clients.py
12.5 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
"""HTTP clients for search API, reranker, and DashScope chat (relevance labeling)."""
from __future__ import annotations
import io
import json
import logging
import threading
import time
import uuid
from typing import Any, Dict, List, Optional, Sequence, Tuple
import requests
from .constants import EVAL_VERBOSE_LOG_FILE, VALID_LABELS
from .logging_setup import setup_eval_logging
from .prompts import classify_prompt, intent_analysis_prompt
from .utils import build_label_doc_line, extract_json_blob, safe_json_dumps
_VERBOSE_LOGGER_LOCK = threading.Lock()
_eval_llm_verbose_logger_singleton: logging.Logger | None = None
_eval_llm_verbose_path_logged = False
def _get_eval_llm_verbose_logger() -> logging.Logger:
"""File logger for full LLM prompts/responses → ``logs/verbose/eval_verbose.log``."""
setup_eval_logging()
global _eval_llm_verbose_logger_singleton, _eval_llm_verbose_path_logged
with _VERBOSE_LOGGER_LOCK:
if _eval_llm_verbose_logger_singleton is not None:
return _eval_llm_verbose_logger_singleton
log_path = EVAL_VERBOSE_LOG_FILE
log_path.parent.mkdir(parents=True, exist_ok=True)
lg = logging.getLogger("search_eval.verbose_llm")
lg.setLevel(logging.INFO)
if not lg.handlers:
handler = logging.FileHandler(log_path, encoding="utf-8")
handler.setFormatter(logging.Formatter("%(asctime)s - %(levelname)s - %(message)s"))
lg.addHandler(handler)
lg.propagate = False
_eval_llm_verbose_logger_singleton = lg
if not _eval_llm_verbose_path_logged:
_eval_llm_verbose_path_logged = True
logging.getLogger("search_eval").info(
"LLM verbose I/O log (full prompt + response): %s",
log_path.resolve(),
)
return lg
def _log_eval_llm_verbose(
*,
phase: str,
model: str,
prompt: str,
assistant_text: str,
raw_response: str,
) -> None:
log = _get_eval_llm_verbose_logger()
sep = "=" * 80
log.info("\n%s", sep)
log.info("phase=%s model=%s", phase, model)
log.info("%s\nFULL PROMPT (user message)\n%s", sep, prompt)
log.info("%s\nASSISTANT CONTENT (parsed)\n%s", sep, assistant_text)
log.info("%s\nRAW RESPONSE (JSON string)\n%s", sep, raw_response)
log.info("%s\n", sep)
def _canonicalize_judge_label(raw: str) -> str | None:
s = str(raw or "").strip().strip('"').strip("'")
if s in VALID_LABELS:
return s
low = s.lower()
for v in VALID_LABELS:
if v.lower() == low:
return v
return None
class SearchServiceClient:
def __init__(self, base_url: str, tenant_id: str):
self.base_url = base_url.rstrip("/")
self.tenant_id = str(tenant_id)
self.session = requests.Session()
def search(self, query: str, size: int, from_: int = 0, language: str = "en", *, debug: bool = False) -> Dict[str, Any]:
payload: Dict[str, Any] = {
"query": query,
"size": size,
"from": from_,
"language": language,
}
if debug:
payload["debug"] = True
response = self.session.post(
f"{self.base_url}/search/",
headers={"Content-Type": "application/json", "X-Tenant-ID": self.tenant_id},
json=payload,
timeout=120,
)
response.raise_for_status()
return response.json()
class RerankServiceClient:
def __init__(self, service_url: str):
self.service_url = service_url.rstrip("/")
self.session = requests.Session()
def rerank(self, query: str, docs: Sequence[str], normalize: bool = False, top_n: Optional[int] = None) -> Tuple[List[float], Dict[str, Any]]:
payload: Dict[str, Any] = {
"query": query,
"docs": list(docs),
"normalize": normalize,
}
if top_n is not None:
payload["top_n"] = int(top_n)
response = self.session.post(self.service_url, json=payload, timeout=180)
response.raise_for_status()
data = response.json()
return list(data.get("scores") or []), dict(data.get("meta") or {})
class DashScopeLabelClient:
"""DashScope OpenAI-compatible chat: synchronous or Batch File API (JSONL job).
Batch flow: https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/
Some regional endpoints (e.g. ``dashscope-us`` compatible-mode) do not implement ``/batches``;
on HTTP 404 from batch calls we fall back to synchronous ``/chat/completions`` and stop using batch
for subsequent requests on this client.
"""
def __init__(
self,
model: str,
base_url: str,
api_key: str,
batch_size: int = 40,
*,
batch_completion_window: str = "24h",
batch_poll_interval_sec: float = 10.0,
enable_thinking: bool = True,
use_batch: bool = False,
):
self.model = model
self.base_url = base_url.rstrip("/")
self.api_key = api_key
self.batch_size = int(batch_size)
self.batch_completion_window = str(batch_completion_window)
self.batch_poll_interval_sec = float(batch_poll_interval_sec)
self.enable_thinking = bool(enable_thinking)
self.use_batch = bool(use_batch)
self.session = requests.Session()
def _auth_headers(self) -> Dict[str, str]:
return {"Authorization": f"Bearer {self.api_key}"}
def _completion_body(self, prompt: str) -> Dict[str, Any]:
body: Dict[str, Any] = {
"model": self.model,
"messages": [{"role": "user", "content": prompt}],
"temperature": 0,
"top_p": 0.1,
"enable_thinking": self.enable_thinking,
}
return body
def _chat_sync(self, prompt: str) -> Tuple[str, str]:
response = self.session.post(
f"{self.base_url}/chat/completions",
headers={**self._auth_headers(), "Content-Type": "application/json"},
json=self._completion_body(prompt),
timeout=180,
)
response.raise_for_status()
data = response.json()
content = str(((data.get("choices") or [{}])[0].get("message") or {}).get("content") or "").strip()
return content, safe_json_dumps(data)
def _chat_batch(self, prompt: str) -> Tuple[str, str]:
"""One chat completion via Batch File API (single-line JSONL job)."""
