mirror of
https://github.com/ZhuLinsen/daily_stock_analysis.git
synced 2026-10-06 14:43:29 +08:00
feat: 扩展 A 股指数注册表与 CSI provider 路由 (#2267)
* feat: add JSON-backed index registry * fix: address index registry review feedback * fix: validate raw duplicate index aliases * fix: converge CSI task identities and registry fallback * docs: align index identity terminology
This commit is contained in:
Elvis Wang
@@ -31676,5 +31676,36 @@
|
||||
["018260.KS","018260.KS","三星SDS","sanxingSDS","sxSDS",["Samsung SDS","삼성에스디에스"],"KR","stock",true,100],
|
||||
["011200.KS","011200.KS","HMM","HMM","HMM",["现代商船"],"KR","stock",true,100],
|
||||
["090430.KS","090430.KS","爱茉莉太平洋","aimolitaipingyang","amltpy",["Amorepacific","아모레퍼시픽"],"KR","stock",true,100],
|
||||
["086520.KQ","086520.KQ","Ecopro","Ecopro","Ecopro",["에코프로"],"KR","stock",true,100]
|
||||
["086520.KQ","086520.KQ","Ecopro","Ecopro","Ecopro",["에코프로"],"KR","stock",true,100],
|
||||
["csi930955","930955.CSI","红利低波100","honglidibo100","hldb100",[],"CN","index",true,100],
|
||||
["csi931052","931052.CSI","国信价值","guoxinjiazhi","gxjz",[],"CN","index",true,100],
|
||||
["csi931446","931446.CSI","东证红利低波","dongzhenghonglidibo","dzhldb",[],"CN","index",true,100],
|
||||
["csi931643","931643.CSI","科创创业50","kechuangchuangye50","kccy50",[],"CN","index",true,100],
|
||||
["csi932365","932365.CSI","中证现金流","zhongzhengxianjinliu","zzxjl",[],"CN","index",true,100],
|
||||
["csi932366","932366.CSI","300现金流","300xianjinliu","300xjl",[],"CN","index",true,100],
|
||||
["sh000001","sh000001","上证指数","shangzhengzhishu","szzs",["000001.SH"],"CN","index",true,100],
|
||||
["sh000009","sh000009","上证380","shangzheng380","sz380",["000009.SH"],"CN","index",true,100],
|
||||
["sh000010","sh000010","上证180","shangzheng180","sz180",["000010.SH"],"CN","index",true,100],
|
||||
["sh000015","sh000015","红利指数","honglizhishu","hlzs",["000015.SH"],"CN","index",true,100],
|
||||
["sh000016","sh000016","上证50","shangzheng50","sz50",["000016.SH"],"CN","index",true,100],
|
||||
["sh000300","sh000300","沪深300","hushen300","hs300",["sz399300","000300.SH","000300.CSI"],"CN","index",true,100],
|
||||
["sh000688","sh000688","科创50","kechuang50","kc50",["000688.SH"],"CN","index",true,100],
|
||||
["sh000852","sh000852","中证1000","zhongzheng1000","zz1000",["000852.SH"],"CN","index",true,100],
|
||||
["sh000905","sh000905","中证500","zhongzheng500","zz500",["sz399905","000905.SH","000905.CSI"],"CN","index",true,100],
|
||||
["sh000906","sh000906","中证800","zhongzheng800","zz800",["000906.SH"],"CN","index",true,100],
|
||||
["sh000922","sh000922","中证红利","zhongzhenghongli","zzhl",["sz399922","000922.SH","000922.CSI"],"CN","index",true,100],
|
||||
["sh000941","sh000941","新能源","xinnengyuan","xny",["000941.SH"],"CN","index",true,100],
|
||||
["sz399001","sz399001","深证成指","shenzhengchengzhi","szcz",["399001.SZ"],"CN","index",true,100],
|
||||
["sz399005","sz399005","中小100","zhongxiao100","zx100",["399005.SZ"],"CN","index",true,100],
|
||||
["sz399006","sz399006","创业板指","chuangyebanzhi","cybz",["399006.SZ"],"CN","index",true,100],
|
||||
["sz399296","sz399296","创成长","chuangchengzhang","ccz",["399296.SZ"],"CN","index",true,100],
|
||||
["sz399303","sz399303","国证2000","guozheng2000","gz2000",["399303.SZ"],"CN","index",true,100],
|
||||
["sz399324","sz399324","深证红利","shenzhenghongli","szhl",["399324.SZ"],"CN","index",true,100],
|
||||
["sz399330","sz399330","深证100","shenzheng100","sz100",["399330.SZ"],"CN","index",true,100],
|
||||
["sz399967","sz399967","中证军工","zhongzhengjungong","zzjg",["399967.SZ"],"CN","index",true,100],
|
||||
["sz399975","sz399975","证券公司","zhengquangongsi","zqgs",["399975.SZ"],"CN","index",true,100],
|
||||
["sz399986","sz399986","中证银行","zhongzhengyinhang","zzyh",["399986.SZ"],"CN","index",true,100],
|
||||
["sz399989","sz399989","中证医疗","zhongzhengyiliao","zzyl",["399989.SZ"],"CN","index",true,100],
|
||||
["sz399997","sz399997","中证白酒","zhongzhengbaijiu","zzbj",["399997.SZ"],"CN","index",true,100],
|
||||
["sz980092","sz980092","自由现金流","ziyouxianjinliu","zyxjl",["980092.SZ"],"CN","index",true,100]
|
||||
]
|
||||
|
||||
@@ -187,6 +187,56 @@ describe('stockIndexLoader', () => {
|
||||
const fetchCallArgs = mockFetch.mock.calls[0][0];
|
||||
expect(fetchCallArgs).toContain('?_t=');
|
||||
});
|
||||
|
||||
test('filters out assetType=index rows from the returned data', async () => {
|
||||
const withIndex = [
|
||||
...mockIndexData,
|
||||
{
|
||||
canonicalCode: 'sh000300',
|
||||
displayCode: 'sh000300',
|
||||
nameZh: '沪深300',
|
||||
pinyinFull: 'hushen300',
|
||||
pinyinAbbr: 'hs300',
|
||||
aliases: ['000300.SH'],
|
||||
market: 'CN',
|
||||
assetType: 'index',
|
||||
active: true,
|
||||
popularity: 100,
|
||||
},
|
||||
];
|
||||
|
||||
mockFetch.mockResolvedValueOnce({
|
||||
ok: true,
|
||||
json: async () => withIndex,
|
||||
} as unknown as Response);
|
||||
|
||||
const result = await loadStockIndex();
|
||||
|
||||
expect(result.loaded).toBe(true);
|
||||
expect(result.fallback).toBe(false);
|
||||
// Index rows are hidden from the current consumers.
|
||||
expect(result.data.some(item => item.assetType === 'index')).toBe(false);
|
||||
// Stock rows are preserved.
|
||||
expect(result.data).toHaveLength(mockIndexData.length);
|
||||
});
|
||||
|
||||
test('filters index rows from compressed tuple payload', async () => {
|
||||
const compressedWithIndex = [
|
||||
['600519.SH', '600519', '贵州茅台', 'guizhoumaotai', 'gzmt', ['茅台'], 'CN', 'stock', true, 100],
|
||||
['sh000300', 'sh000300', '沪深300', 'hushen300', 'hs300', ['000300.SH'], 'CN', 'index', true, 100],
|
||||
];
|
||||
|
||||
mockFetch.mockResolvedValueOnce({
|
||||
ok: true,
|
||||
json: async () => compressedWithIndex,
|
||||
} as unknown as Response);
|
||||
|
||||
const result = await loadStockIndex();
|
||||
|
||||
expect(result.loaded).toBe(true);
|
||||
expect(result.data).toHaveLength(1);
|
||||
expect(result.data[0].canonicalCode).toBe('600519.SH');
|
||||
});
|
||||
});
|
||||
|
||||
describe('compressIndex - Compress index', () => {
|
||||
|
||||
@@ -39,8 +39,14 @@ export async function loadStockIndex(): Promise<IndexLoadResult> {
|
||||
? unpackTuples(data as StockIndexTuple[])
|
||||
: data as StockIndexItem[];
|
||||
|
||||
// The shared payload may now carry ``assetType=index`` rows, but the
|
||||
// current autocomplete/popular/group consumers must not see them. Filter
|
||||
// index rows out before constructing the successful result so stock/ETF
|
||||
// behaviour is unchanged.
|
||||
const visibleItems = items.filter(item => item.assetType !== 'index');
|
||||
|
||||
return {
|
||||
data: items,
|
||||
data: visibleItems,
|
||||
loaded: true,
|
||||
fallback: false,
|
||||
};
|
||||
|
||||
@@ -886,6 +886,13 @@ class DataFetcherManager:
|
||||
if entry is None:
|
||||
return ""
|
||||
exchange = entry.exchange.upper()
|
||||
if exchange == "CSI":
|
||||
# CSI indices are only supported by AkShare (``csi{code}``); the
|
||||
# other providers in the fixed daily chain return an empty symbol
|
||||
# so the caller records an ``unsupported`` provider-run and skips.
|
||||
if fetcher_name == "AkshareFetcher":
|
||||
return f"csi{entry.bare_code}"
|
||||
return ""
|
||||
if exchange not in {"SH", "SZ"}:
|
||||
return ""
|
||||
if fetcher_name in {"TencentFetcher", "AkshareFetcher"}:
|
||||
@@ -1637,6 +1644,9 @@ class DataFetcherManager:
|
||||
raw_stock_code = (stock_code or "").strip()
|
||||
target = parse_analysis_target(raw_stock_code)
|
||||
self._warn_bare_index_conflict(target)
|
||||
if target.asset_type == ParseStatus.UNSUPPORTED:
|
||||
reason = target.unsupported_reason or "unsupported analysis target"
|
||||
raise DataFetchError(f"{raw_stock_code}: {reason}")
|
||||
if target.asset_type == ParseStatus.INDEX:
|
||||
return self._get_cn_index_daily_data(
|
||||
target,
|
||||
@@ -2625,6 +2635,13 @@ class DataFetcherManager:
|
||||
raw_stock_code = (stock_code or "").strip()
|
||||
target = parse_analysis_target(raw_stock_code)
|
||||
self._warn_bare_index_conflict(target)
|
||||
if target.asset_type == ParseStatus.UNSUPPORTED:
|
||||
logger.warning(
|
||||
"[股票名称] 跳过不支持的输入 %s: %s",
|
||||
raw_stock_code,
|
||||
target.unsupported_reason or "unsupported analysis target",
|
||||
)
|
||||
return ""
|
||||
if target.asset_type == ParseStatus.INDEX:
|
||||
return self._get_cn_index_name(target)
|
||||
|
||||
@@ -2761,6 +2778,13 @@ class DataFetcherManager:
|
||||
normalized_codes: List[str] = []
|
||||
for code in stock_codes:
|
||||
target = parse_analysis_target(code)
|
||||
if target.asset_type == ParseStatus.UNSUPPORTED:
|
||||
logger.warning(
|
||||
"[股票名称] 预取跳过不支持的输入 %s: %s",
|
||||
code,
|
||||
target.unsupported_reason or "unsupported analysis target",
|
||||
)
|
||||
continue
|
||||
normalized_codes.append(
|
||||
target.canonical_id
|
||||
if target.asset_type == ParseStatus.INDEX
|
||||
|
||||
@@ -14,6 +14,17 @@ and this project adheres to [Semantic Versioning](https://semver.org/).
|
||||
<!-- 新条目格式：- [类型] 描述（类型取值：新功能/改进/修复/文档/测试/chore）-->
|
||||
<!-- 每条独立一行追加到本段末尾，无需分类标题，合并时冲突最小 -->
|
||||
|
||||
- [新功能] 建立唯一、可生成、可校验、可降级的指数身份注册表：由 `scripts/stock_index_seeds/index_registry.csv` 的 31 项 manifest 确定性合并进 `apps/dsa-web/public/stocks.index.json`，运行时唯一真源为 JSON 中通过校验的 `active=true`/`assetType=index` 行，移除 `stock_list_parser` 的 5 项硬编码白名单；支持 `--index-only` 生成与字节稳定输出。
|
||||
- [新功能] 补齐显式 SH/SZ/CSI 指数 alias 收敛与 CSI 身份：`sh000300`/`000300.SH`/`sz399300`/`399300.SZ`/`000300.CSI` 均解析到 `sh000300`，`csi930955`/`930955.CSI` 解析到 `csi930955`；未登记 `.CSI` 输入返回 `unsupported`；裸数字恒为 stock 并仅通过 `matched_index` 暴露歧义。
|
||||
- [新功能] 数据管理器按 SH/SZ/CSI 支持矩阵映射 provider symbol：CSI 仅 AkShare 支持（`csi{code}`），Tencent/TickFlow/Yahoo 返回空 symbol 并记录 `unsupported` provider-run，不触发指数健康熔断。
|
||||
- [改进] 存储层 `_derive_canonical_id` 统一为 parser 推导（裸码=stock、显式指数=index），并新增幂等分批修复历史裸码错误 canonical 串桶（`000001`/`000016`/`000688`/`930955` 等），显式指数行与正确 stock 行不受影响，registry 为空时修复 no-op。
|
||||
- [改进] Web loader 完整解压含 index 的共享 payload，但在返回给 autocomplete/popular/group 消费面前过滤 `assetType=index`，股票与 ETF 行为保持不变。
|
||||
- [测试] 为生成器、loader、parser、provider 路由、存储修复与 Web 门槛补充 TDD 回归锚点。
|
||||
- [修复] PR #2267 review 收敛 CSI 显式身份：将 `csi` prefix（canonical）与 `.CSI` suffix（显式 alias）在 parser/build/runtime 规范化器中分离，未登记显式 `csiNNNNNN`/`NNNNNN.CSI` 一律返回 `unsupported`（不再落入美股或猜测 SH/SZ），并防止未登记 `csi000300` 被等价成已登记 `000300.CSI` alias；存储 `_derive_canonical_id` 对 unsupported 输入返回 NULL，避免进入持久化 canonical 桶。
|
||||
- [修复] 在 seed、build entry 与 runtime candidate 三层严格校验 alias 唯一性与整数 popularity：NFKC/casefold 等价 alias 跨条目冲突被拒绝（无静默覆盖），非负整数之外（小数/布尔/负值/字符串）popularity 一律拒绝，整数 `100` 保持有效。
|
||||
- [修复] 收敛已登记 CSI 显式身份在 resolver、任务去重键与历史候选中的分裂：`csi930955`/`930955.CSI`/`CSI930955` 统一解析为 parser canonical `csi930955`，未登记 `csi930956`/`930956.CSI` 保持既有降级语义；`is_code_like()`、REST/watchlist 输入边界与完整 Pipeline 透传不变。
|
||||
- [修复] 阻止任意更新的非 bundled 指数候选（含 legacy `static` 子集）在 remote 缺失/损坏时以 active-index 子集覆盖 bundled baseline：所有非 bundled 候选必须为 bundled active-index canonical 集合的合法超集，否则回退 bundled 并记录 WARNING。
|
||||
|
||||
## [3.31.0] - 2026-08-23
|
||||
|
||||
### 发布亮点
|
||||
|
||||
@@ -11,14 +11,17 @@ Input:
|
||||
Output: apps/dsa-web/public/stocks.index.json
|
||||
|
||||
Usage:
|
||||
python3 scripts/generate_index_from_csv.py # 默认使用 Tushare
|
||||
python3 scripts/generate_index_from_csv.py --source akshare
|
||||
python3 scripts/generate_index_from_csv.py --test # 测试模式
|
||||
python scripts/generate_index_from_csv.py # 默认使用 Tushare
|
||||
python scripts/generate_index_from_csv.py --source akshare
|
||||
python scripts/generate_index_from_csv.py --test # 测试模式
|
||||
python scripts/generate_index_from_csv.py --index-only --test # 仅合并指数 seed
|
||||
"""
|
||||
|
||||
import argparse
|
||||
import csv
|
||||
import json
|
||||
import math
|
||||
import os
|
||||
import re
|
||||
import sys
|
||||
import unicodedata
|
||||
@@ -28,6 +31,8 @@ from typing import List, Dict, Any, Optional
|
||||
# Add the project root to sys.path.
|
||||
sys.path.insert(0, str(Path(__file__).parent.parent))
|
||||
|
||||
from src.services.stock_index_remote_service import validate_stock_index_payload
|
||||
|
||||
try:
|
||||
from pypinyin import lazy_pinyin, Style
|
||||
PYPINYIN_AVAILABLE = True
|
||||
@@ -616,6 +621,355 @@ def compress_index(index: List[Dict[str, Any]]) -> List[List]:
|
||||
return compressed
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Index registry seed — build-time manifest merge.
|
||||
# ---------------------------------------------------------------------------
|
||||
_INDEX_REGISTRY_SEED_PATH = Path(__file__).parent / "stock_index_seeds" / "index_registry.csv"
|
||||
_INDEX_NAMESPACE_RE = re.compile(r"^(sh|sz|csi)\d{6}$")
|
||||
_EXPLICIT_INDEX_ALIAS_RE = re.compile(
|
||||
r"^(?:(?:sh|sz|csi)\d{6}|\d{6}\.(?:sh|sz|csi))$"
|
||||
)
|
||||
|
||||
|
||||
def load_index_registry_seed(seed_path: Optional[Path] = None) -> List[Dict[str, Any]]:
|
||||
"""Load the approved index-registry seed CSV into raw row dicts.
|
||||
|
||||
Columns: ``canonical_code,display_code,name_zh,aliases,name_source,popularity``.
|
||||
``aliases`` uses the existing ``|``-separated ``parse_aliases()`` convention.
|
||||
"""
|
||||
path = seed_path or _INDEX_REGISTRY_SEED_PATH
|
||||
if not path.is_file():
|
||||
raise FileNotFoundError(f"index registry seed not found: {path}")
|
||||
|
||||
rows: List[Dict[str, Any]] = []
|
||||
# Normalized identity keys must not map to more than one canonical within
|
||||
# the seed. A NFKC/casefold-equivalent duplicate alias owned by two entries
|
||||
# (e.g. ``csi930955`` and ``CSI930955`` split across rows) would otherwise
|
||||
# silently overwrite one identity, so it is rejected at the build-time
|
||||
# boundary instead of at runtime. A key that equals its own row's canonical
|
||||
# (e.g. alias ``000300.SH`` on row ``sh000300``) is legitimate and skipped.
|
||||
seen_identity_keys: Dict[str, str] = {}
|
||||
with open(path, "r", encoding="utf-8-sig") as f:
|
||||
reader = csv.DictReader(f)
|
||||
for row in reader:
|
||||
canonical = (row.get("canonical_code") or "").strip()
|
||||
display = (row.get("display_code") or "").strip()
|
||||
name = (row.get("name_zh") or "").strip()
|
||||
if not canonical or not display or not name:
|
||||
raise ValueError(f"index registry seed row missing required field: {row}")
|
||||
|
||||
raw_aliases = [
|
||||
alias.strip()
|
||||
for alias in str(row.get("aliases") or "").split("|")
|
||||
if alias.strip()
|
||||
]
|
||||
_validate_unique_index_aliases(raw_aliases, canonical)
|
||||
aliases = parse_aliases(row)
|
||||
_validate_unique_index_aliases(aliases, canonical)
|
||||
popularity_raw = (row.get("popularity") or "100").strip() or "100"
|
||||
try:
|
||||
popularity = int(popularity_raw)
|
||||
except (TypeError, ValueError) as exc:
|
||||
raise ValueError(
|
||||
f"index registry seed popularity must be a plain integer, "
|
||||
f"got {popularity_raw!r} for canonical {canonical!r}"
|
||||
) from exc
|
||||
if popularity < 0:
|
||||
raise ValueError(
|
||||
f"index registry seed popularity must be non-negative: "
|
||||
f"{popularity!r} for canonical {canonical!r}"
|
||||
)
|
||||
|
||||
# Reject a normalized identity key that maps to a different canonical.
|
||||
for key in [canonical, display] + aliases:
|
||||
norm_key = _normalize_index_key(key)
|
||||
if not norm_key:
|
||||
continue
|
||||
existing = seen_identity_keys.get(norm_key)
|
||||
if existing is not None and existing != canonical:
|
||||
raise ValueError(
|
||||
f"seed identity key {key!r} normalizes to {norm_key!r} "
|
||||
f"already owned by canonical {existing!r}"
|
||||
)
|
||||
seen_identity_keys[norm_key] = canonical
|
||||
|
||||
rows.append({
|
||||
"canonical_code": canonical,
|
||||
"display_code": display,
|
||||
"name_zh": name,
|
||||
"aliases": aliases,
|
||||
"name_source": (row.get("name_source") or "").strip(),
|
||||
"popularity": popularity,
|
||||
})
|
||||
return rows
|
||||
|
||||
|
||||
def _normalize_index_key(value: str) -> str:
|
||||
"""Normalize resolver keys while keeping CSI suffix aliases distinct."""
|
||||
normalized = unicodedata.normalize(
|
||||
"NFKC", str(value or "")
|
||||
).strip().casefold()
|
||||
prefix_match = re.fullmatch(r"(sh|sz)(\d{6})", normalized)
|
||||
if prefix_match:
|
||||
return f"{prefix_match.group(1)}{prefix_match.group(2)}"
|
||||
suffix_match = re.fullmatch(r"(\d{6})\.(sh|sz)", normalized)
|
||||
if suffix_match:
|
||||
return f"{suffix_match.group(2)}{suffix_match.group(1)}"
|
||||
return normalized
|
||||
|
||||
|
||||
def _validate_unique_index_aliases(aliases: Any, canonical: str) -> None:
|
||||
"""Reject duplicate aliases after NFKC/case-insensitive normalization."""
|
||||
if not isinstance(aliases, list):
|
||||
raise ValueError(f"index aliases must be a list: {canonical!r}")
|
||||
seen: set[str] = set()
|
||||
for alias in aliases:
|
||||
if not isinstance(alias, str):
|
||||
raise ValueError(f"index alias must be a string: {canonical!r}")
|
||||
normalized = _normalize_index_key(alias)
|
||||
if normalized in seen:
|
||||
raise ValueError(
|
||||
f"duplicate index alias after normalization for {canonical!r}: {alias!r}"
|
||||
)
|
||||
seen.add(normalized)
|
||||
|
||||
|
||||
def build_index_entries_from_seed(seed_rows: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
|
||||
"""Convert seed rows into the 10-column index tuple dicts.
