YG-Soft/X-Financial
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(server): 票据文件夹资产缓存与文档预览统一生成

Browse Source 
- 新增 document_preview 模块，DocumentPreviewAssets 统一处理 data URL 解码、pdftoppm PNG 预览生成（poppler-data 编码）、renderer_id 标识
- receipt_folder 服务复用预览生成，缓存票据资产并提供清理；删除票据时保留已关联报销单的附件副本
- document_intelligence 新增票据预览/资产缓存接入与字段提取增强；ocr 抽取复用预览工具，附件分析/文档/操作/展示四个子模块同步适配
- receipt_folder 端点补充资产缓存头，补/扩 document_intelligence、ocr_endpoints、ocr_service、receipt_folder_service、reimbursement_endpoints 测试，新增 attachment_analysis 回归测试
This commit is contained in:
caoxiaozhu
2026-06-23 09:42:00 +08:00
parent bc743adef3
commit 84a8998e59
15 changed files with 1076 additions and 79 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
server/tests/test_document_intelligence.py
View File

@@ -84,6 +84,35 @@ def test_document_intelligence_prefers_train_ticket_for_railway_e_ticket_invoice
assert any(field.label == "金额" and field.value == "354元" for field in insight.fields)
def test_document_intelligence_recovers_train_ticket_from_english_station_ocr_text() -> None:
insight = build_document_insight(
filename="2月20_武汉-上海.pdf",
summary=":26429165800002785705；:2026 05 18；Wuhan Shanghaihongqiao G458",
text=(
":26429165800002785705\n"
":2026 05 18\n"
"G458\n"
"Wuhan\n"
"Shanghaihongqiao\n"
"2026 02 20 07:55\n"
"06 01B\n"
": 354.00\n"
"4201061987****1615\n"
":6580061086021391007342026\n"
"12306 95306"
),
)
assert insight.document_type == "train_ticket"
assert insight.document_type_label == "火车/高铁票"
assert insight.scene_code == "travel"
fields = {field.label: field.value for field in insight.fields}
assert fields["金额"] == "354元"
assert fields["列车出发时间"] == "2026-02-20 07:55"
assert fields["车次/航班"] == "G458"
assert fields["行程"] == "武汉-上海"
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",

169
server/tests/test_expense_claim_attachment_analysis_regression.py Normal file
View File

