from app import categorization
with db.session() as conn:
batch_id = db.create_batch(conn)
with patch("app.database.list_categorias", side_effect=RuntimeError("db explodiu")):
result = categorization.categorize_supplier("Fornecedor Y", conn, batch_id)
self.assertEqual(result, db.RESERVED_CATEGORIA_ID)
def test_repeated_suppliers_call_ai_once_each_and_respect_batch_limit(self):
"""Equivalente automatizado da validação manual 6.5: lote com fornecedores
repetidos chama a IA no máximo uma vez por fornecedor distinto, e respeita
o limite por lote (observável via logs)."""
from app import database as db
from app import categorization
os.environ["OPENAI_API_KEY"] = "test-key"
os.environ["CATEGORIZATION_MAX_AI_CALLS_PER_BATCH"] = "2"
from app.config import get_settings
get_settings.cache_clear()
with db.session() as conn:
db.create_categoria(conn, "Alimentação", "MERCADO")
batch_id = db.create_batch(conn)
suppliers = ["Fornecedor A", "Fornecedor A", "Fornecedor B", "Mercado C", "Fornecedor A"]
with patch("app.categorization._call_ai", return_value=None) as mocked, \
self.assertLogs("lernotafiscal.categorization", level="INFO") as logs:
for name in suppliers:
categorization.categorize_supplier(name, conn, batch_id)
# 2 fornecedores distintos com IA tentada (A, B) antes do limite=2 ser atingido
self.assertEqual(mocked.call_count, 2)
self.assertTrue(any("limite" in msg for msg in logs.output))
class CategoriaCrudTests(_DbTestCase):
def test_create_get_list_update(self):
from app import database as db
with db.session() as conn:
cat_id = db.create_categoria(conn, "Lazer", "CINEMA")
row = db.get_categoria(conn, cat_id)
self.assertEqual(row["categoria"], "Lazer")
db.update_categoria(conn, cat_id, "Entretenimento", "CINEMA")
row = db.get_categoria(conn, cat_id)
self.assertEqual(row["categoria"], "Entretenimento")
rows = db.list_categorias(conn)
self.assertTrue(any(r["id"] == cat_id for r in rows))
# linha reservada sempre presente
self.assertTrue(any(r["id"] == 1 for r in rows))
def test_delete_reassigns_referencing_documents_to_reserved(self):
from app import database as db
with db.session() as conn:
cat_id = db.create_categoria(conn, "Lazer", "CINEMA")
doc_id = db.create_fiscal(conn, mes=7, ano=2026, supplier_name="Cinema X", total_paid=30.0)
conn.execute("UPDATE fiscal_documents SET categoria_id = ? WHERE id = ?", (cat_id, doc_id))
conn.commit()
deleted = db.delete_categoria(conn, cat_id)
self.assertTrue(deleted)
self.assertIsNone(db.get_categoria(conn, cat_id))
self.assertEqual(db.get_fiscal(conn, doc_id)["categoria_id"], db.RESERVED_CATEGORIA_ID)
def test_delete_reserved_row_rejected(self):
from app import database as db
with db.session() as conn:
deleted = db.delete_categoria(conn, db.RESERVED_CATEGORIA_ID)
self.assertFalse(deleted)
self.assertIsNotNone(db.get_categoria(conn, db.RESERVED_CATEGORIA_ID))
def test_sem_categoria_group_and_filter_cover_legacy_null_rows(self):
"""Cobre o grupo/filtro 'Sem categoria' (categoria_id IS NULL, documento
legado nunca reprocessado), distinto da categoria reservada 'Não Encontrado'
(categoria_id = 1, usada pelo fluxo de categorização ativo)."""
from app import database as db
with db.session() as conn:
# Legado: create_fiscal nunca seta categoria_id -> permanece NULL.
legacy_id = db.create_fiscal(
conn, mes=7, ano=2026, supplier_name="Doc Legado", total_paid=40.0
)
self.assertIsNone(db.get_fiscal(conn, legacy_id)["categoria_id"])
# Documento processado pelo fluxo, sem match -> categoria_id = 1 (reservado).
found_id = db.create_fiscal(
conn, mes=7, ano=2026, supplier_name="Doc Não Encontrado", total_paid=15.0
)
conn.execute(
"UPDATE fiscal_documents SET categoria_id = ? WHERE id = ?",
(db.RESERVED_CATEGORIA_ID, found_id),
)
conn.commit()
totals = {r["categoria"]: r for r in db.category_totals(conn)}
self.assertIn("Sem categoria", totals)
self.assertEqual(totals["Sem categoria"]["total"], 40.0)
self.assertIsNone(totals["Sem categoria"]["categoria_id"])
self.assertIn("Não Encontrado", totals)
self.assertEqual(totals["Não Encontrado"]["total"], 15.0)
none_filtered = db.list_fiscal(conn, category="none")
self.assertEqual(len(none_filtered), 1)
self.assertEqual(none_filtered[0]["id"], legacy_id)
reserved_filtered = db.list_fiscal(conn, category=str(db.RESERVED_CATEGORIA_ID))
self.assertEqual(len(reserved_filtered), 1)
self.assertEqual(reserved_filtered[0]["id"], found_id)
self.assertEqual(db.overall_totals(conn, category="none")["count"], 1)
self.assertEqual(db.monthly_totals(conn, category="none")[0]["total"], 40.0)
self.assertEqual(db.supplier_totals(conn, category="none")[0]["supplier_name"], "Doc Legado")
class IngestionCategorizationIntegrationTests(_DbTestCase):
"""6.3: fluxo de ingestão ponta a ponta (confirm_batch) resulta em categoria_id
correto para os cenários de IA, palavra-chave e 'Não Encontrado'."""