|
||||
|
||||
``display_code`` is honored from the seed (SH/SZ display equals canonical;
|
||||
CSI display is ``{code}.CSI``). All rows are ``market=CN``,
|
||||
``assetType=index``, ``active=True``, ``popularity`` from seed.
|
||||
"""
|
||||
entries: List[Dict[str, Any]] = []
|
||||
for row in seed_rows:
|
||||
canonical = row["canonical_code"]
|
||||
_validate_unique_index_aliases(row.get("aliases"), canonical)
|
||||
popularity = row.get("popularity", 100)
|
||||
if (
|
||||
isinstance(popularity, bool)
|
||||
or not isinstance(popularity, int)
|
||||
or popularity < 0
|
||||
):
|
||||
raise ValueError(
|
||||
f"index popularity must be a non-negative integer: {canonical!r}"
|
||||
)
|
||||
display = (row.get("display_code") or "").strip() or canonical
|
||||
pinyin_full, pinyin_abbr = generate_pinyin(row["name_zh"])
|
||||
entries.append({
|
||||
"canonicalCode": canonical,
|
||||
"displayCode": display,
|
||||
"nameZh": row["name_zh"],
|
||||
"pinyinFull": pinyin_full,
|
||||
"pinyinAbbr": pinyin_abbr,
|
||||
"aliases": list(row["aliases"]),
|
||||
"market": "CN",
|
||||
"assetType": "index",
|
||||
"active": True,
|
||||
"popularity": popularity,
|
||||
})
|
||||
return entries
|
||||
|
||||
|
||||
def validate_index_registry(
|
||||
entries: List[Dict[str, Any]],
|
||||
non_index_rows: Optional[List[List[Any]]] = None,
|
||||
) -> None:
|
||||
"""Semantic validation for the index registry (build-time and candidates).
|
||||
|
||||
Rules (implementation-contracts.md §Semantic Validation):
|
||||
1. canonical matches ``^(sh|sz|csi)\\d{6}$``; SH/SZ display == canonical,
|
||||
CSI display == ``{code}.CSI``.
|
||||
2. market=CN, assetType=index, active=True, non-empty name, valid pinyin,
|
||||
finite numeric popularity.
|
||||
3. canonical/display/alias normalize to exactly one canonical within the set.
|
||||
4. Index explicit keys must not collide with active stock/ETF keys; bare
|
||||
numeric display/alias rejected.
|
||||
5. Text aliases rejected from identity resolver seed.
|
||||
6. Each namespace has at least one daily provider in the manifest matrix.
|
||||
|
||||
``non_index_rows`` carries the active stock/ETF compressed tuples from the
|
||||
same payload so rule 4 can reject an index canonical/display/alias that
|
||||
collides with a stock/ETF identity after normalization.
|
||||
"""
|
||||
if not entries:
|
||||
return
|
||||
|
||||
canonical_map: Dict[str, str] = {}
|
||||
resolver_map: Dict[str, str] = {}
|
||||
bare_conflicts: Dict[str, str] = {}
|
||||
|
||||
# Active stock/ETF identity keys (canonical/display/aliases) that an index
|
||||
# explicit key must never collide with after normalization.
|
||||
stock_keys: Dict[str, str] = {}
|
||||
for row in non_index_rows or []:
|
||||
if not isinstance(row, list) or len(row) < 10:
|
||||
continue
|
||||
if str(row[7] or "").strip() == "index":
|
||||
continue
|
||||
if row[8] is not True:
|
||||
continue
|
||||
stock_canonical = str(row[0] or "").strip()
|
||||
for key in [row[0], row[1]] + list(row[5] if isinstance(row[5], list) else []):
|
||||
norm_key = _normalize_index_key(key)
|
||||
if norm_key:
|
||||
stock_keys.setdefault(norm_key, stock_canonical)
|
||||
|
||||
for entry in entries:
|
||||
canonical = str(entry["canonicalCode"] or "").strip()
|
||||
display = str(entry["displayCode"] or "").strip()
|
||||
name = str(entry["nameZh"] or "").strip()
|
||||
market = str(entry["market"] or "").strip()
|
||||
asset_type = str(entry["assetType"] or "").strip()
|
||||
active = entry["active"]
|
||||
popularity = entry["popularity"]
|
||||
aliases = entry.get("aliases")
|
||||
|
||||
if not _INDEX_NAMESPACE_RE.match(canonical):
|
||||
raise ValueError(f"index canonical must match ^(sh|sz|csi)\\d{{6}}$: {canonical!r}")
|
||||
namespace = canonical[:3] if canonical.startswith("csi") else canonical[:2]
|
||||
if namespace in {"sh", "sz"}:
|
||||
if display != canonical:
|
||||
raise ValueError(f"SH/SZ index display must equal canonical: {canonical!r} != {display!r}")
|
||||
elif namespace == "csi":
|
||||
expected_display = f"{canonical[3:]}.CSI"
|
||||
if display != expected_display:
|
||||
raise ValueError(f"CSI index display must be {expected_display!r}, got {display!r}")
|
||||
|
||||
if market != "CN":
|
||||
raise ValueError(f"index market must be CN: {canonical!r}")
|
||||
if asset_type != "index":
|
||||
raise ValueError(f"index asset_type must be index: {canonical!r}")
|
||||
if active is not True:
|
||||
raise ValueError(f"index active must be True: {canonical!r}")
|
||||
if not name:
|
||||
raise ValueError(f"index name must be non-empty: {canonical!r}")
|
||||
pinyin_full = entry.get("pinyinFull")
|
||||
pinyin_abbr = entry.get("pinyinAbbr")
|
||||
if (
|
||||
not isinstance(pinyin_full, str)
|
||||
or not pinyin_full.strip()
|
||||
or not isinstance(pinyin_abbr, str)
|
||||
or not pinyin_abbr.strip()
|
||||
):
|
||||
raise ValueError(f"index pinyin fields must be non-empty: {canonical!r}")
|
||||
if not isinstance(aliases, list):
|
||||
raise ValueError(f"index aliases must be a list: {canonical!r}")
|
||||
_validate_unique_index_aliases(aliases, canonical)
|
||||
# Popularity must be a plain non-negative integer. Fractional
|
||||
# (``1.5``), boolean (``True``) and negative values are rejected
|
||||
# without truncation — only an integer value like ``100`` is valid.
|
||||
if (
|
||||
isinstance(popularity, bool)
|
||||
or not isinstance(popularity, int)
|
||||
or not math.isfinite(float(popularity))
|
||||
or popularity < 0
|
||||
):
|
||||
raise ValueError(f"index popularity must be a non-negative integer: {canonical!r}")
|
||||
|
||||
if namespace not in {"sh", "sz", "csi"}:
|
||||
raise ValueError(f"index namespace has no provider mapping: {namespace!r}")
|
||||
|
||||
# canonical uniqueness
|
||||
norm_canonical = _normalize_index_key(canonical)
|
||||
if norm_canonical in canonical_map:
|
||||
raise ValueError(f"duplicate index canonical: {canonical!r}")
|
||||
canonical_map[norm_canonical] = canonical
|
||||
|
||||
for alias in aliases:
|
||||
norm_alias = _normalize_index_key(alias)
|
||||
if norm_alias.isdigit():
|
||||
raise ValueError(f"bare numeric display/alias rejected for index: {alias!r}")
|
||||
if not _EXPLICIT_INDEX_ALIAS_RE.fullmatch(norm_alias):
|
||||
raise ValueError(
|
||||
f"index aliases must use an explicit code form: {alias!r}"
|
||||
)
|
||||
|
||||
# canonical + display + aliases must resolve to exactly one canonical
|
||||
for key in [canonical, display] + aliases:
|
||||
norm_key = _normalize_index_key(key)
|
||||
if not norm_key:
|
||||
continue
|
||||
if norm_key.isdigit():
|
||||
raise ValueError(f"bare numeric display/alias rejected for index: {key!r}")
|
||||
if norm_key in resolver_map and resolver_map[norm_key] != canonical:
|
||||
raise ValueError(
|
||||
f"index resolver key {key!r} maps to multiple canonicals "
|
||||
f"({resolver_map[norm_key]} vs {canonical})"
|
||||
)
|
||||
if norm_key in stock_keys:
|
||||
raise ValueError(
|
||||
f"index resolver key {key!r} collides with active stock/ETF "
|
||||
f"identity {stock_keys[norm_key]!r}"
|
||||
)
|
||||
resolver_map[norm_key] = canonical
|
||||
|
||||
# bare-conflict map: numeric base of explicit aliases, for matched_index
|
||||
for alias in aliases:
|
||||
base = "".join(ch for ch in alias if ch.isdigit())
|
||||
if base and base.isdigit() and len(base) == 6:
|
||||
bare_conflicts.setdefault(base, canonical)
|
||||
|
||||
return
|
||||
|
||||
|
||||
def _load_existing_payload(output_path: Path) -> List[List[Any]]:
|
||||
"""Load the existing compressed JSON payload (must be a list)."""
|
||||
with open(output_path, "r", encoding="utf-8") as f:
|
||||
payload = json.load(f)
|
||||
if not isinstance(payload, list):
|
||||
raise ValueError(f"existing payload is not a list: {output_path}")
|
||||
return payload
|
||||
|
||||
|
||||
def _atomic_write_json(output_path: Path, compressed: List[List[Any]]) -> None:
|
||||
"""Write the compressed payload atomically (temp file + os.replace)."""
|
||||
output_path.parent.mkdir(parents=True, exist_ok=True)
|
||||
temp_path = output_path.with_name(f".{output_path.name}.{os.getpid()}.tmp")
|
||||
try:
|
||||
with open(temp_path, "w", encoding="utf-8") as f:
|
||||
f.write("[\n")
|
||||
for i, item in enumerate(compressed):
|
||||
json.dump(item, f, ensure_ascii=False, separators=(",", ":"))
|
||||
if i < len(compressed) - 1:
|
||||
f.write(",\n")
|
||||
else:
|
||||
f.write("\n")
|
||||
f.write("]\n")
|
||||
os.replace(temp_path, output_path)
|
||||
finally:
|
||||
try:
|
||||
temp_path.unlink()
|
||||
except FileNotFoundError:
|
||||
pass
|
||||
|
||||
|
||||
def run_index_only(output_path: Path, *, test: bool = False) -> List[List[Any]]:
|
||||
"""Merge the approved index-registry seed into the existing compressed JSON.
|
||||
|
||||
Preserves all non-index tuples in their original order, removes any old
|
||||
index tuples, appends seed-generated index rows sorted by canonicalCode,
|
||||
validates, and atomically replaces the output (unless ``test``).
|
||||
"""
|
||||
seed_rows = load_index_registry_seed()
|
||||
index_entries = build_index_entries_from_seed(seed_rows)
|
||||
|
||||
existing = _load_existing_payload(output_path)
|
||||
validate_stock_index_payload(existing, min_items=0)
|
||||
non_index = [item for item in existing if not (len(item) > 7 and item[7] == "index")]
|
||||
# Validate the seed index rows against the existing active stock/ETF rows so
|
||||
# an index identity that collides with a stock/ETF key is rejected.
|
||||
validate_index_registry(index_entries, non_index_rows=non_index)
|
||||
|
||||
index_compressed = compress_index(index_entries)
|
||||
index_compressed.sort(key=lambda item: str(item[0]))
|
||||
|
||||
merged = non_index + index_compressed
|
||||
if not test:
|
||||
_atomic_write_json(output_path, merged)
|
||||
return merged
|
||||
|
||||
|
||||
def main():
|
||||
"""主函数"""
|
||||
parser = argparse.ArgumentParser(description='从 CSV 生成股票自动补全索引')
|
||||
@@ -625,6 +979,11 @@ def main():
|
||||
default='tushare',
|
||||
help='数据源选择（默认: tushare）'
|
||||
)
|
||||
parser.add_argument(
|
||||
'--index-only',
|
||||
action='store_true',
|
||||
help='仅合并指数注册表 seed 到现有压缩 JSON，不重建股票索引'
|
||||
)
|
||||
parser.add_argument(
|
||||
'--test', '-t',
|
||||
action='store_true',
|
||||
@@ -635,11 +994,39 @@ def main():
|
||||
print("=" * 60)
|
||||
print("股票索引生成工具（从 CSV）")
|
||||
print("=" * 60)
|
||||
print(f"数据源：{args.source}")
|
||||
|
||||
if not require_pypinyin():
|
||||
return 1
|
||||
|
||||
# 输出路径
|
||||
output_path = (
|
||||
Path(__file__).parent.parent / "apps" / "dsa-web" / "public" / "stocks.index.json"
|
||||
)
|
||||
output_path.parent.mkdir(parents=True, exist_ok=True)
|
||||
|
||||
# --index-only: 只合并指数 seed，不重建股票索引。
|
||||
if args.index_only:
|
||||
print(f"数据源：index-only（合并指数注册表 seed）")
|
||||
print("\n[1/3] 读取指数注册表 seed...")
|
||||
merged = run_index_only(output_path, test=args.test)
|
||||
print(f" 合并后共 {len(merged)} 条记录")
|
||||
index_rows = [item for item in merged if len(item) > 7 and item[7] == "index"]
|
||||
print(f" 其中指数 {len(index_rows)} 条")
|
||||
if args.test:
|
||||
print("\n[2/3] 测试模式：跳过写入文件")
|
||||
else:
|
||||
print(f"\n[2/3] 写入文件：{output_path}")
|
||||
file_size = output_path.stat().st_size
|
||||
print(f" 文件大小：{file_size / 1024:.2f} KB")
|
||||
print("\n[3/3] 验证合并结果...")
|
||||
# In test mode the output file is untouched, so validate/report the
|
||||
# would-be merged payload returned by ``run_index_only`` rather than
|
||||
# reopening the unchanged file.
|
||||
print(f" 验证通过：{len(merged)} 条记录")
|
||||
return 0
|
||||
|
||||
print(f"数据源：{args.source}")
|
||||
|
||||
# 加载数据
|
||||
print("\n[1/5] 读取 CSV 数据...")
|
||||
if args.source == 'tushare':
|
||||
@@ -661,11 +1048,20 @@ def main():
|
||||
print("\n[2/5] 生成索引数据...")
|
||||
index = build_stock_index(stocks)
|
||||
|
||||
# 输出路径
|
||||
output_path = (
|
||||
Path(__file__).parent.parent / "apps" / "dsa-web" / "public" / "stocks.index.json"
|
||||
# 合并指数注册表 seed，防止后续重建擦除 index 行。
|
||||
print("\n[2.5/5] 合并指数注册表 seed...")
|
||||
seed_rows = load_index_registry_seed()
|
||||
index_entries = build_index_entries_from_seed(seed_rows)
|
||||
# Validate the seed index rows against the freshly built stock/ETF rows so
|
||||
# an index identity that collides with a stock/ETF key is rejected.
|
||||
validate_index_registry(
|
||||
index_entries,
|
||||
non_index_rows=compress_index(index),
|
||||
)
|
||||
output_path.parent.mkdir(parents=True, exist_ok=True)
|
||||
# Canonical-sort the index rows so the full rebuild is byte-stable and
|
||||
# matches the ``--index-only`` ordering; non-index rows keep build order.
|
||||
index_entries.sort(key=lambda entry: str(entry["canonicalCode"]))
|
||||
index.extend(index_entries)
|
||||
|
||||
print("\n[3/5] 压缩索引数据...")
|
||||
compressed = compress_index(index)
|
||||
@@ -686,15 +1082,7 @@ def main():
|
||||
print(f" {i + 1}. {item}")
|
||||
else:
|
||||
print(f"\n[4/5] 写入文件：{output_path}")
|
||||
with open(output_path, 'w', encoding='utf-8') as f:
|
||||
f.write('[\n')
|
||||
for i, item in enumerate(compressed):
|
||||
json.dump(item, f, ensure_ascii=False, separators=(',', ':'))
|
||||
if i < len(compressed) - 1:
|
||||
f.write(',\n')
|
||||
else:
|
||||
f.write('\n')
|
||||
f.write(']\n')
|
||||
_atomic_write_json(output_path, compressed)
|
||||
|
||||
file_size = output_path.stat().st_size
|
||||
print(f" 文件大小：{file_size / 1024:.2f} KB")
|
||||
|
||||
@@ -0,0 +1,32 @@
|
||||
canonical_code,display_code,name_zh,aliases,name_source,popularity
|
||||
sh000001,sh000001,上证指数,000001.SH,腾讯,100
|
||||
sh000009,sh000009,上证380,000009.SH,腾讯,100
|
||||
sh000010,sh000010,上证180,000010.SH,腾讯,100
|
||||
sh000016,sh000016,上证50,000016.SH,腾讯,100
|
||||
sh000300,sh000300,沪深300,sz399300|000300.SH|000300.CSI,腾讯,100
|
||||
sh000688,sh000688,科创50,000688.SH,腾讯,100
|
||||
sh000852,sh000852,中证1000,000852.SH,腾讯,100
|
||||
sh000905,sh000905,中证500,sz399905|000905.SH|000905.CSI,腾讯,100
|
||||
sh000906,sh000906,中证800,000906.SH,腾讯,100
|
||||
sz399001,sz399001,深证成指,399001.SZ,腾讯,100
|
||||
sz399005,sz399005,中小100,399005.SZ,腾讯,100
|
||||
sz399006,sz399006,创业板指,399006.SZ,腾讯,100
|
||||
sz399303,sz399303,国证2000,399303.SZ,腾讯,100
|
||||
sz399330,sz399330,深证100,399330.SZ,腾讯,100
|
||||
sh000015,sh000015,红利指数,000015.SH,腾讯,100
|
||||
sh000922,sh000922,中证红利,sz399922|000922.SH|000922.CSI,腾讯,100
|
||||
sz399324,sz399324,深证红利,399324.SZ,腾讯,100
|
||||
sh000941,sh000941,新能源,000941.SH,腾讯,100
|
||||
sz399296,sz399296,创成长,399296.SZ,腾讯,100
|
||||
sz399967,sz399967,中证军工,399967.SZ,腾讯,100
|
||||
sz399975,sz399975,证券公司,399975.SZ,腾讯,100
|
||||
sz399986,sz399986,中证银行,399986.SZ,腾讯,100
|
||||
sz399989,sz399989,中证医疗,399989.SZ,腾讯,100
|
||||
sz399997,sz399997,中证白酒,399997.SZ,腾讯,100
|
||||
sz980092,sz980092,自由现金流,980092.SZ,腾讯,100
|
||||
csi932365,932365.CSI,中证现金流,,东财 2.932365,100
|
||||
csi930955,930955.CSI,红利低波100,,东财 2.930955,100
|
||||
csi931052,931052.CSI,国信价值,,东财 2.931052,100
|
||||
csi931446,931446.CSI,东证红利低波,,东财 2.931446,100
|
||||
csi931643,931643.CSI,科创创业50,,东财 2.931643,100
|
||||
csi932366,932366.CSI,300现金流,,东财 2.932366,100
|
||||
|
@@ -3,6 +3,9 @@ from __future__ import annotations
|
||||
|
||||
import json
|
||||
import logging
|
||||
import math
|
||||
import re
|
||||
import unicodedata
|
||||
from pathlib import Path
|
||||
from threading import RLock
|
||||
from typing import Dict, Iterable, Optional
|
||||
@@ -18,9 +21,34 @@ from src.services.stock_index_remote_service import (
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
_STOCK_INDEX_FILENAME = "stocks.index.json"
|
||||
_EXPLICIT_INDEX_ALIAS_RE = re.compile(
|
||||
r"^(?:(?:sh|sz|csi)\d{6}|\d{6}\.(?:sh|sz|csi))$"
|
||||
)
|
||||
|
||||
|
||||
def _normalize_index_identity_key(value: object) -> str:
|
||||
"""Normalize an identity key to one resolver identity form.
|
||||
|
||||
``sh``/``sz`` prefix and ``{code}.SH`` / ``{code}.SZ`` suffix collapse to
|
||||
the canonical lowercase-prefixed key. CSI is deliberately **not**
|
||||
collapsed: ``csi{code}`` (prefix) and ``{code}.CSI`` (suffix) are kept
|
||||
distinct so an unregistered ``csi`` prefix never conflates with a
|
||||
registered ``{code}.CSI`` alias — mirroring ``stock_list_parser``.
|
||||
"""
|
||||
normalized = unicodedata.normalize(
|
||||
"NFKC", str(value or "")
|
||||
).strip().casefold()
|
||||
prefix_match = re.fullmatch(r"(sh|sz)(\d{6})", normalized)
|
||||
if prefix_match:
|
||||
return f"{prefix_match.group(1)}{prefix_match.group(2)}"
|
||||
suffix_match = re.fullmatch(r"(\d{6})\.(sh|sz)", normalized)
|
||||
if suffix_match:
|
||||
return f"{suffix_match.group(2)}{suffix_match.group(1)}"
|
||||
return normalized
|
||||
_STOCK_INDEX_CACHE: Dict[str, str] | None = None
|
||||
_STOCK_CODE_LOOKUP_CACHE: Dict[str, str] | None = None
|
||||
_STOCK_CODE_CANDIDATES_CACHE: Dict[str, tuple[str, ...]] | None = None
|
||||
_ACTIVE_INDEX_ROWS_CACHE: list | None = None
|
||||
_REMOTE_INDEX_VALIDITY_CACHE: tuple[Path, float, int, bool] | None = None
|
||||
_STOCK_INDEX_CACHE_LOCK = RLock()
|
||||
|
||||
@@ -373,14 +401,266 @@ def get_stock_code_index_map() -> Dict[str, str]:
|
||||
return _STOCK_CODE_LOOKUP_CACHE
|
||||
|
||||
|
||||
def _load_active_index_rows() -> list:
|
||||
"""Load and cache the active ``assetType=index`` rows from the best candidate.