@@ -0,0 +1,169 @@
from __future__ import annotations
import json
from decimal import Decimal
from app.schemas.ocr import OcrRecognizeBatchRead, OcrRecognizeDocumentRead
from app.services.expense_claim_attachment_storage import ExpenseClaimAttachmentStorage
from app.services.ocr import OcrService
from test_reimbursement_endpoints import build_client, seed_claim
def test_train_ticket_attachment_with_structured_fields_is_not_flagged_as_unreadable(
monkeypatch,
tmp_path,
) -> None:
def fake_recognize(
self,
files: list[tuple[str, bytes, str | None]],
) -> OcrRecognizeBatchRead:
return OcrRecognizeBatchRead(
total_file_count=1,
success_count=1,
documents=[
OcrRecognizeDocumentRead(
filename="2月20_武汉-上海.pdf",
media_type="application/pdf",
text=(
":26429165800002785705\n"
":2026 05 18\n"
"G458\n"
"Wuhan\n"
"Shanghaihongqiao\n"
"2026 02 20 07:55\n"
"06 01B\n"
": 354.00\n"
"4201061987****1615\n"
":6580061086021391007342026\n"
"12306 95306"
),
summary="Wuhan Shanghaihongqiao G458 354.00",
avg_score=0.0,
line_count=0,
page_count=1,
warnings=[],
)
],
)
monkeypatch.setattr(OcrService, "recognize_files", fake_recognize)
monkeypatch.setattr(ExpenseClaimAttachmentStorage, "root", lambda self: tmp_path)
client, session_factory = build_client()
with session_factory() as db:
claim, item = seed_claim(db)
claim.expense_type = "travel"
claim.reason = "武汉-上海差旅"
claim.location = "上海"
claim.amount = Decimal("354.00")
item.item_type = "train_ticket"
item.item_reason = "武汉-上海"
item.item_location = "上海"
item.item_amount = Decimal("354.00")
db.commit()
claim_id = claim.id
item_id = item.id
upload_response = client.post(
f"/api/v1/reimbursements/claims/{claim_id}/items/{item_id}/attachment",
headers={"x-auth-username": "emp-1", "x-auth-name": "Zhang San"},
files=[("file", ("2月20_武汉-上海.pdf", b"%PDF-1.4 fake", "application/pdf"))],
)
assert upload_response.status_code == 200
attachment = upload_response.json()["attachment"]
analysis = attachment["analysis"]
points = analysis["points"]
assert attachment["document_info"]["document_type"] == "train_ticket"
assert analysis["severity"] == "pass"
assert not any("未识别到有效文字" in point for point in points)
assert not any("未识别到列车出发时间" in point for point in points)
def test_attachment_meta_read_repairs_stale_unreadable_train_ticket_analysis(
monkeypatch,
tmp_path,
) -> None:
def fake_recognize(
self,
files: list[tuple[str, bytes, str | None]],
) -> OcrRecognizeBatchRead:
return OcrRecognizeBatchRead(
total_file_count=1,
success_count=1,
documents=[
OcrRecognizeDocumentRead(
filename="2月20_武汉-上海.pdf",
media_type="application/pdf",
text=(
":26429165800002785705 :2026 05 18\n"
"G458\n"
"Wuhan Shanghaihongqiao\n"
"2026 02 20 07:55 06 01B\n"
": 354.00\n"
"4201061987****1615\n"
":6580061086021391007342026\n"
"12306 95306"
),
summary="Wuhan Shanghaihongqiao G458 354.00",
avg_score=0.0,
line_count=0,
page_count=1,
warnings=[],
)
],
)
monkeypatch.setattr(OcrService, "recognize_files", fake_recognize)
monkeypatch.setattr(ExpenseClaimAttachmentStorage, "root", lambda self: tmp_path)
client, session_factory = build_client()
with session_factory() as db:
claim, item = seed_claim(db)
claim.expense_type = "travel"
claim.reason = "武汉-上海差旅"
claim.location = "上海"
claim.amount = Decimal("354.00")
item.item_type = "train_ticket"
item.item_reason = "武汉-上海"
item.item_location = "上海"
item.item_amount = Decimal("354.00")
db.commit()
claim_id = claim.id
item_id = item.id
upload_response = client.post(
f"/api/v1/reimbursements/claims/{claim_id}/items/{item_id}/attachment",
headers={"x-auth-username": "emp-1", "x-auth-name": "Zhang San"},
files=[("file", ("2月20_武汉-上海.pdf", b"%PDF-1.4 fake", "application/pdf"))],
)
assert upload_response.status_code == 200
meta_path = next(tmp_path.rglob("*.meta.json"))
meta = json.loads(meta_path.read_text(encoding="utf-8"))
meta["analysis"] = {
"severity": "high",
"label": "高风险",
"headline": "AI提示：附件不符合票据校验条件",
"summary": "当前附件存在明显异常，票据类型与当前费用场景不匹配，或无法作为有效报销材料。",
"points": [
"附件内容：未识别到有效文字，当前附件更像普通图片或内容过于模糊。",
"日期字段：未识别到列车出发时间或乘车日期。",
],
"rule_basis": [],
"suggestion": "建议过滤当前不匹配的票据，重新上传符合当前费用场景的清晰原件。",
}
meta_path.write_text(json.dumps(meta, ensure_ascii=False), encoding="utf-8")
meta_response = client.get(
f"/api/v1/reimbursements/claims/{claim_id}/items/{item_id}/attachment/meta",
headers={"x-auth-username": "emp-1", "x-auth-name": "Zhang San"},
)
assert meta_response.status_code == 200
analysis = meta_response.json()["analysis"]
points = analysis["points"]
assert analysis["severity"] == "pass"
assert not any("未识别到有效文字" in point for point in points)
assert not any("未识别到列车出发时间" in point for point in points)

