sino-bc-cli 1.8.38 → 1.8.39

package/dist/skills/biz-line-strategy-generate/scripts/detail_store.py

@@ -0,0 +1,82 @@
+"""本条任务的明细落盘协议；多次导出和校验按行重读，不保留完整列表。"""
+import json
+from pathlib import Path
+from uuid import uuid4
+
+
+def _check_record(record):
+    if not isinstance(record, dict) or any(
+            not isinstance(key, str) or not isinstance(value, (str, int, float, bool, type(None)))
+            for key, value in record.items()):
+        raise ValueError('明细记录必须是字段到单元格值的对象')
+
+
+class DetailRows:
+    """只保存路径和行数；每次遍历重新打开，不能复用一次性迭代器。"""
+
+    def __init__(self, path, count):
+        if type(count) is not int or count < 0:
+            raise ValueError('明细行数必须是非负整数')
+        self.path = Path(path)
+        self.count = count
+
+    def __len__(self):
+        return self.count
+
+    def __iter__(self):
+        count = 0
+        with self.path.open(encoding='utf-8') as stream:
+            for line in stream:
+                count += 1
+                if count > self.count:
+                    raise ValueError('明细实际行数超过登记数量')
+                record = json.loads(line)
+                _check_record(record)
+                yield record
+        if count != self.count:
+            raise ValueError('明细实际行数少于登记数量')
+
+    def reference(self, root):
+        return {'version': 1, 'path': str(self.path.relative_to(root)), 'count': self.count}
+
+
+def write_detail_rows(root, records):
+    """完成全部记录后才发布引用；失败时关闭生成器并移除半成品。"""
+    final = Path(root) / f'details-{uuid4().hex}.jsonl'
+    partial = final.with_suffix('.part')
+    count = 0
+    try:
+        with partial.open('x', encoding='utf-8') as stream:
+            for record in records:
+                _check_record(record)
+                stream.write(json.dumps(record, ensure_ascii=False) + '\n')
+                count += 1
+        partial.replace(final)
+        return DetailRows(final, count)
+    finally:
+        close = getattr(records, 'close', None)
+        try:
+            if close is not None:
+                close()
+        finally:
+            partial.unlink(missing_ok=True)
+
+
+def details_from_payload(payload, root):
+    """新引用只读取本条目录；旧诊断数组保持兼容，不将损坏引用当空明细。"""
+    if 'detailSource' not in payload:
+        details = payload.get('details', [])
+        if not isinstance(details, list):
+            raise ValueError('payload.details 必须是数组')
+        return details
+    source = payload['detailSource']
+    if 'details' in payload or not isinstance(source, dict) or set(source) != {'version', 'path', 'count'}:
+        raise ValueError('明细引用格式无效或与内嵌明细冲突')
+    if type(source['version']) is not int or source['version'] != 1 or not isinstance(source['path'], str):
+        raise ValueError('明细引用版本或路径无效')
+    relative = Path(source['path'])
+    root = Path(root).resolve()
+    path = (root / relative).resolve()
+    if relative.is_absolute() or '..' in relative.parts or not path.is_relative_to(root) or path == root:
+        raise ValueError('明细文件必须位于本条任务目录内')
+    return DetailRows(path, source['count'])

package/dist/skills/biz-line-strategy-generate/scripts/summarize.py

@@ -14,6 +14,7 @@ SCRIPTS = Path(__file__).resolve().parent
 if str(SCRIPTS) not in sys.path:
     sys.path.insert(0, str(SCRIPTS))
 