|
||||
|
||||
Returns raw compressed tuples (no parser type dependency) so the loader
|
||||
stays free of a circular import with ``stock_list_parser``. Each candidate
|
||||
must pass base + semantic validation before its index rows are used.
|
||||
|
||||
Non-regression guard: when the bundled candidate is a valid baseline, every
|
||||
OTHER candidate — the remote cache AND the legacy ``static`` fallback — must
|
||||
be a legal superset of the bundled active-index canonical set to be selected.
|
||||
A candidate that drops any bundled active index canonical is skipped
|
||||
(WARNING) in favour of the bundled candidate, so a stale or partial
|
||||
non-bundled file can never bypass the bundled baseline when the remote cache
|
||||
is missing/invalid.
|
||||
"""
|
||||
global _ACTIVE_INDEX_ROWS_CACHE
|
||||
|
||||
if _ACTIVE_INDEX_ROWS_CACHE is not None:
|
||||
return _ACTIVE_INDEX_ROWS_CACHE
|
||||
|
||||
with _STOCK_INDEX_CACHE_LOCK:
|
||||
if _ACTIVE_INDEX_ROWS_CACHE is not None:
|
||||
return _ACTIVE_INDEX_ROWS_CACHE
|
||||
|
||||
remote_path = get_remote_stock_index_cache_path()
|
||||
candidate_paths = get_stock_index_candidate_paths()
|
||||
candidates = _get_fresh_stock_index_candidates(candidate_paths, remote_path)
|
||||
|
||||
# First pass: find the bundled candidate's active index canonical set
|
||||
# as the non-regression baseline. The bundled candidate is the
|
||||
# ``apps/dsa-web/public/stocks.index.json`` path (identified from the
|
||||
# declared candidate order, not the first non-remote candidate ordered
|
||||
# by mtime), and its index rows must pass semantic validation before
|
||||
# they become the baseline.
|
||||
bundled_path = _get_bundled_stock_index_path(candidate_paths, remote_path)
|
||||
bundled_rows: list | None = None
|
||||
bundled_canonicals: set[str] | None = None
|
||||
if bundled_path is not None:
|
||||
try:
|
||||
raw_items = _load_stock_index_payload(bundled_path)
|
||||
validate_stock_index_payload(raw_items, min_items=0)
|
||||
rows = _extract_active_index_rows(raw_items)
|
||||
if rows:
|
||||
_validate_index_rows_semantics(
|
||||
rows, _extract_active_non_index_rows(raw_items)
|
||||
)
|
||||
bundled_rows = rows
|
||||
bundled_canonicals = {str(r[0]) for r in rows}
|
||||
except (OSError, TypeError, ValueError) as exc:
|
||||
logger.debug("[股票索引] 解析 bundled 指数行失败 %s: %s", bundled_path, exc)
|
||||
|
||||
# Second pass: pick the best candidate (remote preferred when it is a
|
||||
# legal superset of the bundled baseline).
|
||||
for index_path in candidates:
|
||||
try:
|
||||
raw_items = _load_stock_index_payload(index_path)
|
||||
if _same_path(index_path, remote_path):
|
||||
validate_stock_index_payload(raw_items)
|
||||
else:
|
||||
validate_stock_index_payload(raw_items, min_items=0)
|
||||
rows = _extract_active_index_rows(raw_items)
|
||||
if not rows:
|
||||
continue
|
||||
_validate_index_rows_semantics(
|
||||
rows, _extract_active_non_index_rows(raw_items)
|
||||
)
|
||||
# Non-regression guard: when the bundled candidate is a valid
|
||||
# baseline, every OTHER candidate (remote cache or legacy
|
||||
# ``static`` fallback) must be a legal superset of the bundled
|
||||
# active-index canonical set. A candidate that drops any
|
||||
# bundled baseline canonical is skipped (WARNING) so a stale
|
||||
# or partial legacy-static file can never bypass the bundled
|
||||
# baseline when the remote cache is missing/invalid.
|
||||
if bundled_canonicals and bundled_path is not None and not _same_path(index_path, bundled_path):
|
||||
candidate_canonicals = {str(r[0]) for r in rows}
|
||||
missing = bundled_canonicals - candidate_canonicals
|
||||
if missing:
|
||||
logger.warning(
|
||||
"[股票索引] 指数候选 %s 缺少 bundled baseline canonical: %s — 回退 bundled",
|
||||
index_path,
|
||||
sorted(missing),
|
||||
)
|
||||
continue
|
||||
_ACTIVE_INDEX_ROWS_CACHE = rows
|
||||
return rows
|
||||
except (OSError, TypeError, ValueError) as exc:
|
||||
logger.debug("[股票索引] 解析指数行失败 %s: %s", index_path, exc)
|
||||
|
||||
# All candidates failed or were skipped — empty registry + WARNING.
|
||||
logger.warning("[股票索引] 所有指数候选均失败，指数注册表为空")
|
||||
_ACTIVE_INDEX_ROWS_CACHE = []
|
||||
return _ACTIVE_INDEX_ROWS_CACHE
|
||||
|
||||
|
||||
def _get_bundled_stock_index_path(
|
||||
candidate_paths: Iterable[Path],
|
||||
remote_path: Path,
|
||||
) -> Path | None:
|
||||
"""Return the bundled ``apps/dsa-web/public/stocks.index.json`` path.
|
||||
|
||||
The bundled candidate is the deterministic non-remote baseline for the
|
||||
remote non-regression check, independent of mtime ordering. It is the
|
||||
declared ``apps/dsa-web/public`` candidate (the non-remote, non-legacy
|
||||
``static`` candidate in the declared candidate order), not the remote cache
|
||||
and not the legacy ``static`` fallback.
|
||||
"""
|
||||
for candidate in candidate_paths:
|
||||
if _same_path(candidate, remote_path):
|
||||
continue
|
||||
if candidate.as_posix().endswith("static/stocks.index.json"):
|
||||
continue
|
||||
if candidate.is_file():
|
||||
return candidate
|
||||
return None
|
||||
|
||||
|
||||
def _extract_active_index_rows(raw_items: list) -> list:
|
||||
"""Return the active ``assetType=index`` rows from a raw payload."""
|
||||
rows = []
|
||||
for item in raw_items:
|
||||
if not isinstance(item, list) or len(item) < 10:
|
||||
continue
|
||||
if item[7] != "index":
|
||||
continue
|
||||
if item[8] is not True:
|
||||
continue
|
||||
rows.append(item)
|
||||
return rows
|
||||
|
||||
|
||||
def _extract_active_non_index_rows(raw_items: list) -> list:
|
||||
"""Return the active stock/ETF rows from a raw payload (for collision checks)."""
|
||||
rows = []
|
||||
for item in raw_items:
|
||||
if not isinstance(item, list) or len(item) < 10:
|
||||
continue
|
||||
if str(item[7] or "").strip() == "index":
|
||||
continue
|
||||
if item[8] is not True:
|
||||
continue
|
||||
rows.append(item)
|
||||
return rows
|
||||
|
||||
|
||||
def _validate_index_rows_semantics(rows: list, non_index_rows: list | None = None) -> None:
|
||||
"""Semantic validation for active index rows (implementation-contracts §Semantic Validation).
|
||||
|
||||
Raises ``ValueError`` on any identity/alias/stock-index conflict so a
|
||||
malformed candidate is rejected rather than silently loaded. ``non_index_rows``
|
||||
carries the active stock/ETF rows from the same payload so an index
|
||||
canonical/display/alias that collides with a stock/ETF identity after
|
||||
normalization is rejected.
|
||||
"""
|
||||
canonical_map: dict[str, str] = {}
|
||||
resolver_map: dict[str, str] = {}
|
||||
|
||||
# Active stock/ETF identity keys an index explicit key must never collide with.
|
||||
stock_keys: dict[str, str] = {}
|
||||
for row in non_index_rows or []:
|
||||
stock_canonical = str(row[0] or "").strip()
|
||||
for key in [row[0], row[1]] + list(row[5] if isinstance(row[5], list) else []):
|
||||
norm_key = _normalize_index_identity_key(key)
|
||||
if norm_key:
|
||||
stock_keys.setdefault(norm_key, stock_canonical)
|
||||
|
||||
for row in rows:
|
||||
canonical = str(row[0] or "").strip()
|
||||
display = str(row[1] or "").strip()
|
||||
name = str(row[2] or "").strip()
|
||||
pinyin_full = row[3]
|
||||
pinyin_abbr = row[4]
|
||||
market = str(row[6] or "").strip()
|
||||
asset_type = str(row[7] or "").strip()
|
||||
active = row[8]
|
||||
popularity = row[9]
|
||||
aliases = row[5]
|
||||
|
||||
if not re.fullmatch(r"(sh|sz|csi)\d{6}", canonical):
|
||||
raise ValueError(f"index canonical must match ^(sh|sz|csi)\\d{{6}}$: {canonical!r}")
|
||||
namespace = canonical[:3] if canonical.startswith("csi") else canonical[:2]
|
||||
if namespace in {"sh", "sz"}:
|
||||
if display != canonical:
|
||||
raise ValueError(f"SH/SZ index display must equal canonical: {canonical!r}")
|
||||
elif namespace == "csi":
|
||||
if display != f"{canonical[3:]}.CSI":
|
||||
raise ValueError(f"CSI index display must be {canonical[3:]}.CSI: {display!r}")
|
||||
if market != "CN":
|
||||
raise ValueError(f"index market must be CN: {canonical!r}")
|
||||
if asset_type != "index":
|
||||
raise ValueError(f"index asset_type must be index: {canonical!r}")
|
||||
if active is not True:
|
||||
raise ValueError(f"index active must be True: {canonical!r}")
|
||||
if not name:
|
||||
raise ValueError(f"index name must be non-empty: {canonical!r}")
|
||||
if (
|
||||
not isinstance(pinyin_full, str)
|
||||
or not pinyin_full.strip()
|
||||
or not isinstance(pinyin_abbr, str)
|
||||
or not pinyin_abbr.strip()
|
||||
):
|
||||
raise ValueError(f"index pinyin fields must be non-empty: {canonical!r}")
|
||||
if not isinstance(aliases, list):
|
||||
raise ValueError(f"index aliases must be a list: {canonical!r}")
|
||||
if (
|
||||
isinstance(popularity, bool)
|
||||
or not isinstance(popularity, int)
|
||||
or not math.isfinite(float(popularity))
|
||||
or popularity < 0
|
||||
):
|
||||
raise ValueError(f"index popularity must be a non-negative integer: {canonical!r}")
|
||||
|
||||
norm_canonical = _normalize_index_identity_key(canonical)
|
||||
if norm_canonical in canonical_map:
|
||||
raise ValueError(f"duplicate index canonical: {canonical!r}")
|
||||
canonical_map[norm_canonical] = canonical
|
||||
|
||||
seen_aliases: set[str] = set()
|
||||
for alias in aliases:
|
||||
norm_alias = _normalize_index_identity_key(alias)
|
||||
if norm_alias in seen_aliases:
|
||||
raise ValueError(
|
||||
f"duplicate index alias after normalization for {canonical!r}: {alias!r}"
|
||||
)
|
||||
seen_aliases.add(norm_alias)
|
||||
if norm_alias.isdigit():
|
||||
raise ValueError(f"bare numeric display/alias rejected for index: {alias!r}")
|
||||
if not _EXPLICIT_INDEX_ALIAS_RE.fullmatch(norm_alias):
|
||||
raise ValueError(
|
||||
f"index aliases must use an explicit code form: {alias!r}"
|
||||
)
|
||||
|
||||
for key in [canonical, display] + aliases:
|
||||
norm_key = _normalize_index_identity_key(key)
|
||||
if not norm_key:
|
||||
continue
|
||||
if norm_key.isdigit():
|
||||
raise ValueError(f"bare numeric display/alias rejected for index: {key!r}")
|
||||
if norm_key in resolver_map and resolver_map[norm_key] != canonical:
|
||||
raise ValueError(
|
||||
f"index resolver key {key!r} maps to multiple canonicals "
|
||||
f"({resolver_map[norm_key]} vs {canonical})"
|
||||
)
|
||||
if norm_key in stock_keys:
|
||||
raise ValueError(
|
||||
f"index resolver key {key!r} collides with active stock/ETF "
|
||||
f"identity {stock_keys[norm_key]!r}"
|
||||
)
|
||||
resolver_map[norm_key] = canonical
|
||||
|
||||
|
||||
def clear_stock_index_cache() -> None:
|
||||
"""Clear the in-process stock index lookup cache."""
|
||||
global _REMOTE_INDEX_VALIDITY_CACHE
|
||||
global _STOCK_CODE_CANDIDATES_CACHE, _STOCK_CODE_LOOKUP_CACHE, _STOCK_INDEX_CACHE
|
||||
global _ACTIVE_INDEX_ROWS_CACHE
|
||||
with _STOCK_INDEX_CACHE_LOCK:
|
||||
_STOCK_INDEX_CACHE = None
|
||||
_STOCK_CODE_LOOKUP_CACHE = None
|
||||
_STOCK_CODE_CANDIDATES_CACHE = None
|
||||
_ACTIVE_INDEX_ROWS_CACHE = None
|
||||
_REMOTE_INDEX_VALIDITY_CACHE = None
|
||||
|
||||
|
||||
|
||||
@@ -122,6 +122,39 @@ class HistoryService:
|
||||
is_bse_code,
|
||||
normalize_stock_code,
|
||||
)
|
||||
from src.services.stock_code_utils import (
|
||||
_converge_registered_csi_identity,
|
||||
)
|
||||
|
||||
# PR #2267 review remediation: converge registered CSI aliases
|
||||
# (``csi930955`` / ``930955.CSI`` / ``CSI930955``) so a record persisted under any
|
||||
# equivalent form is reachable from every equivalent query input.
|
||||
# This is a *persisted-read* filter path, so the candidate set must
|
||||
# include:
|
||||
# 1. the parser canonical (``csi930955`` — current storage form),
|
||||
# 2. the old resolver's uppercase canonical (``CSI930955`` — how
|
||||
# pre-fix records were saved), and
|
||||
# 3. the IndexEntry's explicit aliases (``930955.CSI``).
|
||||
converged_csi = _converge_registered_csi_identity(raw_code)
|
||||
if converged_csi is not None:
|
||||
from src.data.stock_index_loader import _load_active_index_rows
|
||||
active_rows = _load_active_index_rows()
|
||||
alias_keys: List[str] = []
|
||||
display_keys: List[str] = []
|
||||
for row in active_rows:
|
||||
if row and str(row[0] or "").strip() == converged_csi:
|
||||
display_keys = [str(row[1] or "").strip()]
|
||||
alias_keys = [
|
||||
str(a) for a in (row[5] if isinstance(row[5], list) else [])
|
||||
if str(a).strip()
|
||||
]
|
||||
break
|
||||
canonical_upper = canonical_stock_code(converged_csi) or converged_csi.upper()
|
||||
add_keys = [converged_csi, canonical_upper] + display_keys + alias_keys
|
||||
for key in add_keys:
|
||||
if key and key not in candidates:
|
||||
candidates.append(key)
|
||||
return candidates
|
||||
|
||||
raw_canonical = canonical_stock_code(raw_code)
|
||||
normalized = canonical_stock_code(normalize_stock_code(raw_canonical))
|
||||
|
||||
@@ -6,6 +6,7 @@ Shared stock code utilities.
|
||||
from __future__ import annotations
|
||||
|
||||
import re
|
||||
import unicodedata
|
||||
from importlib import import_module
|
||||
from dataclasses import dataclass
|
||||
from typing import List, Optional
|
||||
@@ -488,6 +489,20 @@ def resolve_index_stock_code_for_analysis(raw: str) -> str:
|
||||
if not text:
|
||||
return ""
|
||||
|
||||
# PR #2267 review remediation: converge registered CSI aliases
|
||||
# (``csi930955`` / ``930955.CSI`` / ``CSI930955``) to the parser canonical so
|
||||
# the resolver, task dedupe key and history candidates do not split the
|
||||
# same index into distinct keys. The parser canonical is returned verbatim
|
||||
# (lowercase ``csi{code}``) rather than re-uppercased by
|
||||
# ``canonical_stock_code`` below.
|
||||
normalized_csi = unicodedata.normalize("NFKC", text).strip().casefold()
|
||||
if re.fullmatch(r"csi\d{6}", normalized_csi) or re.fullmatch(r"\d{6}\.csi", normalized_csi):
|
||||
converged = _converge_registered_csi_identity(text)
|
||||
# Return the parser canonical verbatim (lowercase ``csi{code}``) for
|
||||
# every registered CSI form, including the already-canonical input.
|
||||
if converged is not None:
|
||||
return converged
|
||||
|
||||
if is_code_like(text) or (text.isdigit() and len(text) == 4):
|
||||
from src.data.stock_index_loader import resolve_index_stock_code
|
||||
|
||||
@@ -496,3 +511,35 @@ def resolve_index_stock_code_for_analysis(raw: str) -> str:
|
||||
return canonical_stock_code(resolved)
|
||||
|
||||
return canonical_stock_code(text)
|
||||
|
||||
|
||||
def _converge_registered_csi_identity(raw: str) -> Optional[str]:
|
||||
"""Converge a registered CSI explicit identity to its parser canonical.
|
||||
|
||||
PR #2267 review remediation: the resolver, task dedupe key and history
|
||||
candidate builders must treat ``csi930955`` / ``930955.CSI`` /
|
||||
``CSI930955`` as the same registered CSI
|
||||
index identity, so they do not split into distinct task keys or history
|
||||
candidates. Returns the parser canonical (lowercase ``csi{code}``) when the
|
||||
input is a registered CSI explicit identity, or ``None`` when it is not
|
||||
(so callers keep their existing degradation semantics for unregistered
|
||||
``csi930956`` / ``930956.CSI``).
|
||||
"""
|
||||
text = (raw or "").strip()
|
||||
if not text:
|
||||
return None
|
||||
|
||||
normalized = unicodedata.normalize("NFKC", text).strip().casefold()
|
||||
is_csi_form = bool(
|
||||
re.fullmatch(r"csi\d{6}", normalized)
|
||||
or re.fullmatch(r"\d{6}\.csi", normalized)
|
||||
)
|
||||
if not is_csi_form:
|
||||
return None
|
||||
|
||||
from src.services.stock_list_parser import parse_analysis_target
|
||||
|
||||
target = parse_analysis_target(text)
|
||||
if target.asset_type == "index" and target.canonical_id:
|
||||
return target.canonical_id
|
||||
return None
|
||||
|
||||
@@ -32,6 +32,7 @@ from __future__ import annotations
|
||||
|
||||
import json
|
||||
import re
|
||||
import unicodedata
|
||||
from dataclasses import dataclass, field
|
||||
from pathlib import Path
|
||||
from typing import Dict, Iterable, List, Optional, Tuple
|
||||
@@ -49,6 +50,41 @@ __all__ = [
|
||||
|
||||
|
||||
_STOCK_LIST_SEPARATOR_RE = re.compile(r"[\s,;\uFF0C\u3001\uFF1B]+")
|
||||
_EXPLICIT_INDEX_ALIAS_RE = re.compile(
|
||||
r"^(?:(?:sh|sz|csi)\d{6}|\d{6}\.(?:sh|sz|csi))$"
|
||||
)
|
||||
# An unregistered numeric CSI form must surface as ``unsupported`` rather than
|
||||
# degrade into a US ticker. Registered six-digit forms resolve before this
|
||||
# guard; malformed lengths are rejected here without blocking tickers such as
|
||||
# ``CSIQ``.
|
||||
_EXPLICIT_CSI_FORM_RE = re.compile(r"^(?:csi\d+|\d+\.csi)$")
|
||||
|
||||
|
||||
def _normalize_index_key(value: str) -> str:
|
||||
"""Normalize an identity key to one resolver identity form.
|
||||
|
||||
``sh``/``sz`` prefix and ``{code}.SH`` / ``{code}.SZ`` suffix are
|
||||
interchangeable exchange identities, so they collapse to the canonical
|
||||
lowercase-prefixed key (``sh000300`` == ``000300.SH``).
|
||||
|
||||
CSI is deliberately **not** collapsed: ``csi{code}`` (prefix) and
|
||||
``{code}.CSI`` (suffix) are kept distinct. The ``csi`` prefix is a canonical
|
||||
identity that only resolves when a manifest row owns that exact ``csi{code}``
|
||||
key, while a ``{code}.CSI`` suffix is an *explicit alias* that belongs to the
|
||||
entry that registered it (e.g. ``000300.CSI`` is an alias of ``sh000300``).
|
||||
Keeping them separate prevents ``csi000300`` (unregistered) from being
|
||||
conflated with the registered ``000300.CSI`` alias of ``sh000300``.
|
||||
"""
|
||||
normalized = unicodedata.normalize(
|
||||
"NFKC", str(value or "")
|
||||
).strip().casefold()
|
||||
prefix_match = re.fullmatch(r"(sh|sz)(\d{6})", normalized)
|
||||
if prefix_match:
|
||||
return f"{prefix_match.group(1)}{prefix_match.group(2)}"
|
||||
suffix_match = re.fullmatch(r"(\d{6})\.(sh|sz)", normalized)
|
||||
if suffix_match:
|
||||
return f"{suffix_match.group(2)}{suffix_match.group(1)}"
|
||||
return normalized
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
@@ -208,7 +244,7 @@ class IndexEntry:
|
||||
"""A single known index in :class:`IndexRegistry`."""
|
||||
|
||||
bare_code: str
|
||||
exchange: str # 'SH' or 'SZ'
|
||||
exchange: str # 'SH' / 'SZ' / 'CSI'
|
||||
canonical_id: str
|
||||
display_name: str
|
||||
aliases: Tuple[str, ...] = ()
|
||||
@@ -240,20 +276,64 @@ class AnalysisTarget:
|
||||
class IndexRegistry:
|
||||
"""Authoritative source for ``asset_type=index`` resolution.
