stock-analysis-plugin 0.2.10 → 0.2.12
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +237 -381
- package/docs/mcp.md +60 -0
- package/hermes/__init__.py +35 -0
- package/hermes/plugin.yaml +131 -0
- package/hermes/schemas.py +6 -0
- package/hermes/tools.py +630 -0
- package/mcp_host/__init__.py +1 -0
- package/mcp_host/__main__.py +3 -0
- package/mcp_host/server.py +222 -0
- package/package.json +17 -10
- package/pi/index.ts +53 -2887
- package/scripts/cli-json.mjs +23 -0
- package/scripts/exec-process.mjs +89 -0
- package/scripts/setup-python.mjs +37 -4
- package/scripts/tool-calls.mjs +489 -0
- package/skills/event-driven/SKILL.md +2 -2
- package/skills/growth-quality/SKILL.md +1 -1
- package/skills/stock-analysis/SKILL.md +12 -7
- package/skills/stock-analysis/references/stateless-analysis.md +9 -1
- package/tools/_cache.py +13 -0
- package/tools/_subproc.py +46 -0
- package/tools/alert_rules.py +65 -16
- package/tools/backtest.py +3 -3
- package/tools/capabilities.py +32 -9
- package/tools/cli_budgets.json +2 -0
- package/tools/diagnostics.py +3 -4
- package/tools/gather.py +2 -6
- package/tools/import_parser.py +3 -4
- package/tools/lixinger.py +119 -27
- package/tools/market_regime.py +5 -2
- package/tools/market_review.py +5 -6
- package/tools/name_resolver.py +82 -21
- package/tools/report_renderer.py +3 -3
- package/tools/requirements.txt +6 -5
- package/tools/risk_screening.py +3 -2
- package/tools/screener.py +6 -4
- package/tools/search_intel.py +81 -12
- package/tools/sec_edgar.py +266 -0
- package/tools/sec_holders.py +6 -0
- package/tools/signal_tracker.py +11 -10
- package/tools/stateless_analysis.py +115 -0
- package/tools/stock_data.py +707 -222
- package/tools/tiaa_disclosures.py +267 -0
- package/tools/tool_schemas.json +2704 -0
- package/tools/trading_calendar.py +65 -30
- package/tools/value_data.py +366 -134
- package/tools/watchlist.py +43 -5
|
@@ -0,0 +1,267 @@
|
|
|
1
|
+
"""Public, login-free TIAA disclosures from the official YouZhiYouXing site."""
|
|
2
|
+
|
|
3
|
+
import re
|
|
4
|
+
import sys
|
|
5
|
+
import threading
|
|
6
|
+
import time
|
|
7
|
+
import xml.etree.ElementTree as ET
|
|
8
|
+
from concurrent.futures import TimeoutError as FutureTimeoutError
|
|
9
|
+
from datetime import datetime, timezone
|
|
10
|
+
from html import unescape
|
|
11
|
+
from html.parser import HTMLParser
|
|
12
|
+
from pathlib import Path
|
|
13
|
+
from urllib.parse import urlparse
|
|
14
|
+
|
|
15
|
+
import requests
|
|
16
|
+
|
|
17
|
+
sys.path.insert(0, str(Path(__file__).resolve().parent))
|
|
18
|
+
from _cache import cached # noqa: E402
|
|
19
|
+
|
|
20
|
+
_SITEMAP = "https://youzhiyouxing.cn/sitemap.xml"
|
|
21
|
+
_ADVISOR_URL = "https://www.csisc.cn/zbscbzw/c100264/list_jjbm.shtml"
|
|
22
|
+
_ARTICLE_URL = re.compile(r"https://youzhiyouxing\.cn/(?:n/)?lite-posts/(\d+)(?:[/?#].*)?")
|
|
23
|
+
_DISCLOSURE_TERMS = ("示例账户", "成分基金", "配置", "业绩", "收益", "调仓", "持仓")
|
|
24
|
+
_MAX_ARTICLES = 40
|
|
25
|
+
_DISCLOSURE_DEADLINE = 30
|
|
26
|
+
|
|
27
|
+
|
|
28
|
+
class _DisclosureTimeout(Exception):
|
|
29
|
+
"""A bounded disclosure attempt failed and must not be cached as an empty result."""
|
|
30
|
+
|
|
31
|
+
|
|
32
|
+
def _until(deadline, fn, *args):
|
|
33
|
+
"""Run a blocking provider call in a daemon thread and stop waiting at the deadline.
