1

native-engine/build-engine.sh

@@ -8,10 +8,19 @@ if [ ! -x "$PYTHON" ]; then
 fi
 export PYINSTALLER_CONFIG_DIR="$ROOT/.build-tools/pyinstaller-cache"
 cd "$ROOT/native-engine"
+SOURCE_HASH="$(bash "$ROOT/native-engine/engine-fingerprint.sh")"
 "$PYTHON" -m PyInstaller --noconfirm --clean --onedir --windowed --name receipt-engine-helper \
     --osx-bundle-identifier test.reimburse.engine \
     --distpath "$ROOT/native-engine/dist" --workpath "$ROOT/native-engine/build" \
     --collect-all rapidocr_onnxruntime --collect-all onnxruntime \
     --collect-all pypdfium2 --collect-all pypdfium2_raw \
     --add-data "personal-expense-template.xlsx:." engine.py
+BINARY="$ROOT/native-engine/dist/receipt-engine-helper.app/Contents/MacOS/receipt-engine-helper"
+RECEIPT_ENGINE_BINARY="$BINARY" "$PYTHON" -m unittest discover -s tests -p test_exports.py -k engine_process -v
+if [ "$SOURCE_HASH" != "$(bash "$ROOT/native-engine/engine-fingerprint.sh")" ]; then
+    echo "error: 打包期间引擎源码发生变化，请重新构建。" >&2
+    exit 1
+fi
+BINARY_HASH="$(shasum -a 256 "$BINARY" | cut -d ' ' -f 1)"
+printf '%s\n%s\n' "$SOURCE_HASH" "$BINARY_HASH" > "$ROOT/native-engine/dist/receipt-engine-helper.source.sha256"
 echo "本地处理引擎已打包，可以在 Xcode 中运行 reimburse。"

native-engine/engine-fingerprint.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+set -euo pipefail
+cd "$(dirname "$0")"
+shasum -a 256 ./*.py ./*.sh requirements*.txt personal-expense-template.xlsx helper.entitlements |
+    shasum -a 256 | cut -d ' ' -f 1

native-engine/engine.py

@@ -90,6 +90,7 @@ def dispatch(request):
         raise ValueError('至少完成一组核对后才能导出')
     destination = Path(request['destination'])
     temporary = destination.with_name('.' + str(uuid.uuid4()) + destination.suffix)
+    metadata = {}
     try:
         if operation in ('ppt', 'approved-ppt'):
             export_ppt(state, temporary, request.get('classified', True))
@@ -97,13 +98,14 @@ def dispatch(request):
             export_travel(state, temporary)
         elif operation == 'expense':
             template = Path(__file__).parent / 'personal-expense-template.xlsx'
-            export_expense(state, temporary, template, request.get('purposes', {}), request.get('signatures', []))
+            metadata = export_expense(state, temporary, template, request.get('purposes', {}), request.get('signatures', []),
+                                      request.get('payee', {}), request.get('categoryPurposes'))
         else:
             raise ValueError('未知操作：' + operation)
         os.replace(temporary, destination)
     finally:
         temporary.unlink(missing_ok=True)
-    return dict(destination=str(destination))
+    return dict(destination=str(destination), **metadata)
 
 
 if __name__ == '__main__':

native-engine/ensure-engine.sh

@@ -0,0 +1,15 @@
+#!/bin/bash
+set -euo pipefail
+ENGINE="$(cd "$(dirname "$0")" && pwd)"
+BINARY="$ENGINE/dist/receipt-engine-helper.app/Contents/MacOS/receipt-engine-helper"
+STAMP="$ENGINE/dist/receipt-engine-helper.source.sha256"
+SOURCE_HASH="$(bash "$ENGINE/engine-fingerprint.sh")"
+if [ -x "$BINARY" ] && [ -f "$STAMP" ]; then
+    BINARY_HASH="$(shasum -a 256 "$BINARY" | cut -d ' ' -f 1)"
+    if [ "$(cat "$STAMP")" = "$(printf '%s\n%s' "$SOURCE_HASH" "$BINARY_HASH")" ]; then
+        echo "本地引擎与当前源码一致。"
+        exit 0
+    fi
+fi
+echo "本地引擎缺失或已过期，正在重新打包并验证导出功能…"
+bash "$ENGINE/build-engine.sh"

native-engine/exports.py

@@ -183,7 +183,37 @@ def export_travel(state, destination):
     workbook.save(destination)
 