|
||||
|
||||
The registry is populated with a small built-in white-list of canonical
|
||||
A-share indices (CSI 300 / SSE 50 / STAR 50 / SZSE Component / ChiNext)
|
||||
that already exist as hard-coded white-lists across
|
||||
``data_provider/*_fetcher.py``. PR1 deliberately keeps the registry
|
||||
in-memory and immutable — later phases of issue #2063 will load the
|
||||
``asset_type=index`` rows from ``apps/dsa-web/public/stocks.index.json``
|
||||
once that file carries index rows; the parser contract won't change.
|
||||
The registry is populated from the validated ``assetType=index`` rows of
|
||||
``apps/dsa-web/public/stocks.index.json`` (via :func:`default_index_registry`).
|
||||
It builds three lookup maps at construction time:
|
||||
|
||||
* canonical map — ``canonical_id -> entry``
|
||||
* resolver-key map — every normalized canonical/display/alias key -> entry
|
||||
* bare-conflict map — the 6-digit numeric base of explicit aliases -> entry
|
||||
|
||||
Construction rejects duplicate or ambiguous keys (identity/alias/stock-index
|
||||
conflicts) by raising ``ValueError`` so a malformed candidate is never
|
||||
silently loaded.
|
||||
"""
|
||||
|
||||
def __init__(self, entries: Iterable[IndexEntry] = ()):
|
||||
self._entries: List[IndexEntry] = list(entries)
|
||||
self._by_canonical: Dict[str, IndexEntry] = {
|
||||
e.canonical_id: e for e in self._entries
|
||||
}
|
||||
self._by_canonical: Dict[str, IndexEntry] = {}
|
||||
self._by_resolver_key: Dict[str, IndexEntry] = {}
|
||||
self._by_bare_conflict: Dict[str, IndexEntry] = {}
|
||||
|
||||
for entry in self._entries:
|
||||
norm_canonical = _normalize_index_key(entry.canonical_id)
|
||||
if norm_canonical in self._by_canonical:
|
||||
raise ValueError(
|
||||
f"duplicate index canonical: {entry.canonical_id!r}"
|
||||
)
|
||||
self._by_canonical[norm_canonical] = entry
|
||||
|
||||
for alias in entry.aliases:
|
||||
norm_alias = _normalize_index_key(alias)
|
||||
if norm_alias.isdigit():
|
||||
raise ValueError(
|
||||
f"bare numeric display/alias rejected for index: {alias!r}"
|
||||
)
|
||||
if not _EXPLICIT_INDEX_ALIAS_RE.fullmatch(norm_alias):
|
||||
raise ValueError(
|
||||
f"index aliases must use an explicit code form: {alias!r}"
|
||||
)
|
||||
|
||||
for key in [entry.canonical_id] + list(entry.aliases):
|
||||
norm_key = _normalize_index_key(key)
|
||||
if not norm_key:
|
||||
continue
|
||||
if norm_key.isdigit():
|
||||
raise ValueError(
|
||||
f"bare numeric display/alias rejected for index: {key!r}"
|
||||
)
|
||||
existing = self._by_resolver_key.get(norm_key)
|
||||
if existing is not None and existing is not entry:
|
||||
raise ValueError(
|
||||
f"index resolver key {key!r} maps to multiple canonicals "
|
||||
f"({existing.canonical_id} vs {entry.canonical_id})"
|
||||
)
|
||||
self._by_resolver_key[norm_key] = entry
|
||||
|
||||
for alias in entry.aliases:
|
||||
base = "".join(ch for ch in alias if ch.isdigit())
|
||||
if base and base.isdigit() and len(base) == 6:
|
||||
self._by_bare_conflict.setdefault(base, entry)
|
||||
|
||||
def __len__(self) -> int:
|
||||
return len(self._entries)
|
||||
@@ -275,11 +355,16 @@ class IndexRegistry:
|
||||
if exchange is None:
|
||||
return None
|
||||
canonical = f"{prefix}{bare_code}"
|
||||
entry = self._by_canonical.get(canonical)
|
||||
entry = self._by_canonical.get(_normalize_index_key(canonical))
|
||||
if entry is not None:
|
||||
return entry
|
||||
# Alias fallback — e.g. user typed ``sz399300`` but registry lists it
|
||||
# under ``sh000300`` via an alias. Walk the registry once.
|
||||
# under ``sh000300`` via an alias. Check the resolver map for the
|
||||
# prefixed form first (``sz399300`` is an explicit alias), then walk
|
||||
# the registry once for bare-code matches.
|
||||
entry = self._by_resolver_key.get(_normalize_index_key(canonical))
|
||||
if entry is not None:
|
||||
return entry
|
||||
for entry in self._entries:
|
||||
if entry.exchange == exchange and entry.matches_code(bare_code):
|
||||
return entry
|
||||
@@ -298,64 +383,86 @@ class IndexRegistry:
|
||||
return entry
|
||||
return None
|
||||
|
||||
def find_by_explicit_key(self, key: str) -> Optional[IndexEntry]:
|
||||
"""Look up an index by an explicit canonical/display/alias key.
|
||||
|
||||
Used by :func:`parse_analysis_target` to resolve ``csi930955`` /
|
||||
``930955.CSI`` and other explicit index forms before generic stock
|
||||
normalization runs.
|
||||
"""
|
||||
return self._by_resolver_key.get(_normalize_index_key(key))
|
||||
|
||||
def find_by_bare_conflict(self, bare_code: str) -> Optional[IndexEntry]:
|
||||
"""Return the index whose explicit alias base equals ``bare_code``."""
|
||||
return self._by_bare_conflict.get(_normalize_index_key(bare_code))
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Default index registry — the built-in white-list.
|
||||
# Default index registry — built from the validated JSON index rows.
|
||||
# ---------------------------------------------------------------------------
|
||||
# Five canonical A-share indices, mirroring the hard-coded lists already
|
||||
# present in ``data_provider/{efinance,akshare,yfinance,tickflow}_fetcher.py``.
|
||||
# Keeping this list in one place + giving it a public ``IndexRegistry`` type
|
||||
# is the whole point of PR1; later phases may move the data into
|
||||
# ``apps/dsa-web/public/stocks.index.json`` and load it, but the parser
|
||||
# contract stays stable.
|
||||
_DEFAULT_INDEX_ENTRIES: Tuple[IndexEntry, ...] = (
|
||||
IndexEntry(
|
||||
bare_code="000300",
|
||||
exchange="SH",
|
||||
canonical_id="sh000300",
|
||||
display_name="沪深300",
|
||||
aliases=("000300.SH", "CSI300", "HS300"),
|
||||
),
|
||||
IndexEntry(
|
||||
bare_code="000016",
|
||||
exchange="SH",
|
||||
canonical_id="sh000016",
|
||||
display_name="上证50",
|
||||
aliases=("000016.SH", "SSE50"),
|
||||
),
|
||||
IndexEntry(
|
||||
bare_code="000688",
|
||||
exchange="SH",
|
||||
canonical_id="sh000688",
|
||||
display_name="科创50",
|
||||
aliases=("000688.SH", "STAR50"),
|
||||
),
|
||||
IndexEntry(
|
||||
bare_code="399001",
|
||||
exchange="SZ",
|
||||
canonical_id="sz399001",
|
||||
display_name="深证成指",
|
||||
aliases=("399001.SZ", "SZSE"),
|
||||
),
|
||||
IndexEntry(
|
||||
bare_code="399006",
|
||||
exchange="SZ",
|
||||
canonical_id="sz399006",
|
||||
display_name="创业板指",
|
||||
aliases=("399006.SZ", "ChiNext"),
|
||||
),
|
||||
)
|
||||
# The unique runtime identity source is the ``active=true`` / ``assetType=index``
|
||||
# rows of the candidate ``stocks.index.json``. The old hard-coded
|
||||
# ``_DEFAULT_INDEX_ENTRIES`` white-list is removed; no second runtime white-list
|
||||
# is kept. When all candidates fail, the registry is empty and every prefixed
|
||||
# input degrades to the stock path (fail-open).
|
||||
|
||||
|
||||
def _index_entry_from_row(row) -> Optional[IndexEntry]:
|
||||
"""Convert one validated active index tuple into an :class:`IndexEntry`."""
|
||||
if not isinstance(row, list) or len(row) < 10:
|
||||
return None
|
||||
canonical = str(row[0] or "").strip()
|
||||
display = str(row[1] or "").strip()
|
||||
name = str(row[2] or "").strip()
|
||||
aliases = row[5] if isinstance(row[5], list) else []
|
||||
if not canonical or not name:
|
||||
return None
|
||||
namespace = canonical[:3] if canonical.startswith("csi") else canonical[:2]
|
||||
if namespace == "csi":
|
||||
exchange = "CSI"
|
||||
bare_code = canonical[3:]
|
||||
elif namespace in {"sh", "sz"}:
|
||||
exchange = _EXCHANGE_PREFIX_TO_CODE.get(namespace, "UNKNOWN")
|
||||
bare_code = canonical[2:]
|
||||
else:
|
||||
return None
|
||||
aliases = [str(a) for a in (row[5] if isinstance(row[5], list) else []) if str(a).strip()]
|
||||
# The display code (e.g. ``930955.CSI``) is a resolver key even when it is
|
||||
# not an explicit alias; add it so ``find_by_explicit_key`` and the
|
||||
# bare-conflict map can resolve it. SH/SZ display equals canonical so it is
|
||||
# already covered.
|
||||
display = str(row[1] or "").strip()
|
||||
if display and display != canonical and display not in aliases:
|
||||
aliases.append(display)
|
||||
return IndexEntry(
|
||||
bare_code=bare_code,
|
||||
exchange=exchange,
|
||||
canonical_id=canonical,
|
||||
display_name=name,
|
||||
aliases=tuple(aliases),
|
||||
)
|
||||
|
||||
|
||||
def default_index_registry() -> IndexRegistry:
|
||||
"""Return the built-in :class:`IndexRegistry` shipped with PR1.
|
||||
"""Return the :class:`IndexRegistry` built from the validated JSON index rows.
|
||||
|
||||
``IndexRegistry`` is intentionally cheap to construct (5 small dataclass
|
||||
entries); callers may rebuild it on every call rather than caching it
|
||||
globally. Once the JSON registry carries ``asset_type=index`` rows this
|
||||
factory should load from disk and stay backward-compatible.
|
||||
Loads the active ``assetType=index`` rows via the stock-index loader's
|
||||
cached active-index-row loader, converts each to a lightweight
|
||||
:class:`IndexEntry`, and constructs a fresh :class:`IndexRegistry`. The
|
||||
registry is intentionally cheap to construct; callers may rebuild it on
|
||||
every call rather than caching it globally. When the loader returns no
|
||||
rows (all candidates failed), the registry is empty and every prefixed
|
||||
input degrades to the stock path.
|
||||
"""
|
||||
return IndexRegistry(_DEFAULT_INDEX_ENTRIES)
|
||||
from src.data.stock_index_loader import _load_active_index_rows
|
||||
|
||||
rows = _load_active_index_rows()
|
||||
entries = []
|
||||
for row in rows:
|
||||
entry = _index_entry_from_row(row)
|
||||
if entry is not None:
|
||||
entries.append(entry)
|
||||
return IndexRegistry(entries)
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
@@ -514,6 +621,56 @@ def parse_analysis_target(
|
||||
if registry is None:
|
||||
registry = default_index_registry()
|
||||
|
||||
# Resolve explicit index keys (canonical/display/alias) BEFORE generic
|
||||
# stock normalization, so ``csi930955`` / ``930955.CSI`` and other explicit
|
||||
# index forms resolve to ``index`` without touching
|
||||
# ``normalize_stock_code()``. An unregistered ``.CSI`` input is surfaced as
|
||||
# ``unsupported`` (never a US ticker or a guessed SH/SZ index).
|
||||
explicit_entry = registry.find_by_explicit_key(raw)
|
||||
if explicit_entry is not None:
|
||||
canonical, display, exchange = _canonicalize_for_index(explicit_entry, raw)
|
||||
# Preserve the prefix/bare split for explicit sh/sz prefixed forms so
|
||||
# the existing ``normalized_prefix``/``normalized_code`` contract stays
|
||||
# intact (review blocker OR-COR-d24a4e9a family). CSI forms carry no
|
||||
# sh/sz prefix.
|
||||
norm_prefix = None
|
||||
norm_code = raw
|
||||
lower = raw.lower()
|
||||
for p in ("sh", "sz"):
|
||||
if lower.startswith(p) and len(raw) > len(p):
|
||||
norm_prefix = p
|
||||
norm_code = raw[len(p):]
|
||||
break
|
||||
return AnalysisTarget(
|
||||
raw_input=raw_input,
|
||||
asset_type=ParseStatus.INDEX,
|
||||
canonical_id=canonical,
|
||||
display_code=display,
|
||||
exchange=exchange,
|
||||
normalized_prefix=norm_prefix,
|
||||
normalized_code=norm_code,
|
||||
matched_index=explicit_entry,
|
||||
)
|
||||
|
||||
# PR #2267 review fix: an explicit CSI form not owned by the registry must
|
||||
# surface as ``unsupported`` — never a US ticker or a guessed SH/SZ index. This
|
||||
# covers both the ``.CSI`` suffix (already handled by the prior check) and
|
||||
# the ``csi`` prefix form (``csi930956`` / ``CSI930956``), which the generic
|
||||
# normalizer would otherwise mis-route into the US-ticker branch.
|
||||
if _EXPLICIT_CSI_FORM_RE.match(unicodedata.normalize("NFKC", raw).strip().casefold()):
|
||||
return AnalysisTarget(
|
||||
raw_input=raw_input,
|
||||
asset_type=ParseStatus.UNSUPPORTED,
|
||||
canonical_id=raw,
|
||||
display_code=raw,
|
||||
exchange="UNKNOWN",
|
||||
unsupported_reason=(
|
||||
f"unregistered CSI index: {raw!r} is not in the index registry"
|
||||
),
|
||||
normalized_prefix=None,
|
||||
normalized_code=raw,
|
||||
)
|
||||
|
||||
# Reuse the repository's existing normalization contract
|
||||
# (``src.services.stock_code_utils._normalize_code_and_exchange``) so
|
||||
# lowercase tickers (``shop``/``hkd``/``aapl``) and suffix-form inputs
|
||||
@@ -905,6 +1062,11 @@ def parse_analysis_target(
|
||||
matched_index = None
|
||||
if prefix is None and bare:
|
||||
candidate = registry.find_by_bare_code(bare)
|
||||
if candidate is None:
|
||||
# Also surface conflicts via the bare-conflict map so a bare code
|
||||
# that is the numeric base of an explicit index alias (e.g.
|
||||
# ``930955`` for ``csi930955``) is advertised as ambiguous.
|
||||
candidate = registry.find_by_bare_conflict(bare)
|
||||
if candidate is not None:
|
||||
matched_index = candidate
|
||||
|
||||
|
||||
+168
-25
@@ -1623,10 +1623,14 @@ class DatabaseManager(metaclass=_DatabaseManagerMeta):
|
||||
SQLite can't add NOT NULL/UNIQUE columns via ALTER).
|
||||
5. Backfill existing rows with ``_derive_canonical_id(code)`` in
|
||||
id-batched chunks (5000/batch), using ``WHERE canonical_id IS NULL``
|
||||
so re-runs are safe. ``_derive_canonical_id`` is index-aware: a
|
||||
bare registered index code (e.g. ``000300``) resolves to the index
|
||||
canonical_id (``sh000300``) rather than the stock-path key
|
||||
(``sz000300``), preventing same-index split across buckets.
|
||||
so re-runs are safe. ``_derive_canonical_id`` mirrors the parser
|
||||
contract: a bare code always derives the stock-path
|
||||
canonical_id (e.g. ``000300`` -> ``sz000300``), while only explicit
|
||||
index forms (``sh000300`` / ``930955.CSI``) derive an index
|
||||
canonical_id. Historical bare rows whose current canonical_id was
|
||||
mis-written as an index identity (e.g. ``sh000300``) are not fixed
|
||||
here — the idempotent ``_backfill_canonical_ids()`` repair corrects
|
||||
that bucket in a separate pass.
|
||||
Function-level lazy import keeps the storage layer free of a
|
||||
circular import on ``src.services.stock_list_parser``.
|
||||
6. Create a plain (non-unique) index ``ix_stock_daily_canonical_id`` so
|
||||
@@ -1667,6 +1671,7 @@ class DatabaseManager(metaclass=_DatabaseManagerMeta):
|
||||
raise
|
||||
|
||||
self._backfill_stock_daily_canonical_id()
|
||||
self._backfill_canonical_ids()
|
||||
self._ensure_stock_daily_canonical_id_index()
|
||||
|
||||
def _ensure_stock_daily_canonical_id_index(self) -> None:
|
||||
@@ -1701,17 +1706,15 @@ class DatabaseManager(metaclass=_DatabaseManagerMeta):
|
||||
)
|
||||
|
||||
def _derive_canonical_id(self, code: str) -> Optional[str]:
|
||||
"""Derive a Phase 1 ``canonical_id`` for ``code``, index-aware.
|
||||
"""Derive a Phase 1 ``canonical_id`` for ``code``.
|
||||
|
||||
Closes review blocker ``OR-COR-4f9ffc38``: when a bare code hits the
|
||||
index registry (``parse_analysis_target(code).matched_index is not
|
||||
None``) we unify to the index's ``canonical_id`` so the same
|
||||
underlying index doesn't split across buckets — e.g. bare ``000300``
|
||||
resolves to ``sh000300`` (the CSI-300 index canonical_id) instead of
|
||||
``sz000300`` (the stock-path canonical_id the classifier would
|
||||
synthesise for a 6-digit ``0``-prefixed code). When no index is
|
||||
matched, the parser's stock/explicit-index ``canonical_id`` is
|
||||
returned unchanged.
|
||||
This method returns ``parse_analysis_target(code).canonical_id`` (or
|
||||
``None``) and never reads ``matched_index.canonical_id`` to override
|
||||
the result. Bare codes always resolve to the stock-path canonical_id
|
||||
(e.g. bare ``000300`` -> ``sz000300``), while explicit index forms
|
||||
(``sh000300`` / ``930955.CSI``) resolve to their index canonical_id.
|
||||
This keeps the storage derivation consistent with the parser contract
|
||||
(bare = stock, explicit index = index).
|
||||
|
||||
Lazy-imports ``parse_analysis_target`` inside the method (the parser
|
||||
module transitively touches the storage layer) and degrades to
|
||||
@@ -1719,7 +1722,10 @@ class DatabaseManager(metaclass=_DatabaseManagerMeta):
|
||||
persist NULL (D1) without raising.
|
||||
"""
|
||||
try:
|
||||
from src.services.stock_list_parser import parse_analysis_target
|
||||
from src.services.stock_list_parser import (
|
||||
ParseStatus,
|
||||
parse_analysis_target,
|
||||
)
|
||||
except Exception as exc:
|
||||
logger.warning(
|
||||
"_derive_canonical_id: cannot import parse_analysis_target "
|
||||
@@ -1728,8 +1734,11 @@ class DatabaseManager(metaclass=_DatabaseManagerMeta):
|
||||
)
|
||||
return None
|
||||
target = parse_analysis_target(code)
|
||||
if target.matched_index is not None:
|
||||
return target.matched_index.canonical_id or None
|
||||
if target.asset_type == ParseStatus.UNSUPPORTED:
|
||||
# An unsupported identity (e.g. an unregistered ``csi930956`` /
|
||||
# ``930956.CSI``) must not enter a persistent canonical bucket.
|
||||
# Return None so the caller persists NULL instead.
|
||||
return None
|
||||
return target.canonical_id or None
|
||||
|
||||
def _backfill_stock_daily_canonical_id(self) -> None:
|
||||
@@ -1745,9 +1754,10 @@ class DatabaseManager(metaclass=_DatabaseManagerMeta):
|
||||
skipped past in this run and naturally retried on next startup while
|
||||
its ``canonical_id`` is still NULL.
|
||||
|
||||
Derivation is index-aware via :meth:`_derive_canonical_id` so a bare
|
||||
index code (e.g. ``000300``) backfills to the index canonical_id
|
||||
(``sh000300``) rather than the stock-path canonical_id (``sz000300``).
|
||||
Derivation follows the parser contract: a bare code that collides with
|
||||
an index identity (e.g. ``000300``) backfills to the stock-path
|
||||
canonical_id (``sz000300``); only explicit index forms
|
||||
(``sh000300`` / ``930955.CSI``) derive to an index canonical_id.
|
||||
"""
|
||||
|
||||
_BATCH_SIZE = 5000
|
||||
@@ -1825,6 +1835,135 @@ class DatabaseManager(metaclass=_DatabaseManagerMeta):
|
||||
total_skipped,
|
||||
)
|
||||
|
||||
def _backfill_canonical_ids(self) -> None:
|
||||
"""Idempotent, batched repair of historical bare-code canonical_id buckets.
|
||||
|
||||
The previous ``_derive_canonical_id`` behavior preferred
|
||||
``matched_index.canonical_id``, so a bare six-digit code that collided
|
||||
with an index identity (e.g. ``000001`` / ``000016`` / ``000688`` /
|
||||
``930955``) was written with the *index* canonical_id (``sh000001`` /
|
||||
``csi930955``) instead of the stock-path canonical_id (``sz000001`` /
|
||||
``bj930955``). This method repairs those rows.
|
||||
|
||||
Only rows matching ALL of the following are touched:
|
||||
* ``code`` is a bare 6-digit numeric string (no explicit prefix/suffix)
|
||||
* current ``canonical_id`` is non-NULL and hits an active index identity
|
||||
* the parser's stock canonical_id differs from the current value
|
||||
|
||||
Explicit ``sh``/``sz``/``csi`` prefix and ``.SH``/``.SZ``/``.CSI``
|
||||
suffix rows are never modified. The repair uses a monotonic ``id``
|
||||
cursor and a conditional UPDATE (guarded by the current value) so it is
|
||||
idempotent and safe under concurrent writers. When the registry is
|
||||
empty the repair is a no-op and logs a WARNING.