|
|
34
|
+
|
|
35
|
+
A provider that ignores its own timeout can then neither stretch the request nor keep
|
|
36
|
+
the CLI process alive after it has answered.
|
|
37
|
+
"""
|
|
38
|
+
box = {}
|
|
39
|
+
|
|
40
|
+
def run():
|
|
41
|
+
try:
|
|
42
|
+
box["value"] = fn(*args)
|
|
43
|
+
except BaseException as exc: # re-raised in the caller's thread
|
|
44
|
+
box["error"] = exc
|
|
45
|
+
|
|
46
|
+
worker = threading.Thread(target=run, daemon=True)
|
|
47
|
+
worker.start()
|
|
48
|
+
worker.join(max(0.0, deadline - time.monotonic()))
|
|
49
|
+
if worker.is_alive():
|
|
50
|
+
raise _DisclosureTimeout
|
|
51
|
+
if "error" in box:
|
|
52
|
+
raise box["error"]
|
|
53
|
+
return box.get("value")
|
|
54
|
+
|
|
55
|
+
|
|
56
|
+
class _ArticleParser(HTMLParser):
|
|
57
|
+
def __init__(self):
|
|
58
|
+
super().__init__(convert_charrefs=True)
|
|
59
|
+
self.meta = {}
|
|
60
|
+
self.parts = []
|
|
61
|
+
self.ignored = 0
|
|
62
|
+
|
|
63
|
+
def handle_starttag(self, tag, attrs):
|
|
64
|
+
if tag in ("script", "style"):
|
|
65
|
+
self.ignored += 1
|
|
66
|
+
if tag == "meta":
|
|
67
|
+
values = dict(attrs)
|
|
68
|
+
if values.get("property") in ("og:title", "og:description"):
|
|
69
|
+
self.meta[values["property"]] = values.get("content", "")
|
|
70
|
+
if tag in ("p", "div", "h1", "h2", "h3", "li", "br"):
|
|
71
|
+
self.parts.append("\n")
|
|
72
|
+
|
|
73
|
+
def handle_endtag(self, tag):
|
|
74
|
+
if tag in ("script", "style") and self.ignored:
|
|
75
|
+
self.ignored -= 1
|
|
76
|
+
if tag in ("p", "div", "h1", "h2", "h3", "li"):
|
|
77
|
+
self.parts.append("\n")
|
|
78
|
+
|
|
79
|
+
def handle_data(self, data):
|
|
80
|
+
if not self.ignored:
|
|
81
|
+
self.parts.append(data)
|
|
82
|
+
|
|
83
|
+
def text(self):
|
|
84
|
+
lines = []
|
|
85
|
+
for line in "".join(self.parts).splitlines():
|
|
86
|
+
clean = re.sub(r"\s+", " ", line).strip()
|
|
87
|
+
if clean:
|
|
88
|
+
lines.append(clean)
|
|
89
|
+
title = self.meta.get("og:title")
|
|
90
|
+
title_lines = [index for index, line in enumerate(lines) if line == title]
|
|
91
|
+
if title_lines:
|
|
92
|
+
lines = lines[title_lines[-1] + 1 :]
|
|
93
|
+
comments = next((index for index, line in enumerate(lines) if line.endswith("条回应")), len(lines))
|
|
94
|
+
return "\n".join(lines[:comments])
|
|
95
|
+
|
|
96
|
+
|
|
97
|
+
def _sitemap_urls():
|
|
98
|
+
response = requests.get(_SITEMAP, headers={"User-Agent": "Mozilla/5.0"}, timeout=(2, 8))
|
|
99
|
+
response.raise_for_status()
|
|
100
|
+
root = ET.fromstring(response.content)
|
|
101
|
+
namespace = {"s": "http://www.sitemaps.org/schemas/sitemap/0.9"}
|
|
102
|
+
entries = []
|
|
103
|
+
for node in root.findall("s:url", namespace):
|
|
104
|
+
location = node.findtext("s:loc", default="", namespaces=namespace)
|
|
105
|
+
match = _ARTICLE_URL.fullmatch(location)
|
|
106
|
+
if match:
|
|
107
|
+
entries.append((node.findtext("s:lastmod", default="", namespaces=namespace), match.group(1)))
|
|
108
|
+
return [
|
|
109
|
+
f"https://youzhiyouxing.cn/n/lite-posts/{article_id}" for _, article_id in sorted(entries, reverse=True)[:40]
|
|
110
|
+
]
|
|
111
|
+
|
|
112
|
+
|
|
113
|
+
def _search_urls(portfolio_name, budget_seconds=45):
|
|
114
|
+
try:
|
|
115
|
+
from search_intel import search_news
|
|
116
|
+
|
|
117
|
+
query = f"site:youzhiyouxing.cn/n/lite-posts/ {portfolio_name} 示例账户 成分基金"
|
|
118
|
+
result = search_news(query, max_results=10, budget_seconds=budget_seconds)