 
-def export_expense(state, destination, template, purposes, signatures):
+def expense_rows(matches, purposes, category_purposes=None):
+    grouped = {}
+    for match in matches:
+        category = match.get('category') or '其他'
+        row = grouped.setdefault(category, dict(category=category, count=0, amount=Decimal(0), types=[], purposes=[]))
+        row['count'] += len(match['invoices'])
+        purpose = purposes.get(match['id'], '').strip()
+        if purpose and purpose not in row['purposes']:
+            row['purposes'].append(purpose)
+        for invoice in match['invoices']:
+            content = text(invoice)
+            if '专用发票' in content:
+                invoice_type = '专票'
+            elif 'invoice' in content.lower():
+                invoice_type = 'Invoice'
+            elif '押金' in content and '收据' in content:
+                invoice_type = '押金收据'
+            else:
+                invoice_type = '普票'
+            if invoice_type not in row['types']:
+                row['types'].append(invoice_type)
+            row['amount'] += max((abs(Decimal(value.replace(',', ''))) for value in invoice['ocr']['amounts']), default=Decimal(0))
+    for row in grouped.values():
+        if category_purposes is not None:
+            row['purpose'] = category_purposes.get(row['category'], '').strip() or row['category']
+        else:
+            row['purpose'] = '；'.join(row['purposes']) or row['category']
+    return list(grouped.values())
+
+
+def export_expense(state, destination, template, purposes, signatures, payee=None, category_purposes=None):
     from lxml import etree as ET
     namespace = 'http://schemas.openxmlformats.org/spreadsheetml/2006/main'
     relationships = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships'
@@ -207,7 +237,10 @@ def export_expense(state, destination, template, purposes, signatures):
         names.clear()
     sheets.clear()
     signatures = signatures or DEFAULT_SIGNATURES
-    page_count = math.ceil(len(state['matches']) / 13)
+    rows = expense_rows(state['matches'], purposes, category_purposes)
+    if not rows:
+        raise ValueError('请先勾选要导出的已核对材料')
+    page_count = math.ceil(len(rows) / 13)
 
     def set_cell(document, reference, value):
         cell = document.find('.//' + tag('c') + '[@r="' + reference + '"]')
@@ -229,25 +262,18 @@ def export_expense(state, destination, template, purposes, signatures):
 
     for page in range(page_count):
         document = ET.fromstring(sheet_bytes)
+        profile = payee or {}
+        for reference, field in [('C23', 'recipient'), ('C24', 'bankName'), ('C25', 'accountNumber'), ('H6', 'preparer')]:
+            value = str(profile.get(field, '') or '').strip()
+            if field == 'preparer' and not value:
+                value = str(profile.get('recipient', '') or '').strip()
+            set_cell(document, reference, value)
         for row_number in range(9, 22):
             for column in ['B', 'C', 'G', 'H', 'I']:
                 set_cell(document, column + str(row_number), '')
             set_cell(document, 'A' + str(row_number), row_number - 8)
-        for index, match in enumerate(state['matches'][page * 13:(page + 1) * 13]):
-            invoice_type = '普票'
-            for invoice in match['invoices']:
-                content = text(invoice)
-                if '专用发票' in content:
-                    invoice_type = '专票'
-                    break
-                if 'invoice' in content.lower():
-                    invoice_type = 'Invoice'
-                    break
-                if '押金' in content and '收据' in content:
-                    invoice_type = '押金收据'
-                    break
-            total = sum(max((abs(Decimal(value.replace(',', ''))) for value in item['ocr']['amounts']), default=Decimal(0)) for item in match['invoices'])
-            for column, value in [('B', invoice_type), ('C', purposes.get(match['id'], '').strip()), ('G', len(match['invoices'])), ('H', total)]:
+        for index, row in enumerate(rows[page * 13:(page + 1) * 13]):
+            for column, value in [('B', '/'.join(row['types'])), ('C', row['purpose']), ('G', row['count']), ('H', row['amount'])]:
                 set_cell(document, column + str(index + 9), value)
         for index, cell in enumerate(['A30', 'D30', 'A31', 'D31', 'A32', 'D32']):
             value = signatures[index].strip() if index < len(signatures) else ''
@@ -277,3 +303,4 @@ def export_expense(state, destination, template, purposes, signatures):
     with zipfile.ZipFile(destination, 'w', zipfile.ZIP_DEFLATED) as output:
         for name, content in parts.items():
             output.writestr(name, content)
+    return dict(expenseGrouping='category-v1', expenseRowCount=len(rows))