|
||||
"""
|
||||
if not self._is_sqlite_engine:
|
||||
return
|
||||
if not inspect(self._engine).has_table(StockDaily.__tablename__):
|
||||
return
|
||||
|
||||
try:
|
||||
from src.services.stock_list_parser import parse_analysis_target
|
||||
from src.data.stock_index_loader import _load_active_index_rows
|
||||
except Exception as exc:
|
||||
logger.warning(
|
||||
"[StockDaily] canonical_id repair cannot import parser/loader: %s — skipping",
|
||||
exc,
|
||||
)
|
||||
return
|
||||
|
||||
active_rows = _load_active_index_rows()
|
||||
if not active_rows:
|
||||
logger.warning("[StockDaily] canonical_id repair skipped: index registry is empty")
|
||||
return
|
||||
|
||||
# Build the set of active index canonical_ids to detect mis-bucketed rows.
|
||||
active_index_canonicals = {
|
||||
str(row[0]).strip() for row in active_rows if row and str(row[0]).strip()
|
||||
}
|
||||
if not active_index_canonicals:
|
||||
logger.warning("[StockDaily] canonical_id repair skipped: no active index canonicals")
|
||||
return
|
||||
|
||||
_BATCH_SIZE = 5000
|
||||
total_repaired = 0
|
||||
total_skipped = 0
|
||||
last_id = 0
|
||||
|
||||
# Only rows whose current canonical_id hits an active index identity are
|
||||
# candidates for repair. Push this filter into the SQL so repeated
|
||||
# startup repairs do not scan the full table.
|
||||
canonical_placeholders = ", ".join(f":c{i}" for i in range(len(active_index_canonicals)))
|
||||
canonical_params = {f"c{i}": c for i, c in enumerate(sorted(active_index_canonicals))}
|
||||
|
||||
while True:
|
||||
with self._engine.begin() as connection:
|
||||
rows = connection.execute(
|
||||
text(
|
||||
f"SELECT id, code, canonical_id FROM {StockDaily.__tablename__} "
|
||||
"WHERE id > :last_id "
|
||||
f"AND canonical_id IN ({canonical_placeholders}) "
|
||||
"AND length(code) = 6 "
|
||||
"AND code NOT GLOB '*[^0-9]*' "
|
||||
f"ORDER BY id LIMIT {_BATCH_SIZE}"
|
||||
),
|
||||
{"last_id": last_id, **canonical_params},
|
||||
).fetchall()
|
||||
|
||||
if not rows:
|
||||
break
|
||||
|
||||
with self._engine.begin() as connection:
|
||||
for row_id, code, current_canonical in rows:
|
||||
last_id = row_id
|
||||
code_str = str(code or "").strip()
|
||||
current = str(current_canonical or "").strip()
|
||||
|
||||
# Only bare 6-digit numeric codes are candidates for repair.
|
||||
if not (code_str.isdigit() and len(code_str) == 6):
|
||||
total_skipped += 1
|
||||
continue
|
||||
|
||||
try:
|
||||
target = parse_analysis_target(code_str)
|
||||
except Exception as exc:
|
||||
logger.warning(
|
||||
"[StockDaily] canonical_id repair derivation failed for "
|
||||
"code=%r (id=%s): %s — skipping",
|
||||
code_str, row_id, exc,
|
||||
)
|
||||
total_skipped += 1
|
||||
continue
|
||||
|
||||
derived = target.canonical_id or None
|
||||
if not derived or derived == current:
|
||||
total_skipped += 1
|
||||
continue
|
||||
|
||||
result = connection.execute(
|
||||
text(
|
||||
f"UPDATE {StockDaily.__tablename__} "
|
||||
"SET canonical_id = :derived "
|
||||
"WHERE id = :row_id AND canonical_id = :current"
|
||||
),
|
||||
{"derived": derived, "row_id": row_id, "current": current},
|
||||
)
|
||||
if result.rowcount == 1:
|
||||
total_repaired += 1
|
||||
elif result.rowcount == 0:
|
||||
# Lost the race with a concurrent writer — safe to skip.
|
||||
total_skipped += 1
|
||||
else:
|
||||
raise RuntimeError(
|
||||
"canonical_id repair updated an unexpected number "
|
||||
f"of rows for id={row_id}: {result.rowcount}"
|
||||
)
|
||||
|
||||
logger.info(
|
||||
"[StockDaily] canonical_id repair stats: repaired_count=%s skipped_count=%s",
|
||||
total_repaired,
|
||||
total_skipped,
|
||||
)
|
||||
|
||||
def _ensure_intelligence_items_unique_index(self) -> None:
|
||||
if not self._is_sqlite_engine:
|
||||
return
|
||||
@@ -3168,8 +3307,10 @@ class DatabaseManager(metaclass=_DatabaseManagerMeta):
|
||||
- SQLite 分支按 chunk 写入以避免绑定参数上限
|
||||
- ``canonical_id``（Expand-Contract PR2）：显式传入则双写；
|
||||
未传（``None``）时用 ``_derive_canonical_id(code)`` 延迟推导
|
||||
（index-aware：裸指数码命中注册表时统一到指数 canonical_id），
|
||||
推导失败写 NULL 降级（D1）。
|
||||
（裸码恒为 stock canonical，如 ``000300`` -> ``sz000300``；
|
||||
仅显式指数形式如 ``sh000300`` / ``930955.CSI`` 推导为指数
|
||||
canonical_id；历史被错误写成指数桶的裸码行由
|
||||
``_backfill_canonical_ids()`` 幂等修复），推导失败写 NULL 降级（D1）。
|
||||
|
||||
Args:
|
||||
df: 包含日线数据的 DataFrame
|
||||
@@ -3188,9 +3329,11 @@ class DatabaseManager(metaclass=_DatabaseManagerMeta):
|
||||
# D1: canonical_id=None → derive via the Phase 1 parser; on failure
|
||||
# degrade to NULL (do NOT raise — read path still uses ``code``).
|
||||
# Empty string is treated like None (never persisted as a key).
|
||||
# Derivation is index-aware (OR-COR-4f9ffc38): a bare index code
|
||||
# (e.g. ``000300``) resolves to the index canonical_id (``sh000300``)
|
||||
# rather than the stock-path canonical_id (``sz000300``).
|
||||
# Parser contract (OR-COR-4f9ffc38): a bare code always derives the
|
||||
# stock canonical_id (e.g. ``000300`` -> ``sz000300``); only explicit
|
||||
# index forms (``sh000300`` / ``930955.CSI``) derive an index canonical_id.
|
||||
# Historical bare rows mis-written as an index identity are repaired
|
||||
# separately by the idempotent ``_backfill_canonical_ids()``.
|
||||
if not canonical_id:
|
||||
try:
|
||||
canonical_id = self._derive_canonical_id(code)
|
||||
|
||||
@@ -5,7 +5,12 @@ from unittest.mock import patch
|
||||
import pandas as pd
|
||||
import pytest
|
||||
|
||||
from data_provider.base import BaseFetcher, DataFetcherManager, STANDARD_COLUMNS
|
||||
from data_provider.base import (
|
||||
BaseFetcher,
|
||||
DataFetchError,
|
||||
DataFetcherManager,
|
||||
STANDARD_COLUMNS,
|
||||
)
|
||||
from src.services.stock_list_parser import (
|
||||
AnalysisTarget,
|
||||
IndexEntry,
|
||||
@@ -78,6 +83,41 @@ def _manager_without_fetchers() -> DataFetcherManager:
|
||||
return manager
|
||||
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
"stock_code",
|
||||
["csi930956", "CSI930956", "930956.CSI", "csi93095", "93095.CSI"],
|
||||
)
|
||||
def test_unregistered_csi_daily_is_rejected_before_provider_calls(stock_code) -> None:
|
||||
fetcher = _FakeFetcher("YfinanceFetcher", priority=1, daily_result=_daily_frame())
|
||||
manager = DataFetcherManager(fetchers=[fetcher])
|
||||
|
||||
with pytest.raises(DataFetchError, match="unregistered CSI index"):
|
||||
manager.get_daily_data(stock_code)
|
||||
|
||||
assert fetcher.daily_calls == []
|
||||
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
"stock_code",
|
||||
["csi930956", "CSI930956", "930956.CSI", "csi93095", "93095.CSI"],
|
||||
)
|
||||
def test_unregistered_csi_name_is_rejected_before_provider_calls(stock_code) -> None:
|
||||
fetcher = _FakeFetcher("YfinanceFetcher", priority=1, name_result="wrong")
|
||||
manager = DataFetcherManager(fetchers=[fetcher])
|
||||
|
||||
assert manager.get_stock_name(stock_code) == ""
|
||||
assert fetcher.name_calls == []
|
||||
|
||||
|
||||
def test_prefetch_stock_names_skips_unregistered_csi() -> None:
|
||||
fetcher = _FakeFetcher("YfinanceFetcher", priority=1, name_result="wrong")
|
||||
manager = DataFetcherManager(fetchers=[fetcher])
|
||||
|
||||
manager.prefetch_stock_names(["csi930956", "930956.CSI"])
|
||||
|
||||
assert fetcher.name_calls == []
|
||||
|
||||
|
||||
@pytest.fixture(autouse=True)
|
||||
def _reset_daily_source_health():
|
||||
DataFetcherManager.reset_daily_source_health()
|
||||
@@ -470,7 +510,6 @@ def test_index_name_static_fallback_uses_only_canonical_key() -> None:
|
||||
"000016.SH",
|
||||
"000016.SS",
|
||||
"000016.SZ",
|
||||
"SSE50",
|
||||
],
|
||||
),
|
||||
("sz399001", ["sz399001", "399001", "399001.SZ"]),
|
||||
@@ -563,3 +602,83 @@ def test_index_and_same_bare_stock_names_use_isolated_cache_keys() -> None:
|
||||
"000016": "深康佳A",
|
||||
"sh000016": "上证50",
|
||||
}
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# CSI provider symbol routing.
|
||||
# ---------------------------------------------------------------------------
|
||||
def test_csi_provider_symbols_follow_manifest_matrix() -> None:
|
||||
csi_target = parse_analysis_target("csi930955")
|
||||
assert csi_target.asset_type == ParseStatus.INDEX
|
||||
assert csi_target.exchange == "CSI"
|
||||
|
||||
# AkShare is the only supported CSI daily provider.
|
||||
assert DataFetcherManager._cn_index_provider_symbol(
|
||||
csi_target, "AkshareFetcher"
|
||||
) == "csi930955"
|
||||
# Tencent / TickFlow / Yahoo are unsupported for CSI.
|
||||
assert DataFetcherManager._cn_index_provider_symbol(
|
||||
csi_target, "TencentFetcher"
|
||||
) == ""
|
||||
assert DataFetcherManager._cn_index_provider_symbol(
|
||||
csi_target, "TickFlowFetcher"
|
||||
) == ""
|
||||
assert DataFetcherManager._cn_index_provider_symbol(
|
||||
csi_target, "YfinanceFetcher"
|
||||
) == ""
|
||||
|
||||
|
||||
def test_csi_daily_route_skips_unsupported_providers_without_health_failure() -> None:
|
||||
tencent = _FakeFetcher("TencentFetcher", priority=0, daily_result=_daily_frame())
|
||||
akshare = _FakeFetcher("AkshareFetcher", priority=1, daily_result=_daily_frame())
|
||||
tickflow = _FakeFetcher("TickFlowFetcher", priority=2, daily_result=_daily_frame())
|
||||
yfinance = _FakeFetcher("YfinanceFetcher", priority=3, daily_result=_daily_frame())
|
||||
manager = DataFetcherManager(fetchers=[tencent, akshare, tickflow, yfinance])
|
||||
|
||||
with patch("data_provider.base.record_provider_run") as record_run, patch.object(
|
||||
DataFetcherManager, "_record_daily_source_failure"
|
||||
) as record_health_failure:
|
||||
df, source = manager.get_daily_data("csi930955")
|
||||
|
||||
# AkShare succeeds with the csi symbol.
|
||||
assert not df.empty
|
||||
assert source == "AkshareFetcher"
|
||||
assert tencent.daily_calls == []
|
||||
assert akshare.daily_calls == ["csi930955"]
|
||||
assert tickflow.daily_calls == []
|
||||
assert yfinance.daily_calls == []
|
||||
# Unsupported providers recorded as unsupported, no health failure.
|
||||
# AkShare's successful call short-circuits the loop, so only Tencent
|
||||
# (unsupported) and AkShare (success) are recorded.
|
||||
assert [item.kwargs.get("error_type") for item in record_run.call_args_list] == [
|
||||
"unsupported",
|
||||
None,
|
||||
]
|
||||
record_health_failure.assert_not_called()
|
||||
|
||||
|
||||
def test_csi_daily_route_returns_empty_when_akshare_fails() -> None:
|
||||
tencent = _FakeFetcher("TencentFetcher", priority=0, daily_result=_daily_frame())
|
||||
akshare = _FakeFetcher(
|
||||
"AkshareFetcher", priority=1, daily_result=RuntimeError("akshare failed")
|
||||
)
|
||||
tickflow = _FakeFetcher("TickFlowFetcher", priority=2, daily_result=_daily_frame())
|
||||
yfinance = _FakeFetcher("YfinanceFetcher", priority=3, daily_result=_daily_frame())
|
||||
manager = DataFetcherManager(fetchers=[tencent, akshare, tickflow, yfinance])
|
||||
|
||||
with patch("data_provider.base.record_provider_run") as record_run:
|
||||
df, source = manager.get_daily_data("csi930955")
|
||||
|
||||
assert df.empty
|
||||
assert source == ""
|
||||
assert tencent.daily_calls == []
|
||||
assert akshare.daily_calls == ["csi930955"]
|
||||
assert tickflow.daily_calls == []
|
||||
assert yfinance.daily_calls == []
|
||||
# Tencent/TickFlow/Yahoo are unsupported (no network), AkShare failed.
|
||||
assert [item.kwargs["error_type"] for item in record_run.call_args_list] == [
|
||||
"unsupported",
|
||||
"RuntimeError",
|
||||
"unsupported",
|
||||
"unsupported",
|
||||
]
|
||||
|
||||
@@ -52,6 +52,35 @@ from src.services.history_service import HistoryService
|
||||
import src.auth as auth
|
||||
|
||||
|
||||
class TestHistoryCsiCandidateConvergence(unittest.TestCase):
|
||||
"""PR #2267 review remediation: registered CSI explicit identities must
|
||||
converge in history filter candidates so a record saved under any
|
||||
equivalent form is reachable from every equivalent query input."""
|
||||
|
||||
def test_registered_csi_forms_include_canonical_uppercase_and_aliases(self):
|
||||
"""A registered CSI identity is a *persisted-read* filter path: the
|
||||
candidate set must include the parser canonical (``csi930955``), the
|
||||
old resolver's uppercase canonical (``CSI930955`` — how pre-fix records
|
||||
were saved) and the IndexEntry's explicit aliases (``930955.CSI``) so a
|
||||
record stored under any of them is hit by any equivalent input."""
|
||||
for code in ("csi930955", "930955.CSI", "CSI930955", " csi930955 "):
|
||||
candidates = HistoryService._history_code_filter_candidates(code)
|
||||
self.assertEqual(
|
||||
set(candidates),
|
||||
{"csi930955", "CSI930955", "930955.CSI"},
|
||||
)
|
||||
self.assertEqual(len(candidates), len(set(candidates)))
|
||||
|
||||
def test_bare_csi_base_remains_stock_candidates(self):
|
||||
candidates = HistoryService._history_code_filter_candidates("930955")
|
||||
self.assertIn("930955", candidates)
|
||||
self.assertNotIn("csi930955", candidates)
|
||||
|
||||
def test_unregistered_csi_form_is_not_converged(self):
|
||||
candidates = HistoryService._history_code_filter_candidates("csi930956")
|
||||
self.assertNotIn("csi930956", candidates)
|
||||
|
||||
|
||||
def _analysis_context_pack_overview() -> dict:
|
||||
return {
|
||||
"pack_version": "1.0",
|
||||
|
||||
@@ -211,6 +211,22 @@ class TestTaskQueue:
|
||||
assert len(dups2) == 1
|
||||
assert isinstance(dups2[0], DuplicateTaskError)
|
||||
|
||||
def test_task_queue_csi_equivalent_forms_dedupe_to_one_task(self):
|
||||
"""Real task-layer regression: submitting both the parser canonical
|
||||
``csi930955`` and its explicit alias ``930955.CSI`` in one batch must
|
||||
yield exactly ONE accepted task (canonicalised to ``csi930955``) and
|
||||
ONE duplicate — not two distinct tasks. This guards the task dedupe key
|
||||
against splitting a registered CSI identity into distinct keys."""
|
||||
queue = self._build_queue()
|
||||
accepted, duplicates = queue.submit_tasks_batch(
|
||||
stock_codes=["csi930955", "930955.CSI"],
|
||||
)
|
||||
assert len(accepted) == 1
|
||||
assert accepted[0].stock_code == "csi930955"
|
||||
assert len(duplicates) == 1
|
||||
assert isinstance(duplicates[0], DuplicateTaskError)
|
||||
assert duplicates[0].stock_code == "csi930955"
|
||||
|
||||
|
||||
class TestIntegration:
|
||||
"""Integration Tests"""
|
||||
|
||||
@@ -29,6 +29,11 @@ from generate_index_from_csv import (
|
||||
build_stock_index,
|
||||
load_tushare_data,
|
||||
load_akshare_data,
|
||||
load_index_registry_seed,
|
||||
build_index_entries_from_seed,
|
||||
validate_index_registry,
|
||||
run_index_only,
|
||||
_normalize_index_key,
|
||||
)
|
||||
|
||||
|
||||
@@ -654,3 +659,346 @@ class TestPinyin:
|
||||
monkeypatch.setattr(sys, 'argv', ['generate_index_from_csv.py'])
|
||||
|
||||
assert main() == 1
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Index registry seed generation
|
||||
# ---------------------------------------------------------------------------
|
||||
class TestIndexRegistrySeed:
|
||||
"""Seed generates exactly 31 index rows; index-only merge is stable."""
|
||||
|
||||
def test_seed_loads_31_rows(self):
|
||||
rows = load_index_registry_seed()
|
||||
assert len(rows) == 31
|
||||
|
||||
def test_seed_entries_build_valid_index_tuples(self):
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
assert len(entries) == 31
|
||||
for entry in entries:
|
||||
assert entry["market"] == "CN"
|
||||
assert entry["assetType"] == "index"
|
||||
assert entry["active"] is True
|
||||
assert entry["popularity"] == 100
|
||||
assert entry["nameZh"]
|
||||
assert entry["pinyinFull"]
|
||||
assert entry["pinyinAbbr"]
|
||||
|
||||
def test_seed_canonical_set_matches_manifest(self):
|
||||
rows = load_index_registry_seed()
|
||||
canonicals = {row["canonical_code"] for row in rows}
|
||||
assert len(canonicals) == 31
|
||||
# Spot-check the 5 original + CSI entries.
|
||||
assert {"sh000300", "sh000016", "sh000688", "sz399001", "sz399006"} <= canonicals
|
||||
assert {"csi930955", "csi932365", "csi931052"} <= canonicals
|
||||
|
||||
def test_seed_csi_display_is_code_dot_csi(self):
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
csi = {e["canonicalCode"]: e["displayCode"] for e in entries if e["canonicalCode"].startswith("csi")}
|
||||
assert csi["csi930955"] == "930955.CSI"
|
||||
assert csi["csi932365"] == "932365.CSI"
|
||||
|
||||
def test_validate_index_registry_accepts_valid_seed(self):
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
validate_index_registry(entries) # should not raise
|
||||
|
||||
def test_validate_index_registry_rejects_duplicate_canonical(self):
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
entries.append(dict(entries[0]))
|
||||
with pytest.raises(ValueError, match="duplicate index canonical"):
|
||||
validate_index_registry(entries)
|
||||
|
||||
def test_validate_index_registry_rejects_bare_numeric_alias(self):
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
entries[0]["aliases"] = ["000300"]
|
||||
with pytest.raises(ValueError, match="bare numeric"):
|
||||
validate_index_registry(entries)
|
||||
|
||||
def test_validate_index_registry_rejects_text_alias(self):
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
entries[0]["aliases"] = ["CSI300"]
|
||||
with pytest.raises(ValueError, match="explicit code form"):
|
||||
validate_index_registry(entries)
|
||||
|
||||
def test_validate_index_registry_rejects_unknown_namespace(self):
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
entries[0]["canonicalCode"] = "xx000300"
|
||||
with pytest.raises(ValueError, match="canonical must match"):
|
||||
validate_index_registry(entries)
|
||||
|
||||
def test_index_only_preserves_non_index_rows_and_appends_31(self, tmp_path):
|
||||
output = tmp_path / "stocks.index.json"
|
||||
output.write_text(
|
||||
json.dumps(
|
||||
[
|
||||
["000001.SZ", "000001", "平安银行", "payh", "payh", [], "CN", "stock", True, 100],
|
||||
["600519.SH", "600519", "贵州茅台", "gzmt", "gzmt", [], "CN", "stock", True, 100],
|
||||
],
|
||||
ensure_ascii=False,
|
||||
),
|
||||
encoding="utf-8",
|
||||
)
|
||||
merged = run_index_only(output, test=True)
|
||||
non_index = [x for x in merged if not (len(x) > 7 and x[7] == "index")]
|
||||
index_rows = [x for x in merged if len(x) > 7 and x[7] == "index"]
|
||||
# Non-index rows preserved in order.
|
||||
assert [x[0] for x in non_index] == ["000001.SZ", "600519.SH"]
|
||||
# Exactly 31 index rows appended.
|
||||
assert len(index_rows) == 31
|
||||
# Index rows sorted by canonical.
|
||||
canonicals = [x[0] for x in index_rows]
|
||||
assert canonicals == sorted(canonicals)
|
||||
|
||||
def test_index_only_is_byte_stable(self, tmp_path):
|
||||
output = tmp_path / "stocks.index.json"
|
||||
output.write_text(
|
||||
json.dumps(
|
||||
[["000001.SZ", "000001", "平安银行", "payh", "payh", [], "CN", "stock", True, 100]],
|
||||
ensure_ascii=False,
|
||||
),
|
||||
encoding="utf-8",
|
||||
)
|
||||
run_index_only(output, test=False)
|
||||
first = output.read_bytes()
|
||||
run_index_only(output, test=False)
|
||||
second = output.read_bytes()
|
||||
assert first == second
|
||||
|
||||
def test_index_only_test_mode_does_not_write(self, tmp_path):
|
||||
output = tmp_path / "stocks.index.json"
|
||||
output.write_text(
|
||||
json.dumps(
|
||||
[["000001.SZ", "000001", "平安银行", "payh", "payh", [], "CN", "stock", True, 100]],
|
||||
ensure_ascii=False,
|
||||
),
|
||||
encoding="utf-8",
|
||||
)
|
||||
before = output.read_bytes()
|
||||
run_index_only(output, test=True)
|
||||
assert output.read_bytes() == before
|
||||
|
||||
def test_full_path_merge_includes_31_index_rows(self, tmp_path, monkeypatch):
|
||||
"""The full rebuild path (``main()``) merges the same 31 index
|
||||
rows before compression, so a full stock-index rebuild never erases the
|
||||
index registry entries."""
|
||||
import generate_index_from_csv as gen
|
||||
|
||||
# A tiny stock list so the full path runs without network.
|
||||
stocks = [
|
||||
{"ts_code": "000001.SZ", "symbol": "000001", "name": "平安银行"},
|
||||
{"ts_code": "600519.SH", "symbol": "600519", "name": "贵州茅台"},
|
||||
]
|
||||
monkeypatch.setattr(gen, "load_tushare_data", lambda data_dir: stocks)
|
||||
monkeypatch.setattr(gen, "require_pypinyin", lambda: True)
|
||||
monkeypatch.setattr(
|
||||
sys, "argv", ["generate_index_from_csv.py", "--source", "tushare", "--test"]
|
||||
)
|
||||
|
||||
assert main() == 0
|
||||
|
||||
# Rebuild the same pipeline in-process to inspect the merged index rows.
|
||||
index = build_stock_index(stocks)
|
||||
seed_rows = load_index_registry_seed()
|
||||
index_entries = build_index_entries_from_seed(seed_rows)
|
||||
validate_index_registry(index_entries)
|
||||
index.extend(index_entries)
|
||||
compressed = compress_index(index)
|
||||
index_rows = [item for item in compressed if len(item) > 7 and item[7] == "index"]
|
||||
assert len(index_rows) == 31
|
||||
# Stock rows are preserved alongside the index rows.
|
||||
stock_rows = [item for item in compressed if len(item) > 7 and item[7] == "stock"]
|
||||
assert len(stock_rows) == 2
|
||||
|
||||
def test_validate_index_registry_rejects_stock_key_collision(self):
|
||||
"""Gap 3: an index canonical/display/alias that collides with an active
|
||||
stock/ETF key after normalization is rejected."""
