112 lines
4.2 KiB
Python
112 lines
4.2 KiB
Python
from __future__ import annotations
|
|
|
|
from sqlalchemy import create_engine
|
|
from sqlalchemy.orm import sessionmaker
|
|
from sqlalchemy.pool import StaticPool
|
|
|
|
from app.services.document_intelligence import DocumentIntelligenceService, build_document_insight
|
|
|
|
|
|
def test_build_document_insight_prefers_transport_for_didi_text_with_hotel_noise() -> None:
|
|
insight = build_document_insight(
|
|
filename="didi-trip.png",
|
|
summary="滴滴出行行程单",
|
|
text="滴滴出行电子发票 订单号 12345678 上车点 深圳湾 下车点 后海 全季酒店 里程 12.4 公里 金额 48 元",
|
|
)
|
|
|
|
assert insight.document_type == "taxi_receipt"
|
|
assert insight.document_type_label == "出租车/网约车票据"
|
|
assert insight.scene_code == "transport"
|
|
assert any(field.label == "金额" and field.value == "48元" for field in insight.fields)
|
|
|
|
|
|
def test_document_intelligence_service_uses_rule_result_when_preview_available() -> None:
|
|
engine = create_engine(
|
|
"sqlite+pysqlite:///:memory:",
|
|
connect_args={"check_same_thread": False},
|
|
poolclass=StaticPool,
|
|
)
|
|
session = sessionmaker(bind=engine, autoflush=False, autocommit=False)()
|
|
try:
|
|
insight = DocumentIntelligenceService(session).build_document_insight(
|
|
filename="mixed-noise.png",
|
|
summary="OCR 混入酒店名称",
|
|
text="全季酒店 滴滴出行 订单号 12345678 上车 下车 金额 52 元",
|
|
preview_data_url="data:image/png;base64,ZmFrZQ==",
|
|
)
|
|
finally:
|
|
session.close()
|
|
|
|
assert insight.document_type == "taxi_receipt"
|
|
assert insight.classification_source == "rule"
|
|
|
|
|
|
def test_document_intelligence_extracts_larger_decimal_amount_from_multiple_candidates() -> None:
|
|
insight = build_document_insight(
|
|
filename="taxi-amount.png",
|
|
summary="滴滴出行电子行程单",
|
|
text="滴滴出行 支付金额 1 元，实付 13.4 元，订单号 12345678",
|
|
)
|
|
|
|
assert any(field.label == "金额" and field.value == "13.4元" for field in insight.fields)
|
|
|
|
|
|
def test_document_intelligence_prefers_train_ticket_for_railway_e_ticket_invoice_text() -> None:
|
|
insight = build_document_insight(
|
|
filename="铁路电子客票.pdf",
|
|
summary="电子发票（铁路电子客票）",
|
|
text=(
|
|
"电子发票（铁路电子客票）\n"
|
|
"发票号码:26319166100006175398\n"
|
|
"上海虹桥站\n"
|
|
"武汉站\n"
|
|
"G456\n"
|
|
"二等座\n"
|
|
"票价:￥354.00"
|
|
),
|
|
)
|
|
|
|
assert insight.document_type == "train_ticket"
|
|
assert insight.document_type_label == "火车/高铁票"
|
|
assert insight.scene_code == "travel"
|
|
assert any(field.label == "金额" and field.value == "354元" for field in insight.fields)
|
|
|
|
|
|
def test_document_intelligence_labels_train_ticket_date_as_train_departure_time() -> None:
|
|
insight = build_document_insight(
|
|
filename="铁路电子客票.pdf",
|
|
summary="铁路电子客票",
|
|
text=(
|
|
"中国铁路电子客票 开票日期 2026-02-18 "
|
|
"G456 上海虹桥-武汉 2026-02-20 08:30开 票价:￥354.00"
|
|
),
|
|
)
|
|
|
|
assert insight.document_type == "train_ticket"
|
|
assert any(
|
|
field.key == "date" and field.label == "列车出发时间" and field.value == "2026-02-20 08:30"
|
|
for field in insight.fields
|
|
)
|
|
assert not any(field.label == "开票日期" for field in insight.fields)
|
|
|
|
|
|
def test_document_intelligence_service_keeps_rule_fields_without_model_correction() -> None:
|
|
engine = create_engine(
|
|
"sqlite+pysqlite:///:memory:",
|
|
connect_args={"check_same_thread": False},
|
|
poolclass=StaticPool,
|
|
)
|
|
session = sessionmaker(bind=engine, autoflush=False, autocommit=False)()
|
|
try:
|
|
insight = DocumentIntelligenceService(session).build_document_insight(
|
|
filename="didi-corrected.png",
|
|
summary="滴滴出行电子行程单",
|
|
text="滴滴出行 支付金额 1 元 订单号 12345678",
|
|
preview_data_url="data:image/png;base64,ZmFrZQ==",
|
|
)
|
|
finally:
|
|
session.close()
|
|
|
|
assert any(field.label == "金额" and field.value == "1元" for field in insight.fields)
|
|
assert insight.warnings == ()
|