def _stage(self, conn, batch_id, upload_id, supplier_name, total_paid):
from app import database as db
return db.insert_detected(
conn, upload_id=upload_id, batch_id=batch_id, source_file_name="f.pdf",
source_page=None, source_location="f.pdf", raw_text="",
mes=7, ano=2026, supplier_name=supplier_name, total_paid=total_paid,
confidence="high", field_confidence={}, legible=True, uncertain_fields=[], extractor="local",
)
def test_confirm_batch_assigns_categoria_via_ai_keyword_and_reserved(self):
from app import database as db
os.environ["OPENAI_API_KEY"] = "test-key"
from app.config import get_settings
get_settings.cache_clear()
with db.session() as conn:
saude_id = db.create_categoria(conn, "Saúde", "FARMACIA")
alimentacao_id = db.create_categoria(conn, "Alimentação", "MERCADO")
batch = db.create_batch(conn)
up = db.insert_upload(conn, batch, "f.pdf", Path("f.pdf"), "application/pdf", 1)
self._stage(conn, batch, up, "Drogaria Central", 10.0) # IA acerta -> Saúde
self._stage(conn, batch, up, "Mercado da Esquina", 20.0) # IA falha -> palavra-chave
self._stage(conn, batch, up, "Loja Sem Categoria Nenhuma", 30.0) # nada casa -> reservado
def fake_ai(supplier_name, known_categories):
return "Saúde" if supplier_name == "Drogaria Central" else None
with patch("app.categorization._call_ai", side_effect=fake_ai):
inserted = db.confirm_batch(conn, batch)
self.assertEqual(inserted, 3)
rows = {r["supplier_name"]: r for r in db.list_fiscal(conn)}
self.assertEqual(rows["Drogaria Central"]["categoria_id"], saude_id)
self.assertEqual(rows["Mercado da Esquina"]["categoria_id"], alimentacao_id)
self.assertEqual(rows["Loja Sem Categoria Nenhuma"]["categoria_id"], db.RESERVED_CATEGORIA_ID)
def test_confirm_batch_saves_document_with_reserved_category_when_nothing_matches(self):
"""3.2: nenhum erro bloqueia a criação do fiscal_documents; sem IA/palavra-chave
casando, o documento é salvo com categoria_id=1 ("Não Encontrado")."""
from app import database as db
with db.session() as conn:
batch = db.create_batch(conn)
up = db.insert_upload(conn, batch, "f.pdf", Path("f.pdf"), "application/pdf", 1)
self._stage(conn, batch, up, "Fornecedor Totalmente Novo", 10.0)
inserted = db.confirm_batch(conn, batch)
self.assertEqual(inserted, 1)
row = db.list_fiscal(conn)[0]
self.assertEqual(row["categoria_id"], db.RESERVED_CATEGORIA_ID)
class CategoriaFormValidationRouteTests(_DbTestCase):
"""Cobre, via requisição HTTP real (rota + CSRF + sessão), o cenário de spec
'User submits an invalid categoria form' que não tinha teste automatizado."""
def _login(self, client):
page = client.get("/login")
token = re.search(r'name="csrf_token" value="([^"]+)"', page.text).group(1)
client.post(
"/login",
data={"csrf_token": token, "username": "admin", "password": "test-pass-123"},
)
def _csrf(self, page_html: str) -> str:
return re.search(r'name="csrf_token" value="([^"]+)"', page_html).group(1)
def test_invalid_new_categoria_form_does_not_create_row(self):
from fastapi.testclient import TestClient
from app import database as db
from app.config import get_settings
from app.main import app
os.environ["ADMIN_USERNAME"] = "admin"
os.environ["ADMIN_PASSWORD"] = "test-pass-123"
get_settings.cache_clear()
with TestClient(app) as client:
self._login(client)
with db.session() as conn:
before = len(db.list_categorias(conn))
new_page = client.get("/categorias/new")
resp = client.post(
"/categorias/new",
data={"csrf_token": self._csrf(new_page.text), "categoria": "", "palavra_chave": ""},
)
self.assertIn("Informe categoria e palavra-chave", resp.text)
with db.session() as conn:
after = len(db.list_categorias(conn))
self.assertEqual(before, after)
def test_invalid_edit_categoria_form_does_not_change_row(self):
from fastapi.testclient import TestClient
from app import database as db
from app.config import get_settings
from app.main import app
os.environ["ADMIN_USERNAME"] = "admin"
os.environ["ADMIN_PASSWORD"] = "test-pass-123"
get_settings.cache_clear()
with TestClient(app) as client:
self._login(client)
with db.session() as conn:
cat_id = db.create_categoria(conn, "Lazer", "CINEMA")
edit_page = client.get(f"/categorias/{cat_id}/edit")
resp = client.post(
f"/categorias/{cat_id}/edit",
data={"csrf_token": self._csrf(edit_page.text), "categoria": "", "palavra_chave": ""},
)
self.assertIn("Informe categoria e palavra-chave", resp.text)
with db.session() as conn:
row = db.get_categoria(conn, cat_id)
self.assertEqual(row["categoria"], "Lazer")
self.assertEqual(row["palavra_chave"], "CINEMA")
if __name__ == "__main__":
unittest.main()