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
# A stock row whose canonical ``sh000300`` collides with the index
|
||||
# canonical ``sh000300``.
|
||||
non_index = [["sh000300", "sh000300", "沪深300", "hushen300", "hs300", [], "CN", "stock", True, 100]]
|
||||
with pytest.raises(ValueError, match="collides with active stock/ETF"):
|
||||
validate_index_registry(entries, non_index_rows=non_index)
|
||||
|
||||
def test_validate_index_registry_rejects_stock_alias_collision(self):
|
||||
"""Gap 3: an index alias that collides with an active stock alias is
|
||||
rejected."""
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
# A stock row whose alias ``000300.SH`` collides with the sh000300 index
|
||||
# alias ``000300.SH``.
|
||||
non_index = [["600519.SH", "600519", "贵州茅台", "gzmt", "gzmt", ["000300.SH"], "CN", "stock", True, 100]]
|
||||
with pytest.raises(ValueError, match="collides with active stock/ETF"):
|
||||
validate_index_registry(entries, non_index_rows=non_index)
|
||||
|
||||
def test_validate_index_registry_rejects_csi_canonical_stock_collision(self):
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
non_index = [["csi930955", "930955", "冲突股票", "ctgp", "ctgp", [], "CN", "stock", True, 100]]
|
||||
with pytest.raises(ValueError, match="collides with active stock/ETF"):
|
||||
validate_index_registry(entries, non_index_rows=non_index)
|
||||
|
||||
def test_validate_index_registry_rejects_equivalent_suffix_stock_collision(self):
|
||||
rows = load_index_registry_seed()
|
||||
entry = build_index_entries_from_seed(rows)[0]
|
||||
entry.update({"canonicalCode": "sh600519", "displayCode": "sh600519", "aliases": []})
|
||||
non_index = [["600519.SH", "600519", "贵州茅台", "gzmt", "gzmt", [], "CN", "stock", True, 100]]
|
||||
with pytest.raises(ValueError, match="collides with active stock/ETF"):
|
||||
validate_index_registry([entry], non_index_rows=non_index)
|
||||
|
||||
def test_validate_index_registry_rejects_non_string_pinyin(self):
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
entries[0]["pinyinFull"] = ["hushen300"]
|
||||
with pytest.raises(ValueError, match="pinyin fields"):
|
||||
validate_index_registry(entries)
|
||||
|
||||
def test_index_only_rejects_stock_key_collision(self, tmp_path):
|
||||
"""Gap 3: ``run_index_only`` validates the seed index rows against the
|
||||
existing active stock/ETF rows and rejects a collision."""
|
||||
output = tmp_path / "stocks.index.json"
|
||||
output.write_text(
|
||||
json.dumps(
|
||||
[["sh000300", "sh000300", "沪深300", "hushen300", "hs300", [], "CN", "stock", True, 100]],
|
||||
ensure_ascii=False,
|
||||
),
|
||||
encoding="utf-8",
|
||||
)
|
||||
with pytest.raises(ValueError, match="collides with active stock/ETF"):
|
||||
run_index_only(output, test=True)
|
||||
|
||||
def test_index_only_rejects_malformed_existing_tuple(self, tmp_path):
|
||||
output = tmp_path / "stocks.index.json"
|
||||
output.write_text(json.dumps([["too-short"]]), encoding="utf-8")
|
||||
with pytest.raises(ValueError, match="compressed tuple"):
|
||||
run_index_only(output, test=True)
|
||||
|
||||
def test_build_index_entries_honors_seed_display_code(self):
|
||||
"""Gap 5: ``build_index_entries_from_seed`` honors the seed
|
||||
``display_code`` column rather than recomputing it."""
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
by_canonical = {e["canonicalCode"]: e["displayCode"] for e in entries}
|
||||
# CSI display comes from the seed ``display_code`` column.
|
||||
assert by_canonical["csi930955"] == "930955.CSI"
|
||||
assert by_canonical["csi932365"] == "932365.CSI"
|
||||
# SH/SZ display equals canonical.
|
||||
assert by_canonical["sh000300"] == "sh000300"
|
||||
assert by_canonical["sz399001"] == "sz399001"
|
||||
|
||||
def test_validate_index_registry_rejects_non_finite_popularity(self):
|
||||
"""Gap 5: a non-finite popularity (e.g. NaN) is rejected — it is not a
|
||||
plain integer (NaN is a float), so it fails the integer check."""
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
entries[0]["popularity"] = float("nan")
|
||||
with pytest.raises(ValueError, match="non-negative integer"):
|
||||
validate_index_registry(entries)
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
"bad_popularity",
|
||||
[1.5, True, -1, -100, 1.0, "100"],
|
||||
)
|
||||
def test_validate_index_registry_rejects_non_integer_popularity(
|
||||
self, bad_popularity
|
||||
):
|
||||
"""PR #2267 review fix: only a plain non-negative integer popularity is
|
||||
valid. Fractional (``1.5``), boolean (``True``), negative and
|
||||
string-valued popularities are rejected without truncation."""
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
entries[0]["popularity"] = bad_popularity
|
||||
with pytest.raises(ValueError, match="non-negative integer"):
|
||||
validate_index_registry(entries)
|
||||
|
||||
def test_validate_index_registry_accepts_integer_popularity(self):
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
entries[0]["popularity"] = 100
|
||||
validate_index_registry(entries) # should not raise
|
||||
|
||||
def test_seed_rejects_fractional_popularity(self, tmp_path):
|
||||
seed = tmp_path / "index_registry.csv"
|
||||
seed.write_text(
|
||||
"canonical_code,display_code,name_zh,aliases,name_source,popularity\n"
|
||||
"sh000300,sh000300,沪深300,,腾讯,1.5\n",
|
||||
encoding="utf-8",
|
||||
)
|
||||
with pytest.raises(ValueError, match="plain integer"):
|
||||
load_index_registry_seed(seed)
|
||||
|
||||
def test_seed_rejects_duplicate_normalized_identity_key(self, tmp_path):
|
||||
seed = tmp_path / "index_registry.csv"
|
||||
seed.write_text(
|
||||
"canonical_code,display_code,name_zh,aliases,name_source,popularity\n"
|
||||
"csi930955,930955.CSI,红利低波100,,东财,100\n"
|
||||
"sh000300,sh000300,沪深300,csi930955,腾讯,100\n",
|
||||
encoding="utf-8",
|
||||
)
|
||||
with pytest.raises(ValueError, match="already owned by canonical"):
|
||||
load_index_registry_seed(seed)
|
||||
|
||||
def test_csi_prefix_and_suffix_keep_distinct_resolver_keys(self):
|
||||
assert _normalize_index_key("csi000300") == "csi000300"
|
||||
assert _normalize_index_key("000300.CSI") == "000300.csi"
|
||||
|
||||
def test_seed_rejects_duplicate_aliases_within_one_row(self, tmp_path):
|
||||
seed = tmp_path / "index_registry.csv"
|
||||
seed.write_text(
|
||||
"canonical_code,display_code,name_zh,aliases,name_source,popularity\n"
|
||||
"sh000300,sh000300,沪深300,000300.CSI|０００３００．ＣＳＩ,腾讯,100\n",
|
||||
encoding="utf-8",
|
||||
)
|
||||
with pytest.raises(ValueError, match="duplicate index alias"):
|
||||
load_index_registry_seed(seed)
|
||||
|
||||
def test_build_rejects_duplicate_aliases_when_seed_loader_is_bypassed(self):
|
||||
rows = load_index_registry_seed()
|
||||
rows[0]["aliases"] = ["000001.SH", "０００００１．ＳＨ"]
|
||||
with pytest.raises(ValueError, match="duplicate index alias"):
|
||||
build_index_entries_from_seed(rows)
|
||||
|
||||
def test_build_rejects_fractional_popularity_when_seed_loader_is_bypassed(self):
|
||||
rows = load_index_registry_seed()
|
||||
rows[0]["popularity"] = 1.5
|
||||
with pytest.raises(ValueError, match="non-negative integer"):
|
||||
build_index_entries_from_seed(rows)
|
||||
|
||||
def test_validate_rejects_duplicate_aliases_within_one_entry(self):
|
||||
rows = load_index_registry_seed()
|
||||
entries = build_index_entries_from_seed(rows)
|
||||
entries[0]["aliases"] = ["000001.SH", "０００００１．ＳＨ"]
|
||||
with pytest.raises(ValueError, match="duplicate index alias"):
|
||||
validate_index_registry(entries)
|
||||
|
||||
def test_full_path_merge_canonical_sorts_index_rows(self, tmp_path, monkeypatch):
|
||||
"""Gap 5: the full rebuild path canonical-sorts the index rows so the
|
||||
output is byte-stable and matches ``--index-only`` ordering."""
|
||||
import generate_index_from_csv as gen
|
||||
|
||||
stocks = [
|
||||
{"ts_code": "000001.SZ", "symbol": "000001", "name": "平安银行"},
|
||||
{"ts_code": "600519.SH", "symbol": "600519", "name": "贵州茅台"},
|
||||
]
|
||||
monkeypatch.setattr(gen, "load_tushare_data", lambda data_dir: stocks)
|
||||
monkeypatch.setattr(gen, "require_pypinyin", lambda: True)
|
||||
monkeypatch.setattr(
|
||||
sys, "argv", ["generate_index_from_csv.py", "--source", "tushare", "--test"]
|
||||
)
|
||||
assert main() == 0
|
||||
|
||||
index = build_stock_index(stocks)
|
||||
seed_rows = load_index_registry_seed()
|
||||
index_entries = build_index_entries_from_seed(seed_rows)
|
||||
index_entries.sort(key=lambda entry: str(entry["canonicalCode"]))
|
||||
index.extend(index_entries)
|
||||
compressed = compress_index(index)
|
||||
index_rows = [item for item in compressed if len(item) > 7 and item[7] == "index"]
|
||||
canonicals = [x[0] for x in index_rows]
|
||||
assert canonicals == sorted(canonicals)
|
||||
|
||||
@@ -361,3 +361,36 @@ class TestResolveIndexStockCodeForAnalysis:
|
||||
with patch("src.data.stock_index_loader.resolve_index_stock_code", return_value=None):
|
||||
assert resolve_index_stock_code_for_analysis("005930") == "005930"
|
||||
assert resolve_index_stock_code_for_analysis("AAPL") == "AAPL"
|
||||
|
||||
# ------------------------------------------------------------------
|
||||
# PR #2267 review remediation — registered CSI explicit identity
|
||||
# convergence (resolver / task dedupe key / history candidates).
|
||||
# ------------------------------------------------------------------
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
"code",
|
||||
["csi930955", "930955.CSI", "CSI930955", " csi930955 ", "930955.csi"],
|
||||
)
|
||||
def test_registered_csi_forms_converge_to_parser_canonical(self, code):
|
||||
"""All registered CSI explicit forms resolve to the parser canonical
|
||||
``csi930955`` so the resolver and task dedupe key do not split the same
|
||||
index into distinct keys."""
|
||||
assert resolve_index_stock_code_for_analysis(code) == "csi930955"
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
"code,expected",
|
||||
[
|
||||
("csi930956", "CSI930956"),
|
||||
("930956.CSI", "930956.CSI"),
|
||||
("CSI930956", "CSI930956"),
|
||||
],
|
||||
)
|
||||
def test_unregistered_csi_forms_keep_existing_behavior(self, code, expected):
|
||||
"""An unregistered CSI form is NOT converged; it keeps its existing
|
||||
canonicalized (uppercased) degradation so it never becomes a guessed
|
||||
index identity."""
|
||||
assert resolve_index_stock_code_for_analysis(code) == expected
|
||||
|
||||
def test_bare_csi_base_remains_stock(self):
|
||||
"""A bare numeric base of a CSI index stays a stock (no convergence)."""
|
||||
assert resolve_index_stock_code_for_analysis("930955") == "930955"
|
||||
|
||||
@@ -6,6 +6,8 @@ import unittest
|
||||
from pathlib import Path
|
||||
from unittest.mock import patch
|
||||
|
||||
import pytest
|
||||
|
||||
from src.data import stock_index_loader
|
||||
|
||||
|
||||
@@ -336,6 +338,406 @@ class TestStockIndexLoader(unittest.TestCase):
|
||||
self.assertIsNone(stock_index_loader.resolve_index_stock_code("2330.TW"))
|
||||
self.assertIsNone(stock_index_loader.resolve_index_stock_code("6505.TWO"))
|
||||
|
||||
# ------------------------------------------------------------------
|
||||
# Active index row loader
|
||||
# ------------------------------------------------------------------
|
||||
|
||||
def _index_payload(self, canonicals):
|
||||
rows = []
|
||||
for c in canonicals:
|
||||
display = f"{c[3:]}.CSI" if c.startswith("csi") else c
|
||||
rows.append([c, display, f"指数{c}", "zhishu", "zs", [], "CN", "index", True, 100])
|
||||
return rows
|
||||
|
||||
def _pad_payload(self, rows, size=100):
|
||||
"""Pad a payload with stock rows so it passes the remote min_items check."""
|
||||
padded = list(rows)
|
||||
while len(padded) < size:
|
||||
i = len(padded)
|
||||
padded.append([f"{i:06d}.SZ", f"{i:06d}", f"股票{i}", "gupiao", "gp", [], "CN", "stock", True, 100])
|
||||
return padded
|
||||
|
||||
def test_load_active_index_rows_returns_index_rows(self):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
bundled_path.write_text(
|
||||
json.dumps(
|
||||
self._index_payload(["sh000300", "csi930955"])
|
||||
+ [["000001.SZ", "000001", "平安银行", "payh", "payh", [], "CN", "stock", True, 100]],
|
||||
ensure_ascii=False,
|
||||
),
|
||||
encoding="utf-8",
|
||||
)
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual({r[0] for r in rows}, {"sh000300", "csi930955"})
|
||||
|
||||
def test_load_active_index_rows_remote_superset_wins(self):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
remote_cache = Path(temp_dir) / "cache" / "stocks.index.json"
|
||||
bundled_path = Path(temp_dir) / "apps" / "stocks.index.json"
|
||||
remote_cache.parent.mkdir(parents=True, exist_ok=True)
|
||||
remote_cache.write_text(
|
||||
json.dumps(self._pad_payload(self._index_payload(["sh000300", "sh000016", "csi930955"])), ensure_ascii=False),
|
||||
encoding="utf-8",
|
||||
)
|
||||
bundled_path.parent.mkdir(parents=True, exist_ok=True)
|
||||
bundled_path.write_text(
|
||||
json.dumps(self._index_payload(["sh000300", "sh000016"]), ensure_ascii=False),
|
||||
encoding="utf-8",
|
||||
)
|
||||
os.utime(remote_cache, (2_000, 2_000))
|
||||
os.utime(bundled_path, (1_000, 1_000))
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=remote_cache), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(remote_cache, bundled_path)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual({r[0] for r in rows}, {"sh000300", "sh000016", "csi930955"})
|
||||
|
||||
def test_load_active_index_rows_remote_subset_falls_back_to_bundled(self):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
remote_cache = Path(temp_dir) / "cache" / "stocks.index.json"
|
||||
bundled_path = Path(temp_dir) / "apps" / "stocks.index.json"
|
||||
remote_cache.parent.mkdir(parents=True, exist_ok=True)
|
||||
# Remote drops sh000016 (a bundled baseline canonical).
|
||||
remote_cache.write_text(
|
||||
json.dumps(self._pad_payload(self._index_payload(["sh000300"])), ensure_ascii=False),
|
||||
encoding="utf-8",
|
||||
)
|
||||
bundled_path.parent.mkdir(parents=True, exist_ok=True)
|
||||
bundled_path.write_text(
|
||||
json.dumps(self._index_payload(["sh000300", "sh000016"]), ensure_ascii=False),
|
||||
encoding="utf-8",
|
||||
)
|
||||
os.utime(remote_cache, (2_000, 2_000))
|
||||
os.utime(bundled_path, (1_000, 1_000))
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=remote_cache), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(remote_cache, bundled_path)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
# Falls back to bundled (which has both).
|
||||
self.assertEqual({r[0] for r in rows}, {"sh000300", "sh000016"})
|
||||
|
||||
def test_load_active_index_rows_all_failed_returns_empty(self):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
missing_path = Path(temp_dir) / "missing.json"
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=missing_path), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(missing_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual(rows, [])
|
||||
|
||||
def test_load_active_index_rows_rejects_semantic_conflict(self):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
# Two index rows share the same canonical — semantic conflict.
|
||||
bundled_path.write_text(
|
||||
json.dumps(
|
||||
self._index_payload(["sh000300"]) + self._index_payload(["sh000300"]),
|
||||
ensure_ascii=False,
|
||||
),
|
||||
encoding="utf-8",
|
||||
)
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
# Semantic conflict rejects the candidate → empty registry.
|
||||
self.assertEqual(rows, [])
|
||||
|
||||
def test_clear_stock_index_cache_clears_active_index_rows(self):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
bundled_path.write_text(
|
||||
json.dumps(self._index_payload(["sh000300"]), ensure_ascii=False),
|
||||
encoding="utf-8",
|
||||
)
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
first = stock_index_loader._load_active_index_rows()
|
||||
stock_index_loader.clear_stock_index_cache()
|
||||
second = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual(first, second)
|
||||
|
||||
def test_validate_index_rows_semantics_rejects_malformed_csi_display(self):
|
||||
"""Gap 1: a CSI row whose display is not ``{code}.CSI`` is rejected."""
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
# csi930955 with a wrong display (not ``930955.CSI``).
|
||||
bundled_path.write_text(
|
||||
json.dumps(
|
||||
[["csi930955", "930955", "红利低波100", "honglidibo100", "hldb100", [], "CN", "index", True, 100]],
|
||||
ensure_ascii=False,
|
||||
),
|
||||
encoding="utf-8",
|
||||
)
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
# Malformed CSI display rejects the candidate → empty registry.
|
||||
self.assertEqual(rows, [])
|
||||
|
||||
def test_validate_index_rows_semantics_rejects_stock_key_collision(self):
|
||||
"""Gap 3: an index canonical/display/alias that collides with an active
|
||||
stock/ETF key after normalization is rejected."""