native-engine/tests/test_engine_build.py

@@ -0,0 +1,76 @@
+import os
+import shutil
+import subprocess
+import tempfile
+import unittest
+from pathlib import Path
+
+
+class EngineBuildTests(unittest.TestCase):
+    def setUp(self):
+        self.temporary = tempfile.TemporaryDirectory()
+        self.root = Path(self.temporary.name)
+        engine = Path(__file__).resolve().parents[1]
+        for name in ['engine-fingerprint.sh', 'ensure-engine.sh']:
+            shutil.copyfile(engine / name, self.root / name)
+        for name in ['engine.py', 'exports.py', 'requirements.txt', 'requirements.lock.txt',
+                     'personal-expense-template.xlsx', 'helper.entitlements']:
+            (self.root / name).write_text('fixture\n')
+        self.binary = self.root / 'dist/receipt-engine-helper.app/Contents/MacOS/receipt-engine-helper'
+        (self.root / 'build-engine.sh').write_text('''#!/bin/bash
+set -euo pipefail
+cd "$(dirname "$0")"
+echo build >> builds.log
+mkdir -p dist/receipt-engine-helper.app/Contents/MacOS
+printf 'fixture engine\\n' > dist/receipt-engine-helper.app/Contents/MacOS/receipt-engine-helper
+chmod +x dist/receipt-engine-helper.app/Contents/MacOS/receipt-engine-helper
+SOURCE_HASH="$(bash engine-fingerprint.sh)"
+BINARY_HASH="$(shasum -a 256 dist/receipt-engine-helper.app/Contents/MacOS/receipt-engine-helper | cut -d ' ' -f 1)"
+printf '%s\\n%s\\n' "$SOURCE_HASH" "$BINARY_HASH" > dist/receipt-engine-helper.source.sha256
+''')
+
+    def tearDown(self):
+        self.temporary.cleanup()
+
+    def ensure(self):
+        return subprocess.run(['bash', str(self.root / 'ensure-engine.sh')],
+                              capture_output=True, text=True, timeout=20)
+
+    def build_count(self):
+        return len((self.root / 'builds.log').read_text().splitlines())
+
+    def test_missing_engine_builds_and_unchanged_engine_is_reused(self):
+        self.assertEqual(self.ensure().returncode, 0)
+        self.assertEqual(self.build_count(), 1)
+        self.assertEqual(self.ensure().returncode, 0)
+        self.assertEqual(self.build_count(), 1)
+
+    def test_changed_sources_rebuild_even_when_timestamp_is_unchanged(self):
+        self.assertEqual(self.ensure().returncode, 0)
+        source = self.root / 'exports.py'
+        original = source.stat()
+        source.write_text('new grouping implementation\n')
+        os.utime(source, ns=(original.st_atime_ns, original.st_mtime_ns))
+        self.assertEqual(self.ensure().returncode, 0)
+        self.assertEqual(self.build_count(), 2)
+
+    def test_replaced_binary_rebuilds(self):
+        self.assertEqual(self.ensure().returncode, 0)
+        self.binary.write_text('old engine\n')
+        self.assertEqual(self.ensure().returncode, 0)
+        self.assertEqual(self.build_count(), 2)
+
+    def test_missing_stamp_rebuilds(self):
+        self.assertEqual(self.ensure().returncode, 0)
+        (self.root / 'dist/receipt-engine-helper.source.sha256').unlink()
+        self.assertEqual(self.ensure().returncode, 0)
+        self.assertEqual(self.build_count(), 2)
+
+    def test_failed_rebuild_does_not_accept_old_engine(self):
+        self.assertEqual(self.ensure().returncode, 0)
+        (self.root / 'build-engine.sh').write_text('exit 17\n')
+        self.assertEqual(self.ensure().returncode, 17)
+
+
+if __name__ == '__main__':
+    unittest.main()

native-engine/tests/test_exports.py

@@ -1,4 +1,8 @@
 import copy
+import json
+import os
+import subprocess
+import sys
 import tempfile
 import unittest
 import zipfile
@@ -68,6 +72,8 @@ class ExportTests(unittest.TestCase):
 
     def test_expense_pagination(self):
         self.state['matches'] = [copy.deepcopy(self.state['matches'][0]) for _ in range(14)]
+        for index, match in enumerate(self.state['matches']):
+            match['category'] = f'类别{index}'
         destination = self.root / 'expense.xlsx'
         export_expense(self.state, destination, self.template, {}, [])
         workbook = load_workbook(destination)
@@ -76,6 +82,137 @@ class ExportTests(unittest.TestCase):
         self.assertEqual(workbook.worksheets[1]['H10'].value, '')
         self.assertEqual(workbook.worksheets[1]['A30'].value, '部门长：')
 