|
|
119
|
+
if any("time budget exhausted" in str(error).lower() for error in result.get("engines_errors", {}).values()):
|
|
120
|
+
raise _DisclosureTimeout
|
|
121
|
+
urls = []
|
|
122
|
+
for row in result.get("results", []):
|
|
123
|
+
match = _ARTICLE_URL.fullmatch(row.get("url", ""))
|
|
124
|
+
if match:
|
|
125
|
+
urls.append(f"https://youzhiyouxing.cn/n/lite-posts/{match.group(1)}")
|
|
126
|
+
return urls
|
|
127
|
+
except _DisclosureTimeout:
|
|
128
|
+
raise
|
|
129
|
+
except Exception:
|
|
130
|
+
return []
|
|
131
|
+
|
|
132
|
+
|
|
133
|
+
def _article(url, portfolio_name):
|
|
134
|
+
try:
|
|
135
|
+
response = requests.get(url, headers={"User-Agent": "Mozilla/5.0"}, timeout=(1, 3))
|
|
136
|
+
response.raise_for_status()
|
|
137
|
+
parser = _ArticleParser()
|
|
138
|
+
parser.feed(response.text)
|
|
139
|
+
text = parser.text()
|
|
140
|
+
if portfolio_name not in text or not any(term in text for term in _DISCLOSURE_TERMS):
|
|
141
|
+
return None
|
|
142
|
+
date_match = re.search(r"(20\d{2})\s*年\s*(\d{1,2})\s*月\s*(\d{1,2})\s*日", response.text)
|
|
143
|
+
published_at = None
|
|
144
|
+
if date_match:
|
|
145
|
+
published_at = "-".join([date_match.group(1), *(part.zfill(2) for part in date_match.groups()[1:])])
|
|
146
|
+
return {
|
|
147
|
+
"title": parser.meta.get("og:title", ""),
|
|
148
|
+
"url": url,
|
|
149
|
+
"published_at": published_at,
|
|
150
|
+
"summary": parser.meta.get("og:description", ""),
|
|
151
|
+
"text": text[:3000],
|
|
152
|
+
"source": "youzhiyouxing",
|
|
153
|
+
"partial": True,
|
|
154
|
+
}
|
|
155
|
+
except Exception:
|
|
156
|
+
return None
|
|
157
|
+
|
|
158
|
+
|
|
159
|
+
def advisor_metadata(symbol):
|
|
160
|
+
advisor_code = f"A{symbol[4:7]}"
|
|
161
|
+
|
|
162
|
+
def fetch():
|
|
163
|
+
response = requests.get(_ADVISOR_URL, headers={"User-Agent": "Mozilla/5.0"}, timeout=15)
|
|
164
|
+
response.raise_for_status()
|
|
165
|
+
text = response.content.decode("utf-8-sig")
|
|
166
|
+
rows = re.findall(
|
|
167
|
+
r"<tr\b[^>]*>\s*<td\b[^>]*>\s*\d+\s*</td>\s*"
|
|
168
|
+
r"<td\b[^>]*>\s*(.*?)\s*</td>\s*<td\b[^>]*>\s*(A\d{3})\s*</td>",
|
|
169
|
+
text,
|
|
170
|
+
re.S,
|
|
171
|
+
)
|
|
172
|
+
if not rows:
|
|
173
|
+
raise ValueError("CSISC advisor organization table not found")
|
|
174
|
+
return {
|
|
175
|
+
"organizations": {code: unescape(re.sub(r"<[^>]+>", "", name)).strip() for name, code in rows},
|
|
176
|
+
"retrieved_at": datetime.now(timezone.utc).isoformat(),
|
|
177
|
+
}
|
|
178
|
+
|
|
179
|
+
data = cached("csisc-fund-advisor-organizations", 24 * 60 * 60, fetch)
|
|
180
|
+
return {
|
|
181
|
+
"advisor_code": advisor_code,
|
|
182
|
+
"advisor_org": data["organizations"].get(advisor_code),
|
|
183
|
+
"advisor_mapping_basis": "tiaa_prefix_best_effort",
|
|
184
|
+
"advisor_source_url": _ADVISOR_URL,
|
|
185
|
+
"advisor_retrieved_at": data["retrieved_at"],
|
|
186
|
+
}
|
|
187
|
+
|
|
188
|
+
|
|
189
|
+
def public_disclosures(symbol, portfolio_name):
|
|
190
|
+
def fetch():
|
|
191
|
+
deadline = time.monotonic() + _DISCLOSURE_DEADLINE
|
|
192
|
+
urls = _sitemap_urls()
|
|
193
|
+
search_results = []
|
|
194
|
+
|
|
195
|
+
timeout_state = {"search": False, "articles": False}
|
|
196
|
+
if time.monotonic() < deadline:
|
|
197
|
+
try:
|
|
198
|
+
search_results = _until(
|
|
199
|
+
deadline, _search_urls, portfolio_name, max(1, int(deadline - time.monotonic()))
|
|
200
|
+
)