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
# A stock row whose canonical ``sh000300`` collides with the index
|
||||
# canonical ``sh000300`` in the same candidate.
|
||||
bundled_path.write_text(
|
||||
json.dumps(
|
||||
self._index_payload(["sh000300"])
|
||||
+ [["sh000300", "sh000300", "冲突股票", "ctgp", "ctgp", [], "CN", "stock", True, 100]],
|
||||
ensure_ascii=False,
|
||||
),
|
||||
encoding="utf-8",
|
||||
)
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
# The candidate is rejected because both identities claim one key.
|
||||
self.assertEqual(rows, [])
|
||||
|
||||
def test_validate_index_rows_semantics_rejects_csi_canonical_stock_collision(self):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
bundled_path.write_text(
|
||||
json.dumps(
|
||||
self._index_payload(["csi930955"])
|
||||
+ [["csi930955", "930955", "冲突股票", "ctgp", "ctgp", [], "CN", "stock", True, 100]],
|
||||
ensure_ascii=False,
|
||||
),
|
||||
encoding="utf-8",
|
||||
)
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual(rows, [])
|
||||
|
||||
def test_validate_index_rows_semantics_rejects_text_alias(self):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
payload = self._index_payload(["sh000300"])
|
||||
payload[0][5] = ["CSI300"]
|
||||
bundled_path.write_text(json.dumps(payload, ensure_ascii=False), encoding="utf-8")
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual(rows, [])
|
||||
|
||||
def test_validate_index_rows_semantics_rejects_blank_pinyin(self):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
payload = self._index_payload(["sh000300"])
|
||||
payload[0][3] = ""
|
||||
bundled_path.write_text(json.dumps(payload, ensure_ascii=False), encoding="utf-8")
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual(rows, [])
|
||||
|
||||
def test_validate_index_rows_semantics_rejects_non_string_pinyin(self):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
payload = self._index_payload(["sh000300"])
|
||||
payload[0][3] = ["hushen300"]
|
||||
bundled_path.write_text(json.dumps(payload, ensure_ascii=False), encoding="utf-8")
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual(rows, [])
|
||||
|
||||
def test_validate_index_rows_semantics_rejects_equivalent_suffix_stock_collision(self):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
payload = self._index_payload(["sh600519"])
|
||||
payload += [["600519.SH", "600519", "贵州茅台", "gzmt", "gzmt", [], "CN", "stock", True, 100]]
|
||||
bundled_path.write_text(json.dumps(payload, ensure_ascii=False), encoding="utf-8")
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual(rows, [])
|
||||
|
||||
def test_load_active_index_rows_rejects_mixed_valid_and_short_local_rows(self):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
payload = self._index_payload(["sh000300"]) + [["too-short"]]
|
||||
bundled_path.write_text(json.dumps(payload, ensure_ascii=False), encoding="utf-8")
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual(rows, [])
|
||||
|
||||
def test_validate_index_rows_semantics_rejects_non_integer_popularity(self):
|
||||
"""PR #2267 review fix: only a plain non-negative integer popularity is
|
||||
valid at the runtime candidate boundary; fractional/boolean/negative/
|
||||
string popularities reject the candidate."""
|
||||
for bad_popularity in (1.5, True, -1, 1.0, "100"):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
payload = self._index_payload(["sh000300"])
|
||||
payload[0][9] = bad_popularity
|
||||
bundled_path.write_text(json.dumps(payload, ensure_ascii=False), encoding="utf-8")
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual(rows, [])
|
||||
|
||||
def test_validate_index_rows_semantics_accepts_integer_popularity(self):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
bundled_path.write_text(
|
||||
json.dumps(self._index_payload(["sh000300"]), ensure_ascii=False),
|
||||
encoding="utf-8",
|
||||
)
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual({r[0] for r in rows}, {"sh000300"})
|
||||
|
||||
def test_validate_index_rows_semantics_rejects_cross_entry_duplicate_alias(self):
|
||||
"""PR #2267 review fix: two index rows whose aliases normalize to the
|
||||
same identity key (e.g. ``csi930955`` and ``CSI930955`` as aliases of
|
||||
different canonicals) must reject the candidate — no silent overwrite."""
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
payload = self._index_payload(["sh000300", "sz399001"])
|
||||
payload[0][5] = ["csi930955"]
|
||||
payload[1][5] = ["CSI930955"]
|
||||
bundled_path.write_text(json.dumps(payload, ensure_ascii=False), encoding="utf-8")
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual(rows, [])
|
||||
|
||||
def test_validate_index_rows_semantics_rejects_duplicate_aliases_in_one_row(self):
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
bundled_path = Path(temp_dir) / "stocks.index.json"
|
||||
payload = self._index_payload(["sh000300"])
|
||||
payload[0][5] = ["000300.CSI", "０００３００．ＣＳＩ"]
|
||||
bundled_path.write_text(json.dumps(payload, ensure_ascii=False), encoding="utf-8")
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=Path(temp_dir) / "missing.json"), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(bundled_path,)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual(rows, [])
|
||||
|
||||
def test_load_active_index_rows_bundled_baseline_ignores_newer_legacy_static(self):
|
||||
"""Gap 4: the bundled baseline comes from the declared bundled candidate
|
||||
(``apps/dsa-web/public``), not the first non-remote candidate ordered by
|
||||
mtime. A newer legacy ``static`` candidate must not become the baseline."""
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
remote_cache = Path(temp_dir) / "cache" / "stocks.index.json"
|
||||
bundled_path = Path(temp_dir) / "apps" / "stocks.index.json"
|
||||
legacy_static = Path(temp_dir) / "static" / "stocks.index.json"
|
||||
for p in (remote_cache, bundled_path, legacy_static):
|
||||
p.parent.mkdir(parents=True, exist_ok=True)
|
||||
# Remote is a superset of the bundled baseline.
|
||||
remote_cache.write_text(
|
||||
json.dumps(self._pad_payload(self._index_payload(["sh000300", "sh000016", "csi930955"])), ensure_ascii=False),
|
||||
encoding="utf-8",
|
||||
)
|
||||
bundled_path.write_text(
|
||||
json.dumps(self._index_payload(["sh000300", "sh000016"]), ensure_ascii=False),
|
||||
encoding="utf-8",
|
||||
)
|
||||
# Legacy static is NEWER than bundled but only carries sh000300.
|
||||
legacy_static.write_text(
|
||||
json.dumps(self._index_payload(["sh000300"]), ensure_ascii=False),
|
||||
encoding="utf-8",
|
||||
)
|
||||
os.utime(remote_cache, (3_000, 3_000))
|
||||
os.utime(legacy_static, (2_000, 2_000))
|
||||
os.utime(bundled_path, (1_000, 1_000))
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=remote_cache), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(remote_cache, bundled_path, legacy_static)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
# Remote superset of the bundled baseline wins.
|
||||
self.assertEqual({r[0] for r in rows}, {"sh000300", "sh000016", "csi930955"})
|
||||
|
||||
def test_load_active_index_rows_malformed_bundled_falls_back_to_valid_remote(self):
|
||||
"""Gap 4: a malformed bundled candidate (fails semantic validation) must
|
||||
not become the baseline; a valid remote superset is then accepted."""
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
remote_cache = Path(temp_dir) / "cache" / "stocks.index.json"
|
||||
bundled_path = Path(temp_dir) / "apps" / "stocks.index.json"
|
||||
remote_cache.parent.mkdir(parents=True, exist_ok=True)
|
||||
bundled_path.parent.mkdir(parents=True, exist_ok=True)
|
||||
# Remote is a valid superset.
|
||||
remote_cache.write_text(
|
||||
json.dumps(self._pad_payload(self._index_payload(["sh000300", "sh000016", "csi930955"])), ensure_ascii=False),
|
||||
encoding="utf-8",
|
||||
)
|
||||
# Bundled has a malformed CSI display (not ``{code}.CSI``).
|
||||
bundled_path.write_text(
|
||||
json.dumps(
|
||||
[["csi930955", "930955", "红利低波100", "honglidibo100", "hldb100", [], "CN", "index", True, 100]],
|
||||
ensure_ascii=False,
|
||||
),
|
||||
encoding="utf-8",
|
||||
)
|
||||
os.utime(remote_cache, (2_000, 2_000))
|
||||
os.utime(bundled_path, (1_000, 1_000))
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=remote_cache), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(remote_cache, bundled_path)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
# Malformed bundled is skipped; valid remote superset is loaded.
|
||||
self.assertEqual({r[0] for r in rows}, {"sh000300", "sh000016", "csi930955"})
|
||||
|
||||
def test_load_active_index_rows_legacy_static_subset_cannot_bypass_bundled_baseline(self):
|
||||
"""Review remediation: when the remote cache is missing/invalid, a newer
|
||||
legacy ``static`` candidate that is a SUBSET of the bundled baseline must
|
||||
NOT be selected — the bundled baseline wins so no active index is lost."""
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
remote_cache = Path(temp_dir) / "cache" / "stocks.index.json"
|
||||
bundled_path = Path(temp_dir) / "apps" / "stocks.index.json"
|
||||
legacy_static = Path(temp_dir) / "static" / "stocks.index.json"
|
||||
for p in (remote_cache, bundled_path, legacy_static):
|
||||
p.parent.mkdir(parents=True, exist_ok=True)
|
||||
# Remote cache is invalid (not JSON).
|
||||
remote_cache.write_text("not-json", encoding="utf-8")
|
||||
# Bundled baseline carries sh000300 + sh000016.
|
||||
bundled_path.write_text(
|
||||
json.dumps(self._index_payload(["sh000300", "sh000016"]), ensure_ascii=False),
|
||||
encoding="utf-8",
|
||||
)
|
||||
# Legacy static is NEWER than bundled but only carries sh000300.
|
||||
legacy_static.write_text(
|
||||
json.dumps(self._index_payload(["sh000300"]), ensure_ascii=False),
|
||||
encoding="utf-8",
|
||||
)
|
||||
os.utime(remote_cache, (3_000, 3_000))
|
||||
os.utime(legacy_static, (2_000, 2_000))
|
||||
os.utime(bundled_path, (1_000, 1_000))
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=remote_cache), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(remote_cache, bundled_path, legacy_static)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
# Bundled baseline wins (both canonicals preserved).
|
||||
self.assertEqual({r[0] for r in rows}, {"sh000300", "sh000016"})
|
||||
|
||||
def test_load_active_index_rows_legacy_static_superset_still_wins(self):
|
||||
"""A legacy ``static`` candidate that is a legal SUPERSET of the bundled
|
||||
baseline is still accepted (future supersets are allowed)."""
|
||||
with tempfile.TemporaryDirectory() as temp_dir:
|
||||
remote_cache = Path(temp_dir) / "cache" / "stocks.index.json"
|
||||
bundled_path = Path(temp_dir) / "apps" / "stocks.index.json"
|
||||
legacy_static = Path(temp_dir) / "static" / "stocks.index.json"
|
||||
for p in (remote_cache, bundled_path, legacy_static):
|
||||
p.parent.mkdir(parents=True, exist_ok=True)
|
||||
remote_cache.write_text("not-json", encoding="utf-8")
|
||||
bundled_path.write_text(
|
||||
json.dumps(self._index_payload(["sh000300", "sh000016"]), ensure_ascii=False),
|
||||
encoding="utf-8",
|
||||
)
|
||||
# Legacy static is a superset (adds csi930955).
|
||||
legacy_static.write_text(
|
||||
json.dumps(self._index_payload(["sh000300", "sh000016", "csi930955"]), ensure_ascii=False),
|
||||
encoding="utf-8",
|
||||
)
|
||||
os.utime(remote_cache, (3_000, 3_000))
|
||||
os.utime(legacy_static, (2_000, 2_000))
|
||||
os.utime(bundled_path, (1_000, 1_000))
|
||||
with patch.object(stock_index_loader, "get_remote_stock_index_cache_path", return_value=remote_cache), \
|
||||
patch.object(stock_index_loader, "get_stock_index_candidate_paths", return_value=(remote_cache, bundled_path, legacy_static)):
|
||||
rows = stock_index_loader._load_active_index_rows()
|
||||
self.assertEqual({r[0] for r in rows}, {"sh000300", "sh000016", "csi930955"})
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
|
||||
+191
-14
@@ -138,16 +138,19 @@ class TestContract2BareCodeDefaultsToStock:
|
||||
assert target.matched_index.display_name == "沪深300"
|
||||
|
||||
def test_bare_000001_is_stock(self) -> None:
|
||||
"""Conflict code: ``000001`` is平安银行 (SZ stock) AND the深证成指
|
||||
isn't this — actually 000001.SZ is the stock, the index is sz399001.
|
||||
So bare 000001 → stock, no registry hit.
|
||||
"""Conflict code: ``000001`` is平安银行 (SZ stock) AND the上证指数
|
||||
(``sh000001``) is now in the registry. Per contract #2 bare codes
|
||||
always resolve to stock; the index conflict is surfaced via
|
||||
``matched_index`` without flipping asset_type.
|
||||
"""
|
||||
target = parse_analysis_target("000001")
|
||||
assert target.asset_type == ParseStatus.STOCK
|
||||
assert target.exchange == "SZ"
|
||||
assert target.matched_index is None
|
||||
# canonical_id is round-trippable: sh/sz prefix synthesised from 0/2/3.
|
||||
assert target.canonical_id == "sz000001"
|
||||
# The registry now carries sh000001 (上证指数) whose alias base is 000001.
|
||||
assert target.matched_index is not None
|
||||
assert target.matched_index.canonical_id == "sh000001"
|
||||
|
||||
def test_bare_600519_is_sh_stock(self) -> None:
|
||||
target = parse_analysis_target("600519")
|
||||
@@ -369,14 +372,18 @@ class TestEdgeCases:
|
||||
# Default registry — public API surface.
|
||||
# ---------------------------------------------------------------------------
|
||||
class TestDefaultIndexRegistry:
|
||||
def test_default_registry_has_five_entries(self) -> None:
|
||||
def test_default_registry_has_31_entries(self) -> None:
|
||||
registry = default_index_registry()
|
||||
assert len(registry) == 5
|
||||
assert len(registry) == 31
|
||||
|
||||
def test_default_registry_canonical_ids(self) -> None:
|
||||
registry = default_index_registry()
|
||||
ids = {entry.canonical_id for entry in registry}
|
||||
assert ids == {"sh000300", "sh000016", "sh000688", "sz399001", "sz399006"}
|
||||
assert len(ids) == 31
|
||||
# The 5 original hard-coded indices are preserved.
|
||||
assert {"sh000300", "sh000016", "sh000688", "sz399001", "sz399006"} <= ids
|
||||
# CSI entries are present.
|
||||
assert {"csi930955", "csi932365"} <= ids
|
||||
|
||||
def test_default_registry_find_by_prefixed_code(self) -> None:
|
||||
registry = default_index_registry()
|
||||
@@ -391,6 +398,49 @@ class TestDefaultIndexRegistry:
|
||||
assert registry.find_by_prefixed_code("hk", "000300") is None
|
||||
assert registry.find_by_prefixed_code("us", "000300") is None
|
||||
|
||||
def test_default_registry_find_by_explicit_key_csi(self) -> None:
|
||||
registry = default_index_registry()
|
||||
entry = registry.find_by_explicit_key("csi930955")
|
||||
assert entry is not None
|
||||
assert entry.canonical_id == "csi930955"
|
||||
assert entry.exchange == "CSI"
|
||||
# display form also resolves
|
||||
entry2 = registry.find_by_explicit_key("930955.CSI")
|
||||
assert entry2 is not None
|
||||
assert entry2.canonical_id == "csi930955"
|
||||
|
||||
def test_default_registry_find_by_bare_conflict(self) -> None:
|
||||
registry = default_index_registry()
|
||||
# 930955 is the bare base of the csi930955 alias.
|
||||
entry = registry.find_by_bare_conflict("930955")
|
||||
assert entry is not None
|
||||
assert entry.canonical_id == "csi930955"
|
||||
|
||||
def test_display_name_is_not_an_identity_alias(self) -> None:
|
||||
"""Gap 2: a Chinese display name (e.g. ``沪深300``) must never resolve
|
||||
as an index identity — text names are not identity aliases."""
|
||||
registry = default_index_registry()
|
||||
assert registry.find_by_explicit_key("沪深300") is None
|
||||
assert registry.find_by_explicit_key("上证50") is None
|
||||
# The canonical/display/alias code forms still resolve.
|
||||
assert registry.find_by_explicit_key("sh000300") is not None
|
||||
assert registry.find_by_explicit_key("000300.SH") is not None
|
||||
# Parsing the Chinese name must not elevate to index.
|
||||
target = parse_analysis_target("沪深300")
|
||||
assert target.asset_type != ParseStatus.INDEX
|
||||
|
||||
def test_custom_registry_rejects_text_identity_alias(self) -> None:
|
||||
with pytest.raises(ValueError, match="explicit code form"):
|
||||
IndexRegistry((
|
||||
IndexEntry(
|
||||
bare_code="000300",
|
||||
exchange="SH",
|
||||
canonical_id="sh000300",
|
||||
display_name="沪深300",
|
||||
aliases=("CSI300",),
|
||||
),
|
||||
))
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Batch parsing helper.
|
||||
@@ -425,15 +475,14 @@ class TestMaintainerSpecSamples:
|
||||
assert target.asset_type == ParseStatus.INDEX
|
||||
assert target.canonical_id == "sh000300"
|
||||
|
||||
def test_sz399300_falls_back_to_stock(self) -> None:
|
||||
"""sz399300 is NOT in the canonical 5-index white-list (we only carry
|
||||
sz399001 + sz399006 for the SZ side). Per contract #3 it degrades to
|
||||
stock. This sample guards against accidental future registry expand
|
||||
that would silently flip behaviour.
|
||||
def test_sz399300_resolves_to_index(self) -> None:
|
||||
"""``sz399300`` is a registered cross-market alias of ``sh000300``
|
||||
(沪深300) per the manifest, so it resolves to the index.
|
||||
"""
|
||||
target = parse_analysis_target("sz399300")
|
||||
assert target.asset_type == ParseStatus.STOCK
|
||||
assert target.exchange == "SZ"
|
||||
assert target.asset_type == ParseStatus.INDEX
|
||||
assert target.canonical_id == "sh000300"
|
||||
assert target.exchange == "SH"
|
||||
|
||||
def test_sh600519(self) -> None:
|
||||
target = parse_analysis_target("sh600519")
|
||||
@@ -882,3 +931,131 @@ class TestExplicitExchangeSuffixRejections:
|
||||
assert target.canonical_id == code
|
||||
assert target.unsupported_reason is not None
|
||||
assert expected_exchange in target.unsupported_reason
|
||||
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# Index registry & CSI provider symbol governance.
|
||||
# ---------------------------------------------------------------------------
|
||||
class TestIndexAliasMatrix:
|
||||
"""Canonical/display/alias converge; CSI explicit; bare stays stock."""
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
"code,expected_canonical,expected_exchange",
|
||||
[
|
||||
("sh000300", "sh000300", "SH"),
|
||||
("000300.SH", "sh000300", "SH"),
|
||||
("sz399300", "sh000300", "SH"),
|
||||
("399300.SZ", "sh000300", "SH"),
|
||||
("000300.CSI", "sh000300", "SH"),
|
||||
],
|
||||
)
|
||||
def test_sh000300_alias_forms_resolve_to_index(
|
||||
self, code: str, expected_canonical: str, expected_exchange: str
|
||||
) -> None:
|
||||
target = parse_analysis_target(code)
|
||||
assert target.asset_type == ParseStatus.INDEX
|
||||
assert target.canonical_id == expected_canonical
|
||||
assert target.exchange == expected_exchange
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
"code",
|
||||
["csi930955", "930955.CSI", "CSI930955", " csi930955 "],
|
||||
)
|
||||
def test_csi930955_forms_resolve_to_index(self, code: str) -> None:
|
||||
target = parse_analysis_target(code)
|
||||
assert target.asset_type == ParseStatus.INDEX
|
||||
assert target.canonical_id == "csi930955"
|
||||
assert target.exchange == "CSI"
|
||||
|
||||
def test_unknown_csi_is_unsupported_not_us_stock(self) -> None:
|
||||
target = parse_analysis_target("930956.CSI")
|
||||
assert target.asset_type == ParseStatus.UNSUPPORTED
|
||||
assert target.exchange == "UNKNOWN"
|
||||
assert target.unsupported_reason is not None
|
||||
assert "CSI" in target.unsupported_reason
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
"code",
|
||||
[
|
||||
"csi930956",
|
||||
"CSI930956",
|
||||
"930956.CSI",
|
||||
"csi000300",
|
||||
"CSI000300",
|
||||
"csi93095",
|
||||
"csi9309557",
|
||||
"93095.CSI",
|
||||
"9309557.CSI",
|
||||
],
|
||||
)
|
||||
def test_unregistered_explicit_csi_prefix_and_suffix_is_unsupported(
|
||||
self, code: str
|
||||
) -> None:
|
||||
"""PR #2267 review fix: an unknown explicit ``csi`` prefix (or an
|
||||
unregistered ``.CSI`` suffix) must surface as ``unsupported`` — never a
|
||||
US ticker and never a guessed SH/SZ index. Only a manifest-owned
|
||||
identity may route/persist as an index."""
|
||||
target = parse_analysis_target(code)
|
||||
assert target.asset_type == ParseStatus.UNSUPPORTED
|
||||
assert target.exchange == "UNKNOWN"
|
||||
assert target.canonical_id == code
|
||||
assert target.unsupported_reason is not None
|
||||
assert "CSI" in target.unsupported_reason
|
||||
|
||||
def test_csi_prefix_is_not_overequated_to_csi_suffix_alias(self) -> None:
|
||||
"""PR #2267 review fix: ``000300.CSI`` is the registered alias of
|
||||
``sh000300``, but the bare ``csi000300`` prefix is NOT — it must stay
|
||||
unsupported rather than being promoted to the same-code index."""