+from detail_store import write_detail_rows
 from cli_json import BASE_METRICS, REPORT_METRICS, amount_change, evaluate_learning, metric_fields, metric_formula, learning_text
 from cli_json import (
     bid_adjust_stats,
@@ -380,7 +381,6 @@ def build_dimension_details(bl: Path, active_channels, bids):
         ('广告', 'ad_id+ad_name', tuple(CH_LABEL), 'ad_id'),
         ('创意类型', 'ad_format', tuple(CH_LABEL), None),
     ]
-    details = []
     # 预算按渠道列成效；渠道编码就是此枚举的稳定 ID。
     for ch in active_channels:
         totals = {'curr': {}, 'prev': {}}
@@ -390,11 +390,11 @@ def build_dimension_details(bl: Path, active_channels, bids):
             conflicts[period] = detail_conflicts(bl, stem)
             for row in iter_stem_rows(bl, stem):
                 add_detail_metric_row(totals[period], row)
-        details.append({'kind': '预算', 'action': '持续' if totals['prev'] else '新增',
-                        'channel': CH_LABEL[ch], 'id': ch, 'name': CH_LABEL[ch],
-                        **detail_metric_cells([totals['curr']] if totals['curr'] else [],
-                                              [totals['prev']] if totals['prev'] else [],
-                                              conflicts['curr'], conflicts['prev']), 'extra': ''})
+        yield {'kind': '预算', 'action': '持续' if totals['prev'] else '新增',
+                'channel': CH_LABEL[ch], 'id': ch, 'name': CH_LABEL[ch],
+                **detail_metric_cells([totals['curr']] if totals['curr'] else [],
+                                      [totals['prev']] if totals['prev'] else [],
+                                      conflicts['curr'], conflicts['prev']), 'extra': ''}
 
     for kind, stem, channels, id_key in dimensions:
         key = stem.split('+')[-1] if id_key else stem
@@ -425,12 +425,12 @@ def build_dimension_details(bl: Path, active_channels, bids):
             for ident in sorted(set(names['curr']) | set(names['prev'])):
                 action = ('持续' if ident in names['curr'] and ident in names['prev'] else
                           '新增' if ident in names['curr'] else '暂停')
-                details.append({'kind': kind, 'action': action, 'channel': CH_LABEL[ch],
-                                'id': ident, 'name': names['curr'].get(ident) or names['prev'][ident],
-                                **detail_metric_cells([values['curr'][ident]] if values['curr'][ident] else [],
-                                                      [values['prev'][ident]] if values['prev'][ident] else [],
-                                                      conflicts['curr'], conflicts['prev']),
-                                'extra': ''})
+                yield {'kind': kind, 'action': action, 'channel': CH_LABEL[ch],
+                        'id': ident, 'name': names['curr'].get(ident) or names['prev'][ident],
+                        **detail_metric_cells([values['curr'][ident]] if values['curr'][ident] else [],
+                                              [values['prev'][ident]] if values['prev'][ident] else [],
+                                              conflicts['curr'], conflicts['prev']),
+                        'extra': ''}
 
     # 出价沿用既有上调/下调/持平动作；金额取广告组同 ID 的成效，补充只列首末出价。
     for ch in active_channels:
@@ -451,13 +451,12 @@ def build_dimension_details(bl: Path, active_channels, bids):
             action = item.get('action') or ('上调' if end > start else '下调' if end < start else '期内变化但首末持平')
             changes = '；'.join(f"{field['field']} {field['from']:g} → {field['to']:g}"
                                for field in item.get('fields', []))
-            details.append({'kind': '出价', 'action': action, 'channel': CH_LABEL[ch],
-                            'id': ident, 'name': item.get('name') or ident,
-                            **detail_metric_cells([values['curr'][ident]] if values['curr'][ident] else [],
-                                                  [values['prev'][ident]] if values['prev'][ident] else [],
-                                                  conflicts['curr'], conflicts['prev']),
-                            'extra': changes})
-    return details
+            yield {'kind': '出价', 'action': action, 'channel': CH_LABEL[ch],
+                    'id': ident, 'name': item.get('name') or ident,
+                    **detail_metric_cells([values['curr'][ident]] if values['curr'][ident] else [],
+                                          [values['prev'][ident]] if values['prev'][ident] else [],
+                                          conflicts['curr'], conflicts['prev']),
+                    'extra': changes}
 