|
|
201
|
+
# concurrent.futures.TimeoutError only aliases the builtin from Python 3.11.
|
|
202
|
+
except (TimeoutError, FutureTimeoutError, _DisclosureTimeout):
|
|
203
|
+
timeout_state["search"] = True
|
|
204
|
+
if time.monotonic() >= deadline:
|
|
205
|
+
timeout_state["search"] = True
|
|
206
|
+
else:
|
|
207
|
+
timeout_state["search"] = True
|
|
208
|
+
urls = list(dict.fromkeys([*search_results, *urls]))[:_MAX_ARTICLES]
|
|
209
|
+
items = []
|
|
210
|
+
for url in urls:
|
|
211
|
+
if time.monotonic() >= deadline:
|
|
212
|
+
timeout_state["articles"] = True
|
|
213
|
+
break
|
|
214
|
+
try:
|
|
215
|
+
item = _until(deadline, _article, url, portfolio_name)
|
|
216
|
+
if item:
|
|
217
|
+
items.append(item)
|
|
218
|
+
except _DisclosureTimeout:
|
|
219
|
+
timeout_state["articles"] = True
|
|
220
|
+
break
|
|
221
|
+
except (TimeoutError, FutureTimeoutError):
|
|
222
|
+
timeout_state["articles"] = True
|
|
223
|
+
if time.monotonic() >= deadline and urls:
|
|
224
|
+
timeout_state["articles"] = True
|
|
225
|
+
items.sort(key=lambda item: item["published_at"] or "", reverse=True)
|
|
226
|
+
items = items[:3]
|
|
227
|
+
result = {
|
|
228
|
+
"items": items,
|
|
229
|
+
"as_of": max((item["published_at"] for item in items if item["published_at"]), default=None),
|
|
230
|
+
}
|
|
231
|
+
if any(timeout_state.values()):
|
|
232
|
+
if not items:
|
|
233
|
+
raise _DisclosureTimeout
|
|
234
|
+
result["error"] = "disclosure_timeout"
|
|
235
|
+
return result
|
|
236
|
+
|
|
237
|
+
return cached(f"tiaa-public-disclosures:{symbol}", 6 * 60 * 60, fetch)
|
|
238
|
+
|
|
239
|
+
|
|
240
|
+
def search_public(symbol, portfolio_name):
|
|
241
|
+
def fetch():
|
|
242
|
+
try:
|
|
243
|
+
from search_intel import search_news
|
|
244
|
+
|
|
245
|
+
label = portfolio_name or symbol
|
|
246
|
+
result = search_news(f'"{symbol}" "{label}" 示例账户 成分基金 业绩', max_results=3)
|
|
247
|
+
rows = []
|
|
248
|
+
for row in result.get("results", []):
|
|
249
|
+
parsed = urlparse(row.get("url", ""))
|
|
250
|
+
if parsed.scheme not in ("http", "https") or not parsed.netloc:
|
|
251
|
+
continue
|
|
252
|
+
rows.append(
|
|
253
|
+
{
|
|
254
|
+
"title": row.get("title", ""),
|
|
255
|
+
"url": row["url"],
|
|
256
|
+
"snippet": row.get("snippet", ""),
|
|
257
|
+
"source": row.get("source", ""),
|
|
258
|
+
"publisher": parsed.netloc,
|
|
259
|
+
"verification": "unverified_search_result",
|
|
260
|
+
"partial": True,
|
|
261
|
+
}
|
|
262
|
+
)
|
|
263
|
+
return rows
|
|
264
|
+
except Exception:
|
|
265
|
+
return []
|
|
266
|
+
|
|
267
|
+
return cached(f"tiaa-public-search:{symbol}", 6 * 60 * 60, fetch)
|