70
server/tests/test_ocr_endpoints.py
View File

@@ -176,3 +176,73 @@ def test_ocr_recognize_endpoint_returns_structured_payload(monkeypatch, tmp_path
assert deleted_response.status_code == 404
finally:
get_settings.cache_clear()
def test_ocr_recognize_endpoint_returns_receipt_enriched_train_fields(monkeypatch, tmp_path) -> None:
def fake_recognize(
self,
files: list[tuple[str, bytes, str | None]],
) -> OcrRecognizeBatchRead:
return OcrRecognizeBatchRead(
engine="paddleocr_mobile",
model="PP-OCRv5_mobile",
total_file_count=1,
success_count=1,
documents=[
OcrRecognizeDocumentRead(
filename="2月20_武汉-上海.png",
media_type="image/png",
text=(
":26429165800002785705\n"
"G458\n"
"Wuhan\n"
"Shanghaihongqiao\n"
"2026 02 20 07:55\n"
"06 01B\n"
": 354.00\n"
"4201061987****1615\n"
":6580061086021391007342026\n"
"12306 95306"
),
summary="Wuhan Shanghaihongqiao G458 354.00",
avg_score=0.92,
line_count=0,
page_count=1,
document_type="train_ticket",
document_type_label="火车/高铁票",
scene_code="travel",
scene_label="差旅票据",
document_fields=[
OcrRecognizeFieldRead(key="date", label="列车出发时间", value="2026-02-20 07:55"),
OcrRecognizeFieldRead(key="trip_no", label="车次/航班", value="G458"),
OcrRecognizeFieldRead(key="route", label="行程", value="武汉-上海"),
OcrRecognizeFieldRead(key="amount", label="金额", value="354元"),
],
)
],
)
monkeypatch.setenv("STORAGE_ROOT_DIR", str(tmp_path / "storage"))
get_settings.cache_clear()
monkeypatch.setattr(OcrService, "recognize_files", fake_recognize)
try:
client = build_client()
response = client.post(
"/api/v1/ocr/recognize",
headers={"x-auth-username": "pytest", "x-auth-name": "Py Test"},
files=[("files", ("2月20_武汉-上海.png", b"fake-image", "image/png"))],
)
finally:
get_settings.cache_clear()
assert response.status_code == 200
document = response.json()["documents"][0]
fields = {
item["label"]: item["value"]
for item in document["document_fields"]
}
assert document["receipt_id"]
assert fields["身份证号"] == "4201061987****1615"
assert fields["车厢"] == "06车"
assert fields["座位号"] == "01B"
assert fields["票价"] == "354.00元"

49
server/tests/test_ocr_service.py
View File

@@ -101,6 +101,55 @@ print("__OCR_JSON__=" + json.dumps(payload, ensure_ascii=False))
assert skipped.warnings == ["当前仅支持图片和 PDF 文件进行 OCR。"]
def test_ocr_service_recovers_image_text_from_worker_ocr_text(
monkeypatch,
tmp_path: Path,
) -> None:
def fake_invoke_worker(
self,
*,
python_bin: str,
worker_path: str,
input_paths: list[Path],
) -> dict:
return {
"engine": "paddleocr_mobile",
"model": "PP-OCRv5_mobile",
"documents": [
{
"input_path": str(input_paths[0]),
"engine": "paddleocr_mobile",
"model": "PP-OCRv5_mobile",
"ocr_text": "铁路电子客票 武汉-上海 2026 02 20 07:55 G458 : 354.00 12306 95306",
"avg_score": 0.92,
"line_count": 0,
"page_count": 1,
"warnings": [],
"lines": [],
}
],
}
monkeypatch.setenv("STORAGE_ROOT_DIR", str(tmp_path / "storage"))
monkeypatch.setattr(OcrService, "_resolve_python_bin", lambda self: "python")
monkeypatch.setattr(OcrService, "_resolve_worker_path", lambda self: "worker.py")
monkeypatch.setattr(OcrService, "_invoke_worker", fake_invoke_worker)
OcrService._result_cache.clear()
get_settings.cache_clear()
try:
result = OcrService().recognize_files([("train-ticket.png", b"fake-train-image", "image/png")])
finally:
OcrService._result_cache.clear()
get_settings.cache_clear()
recognized = result.documents[0]
assert "铁路电子客票" in recognized.text
assert recognized.document_type == "train_ticket"
assert any(field.label == "列车出发时间" and field.value == "2026-02-20 07:55" for field in recognized.document_fields)
assert any(field.label == "车次/航班" and field.value == "G458" for field in recognized.document_fields)
assert any(field.label == "金额" and field.value == "354元" for field in recognized.document_fields)
def test_ocr_service_passes_configured_device_to_worker(
monkeypatch,
tmp_path: Path,