 
 def summarize_request(req: Path):
@@ -898,8 +897,8 @@ def summarize_request(req: Path):
         if audience_type_moved:
             evidence['audience'].append('受众类型调整')
     # 旧的覆盖、分布及逐枚举成效说明不再进入 Excel；明细统一按维度和指标重建。
-    details = build_dimension_details(bl, active_channels, bids)
-    payload = {"rows": [row], "details": details}
+    details = write_detail_rows(req, build_dimension_details(bl, active_channels, bids))
+    payload = {"rows": [row], "detailSource": details.reference(req)}
     (req / "payload.json").write_text(json.dumps(payload, ensure_ascii=False, indent=2), encoding="utf-8")
     summary = {
         "bizLineId": biz_id,
@@ -916,7 +915,8 @@ def summarize_request(req: Path):
         "request_dir": str(req),
     }
     (req / "summary.json").write_text(json.dumps(summary, ensure_ascii=False, indent=2), encoding="utf-8")
-    return payload, summary
+    # 内存引用仅包含路径和行数，兼容工作流的多次遍历；磁盘 payload 不嵌入明细。
+    return {**payload, 'details': details}, summary
 
 
 def main() -> int:

package/dist/skills/biz-line-strategy-generate/scripts/test_detail_store.py

@@ -0,0 +1,122 @@
+"""明细落盘协议：完整、可重复读取，失败不发布半成品。"""
+import importlib.util
+import json
+import subprocess
+import sys
+import weakref
+from pathlib import Path
+import tempfile
+import unittest
+from unittest.mock import patch
+
+
+class DetailStoreTests(unittest.TestCase):
+    def store(self):
+        self.assertIsNotNone(importlib.util.find_spec('detail_store'), '尚未提供落盘明细读取器')
+        import detail_store
+        return detail_store
+
+    def test_repeated_iteration_and_serializable_reference(self):
+        store = self.store()
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            rows = [{'kind': '广告', 'id': str(i), 'name': '中文\n名称'} for i in range(3)]
+            source = store.write_detail_rows(root, iter(rows))
+            self.assertEqual(len(source), 3)
+            self.assertEqual(list(source), rows)
+            self.assertEqual(list(source), rows)
+            packet = json.loads(json.dumps({'detailSource': source.reference(root)}))
+            self.assertEqual(list(store.details_from_payload(packet, root)), rows)
+            self.assertFalse(any(root.glob('*.part')))
+            self.assertEqual(list(store.write_detail_rows(root, iter([]))), [])
+
+    def test_count_corruption_and_ambiguous_source_fail(self):
+        store = self.store()
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            source = store.write_detail_rows(root, iter([{'id': '1'}]))
+            for count in (0, 2):
+                with self.assertRaises(ValueError):
+                    list(store.DetailRows(source.path, count))
+            source.path.write_text('{bad}\n', encoding='utf-8')
+            with self.assertRaises(ValueError):
+                list(source)
+            for value in ([], {'version': 2, 'path': source.path.name, 'count': 1},
+                          {'version': 1, 'path': '../outside.jsonl', 'count': 1},
+                          {'version': 1, 'path': source.path.name, 'count': True}):
+                with self.assertRaises(ValueError):
+                    store.details_from_payload({'detailSource': value}, root)
+            with self.assertRaises(ValueError):