custom_id = uuid.uuid4().hex
body = self._completion_body(prompt)
line_obj = {
"custom_id": custom_id,
"method": "POST",
"url": "/v1/chat/completions",
"body": body,
}
jsonl = json.dumps(line_obj, ensure_ascii=False, separators=(",", ":")) + "\n"
auth = self._auth_headers()
up = self.session.post(
f"{self.base_url}/files",
headers=auth,
files={
"file": (
"eval_batch_input.jsonl",
io.BytesIO(jsonl.encode("utf-8")),
"application/octet-stream",
)
},
data={"purpose": "batch"},
timeout=300,
)
up.raise_for_status()
file_id = (up.json() or {}).get("id")
if not file_id:
raise RuntimeError(f"DashScope file upload returned no id: {up.text!r}")
cr = self.session.post(
f"{self.base_url}/batches",
headers={**auth, "Content-Type": "application/json"},
json={
"input_file_id": file_id,
"endpoint": "/v1/chat/completions",
"completion_window": self.batch_completion_window,
},
timeout=120,
)
cr.raise_for_status()
batch_payload = cr.json() or {}
batch_id = batch_payload.get("id")
if not batch_id:
raise RuntimeError(f"DashScope batches.create returned no id: {cr.text!r}")
terminal = frozenset({"completed", "failed", "expired", "cancelled"})
batch: Dict[str, Any] = dict(batch_payload)
status = str(batch.get("status") or "")
while status not in terminal:
time.sleep(self.batch_poll_interval_sec)
br = self.session.get(f"{self.base_url}/batches/{batch_id}", headers=auth, timeout=120)
br.raise_for_status()
batch = br.json() or {}
status = str(batch.get("status") or "")
if status != "completed":
raise RuntimeError(
f"DashScope batch {batch_id} ended with status={status!r} errors={batch.get('errors')!r}"
)
out_id = batch.get("output_file_id")
err_id = batch.get("error_file_id")
row = self._find_batch_line_for_custom_id(out_id, custom_id, auth)
if row is None:
err_row = self._find_batch_line_for_custom_id(err_id, custom_id, auth)
if err_row is not None:
raise RuntimeError(f"DashScope batch request failed: {err_row!r}")
raise RuntimeError(f"DashScope batch output missing custom_id={custom_id!r}")
resp = row.get("response") or {}
sc = resp.get("status_code")
if sc is not None and int(sc) != 200:
raise RuntimeError(f"DashScope batch line error: {row!r}")
data = resp.get("body") or {}
content = str(((data.get("choices") or [{}])[0].get("message") or {}).get("content") or "").strip()
return content, safe_json_dumps(row)
def _chat(self, prompt: str, *, phase: str = "chat") -> Tuple[str, str]:
if not self.use_batch:
content, raw = self._chat_sync(prompt)
else:
try:
content, raw = self._chat_batch(prompt)
except requests.exceptions.HTTPError as e:
resp = getattr(e, "response", None)
if resp is not None and resp.status_code == 404:
self.use_batch = False
content, raw = self._chat_sync(prompt)
else:
raise
_log_eval_llm_verbose(
phase=phase,
model=self.model,
prompt=prompt,
assistant_text=content,
raw_response=raw,
)
return content, raw
def _find_batch_line_for_custom_id(
self,
file_id: Optional[str],
custom_id: str,
auth: Dict[str, str],
) -> Optional[Dict[str, Any]]:
if not file_id or str(file_id) in ("null", ""):
return None
r = self.session.get(f"{self.base_url}/files/{file_id}/content", headers=auth, timeout=300)
r.raise_for_status()
for raw in r.text.splitlines():
raw = raw.strip()
if not raw:
continue
try:
obj = json.loads(raw)
except json.JSONDecodeError:
continue
if str(obj.get("custom_id")) == custom_id:
return obj
return None
def query_intent(self, query: str) -> Tuple[str, str]:
prompt = intent_analysis_prompt(query)
return self._chat(prompt, phase="query_intent")
def classify_batch(
self,
query: str,
docs: Sequence[Dict[str, Any]],
*,
query_intent_block: str = "",
) -> Tuple[List[str], str]:
numbered_docs = [build_label_doc_line(idx + 1, doc) for idx, doc in enumerate(docs)]
prompt = classify_prompt(query, numbered_docs, query_intent_block=query_intent_block)
content, raw_response = self._chat(prompt, phase="relevance_classify")
labels: List[str] = []
for line in str(content or "").splitlines():
canon = _canonicalize_judge_label(line)
if canon is not None:
labels.append(canon)
if len(labels) != len(docs):
payload = extract_json_blob(content)
if isinstance(payload, dict) and isinstance(payload.get("labels"), list):
labels = []
for item in payload["labels"][: len(docs)]:
if isinstance(item, dict):
raw_l = str(item.get("label") or "").strip()
else:
raw_l = str(item).strip()
canon = _canonicalize_judge_label(raw_l)
if canon is not None:
labels.append(canon)
if len(labels) != len(docs) or any(label not in VALID_LABELS for label in labels):
raise ValueError(f"unexpected classify output: {content!r}")
return labels, raw_response