169
server/tests/test_receipt_folder_service.py
View File

@@ -1,8 +1,11 @@
from __future__ import annotations
import base64
from app.api.deps import CurrentUserContext
from app.core.config import get_settings
from app.schemas.ocr import OcrRecognizeDocumentRead
from app.services.document_preview import DocumentPreviewAssets
from app.services.receipt_folder import ReceiptFolderService
@@ -69,6 +72,172 @@ def test_receipt_folder_train_ticket_uses_invoice_date_and_enriches_fields(monke
get_settings.cache_clear()
def test_receipt_folder_pdf_preview_regenerates_stale_cached_image(monkeypatch, tmp_path) -> None:
monkeypatch.setenv("STORAGE_ROOT_DIR", str(tmp_path / "storage"))
get_settings.cache_clear()
try:
current_user = CurrentUserContext(
username="pytest",
name="Py Test",
role_codes=[],
is_admin=False,
)
stale_preview = b"stale-preview"
preview_data_url = f"data:image/png;base64,{base64.b64encode(stale_preview).decode('ascii')}"
service = ReceiptFolderService()
receipt = service.save_receipt(
filename="2月20_武汉-上海.pdf",
content=b"%PDF-1.4 fake",
media_type="application/pdf",
current_user=current_user,
document=OcrRecognizeDocumentRead(
filename="2月20_武汉-上海.pdf",
media_type="application/pdf",
preview_kind="image",
preview_data_url=preview_data_url,
),
)
receipt_dir = next(service.root.glob("pytest/*"))
preview_path = receipt_dir / "preview.png"
assert preview_path.read_bytes() == stale_preview
stale_meta = service._read_meta(receipt_dir)
stale_meta.pop("preview_rendered_with", None)
service._write_meta(receipt_dir, stale_meta)
def fake_render_pdf_first_page(*, pdf_path, preview_path, timeout_seconds):
preview_path.write_bytes(b"refreshed-preview")
return preview_path
monkeypatch.setattr(DocumentPreviewAssets, "render_pdf_first_page", fake_render_pdf_first_page)
resolved_path, media_type, file_name = service.resolve_preview(receipt.id, current_user)
assert resolved_path == preview_path
assert media_type == "image/png"
assert file_name == "preview.png"
assert preview_path.read_bytes() == b"refreshed-preview"
meta = service._read_meta(receipt_dir)
assert meta["preview_rendered_with"] == DocumentPreviewAssets.PDF_RENDERER_ID
finally:
get_settings.cache_clear()
def test_receipt_folder_delete_removes_duplicate_marker(monkeypatch, tmp_path) -> None:
monkeypatch.setenv("STORAGE_ROOT_DIR", str(tmp_path / "storage"))
get_settings.cache_clear()
try:
current_user = CurrentUserContext(
username="pytest",
name="Py Test",
role_codes=[],
is_admin=False,
)
service = ReceiptFolderService()
content = b"%PDF-1.4 same receipt"
receipt = service.save_receipt(
filename="same-receipt.pdf",
content=content,
media_type="application/pdf",
current_user=current_user,
document=OcrRecognizeDocumentRead(
filename="same-receipt.pdf",
media_type="application/pdf",
text="same receipt amount 354",
document_type="other",
document_type_label="其他单据",
scene_code="other",
scene_label="其他票据",
),
)
receipt_dir = service.root / "pytest" / receipt.id
assert receipt_dir.exists()
duplicate = service.find_duplicate_receipt(
filename="same-receipt.pdf",
content=content,
current_user=current_user,
)
assert duplicate is not None
assert duplicate.id == receipt.id
service.delete_receipt(receipt_id=receipt.id, current_user=current_user)
assert not receipt_dir.exists()
assert (
service.find_duplicate_receipt(
filename="same-receipt.pdf",
content=content,
current_user=current_user,
)
is None
)
finally:
get_settings.cache_clear()
def test_receipt_folder_recovers_train_ticket_detail_from_other_english_ocr(monkeypatch, tmp_path) -> None:
monkeypatch.setenv("STORAGE_ROOT_DIR", str(tmp_path / "storage"))
get_settings.cache_clear()
try:
current_user = CurrentUserContext(
username="pytest",
name="Py Test",
role_codes=[],
is_admin=False,
)
service = ReceiptFolderService()
receipt = service.save_receipt(
filename="2月20_武汉-上海.pdf",
content=b"%PDF-1.4 fake",
media_type="application/pdf",
current_user=current_user,
document=OcrRecognizeDocumentRead(
filename="2月20_武汉-上海.pdf",
media_type="application/pdf",
text=(
":26429165800002785705\n"
":2026 05 18\n"
"G458\n"
"Wuhan\n"
"Shanghaihongqiao\n"
"2026 02 20 07:55\n"
"06 01B\n"
": 354.00\n"
"4201061987****1615\n"
":6580061086021391007342026\n"
"12306 95306"
),
summary="Wuhan Shanghaihongqiao G458 354.00",
document_type="other",
document_type_label="其他单据",
scene_code="other",
scene_label="其他票据",
),
)
assert receipt.document_type == "train_ticket"
assert receipt.document_type_label == "火车/高铁票"
assert receipt.scene_code == "travel"
assert receipt.amount == "354.00元"
assert receipt.document_date == "2026-02-20"
assert receipt.merchant_name == "中国铁路"
detail = service.get_receipt(receipt.id, current_user)
fields = {field.label: field.value for field in detail.fields}
assert fields["行程"] == "武汉-上海"
assert fields["车次"] == "G458"
assert fields["列车出发时间"] == "2026-02-20 07:55"
assert fields["票价"] == "354.00元"
assert fields["身份证号"] == "4201061987****1615"
assert fields["车厢"] == "06车"
assert fields["座位号"] == "01B"
assert "乘车人" not in fields
finally:
get_settings.cache_clear()
def test_receipt_folder_unlink_receipts_for_claim_marks_linked_receipts_unlinked(monkeypatch, tmp_path) -> None:
monkeypatch.setenv("STORAGE_ROOT_DIR", str(tmp_path / "storage"))
get_settings.cache_clear()