+                store.details_from_payload({'detailSource': source.reference(root), 'details': []}, root)
+            self.assertEqual(store.details_from_payload({'details': [{'id': 'old'}]}, root), [{'id': 'old'}])
+
+    def test_partial_write_closes_generator_and_publishes_nothing(self):
+        store = self.store()
+        closed = []
+        def records():
+            try:
+                yield {'id': '1'}
+                raise OSError('模拟磁盘写入失败')
+            finally:
+                closed.append(True)
+        with tempfile.TemporaryDirectory() as tmp:
+            with self.assertRaises(OSError):
+                store.write_detail_rows(Path(tmp), records())
+            self.assertEqual(closed, [True])
+            self.assertEqual(list(Path(tmp).iterdir()), [])
+
+    def test_disk_failure_closes_suspended_generator(self):
+        store = self.store()
+        closed = []
+        def records():
+            try:
+                yield {'id': '1'}
+                yield {'id': '2'}
+            finally:
+                closed.append(True)
+        with tempfile.TemporaryDirectory() as tmp:
+            with patch.object(store.json, 'dumps', side_effect=OSError('磁盘故障模拟')):
+                with self.assertRaises(OSError):
+                    store.write_detail_rows(Path(tmp), records())
+            self.assertEqual(closed, [True])
+            self.assertEqual(list(Path(tmp).iterdir()), [])
+
+    def test_previous_source_stays_valid_after_another_summary(self):
+        store = self.store()
+        with tempfile.TemporaryDirectory() as tmp:
+            old = store.write_detail_rows(Path(tmp), iter([{'id': 'old'}]))
+            new = store.write_detail_rows(Path(tmp), iter([{'id': 'new'}]))
+            self.assertEqual(list(old), [{'id': 'old'}])
+            self.assertEqual(list(new), [{'id': 'new'}])
+
+
+    def test_closed_generator_releases_analysis_objects(self):
+        store = self.store()
+        references = []
+        class Aggregate:
+            pass
+        def records():
+            aggregate = Aggregate()
+            references.append(weakref.ref(aggregate))
+            yield {'id': '1'}
+        with tempfile.TemporaryDirectory() as tmp:
+            source = store.write_detail_rows(Path(tmp), records())
+            self.assertIsNone(references[0]())
+            self.assertEqual(set(vars(source)), {'path', 'count'})
+
+    def test_diagnostic_incomplete_source_does_not_publish(self):
+        store = self.store()
+        script = Path(__file__).with_name('build_report.py')
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            source = store.write_detail_rows(root, iter([{'kind': '广告', 'id': '1'}]))
+            source.path.write_text('', encoding='utf-8')
+            packet = {'rows': [{'bizLine': '样本'}], 'detailSource': source.reference(root)}
+            output = root / 'broken.xlsx'
+            result = subprocess.run([sys.executable, '-B', str(script), '--input', '-', '--output', str(output)],
+                                    input=json.dumps(packet), text=True, capture_output=True, cwd=root)
+            self.assertNotEqual(result.returncode, 0)
+            self.assertFalse(output.exists())
+
+if __name__ == '__main__':
+    unittest.main()