+    def test_expense_payee_is_text_on_every_page(self):
+        self.state['matches'] = [copy.deepcopy(self.state['matches'][0]) for _ in range(14)]
+        for index, match in enumerate(self.state['matches']):
+            match['category'] = f'类别{index}'
+        destination = self.root / 'payee.xlsx'
+        profile = dict(recipient='测试收款人', bankName='=测试银行支行',
+                       accountNumber='0012345678901234567890', preparer='测试制单人')
+        export_expense(self.state, destination, self.template, {}, [], profile)
+        for sheet in load_workbook(destination).worksheets:
+            for reference, value in [('C23', profile['recipient']), ('C24', profile['bankName']),
+                                     ('C25', profile['accountNumber']), ('H6', profile['preparer'])]:
+                self.assertEqual(sheet[reference].value, value)
+                self.assertEqual(sheet[reference].data_type, 's')
+            self.assertEqual(sheet['C25'].style_id, load_workbook(self.template)['1个人报销单']['C25'].style_id)
+
+    def test_expense_without_profile_clears_template_identity(self):
+        destination = self.root / 'empty-payee.xlsx'
+        export_expense(self.state, destination, self.template, {}, [])
+        sheet = load_workbook(destination).active
+        for reference in ['C23', 'C24', 'C25', 'H6']:
+            self.assertEqual(sheet[reference].value, '')
+
+    def test_expense_preparer_defaults_to_recipient(self):
+        destination = self.root / 'default-preparer.xlsx'
+        export_expense(self.state, destination, self.template, {}, [], {'recipient': '测试收款人'})
+        self.assertEqual(load_workbook(destination).active['H6'].value, '测试收款人')
+
+    def test_expense_profile_through_engine_process(self):
+        binary = os.environ.get('RECEIPT_ENGINE_BINARY')
+        command = [binary] if binary else [sys.executable, str(Path(__file__).resolve().parents[1] / 'engine.py')]
+        destination = self.root / 'process-expense.xlsx'
+        self.state['matches'] = [copy.deepcopy(self.state['matches'][0]) for _ in range(6)]
+        for match in self.state['matches']:
+            match['category'] = '交通'
+        request = dict(operation='expense', state=self.state, destination=str(destination),
+                       categoryPurposes={'交通': '交通'},
+                       payee=dict(recipient='测试收款人', bankName='测试支行', accountNumber='0001234567890123456789'))
+        process = subprocess.run(command, input=json.dumps(request) + '\n', capture_output=True, text=True, timeout=60)
+        self.assertEqual(process.returncode, 0, process.stdout + process.stderr)
+        events = [json.loads(line) for line in process.stdout.splitlines()]
+        self.assertEqual(events[-1]['event'], 'result')
+        sheet = load_workbook(destination).active
+        self.assertEqual(sheet['C23'].value, '测试收款人')
+        self.assertEqual(sheet['C24'].value, '测试支行')
+        self.assertEqual(sheet['C25'].value, '0001234567890123456789')
+        self.assertEqual(sheet['C25'].data_type, 's')
+        self.assertEqual(sheet['H6'].value, '测试收款人')
+        self.assertEqual(sheet['C9'].value, '交通')
+        self.assertEqual(sheet['G9'].value, 6)
+        self.assertEqual(sheet['H9'].value, 600)
+        self.assertEqual(sheet['H10'].value, '')
+
+    def test_expense_engine_process_groups_selected_traffic_and_office(self):
+        binary = os.environ.get('RECEIPT_ENGINE_BINARY')
+        command = [binary] if binary else [sys.executable, str(Path(__file__).resolve().parents[1] / 'engine.py')]
+        amounts = [['1058.00', '1058.00'], ['140.00'], ['300.00'], ['163.92'], ['253.45'], ['74.49'], ['444.00']]
+        selected = []
+        for index, invoice_amounts in enumerate(amounts):
+            match = copy.deepcopy(self.state['matches'][0])
+            match['id'] = f'selected-{index}'
+            match['category'] = '办公用品' if index == 6 else '交通'
+            match['invoices'] = []
+            for invoice_index, amount in enumerate(invoice_amounts):