|
||||
target = parse_analysis_target("000300.CSI")
|
||||
assert target.asset_type == ParseStatus.INDEX
|
||||
assert target.canonical_id == "sh000300"
|
||||
assert target.exchange == "SH"
|
||||
|
||||
unregistered = parse_analysis_target("csi000300")
|
||||
assert unregistered.asset_type == ParseStatus.UNSUPPORTED
|
||||
assert unregistered.exchange == "UNKNOWN"
|
||||
|
||||
def test_us_ticker_starting_with_csi_remains_stock(self) -> None:
|
||||
target = parse_analysis_target("CSIQ")
|
||||
assert target.asset_type == ParseStatus.STOCK
|
||||
assert target.canonical_id == "CSIQ"
|
||||
|
||||
@pytest.mark.parametrize(
|
||||
"code,expected_canonical",
|
||||
[
|
||||
("000001", "sz000001"),
|
||||
("000016", "sz000016"),
|
||||
("000688", "sz000688"),
|
||||
("000300", "sz000300"),
|
||||
("399300", "sz399300"),
|
||||
("930955", "bj930955"),
|
||||
],
|
||||
)
|
||||
def test_bare_conflict_codes_stay_stock_with_matched_index(
|
||||
self, code: str, expected_canonical: str
|
||||
) -> None:
|
||||
target = parse_analysis_target(code)
|
||||
assert target.asset_type == ParseStatus.STOCK
|
||||
assert target.canonical_id == expected_canonical
|
||||
assert target.matched_index is not None
|
||||
|
||||
def test_unregistered_sh_sz_prefixed_stock_keeps_stock_path(self) -> None:
|
||||
target = parse_analysis_target("sh600519")
|
||||
assert target.asset_type == ParseStatus.STOCK
|
||||
assert target.canonical_id == "sh600519"
|
||||
target2 = parse_analysis_target("sz000001")
|
||||
assert target2.asset_type == ParseStatus.STOCK
|
||||
assert target2.canonical_id == "sz000001"
|
||||
|
||||
def test_old_text_aliases_do_not_resolve_to_index(self) -> None:
|
||||
# CSI300 / HS300 / SSE50 / STAR50 are not in the manifest aliases.
|
||||
for code in ("CSI300", "HS300", "SSE50", "STAR50", "SZSE", "ChiNext"):
|
||||
target = parse_analysis_target(code)
|
||||
assert target.asset_type != ParseStatus.INDEX
|
||||
|
||||
def test_nfkc_and_case_normalization_do_not_create_second_identity(self) -> None:
|
||||
a = parse_analysis_target("csi930955")
|
||||
b = parse_analysis_target("ＣＳＩ９３０９５５") # full-width
|
||||
assert a.asset_type == ParseStatus.INDEX
|
||||
assert b.asset_type == ParseStatus.INDEX
|
||||
assert a.canonical_id == b.canonical_id
|
||||
|
||||
+289
-24
@@ -1196,6 +1196,39 @@ class TestStorage(unittest.TestCase):
|
||||
"CREATE INDEX ix_code_date ON stock_daily (code, date)"
|
||||
)
|
||||
|
||||
@staticmethod
|
||||
def _create_legacy_stock_daily_with_canonical_id(db_path: str) -> None:
|
||||
"""``stock_daily`` schema WITH a ``canonical_id`` column (for repair tests)."""
|
||||
with sqlite3.connect(db_path) as conn:
|
||||
conn.execute(
|
||||
"""CREATE TABLE stock_daily (
|
||||
id INTEGER PRIMARY KEY AUTOINCREMENT,
|
||||
code VARCHAR(10) NOT NULL,
|
||||
date DATE NOT NULL,
|
||||
open FLOAT,
|
||||
high FLOAT,
|
||||
low FLOAT,
|
||||
close FLOAT,
|
||||
volume FLOAT,
|
||||
amount FLOAT,
|
||||
pct_chg FLOAT,
|
||||
ma5 FLOAT,
|
||||
ma10 FLOAT,
|
||||
ma20 FLOAT,
|
||||
volume_ratio FLOAT,
|
||||
data_source VARCHAR(50),
|
||||
created_at DATETIME,
|
||||
updated_at DATETIME,
|
||||
canonical_id VARCHAR(32)
|
||||
)"""
|
||||
)
|
||||
conn.execute(
|
||||
"CREATE UNIQUE INDEX uix_code_date ON stock_daily (code, date)"
|
||||
)
|
||||
conn.execute(
|
||||
"CREATE INDEX ix_code_date ON stock_daily (code, date)"
|
||||
)
|
||||
|
||||
@staticmethod
|
||||
def _make_temp_db_path() -> tuple:
|
||||
"""Return (db_dir, db_path); Windows-safe cleanup via shutil.rmtree."""
|
||||
@@ -1329,14 +1362,13 @@ class TestStorage(unittest.TestCase):
|
||||
Config.reset_instance()
|
||||
self._cleanup_temp_dir(db_dir)
|
||||
|
||||
def test_canonical_id_backfill_bare_index_unifies_to_index_canonical_id(self):
|
||||
"""Index-aware backfill (OR-COR-4f9ffc38): bare ``000300`` hits the
|
||||
index registry (``matched_index`` is non-None) and backfills to the
|
||||
index canonical_id ``sh000300`` — NOT ``sz000300`` (the stock-path
|
||||
canonical_id the classifier would synthesise for a 6-digit
|
||||
``0``-prefixed code). Without this unification the same CSI-300 index
|
||||
would split across two canonical_id buckets depending on whether the
|
||||
caller passed a bare code or an explicit ``sh000300`` prefix."""
|
||||
def test_canonical_id_backfill_bare_code_colliding_with_index_stays_stock(self):
|
||||
"""``_derive_canonical_id`` no longer reads
|
||||
``matched_index.canonical_id``. A bare ``000300`` (which collides with
|
||||
the CSI-300 index) now derives to the stock-path canonical_id
|
||||
``sz000300`` — the parser contract says bare codes are always stock.
|
||||
The index conflict is surfaced via ``matched_index`` only, never used
|
||||
to override the canonical_id."""
|
||||
DatabaseManager.reset_instance()
|
||||
db_dir, db_path = self._make_temp_db_path()
|
||||
|
||||
@@ -1355,10 +1387,8 @@ class TestStorage(unittest.TestCase):
|
||||
"SELECT canonical_id FROM stock_daily WHERE code='000300'"
|
||||
).fetchone()[0]
|
||||
|
||||
# Bare index code unifies to the index canonical_id via
|
||||
# ``matched_index.canonical_id`` so bare ``000300`` and explicit
|
||||
# ``sh000300`` land in the same bucket.
|
||||
self.assertEqual(canonical_id, "sh000300")
|
||||
# Bare code resolves to the stock-path canonical_id.
|
||||
self.assertEqual(canonical_id, "sz000300")
|
||||
finally:
|
||||
DatabaseManager.reset_instance()
|
||||
Config.reset_instance()
|
||||
@@ -1683,11 +1713,10 @@ class TestStorage(unittest.TestCase):
|
||||
DatabaseManager.reset_instance()
|
||||
self._cleanup_temp_dir(db_dir)
|
||||
|
||||
def test_save_daily_data_derives_index_aware_canonical_id_for_bare_index_code(self):
|
||||
"""OR-COR-4f9ffc38: ``save_daily_data(df, code="000300")`` with no
|
||||
explicit canonical_id writes ``sh000300`` (index canonical_id), not
|
||||
``sz000300`` (stock-path canonical_id the classifier would synthesise
|
||||
for a 6-digit ``0``-prefixed code)."""
|
||||
def test_save_daily_data_derives_stock_canonical_id_for_bare_index_collision(self):
|
||||
"""``save_daily_data(df, code="000300")`` with no explicit
|
||||
canonical_id writes ``sz000300`` (stock-path canonical_id), NOT
|
||||
``sh000300`` (the index canonical_id). Bare codes are always stock."""
|
||||
DatabaseManager.reset_instance()
|
||||
db_dir, db_path = self._make_temp_db_path()
|
||||
db = DatabaseManager(db_url=f"sqlite:///{db_path}")
|
||||
@@ -1710,16 +1739,16 @@ class TestStorage(unittest.TestCase):
|
||||
and_(StockDaily.code == "000300", StockDaily.date == date(2026, 4, 5))
|
||||
)
|
||||
).scalar_one()
|
||||
self.assertEqual(row.canonical_id, "sh000300")
|
||||
self.assertEqual(row.canonical_id, "sz000300")
|
||||
finally:
|
||||
DatabaseManager.reset_instance()
|
||||
self._cleanup_temp_dir(db_dir)
|
||||
|
||||
def test_save_daily_data_converges_alias_and_prefix_to_same_canonical_id(self):
|
||||
"""OR-COR-4f9ffc38: bare ``000300``, ``sh000300`` and ``000300.SH``
|
||||
all converge to the same canonical_id ``sh000300`` when written via
|
||||
``save_daily_data`` (no explicit canonical_id), so the same underlying
|
||||
index is never split across buckets by input form."""
|
||||
def test_save_daily_data_explicit_index_forms_derive_index_canonical_id(self):
|
||||
"""Explicit index forms (``sh000300`` / ``000300.SH``)
|
||||
derive to the index canonical_id ``sh000300``, while the bare code
|
||||
``000300`` derives to the stock-path ``sz000300``. Explicit index and
|
||||
bare stock are intentionally different buckets."""
|
||||
DatabaseManager.reset_instance()
|
||||
db_dir, db_path = self._make_temp_db_path()
|
||||
db = DatabaseManager(db_url=f"sqlite:///{db_path}")
|
||||
@@ -1747,7 +1776,7 @@ class TestStorage(unittest.TestCase):
|
||||
self.assertEqual(
|
||||
{row.code: row.canonical_id for row in rows},
|
||||
{
|
||||
"000300": "sh000300",
|
||||
"000300": "sz000300",
|
||||
"sh000300": "sh000300",
|
||||
"000300.SH": "sh000300",
|
||||
},
|
||||
@@ -1797,5 +1826,241 @@ class TestStorage(unittest.TestCase):
|
||||
DatabaseManager.reset_instance()
|
||||
Config.reset_instance()
|
||||
|
||||
# ------------------------------------------------------------------
|
||||
# canonical_id repair
|
||||
# ------------------------------------------------------------------
|
||||
|
||||
def test_derive_canonical_id_bare_conflict_stays_stock(self):
|
||||
"""``_derive_canonical_id`` returns the parser stock
|
||||
canonical for bare conflict codes, never the index canonical."""
|
||||
DatabaseManager.reset_instance()
|
||||
db = DatabaseManager(db_url="sqlite:///:memory:")
|
||||
try:
|
||||
self.assertEqual(db._derive_canonical_id("000016"), "sz000016")
|
||||
self.assertEqual(db._derive_canonical_id("930955"), "bj930955")
|
||||
self.assertEqual(db._derive_canonical_id("sh000016"), "sh000016")
|
||||
self.assertEqual(db._derive_canonical_id("930955.CSI"), "csi930955")
|
||||
finally:
|
||||
DatabaseManager.reset_instance()
|
||||
|
||||
def test_derive_canonical_id_rejects_unregistered_csi(self):
|
||||
"""PR #2267 review fix: an unsupported identity (unregistered
|
||||
``csi`` prefix or ``.CSI`` suffix) must NOT enter a persistent
|
||||
canonical bucket — ``_derive_canonical_id`` returns None so the
|
||||
caller persists NULL instead of the raw token."""
|
||||
DatabaseManager.reset_instance()
|
||||
db = DatabaseManager(db_url="sqlite:///:memory:")
|
||||
try:
|
||||
self.assertIsNone(db._derive_canonical_id("csi930956"))
|
||||
self.assertIsNone(db._derive_canonical_id("CSI930956"))
|
||||
self.assertIsNone(db._derive_canonical_id("930956.CSI"))
|
||||
self.assertIsNone(db._derive_canonical_id("csi000300"))
|
||||
finally:
|
||||
DatabaseManager.reset_instance()
|
||||
|
||||
def test_canonical_id_repair_fixes_bare_misbucketed_rows(self):
|
||||
"""Rows whose bare code has an erroneous index canonical are
|
||||
repaired to the parser stock canonical; explicit index rows and
|
||||
correct stock rows are untouched."""
|
||||
DatabaseManager.reset_instance()
|
||||
db_dir, db_path = self._make_temp_db_path()
|
||||
|
||||
try:
|
||||
self._create_legacy_stock_daily_with_canonical_id(db_path)
|
||||
with sqlite3.connect(db_path) as conn:
|
||||
conn.executemany(
|
||||
"""INSERT INTO stock_daily (code, date, close, canonical_id)
|
||||
VALUES (?, ?, ?, ?)""",
|
||||
[
|
||||
# Mis-bucketed bare codes (old index-aware derivation).
|
||||
("000001", "2026-01-01", 1.0, "sh000001"),
|
||||
("000016", "2026-01-02", 2.0, "sh000016"),
|
||||
("000688", "2026-01-03", 3.0, "sh000688"),
|
||||
("930955", "2026-01-04", 4.0, "csi930955"),
|
||||
# Explicit index rows — must NOT be modified.
|
||||
("sh000016", "2026-01-05", 5.0, "sh000016"),
|
||||
("930955.CSI", "2026-01-06", 6.0, "csi930955"),
|
||||
# Correct stock row — must NOT be modified.
|
||||
("600519", "2026-01-07", 7.0, "sh600519"),
|
||||
# Unrelated row — must NOT be modified.
|
||||
("AAPL", "2026-01-08", 8.0, "AAPL"),
|
||||
],
|
||||
)
|
||||
|
||||
DatabaseManager(db_url=f"sqlite:///{db_path}")
|
||||
|
||||
with sqlite3.connect(db_path) as conn:
|
||||
rows = dict(
|
||||
conn.execute(
|
||||
"SELECT code, canonical_id FROM stock_daily ORDER BY id"
|
||||
).fetchall()
|
||||
)
|
||||
|
||||
self.assertEqual(rows["000001"], "sz000001")
|
||||
self.assertEqual(rows["000016"], "sz000016")
|
||||
self.assertEqual(rows["000688"], "sz000688")
|
||||
self.assertEqual(rows["930955"], "bj930955")
|
||||
# Explicit index rows preserved.
|
||||
self.assertEqual(rows["sh000016"], "sh000016")
|
||||
self.assertEqual(rows["930955.CSI"], "csi930955")
|
||||
# Correct stock + unrelated rows preserved.
|
||||
self.assertEqual(rows["600519"], "sh600519")
|
||||
self.assertEqual(rows["AAPL"], "AAPL")
|
||||
finally:
|
||||
DatabaseManager.reset_instance()
|
||||
Config.reset_instance()
|
||||
self._cleanup_temp_dir(db_dir)
|
||||
|
||||
def test_canonical_id_repair_is_idempotent(self):
|
||||
"""Running the repair a second time repairs 0 rows."""
|
||||
DatabaseManager.reset_instance()
|
||||
db_dir, db_path = self._make_temp_db_path()
|
||||
|
||||
try:
|
||||
self._create_legacy_stock_daily_with_canonical_id(db_path)
|
||||
with sqlite3.connect(db_path) as conn:
|
||||
conn.execute(
|
||||
"INSERT INTO stock_daily (code, date, close, canonical_id) "
|
||||
"VALUES (?, ?, ?, ?)",
|
||||
("000016", "2026-01-01", 2.0, "sh000016"),
|
||||
)
|
||||
|
||||
DatabaseManager(db_url=f"sqlite:///{db_path}")
|
||||
with sqlite3.connect(db_path) as conn:
|
||||
first = conn.execute(
|
||||
"SELECT canonical_id FROM stock_daily WHERE code='000016'"
|
||||
).fetchone()[0]
|
||||
self.assertEqual(first, "sz000016")
|
||||
|
||||
DatabaseManager.reset_instance()
|
||||
with self.assertLogs("src.storage", level="INFO") as logs:
|
||||
DatabaseManager(db_url=f"sqlite:///{db_path}")
|
||||
log_text = "\n".join(logs.output)
|
||||
self.assertIn("repaired_count=0", log_text)
|
||||
finally:
|
||||
DatabaseManager.reset_instance()
|
||||
Config.reset_instance()
|
||||
self._cleanup_temp_dir(db_dir)
|
||||
|
||||
def test_canonical_id_repair_skips_when_registry_empty(self):
|
||||
"""When the index registry is empty the repair is a no-op and
|
||||
logs a WARNING; no rows are modified."""
|
||||
DatabaseManager.reset_instance()
|
||||
db_dir, db_path = self._make_temp_db_path()
|
||||
|
||||
try:
|
||||
self._create_legacy_stock_daily_with_canonical_id(db_path)
|
||||
with sqlite3.connect(db_path) as conn:
|
||||
conn.execute(
|
||||
"INSERT INTO stock_daily (code, date, close, canonical_id) "
|
||||
"VALUES (?, ?, ?, ?)",
|
||||
("000016", "2026-01-01", 2.0, "sh000016"),
|
||||
)
|
||||
|
||||
with patch(
|
||||
"src.data.stock_index_loader._load_active_index_rows",
|
||||
return_value=[],
|
||||
), self.assertLogs("src.storage", level="WARNING") as logs:
|
||||
DatabaseManager(db_url=f"sqlite:///{db_path}")
|
||||
|
||||
with sqlite3.connect(db_path) as conn:
|
||||
canonical_id = conn.execute(
|
||||
"SELECT canonical_id FROM stock_daily WHERE code='000016'"
|
||||
).fetchone()[0]
|
||||
self.assertEqual(canonical_id, "sh000016")
|
||||
self.assertTrue(
|
||||
any("registry is empty" in record.getMessage() for record in logs.records)
|
||||
)
|
||||
finally:
|
||||
DatabaseManager.reset_instance()
|
||||
Config.reset_instance()
|
||||
self._cleanup_temp_dir(db_dir)
|
||||
|
||||
def test_canonical_id_repair_handles_concurrent_rewrite_safely(self):
|
||||
"""A conditional UPDATE that loses the race (row already
|
||||
rewritten) is skipped, not double-counted."""
|
||||
DatabaseManager.reset_instance()
|
||||
db_dir, db_path = self._make_temp_db_path()
|
||||
|
||||
try:
|
||||
self._create_legacy_stock_daily_with_canonical_id(db_path)
|
||||
with sqlite3.connect(db_path) as conn:
|
||||
conn.execute(
|
||||
"INSERT INTO stock_daily (code, date, close, canonical_id) "
|
||||
"VALUES (?, ?, ?, ?)",
|
||||
("000016", "2026-01-01", 2.0, "sh000016"),
|
||||
)
|
||||
|
||||
real_parse = stock_list_parser_module.parse_analysis_target
|
||||
|
||||
def racing_parse(code):
|
||||
# Simulate a concurrent writer that already fixed the row.
|
||||
if code == "000016":
|
||||
with sqlite3.connect(db_path) as conn:
|
||||
conn.execute(
|
||||
"UPDATE stock_daily SET canonical_id='sz000016' "
|
||||
"WHERE code='000016'"
|
||||
)
|
||||
conn.commit()
|
||||
return real_parse(code)
|
||||
|
||||
with patch(
|
||||
"src.services.stock_list_parser.parse_analysis_target",
|
||||
side_effect=racing_parse,
|
||||
):
|
||||
DatabaseManager(db_url=f"sqlite:///{db_path}")
|
||||
|
||||
with sqlite3.connect(db_path) as conn:
|
||||
canonical_id = conn.execute(
|
||||
"SELECT canonical_id FROM stock_daily WHERE code='000016'"
|
||||
).fetchone()[0]
|
||||
self.assertEqual(canonical_id, "sz000016")
|
||||
finally:
|
||||
DatabaseManager.reset_instance()
|
||||
Config.reset_instance()
|
||||
self._cleanup_temp_dir(db_dir)
|
||||
|
||||
def test_canonical_id_repair_queries_only_relevant_rows(self):
|
||||
"""Gap 6: the repair SQL filters to rows whose canonical_id is in the
|
||||
active index set, so rows with unrelated canonical_ids are never
|
||||
scanned/derived."""
|
||||
DatabaseManager.reset_instance()
|
||||
db_dir, db_path = self._make_temp_db_path()
|
||||
|
||||
try:
|
||||
self._create_legacy_stock_daily_with_canonical_id(db_path)
|
||||
with sqlite3.connect(db_path) as conn:
|
||||
conn.executemany(
|
||||
"""INSERT INTO stock_daily (code, date, close, canonical_id)
|
||||
VALUES (?, ?, ?, ?)""",
|
||||
[
|
||||
# Mis-bucketed bare code — must be repaired.
|
||||
("000016", "2026-01-01", 2.0, "sh000016"),
|
||||
# Unrelated canonical — must NOT be touched.
|
||||
("600519", "2026-01-02", 7.0, "sh600519"),
|
||||
("AAPL", "2026-01-03", 8.0, "AAPL"),
|
||||
# Explicit index row — must NOT be touched.
|
||||
("sh000016", "2026-01-04", 5.0, "sh000016"),
|
||||
],
|
||||
)
|
||||
|
||||
DatabaseManager(db_url=f"sqlite:///{db_path}")
|
||||
|
||||
with sqlite3.connect(db_path) as conn:
|
||||
rows = dict(
|
||||
conn.execute(
|
||||
"SELECT code, canonical_id FROM stock_daily ORDER BY id"
|
||||
).fetchall()
|
||||
)
|
||||
# Only the mis-bucketed bare code is repaired.
|
||||
self.assertEqual(rows["000016"], "sz000016")
|
||||
self.assertEqual(rows["600519"], "sh600519")
|
||||
self.assertEqual(rows["AAPL"], "AAPL")
|
||||
self.assertEqual(rows["sh000016"], "sh000016")
|
||||
finally:
|
||||
DatabaseManager.reset_instance()
|
||||
Config.reset_instance()
|
||||
self._cleanup_temp_dir(db_dir)
|
||||
|
||||
if __name__ == '__main__':
|
||||
unittest.main()
|
||||
|
||||
@@ -87,6 +87,35 @@ class TestTaskService(unittest.TestCase):
|
||||
self.assertIn("args", captured)
|
||||
self.assertEqual(captured["args"][1], "005930.KS")
|
||||
|
||||
def test_submit_analysis_passes_parser_canonical_to_executor(self):
|
||||
"""Real task-layer regression: TaskService must hand the executor the
|
||||
parser canonical ``csi930955`` (not the raw alias ``930955.CSI`` or the
|
||||
old uppercase ``CSI930955``) so the analysis pipeline receives one
|
||||
consistent identity for the same registered CSI index."""
|
||||
service = TaskService()
|
||||
service._tasks = {}
|
||||
service._tasks_lock = threading.Lock()
|
||||
captured = {}
|
||||
|
||||
executor = MagicMock()
|
||||
|
||||
def capture_submit(*args, **kwargs):
|
||||
captured["args"] = args
|
||||
return "future"
|
||||
|
||||
executor.submit.side_effect = capture_submit
|
||||
service._executor = executor
|
||||
|
||||
# Use the real resolver (not a mock): ``930955.CSI`` is a registered CSI
|
||||
# explicit identity and must converge to the parser canonical
|
||||
# ``csi930955``, which is what gets handed to the executor.
|
||||
result = service.submit_analysis("930955.CSI", report_type="simple", query_source="cli")
|
||||
|
||||
self.assertEqual(result["code"], "csi930955")
|
||||
self.assertIn("args", captured)
|
||||
# executor.submit(self._run_analysis, code, task_id, ...) — code is arg[1]
|
||||
self.assertEqual(captured["args"][1], "csi930955")
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
import unittest
|
||||
|
||||
Reference in New Issue
Block a user