package/dist/skills/biz-line-strategy-generate/scripts/test_excel_streaming.py

@@ -0,0 +1,223 @@
+import json
+import subprocess
+import sys
+import tempfile
+import unittest
+import zipfile
+from pathlib import Path
+from unittest.mock import patch
+import build_report as report
+from detail_store import write_detail_rows
+
+
+class ExcelStreamingTests(unittest.TestCase):
+    def test_export_and_validation_do_not_materialize_xml(self):
+        with tempfile.TemporaryDirectory() as directory:
+            path = str(Path(directory) / 'test.xlsx')
+            details = [{'id': str(i), 'name': '名称'} for i in range(50)]
+            original_write = zipfile.ZipFile.writestr
+            def write(archive, name, *args, **kwargs):
+                if 'sharedStrings' in name or 'worksheets/sheet' in name:
+                    raise AssertionError('整表 XML 写入')
+                return original_write(archive, name, *args, **kwargs)
+            with patch.object(zipfile.ZipFile, 'writestr', write):
+                report.write_xlsx([{'bizLine': '样本'}], path, details)
+            original = zipfile.ZipFile.read
+            def read(archive, name, *args, **kwargs):
+                if 'sharedStrings' in name or 'worksheets/sheet' in name:
+                    raise AssertionError('整表 XML')
+                return original(archive, name, *args, **kwargs)
+            with patch.object(zipfile.ZipFile, 'read', read):
+                self.assertEqual(report.validate_xlsx(path, [{'bizLine': '样本'}], details)['details'], 50)
+
+    def test_bad_coordinate_is_rejected(self):
+        with tempfile.TemporaryDirectory() as directory:
+            path = Path(directory) / 'test.xlsx'
+            report.write_xlsx([{'bizLine': '样本'}], path, [{'id': '1'}])
+            with zipfile.ZipFile(path) as archive:
+                contents = {name: archive.read(name) for name in archive.namelist()}
+            contents['xl/worksheets/sheet2.xml'] = contents['xl/worksheets/sheet2.xml'].replace(b'r="A2"', b'r="Z9"')
+            with zipfile.ZipFile(path, 'w') as archive:
+                for name, content in contents.items():
+                    archive.writestr(name, content)
+            with self.assertRaises(ValueError):
+                report.validate_xlsx(path, [{'bizLine': '样本'}], [{'id': '1'}])
+
+    def test_index_keeps_format_keys_and_bounded_hot_cache(self):
+        with tempfile.TemporaryDirectory() as directory:
+            with report._string_index(directory) as connection:
+                strings = report._Strings(connection)
+                self.assertEqual(strings.intern('【预算】'), 0)
+                self.assertEqual(strings.intern('【预算】', True), 1)
+                self.assertEqual(strings.intern('【预算】', (('【预算】', True),)), 2)
+                self.assertEqual(strings.intern('【预算】'), 0)
+                for index in range(4000):
+                    strings.intern(str(index))
+                strings.intern('长' * 32767)
+                self.assertLessEqual(len(strings.cache), 512)
+                self.assertTrue(all(isinstance(key[1], bool) and len(key[0]) <= 512 for key in strings.cache))
+            self.assertEqual(list(Path(directory).iterdir()), [])
+
+    def test_hot_string_avoids_json_serialization(self):
+        with tempfile.TemporaryDirectory() as directory, report._string_index(directory) as connection:
+            strings = report._Strings(connection)
+            strings.intern('重复文字')
+            with patch.object(report.json, 'dumps', side_effect=AssertionError('命中缓存仍序列化')):
+                self.assertEqual(strings.intern('重复文字'), 0)
+
+    def test_xml_items_reject_invalid_hierarchy(self):
+        ns = report._NS['s']
+        cases = [
+            ('row', f'<worksheet xmlns="{ns}"><notSheetData><row/></notSheetData></worksheet>'),
+            ('row', f'<wrong xmlns="{ns}"><sheetData><row/></sheetData></wrong>'),
+            ('row', f'<worksheet xmlns="{ns}"><sheetData/><sheetData/></worksheet>'),
+            ('row', f'<worksheet xmlns="{ns}"><extra><sheetData><row/></sheetData></extra></worksheet>'),
+            ('si', f'<wrong xmlns="{ns}"><si><t>文字</t></si></wrong>'),
+            ('si', f'<sst xmlns="{ns}"><extra><si><t>文字</t></si></extra></sst>'),
+        ]
+        for tag, xml in cases:
+            with self.subTest(xml=xml), tempfile.TemporaryDirectory() as directory:
+                path = Path(directory) / 'bad.zip'
+                with zipfile.ZipFile(path, 'w') as archive:
+                    archive.writestr('part.xml', xml)
+                with zipfile.ZipFile(path) as archive:
+                    with self.assertRaises(ValueError):
+                        list(report._xml_items(archive, 'part.xml', tag))
+
+    def test_validation_queries_missing_strings_by_row(self):