+                invoice = copy.deepcopy(self.state['matches'][0]['invoices'][0])
+                invoice['id'] = f'invoice-{index}-{invoice_index}'
+                invoice['ocr']['amounts'] = [amount]
+                match['invoices'].append(invoice)
+            selected.append(match)
+        self.state['matches'] = selected
+        destination = self.root / 'two-categories.xlsx'
+        request = dict(operation='expense', state=self.state, destination=str(destination),
+                       categoryPurposes={'交通': '交通', '办公用品': '办公用品'})
+        process = subprocess.run(command, input=json.dumps(request) + '\n', capture_output=True, text=True, timeout=60)
+        self.assertEqual(process.returncode, 0, process.stdout + process.stderr)
+        events = [json.loads(line) for line in process.stdout.splitlines()]
+        self.assertEqual(events[-1]['result']['expenseGrouping'], 'category-v1')
+        self.assertEqual(events[-1]['result']['expenseRowCount'], 2)
+        workbook = load_workbook(destination)
+        self.assertEqual(workbook.sheetnames, ['个人报销单'])
+        sheet = workbook.active
+        rows = [(sheet[f'C{row}'].value, sheet[f'G{row}'].value, sheet[f'H{row}'].value)
+                for row in range(9, 22) if sheet[f'H{row}'].value not in ('', None)]
+        self.assertEqual(rows, [('交通', 7, 3047.86), ('办公用品', 1, 444)])
+        self.assertEqual(round(sum(row[2] for row in rows), 2), 3491.86)
+        self.assertEqual(sheet['F27'].value, '=SUM(H9:H21)')
+
+    def test_expense_same_category_collapses_before_pagination(self):
+        self.state['matches'] = [copy.deepcopy(self.state['matches'][0]) for _ in range(14)]
+        for match in self.state['matches']:
+            match['category'] = '交通'
+        destination = self.root / 'collapsed.xlsx'
+        export_expense(self.state, destination, self.template, {}, [])
+        workbook = load_workbook(destination)
+        self.assertEqual(workbook.sheetnames, ['个人报销单'])
+        sheet = workbook.active
+        self.assertEqual(sheet['C9'].value, '交通')
+        self.assertEqual(sheet['G9'].value, 14)
+        self.assertEqual(sheet['H9'].value, 1400)
+        self.assertEqual(sheet['H10'].value, '')
+
+    def test_expense_category_totals_counts_and_mixed_invoice_types(self):
+        traffic = [copy.deepcopy(self.state['matches'][0]) for _ in range(6)]
+        for match in traffic:
+            match['category'] = '交通'
+            match['invoices'][0]['ocr']['amounts'] = ['100.10', '2']
+        traffic[0]['invoices'][0]['ocr']['rawText'] = '专用发票'
+        hotel = copy.deepcopy(traffic[0])
+        hotel['category'] = '住宿'
+        hotel['invoices'] *= 2
+        hotel['invoices'][0]['ocr']['amounts'] = ['-1,200.30', '2']
+        self.state['matches'] = [traffic[0], hotel] + traffic[1:]
+        original = copy.deepcopy(self.state)
+        destination = self.root / 'grouped.xlsx'
+        export_expense(self.state, destination, self.template, {}, [], category_purposes={'交通': '=交通用途'})
+        sheet = load_workbook(destination).active
+        self.assertEqual(sheet['C9'].value, '=交通用途')
+        self.assertEqual(sheet['C9'].data_type, 's')
+        self.assertEqual(sheet['B9'].value, '专票/普票')
+        self.assertEqual(sheet['G9'].value, 6)
+        self.assertEqual(sheet['H9'].value, 600.6)
+        self.assertEqual(sheet['C10'].value, '住宿')
+        self.assertEqual(sheet['G10'].value, 2)
+        self.assertEqual(sheet['H10'].value, 2400.6)
+        self.assertEqual(sheet['H11'].value, '')
+        self.assertEqual(self.state, original)
+
+    def test_expense_empty_selection_is_rejected(self):
+        self.state['matches'] = []
+        with self.assertRaisesRegex(ValueError, '勾选'):
+            export_expense(self.state, self.root / 'empty.xlsx', self.template, {}, [])
+
     def test_travel_only_verified_invoices(self):
         travel = self.state['invoices'][0]['ocr']['travel']
         travel.update(type='train', travelerName='张三', departure='北京南', destination='上海虹桥', departureTime='2026-06-24 15:00', transportNumber='G21')