5
server/tests/test_reimbursement_endpoints.py
View File

@@ -1,6 +1,7 @@
from __future__ import annotations
import base64
import json
from collections.abc import Generator
from datetime import UTC, date, datetime
from decimal import Decimal
@@ -19,6 +20,7 @@ from app.models.organization import OrganizationUnit
from app.models.risk_observation import RiskObservation, RiskObservationFeedback
from app.models.role import Role
from app.schemas.ocr import OcrRecognizeBatchRead, OcrRecognizeDocumentRead
from app.services.document_preview import DocumentPreviewAssets
from app.services.expense_claim_attachment_storage import ExpenseClaimAttachmentStorage
from app.services.ocr import OcrService
@@ -686,6 +688,9 @@ def test_claim_item_pdf_attachment_preview_returns_generated_image(monkeypatch,
meta_payload = upload_response.json()["attachment"]
assert meta_payload["preview_kind"] == "image"
assert meta_payload["preview_url"].endswith(f"/reimbursements/claims/{claim_id}/items/{item_id}/attachment/preview")
meta_path = next(tmp_path.rglob("invoice.pdf.meta.json"))
stored_meta = json.loads(meta_path.read_text(encoding="utf-8"))
assert stored_meta["preview_rendered_with"] == DocumentPreviewAssets.PDF_RENDERER_ID
preview_response = client.get(
f"/api/v1/reimbursements/claims/{claim_id}/items/{item_id}/attachment/preview",