+        with tempfile.TemporaryDirectory() as directory:
+            path = Path(directory) / 'rows.xlsx'
+            details = [{key: f'{key}-{i}' for key, _ in report.DETAIL_COLUMNS} for i in range(20)]
+            report.write_detail_workbook(details, path)
+            with zipfile.ZipFile(path) as archive, report._string_index(directory) as connection:
+                report._load_strings(archive, connection)
+                queries = []
+                connection.set_trace_callback(lambda sql: queries.append(sql) if sql.startswith('SELECT') else None)
+                report._validate_sheet(archive, 'xl/worksheets/sheet1.xml', details, report.DETAIL_COLUMNS, connection)
+                self.assertLessEqual(len(queries), len(details) + 1)
+
+    def test_failed_write_removes_partial_file_and_index(self):
+        for exception in (OSError('磁盘写入失败'), KeyboardInterrupt()):
+            with self.subTest(exception=type(exception).__name__), tempfile.TemporaryDirectory() as directory:
+                path = Path(directory) / 'partial.xlsx'
+                with patch.object(report._Strings, 'write', side_effect=exception):
+                    with self.assertRaises(type(exception)):
+                        report.write_detail_workbook([{'id': '1'}], path)
+                self.assertEqual(list(Path(directory).iterdir()), [])
+
+    def test_index_creation_failure_and_corrupt_zip_fail(self):
+        with tempfile.TemporaryDirectory() as directory:
+            path = Path(directory) / 'test.xlsx'
+            with patch.object(report.sqlite3, 'connect', side_effect=report.sqlite3.OperationalError('无法创建索引')):
+                with self.assertRaises(report.sqlite3.OperationalError):
+                    report.write_detail_workbook([{'id': '1'}], path)
+            self.assertEqual(list(Path(directory).iterdir()), [])
+            report.write_detail_workbook([{'id': '1'}], path)
+            path.write_bytes(path.read_bytes()[:-22])
+            with self.assertRaises(zipfile.BadZipFile):
+                report.validate_detail_workbook(path, [{'id': '1'}])
+            self.assertFalse(list(Path(directory).glob('.excel-index-*')))
+
+    def test_late_corruption_and_missing_or_extra_rows(self):
+        cases = {
+            '末行错值': lambda xml: xml.replace(b'<v>48</v>', b'<v>0</v>'),
+            '缺行': lambda xml: __import__('re').sub(rb'<row r="31".*?</row>', b'', xml),
+            '多行': lambda xml: xml.replace(b'</sheetData>', b'<row r="32"></row></sheetData>'),
+            '越界': lambda xml: xml.replace(b'<v>48</v>', b'<v>999999</v>'),
+            '负索引': lambda xml: xml.replace(b'<v>48</v>', b'<v>-1</v>'),
+            'XML末尾损坏': lambda xml: xml[:-3],
+            '错误明细父节点': lambda xml: xml.replace(b'sheetData', b'notSheetData'),
+            '重复明细容器': lambda xml: xml.replace(b'</worksheet>', b'<sheetData/></worksheet>'),
+            '错误根节点': lambda xml: xml.replace(b'worksheet', b'wrongRoot'),
+        }
+        details = [{'id': str(i)} for i in range(30)]
+        for label, mutate in cases.items():
+            with self.subTest(label=label), tempfile.TemporaryDirectory() as directory:
+                path = Path(directory) / 'test.xlsx'
+                report.write_detail_workbook(details, path)
+                with zipfile.ZipFile(path) as archive:
+                    content = {name: archive.read(name) for name in archive.namelist()}
+                original = content['xl/worksheets/sheet1.xml']
+                content['xl/worksheets/sheet1.xml'] = mutate(original)
+                self.assertNotEqual(original, content['xl/worksheets/sheet1.xml'])
+                with zipfile.ZipFile(path, 'w') as archive:
+                    for name, value in content.items():
+                        archive.writestr(name, value)
+                with self.assertRaises((ValueError, report.ET.ParseError)):
+                    report.validate_detail_workbook(path, details)
+
+    def test_ai_read_scans_shared_strings_to_end_without_reading_whole_part(self):
+        ai = '速览 **重点**\n【预算策略】\n预算\n【广告产品策略】\n产品\n【受众策略】\n受众\n【创意策略】\n创意'
+        with tempfile.TemporaryDirectory() as directory:
+            path = Path(directory) / 'ai.xlsx'
+            report.write_xlsx([{'bizLine': '样本'}], path, [{'name': str(i)} for i in range(2000)], ai_strategy=ai)
+            original_read = zipfile.ZipFile.read
+            def read(archive, name, *args, **kwargs):
+                if name == 'xl/sharedStrings.xml':
+                    raise AssertionError('整表共享字符串')
+                return original_read(archive, name, *args, **kwargs)
+            with patch.object(zipfile.ZipFile, 'read', read):
+                self.assertEqual(report.read_ai_sheet(path)['一页速览'], '速览 重点')
+            with zipfile.ZipFile(path) as archive:
