refactor(server): split oversized backend services

This commit is contained in:
caoxiaozhu
2026-05-22 10:42:31 +08:00
parent 2e57702638
commit 222ba0bfdc
84 changed files with 26263 additions and 21898 deletions

View File

@@ -0,0 +1,120 @@
from __future__ import annotations
from app.services.document_intelligence_types import DocumentRule
DEFAULT_RULE = DocumentRule(
document_type="other",
document_type_label="其他单据",
scene_code="other",
scene_label="其他票据",
expense_type="other",
keywords=(),
score_bias=0.0,
)
DOCUMENT_RULES: tuple[DocumentRule, ...] = (
DocumentRule(
document_type="flight_itinerary",
document_type_label="机票/航班行程单",
scene_code="travel",
scene_label="差旅票据",
expense_type="travel",
keywords=("电子行程单", "航班号", "航班", "机票", "登机", "航空", "客票"),
score_bias=0.34,
),
DocumentRule(
document_type="train_ticket",
document_type_label="火车/高铁票",
scene_code="travel",
scene_label="差旅票据",
expense_type="travel",
keywords=("铁路电子客票", "电子客票", "高铁", "火车", "动车", "铁路", "车次", "检票", "二等座", "一等座", "票价"),
score_bias=0.32,
),
DocumentRule(
document_type="hotel_invoice",
document_type_label="酒店住宿票据",
scene_code="hotel",
scene_label="住宿票据",
expense_type="hotel",
keywords=("住宿", "房费", "客房", "入住", "离店", "酒店", "宾馆", "间夜"),
score_bias=0.16,
),
DocumentRule(
document_type="taxi_receipt",
document_type_label="出租车/网约车票据",
scene_code="transport",
scene_label="交通票据",
expense_type="transport",
keywords=("滴滴出行", "滴滴", "网约车", "出租车", "打车", "乘车", "用车", "叫车", "车费", "车资", "的士", "快车", "专车", "订单号", "上车", "下车", "起点", "终点", "里程", "司机"),
score_bias=0.38,
),
DocumentRule(
document_type="parking_toll_receipt",
document_type_label="停车/通行费票据",
scene_code="transport",
scene_label="交通票据",
expense_type="transport",
keywords=("停车费", "通行费", "过路费", "收费站", "停车场", "停车"),
score_bias=0.28,
),
DocumentRule(
document_type="meal_receipt",
document_type_label="餐饮票据",
scene_code="meal",
scene_label="餐饮票据",
expense_type="meal",
keywords=("餐饮", "餐费", "用餐", "饭店", "酒楼", "餐厅", "食品", "外卖", "咖啡"),
score_bias=0.14,
),
DocumentRule(
document_type="office_invoice",
document_type_label="办公用品票据",
scene_code="office",
scene_label="办公用品票据",
expense_type="office",
keywords=("办公用品", "文具", "耗材", "打印纸", "墨盒", "硒鼓", "键盘", "鼠标"),
score_bias=0.14,
),
DocumentRule(
document_type="meeting_invoice",
document_type_label="会议/会务票据",
scene_code="meeting",
scene_label="会务票据",
expense_type="meeting",
keywords=("会议", "会务", "会展", "论坛", "会议室", "会场"),
score_bias=0.12,
),
DocumentRule(
document_type="training_invoice",
document_type_label="培训票据",
scene_code="training",
scene_label="培训票据",
expense_type="training",
keywords=("培训", "课程", "讲师", "教材", "学费", "认证"),
score_bias=0.12,
),
DocumentRule(
document_type="vat_invoice",
document_type_label="增值税发票",
scene_code="other",
scene_label="通用发票",
expense_type="other",
keywords=("发票代码", "发票号码", "价税合计", "增值税", "电子发票"),
score_bias=-0.08,
),
DocumentRule(
document_type="receipt",
document_type_label="一般收据/凭证",
scene_code="other",
scene_label="其他票据",
expense_type="other",
keywords=("收据", "凭证", "票据"),
score_bias=-0.18,
),
)
DOCUMENT_TYPE_RULE_MAP = {rule.document_type: rule for rule in DOCUMENT_RULES}
SUPPORTED_DOCUMENT_TYPES = tuple(DOCUMENT_TYPE_RULE_MAP.keys()) + ("other",)