+                contents = {name: archive.read(name) for name in archive.namelist()}
+            contents['xl/sharedStrings.xml'] += b'<broken'
+            with zipfile.ZipFile(path, 'w') as archive:
+                for name, value in contents.items():
+                    archive.writestr(name, value)
+            with self.assertRaises(report.ET.ParseError):
+                report.read_ai_sheet(path)
+
+    def test_diagnostic_file_and_stdin_support_both_detail_formats(self):
+        with tempfile.TemporaryDirectory() as directory:
+            root = Path(directory)
+            source = write_detail_rows(root, [{'id': '1', 'name': '明细'}])
+            payloads = [
+                {'rows': [{'bizLine': '样本'}], 'details': [{'id': '1', 'name': '明细'}]},
+                {'rows': [{'bizLine': '样本'}], 'detailSource':
+                 {'version': 1, 'path': source.path.name, 'count': 1}},
+            ]
+            for number, payload in enumerate(payloads):
+                for mode in ('file', 'stdin'):
+                    path = root / f'{number}-{mode}.xlsx'
+                    input_path = root / 'payload.json'
+                    input_path.write_text(json.dumps(payload), encoding='utf-8')
+                    result = subprocess.run([sys.executable, '-B', str(Path(report.__file__).resolve()),
+                                             '--input', str(input_path) if mode == 'file' else '-',
+                                             '--output', str(path)], input=json.dumps(payload), text=True,
+                                            capture_output=True, cwd=root)
+                    self.assertEqual(result.returncode, 0, result.stderr)
+                    self.assertEqual(report.validate_xlsx(path, payload['rows'], source)['details'], 1)
+            payloads[1]['detailSource']['version'] = 999
+            payloads[1]['details'] = payloads[0]['details']
+            result = subprocess.run([sys.executable, '-B', str(Path(report.__file__).resolve()),
+                                     '--input', '-', '--output', str(root / 'bad.xlsx')],
+                                    input=json.dumps(payloads[1]), text=True, capture_output=True, cwd=root)
+            self.assertNotEqual(result.returncode, 0)
+            self.assertFalse((root / 'bad.xlsx').exists())
+
+    def test_file_source_index_stays_with_source_and_existing_output_is_preserved(self):
+        with tempfile.TemporaryDirectory() as directory:
+            root = Path(directory)
+            source_root, output_root = root / 'source', root / 'published'
+            source_root.mkdir()
+            output_root.mkdir()
+            source = write_detail_rows(source_root, [{'id': '1'}])
+            output = output_root / 'test.xlsx'
+            original = report._string_index
+            observed = []
+            def index(path):
+                observed.append(Path(path))
+                return original(path)
+            with patch.object(report, '_string_index', index):
+                report.write_detail_workbook(source, output)
+                report.validate_detail_workbook(output, source)
+            self.assertEqual(observed, [source_root, source_root])
+            before = output.read_bytes()
+            with self.assertRaises(FileExistsError):
+                report.write_detail_workbook(source, output)
+            self.assertEqual(output.read_bytes(), before)
+            self.assertEqual(list(source_root.glob('.excel-index-*')), [])
+
+
+if __name__ == '__main__':
+    unittest.main()

package/dist/skills/biz-line-strategy-generate/scripts/test_pipeline.py

@@ -11,6 +11,7 @@ from unittest.mock import patch
 sys.dont_write_bytecode = True
 sys.path.insert(0, str(Path(__file__).resolve().parent))
 from run_workflow import run_batch
+from detail_store import write_detail_rows
 from strategy_output import submit_tag_choices
 from test_workflow import arguments, small_payload
 from test_ai_strategy import AI_TEXT
@@ -27,8 +28,10 @@ class SlidingPipeline(unittest.TestCase):
                         for key in ('budget', 'product', 'audience')}
         def analyze(directory):
             payload = small_payload()
+            details = write_detail_rows(directory, iter(payload.pop('details')))
+            payload['detailSource'] = details.reference(directory)
             (directory / 'payload.json').write_text(json.dumps(payload), encoding='utf-8')
-            return payload, {'detail_count': 1, 'curr_total': 100}
+            return {**payload, 'details': details}, {'detail_count': len(details), 'curr_total': 100}
         with patch('run_workflow.fetch_business', side_effect=fetch), \
              patch('run_workflow.summarize_request', side_effect=analyze), \
              patch('strategy_output.TagCache.get', return_value=dictionaries):