File size: 14,103 Bytes
50efdc6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
"""
src/preprocess/email_preprocessor.py
LangGraph ์ง„์ž… ์ „ ์ด๋ฉ”์ผ ์ „์ฒ˜๋ฆฌ โ€” ์ž์—ฐ์–ด ์ด๋ฉ”์ผ โ†’ ๊ตฌ์กฐํ™”๋œ EmailContext.

๋ชฉ์ :
  ํ˜„์žฌ router / worker_a / worker_b / synthesizer 4๊ฐœ ๋…ธ๋“œ๊ฐ€ ๊ฐ์ž LLM ํ˜ธ์ถœ๋กœ
  ๊ฐ™์€ ์ด๋ฉ”์ผ์„ ๋‹ค์‹œ ํŒŒ์‹ฑํ•œ๋‹ค(๋ˆ„๊ฐ€ ๋ณด๋ƒˆ๋Š”์ง€, ๋ฌด์—‡์„ ์š”์ฒญํ•˜๋Š”์ง€, ์–ด๋–ค ์ˆซ์ž๊ฐ€ ์žˆ๋Š”์ง€โ€ฆ).
  ์ด ์ „์ฒ˜๋ฆฌ๋ฅผ ํ•œ ๋ฒˆ ์ˆ˜ํ–‰ํ•ด์„œ EmailContext์— ๋‹ด์•„๋‘๋ฉด:
    1) ๋‹ค์šด์ŠคํŠธ๋ฆผ ๋…ธ๋“œ์˜ LLM ํ˜ธ์ถœ์ด ์งง์€ hint๋ฅผ ๋ฐ›์•„ ์ •ํ™•๋„/์ผ๊ด€์„ฑ ํ–ฅ์ƒ
    2) ๊ฐ™์€ ์ถ”๋ก ์„ 4๋ฒˆ ๋ฐ˜๋ณตํ•˜๋Š” ๋‚ญ๋น„ ๊ฐ์†Œ
    3) ๋””๋ฒ„๊น… ๊ฐ€์‹œ์„ฑ ํ–ฅ์ƒ (์ด๋ฉ”์ผ ํ•œ ํ†ต์ด ์–ด๋–ป๊ฒŒ ํŒŒ์‹ฑ๋๋Š”์ง€ ํ•œ๊ณณ์—์„œ ํ™•์ธ)

์„ค๊ณ„:
  - LangGraph ์™ธ๋ถ€์—์„œ ๋™์ž‘ (graph_builder / api/server / eval ์ง„์ž…์ ์—์„œ ํ˜ธ์ถœ).
  - ๊ฒฐ๊ณผ๋Š” AgentState["email_context"]์— dict๋กœ ์ ์žฌ. ๋…ธ๋“œ๋“ค์€ ์žˆ์œผ๋ฉด ํ™œ์šฉ, ์—†์œผ๋ฉด
    ๊ธฐ์กด ๋™์ž‘(์›๋ณธ user_input ํŒŒ์‹ฑ)์œผ๋กœ ์ž๋™ fallback โ€” ํ•˜์œ„ํ˜ธํ™˜.
  - ์‹คํŒจํ•ด๋„ ๊ทธ๋ž˜ํ”„ ํ๋ฆ„์€ ์ค‘๋‹จํ•˜์ง€ ๋ง ๊ฒƒ. EmailContext.preprocess_ok=False๋กœ ํ‘œ์‹œ๋งŒ.
"""

from __future__ import annotations

import logging
import os
import re
from typing import Optional

from dotenv import load_dotenv
load_dotenv()

from langchain_core.prompts import ChatPromptTemplate
from langchain_openai import ChatOpenAI
from pydantic import BaseModel, Field

from src.config import get_config

logger = logging.getLogger(__name__)


# ---------------------------------------------------------------------------
# ๊ตฌ์กฐํ™”๋œ ๊ฒฐ๊ณผ ์Šคํ‚ค๋งˆ
# ---------------------------------------------------------------------------

class EmailContext(BaseModel):
    """์ „์ฒ˜๋ฆฌ ๊ฒฐ๊ณผ โ€” LangGraph ๋…ธ๋“œ๋“ค์ด user_input๊ณผ ํ•จ๊ป˜ ์ฐธ๊ณ ํ•˜๋Š” ๊ตฌ์กฐํ™” ์ •๋ณด."""

    # ์›๋ฌธ (๊ทธ๋Œ€๋กœ ๋ณด์กด โ€” ๋‹ค์šด์ŠคํŠธ๋ฆผ ๋…ธ๋“œ๊ฐ€ ํ•„์š” ์‹œ ์ฐธ๊ณ )
    raw_text: str = Field(..., description="์›๋ณธ ์ด๋ฉ”์ผ ํ…์ŠคํŠธ (์ˆ˜์ • ์—†์Œ)")

    # ๋ฉ”ํƒ€๋ฐ์ดํ„ฐ (LLM์ด ์ถ”์ถœ, ์—†์œผ๋ฉด ๋นˆ ๋ฌธ์ž์—ด)
    sender_name:    str = Field("", description="๋ฐœ์‹ ์ž ์ด๋ฆ„ (์„œ๋ช…/From ๋ผ์ธ ๋“ฑ์—์„œ ์ถ”์ถœ). ๋ชจ๋ฅด๋ฉด ๋นˆ ๋ฌธ์ž์—ด.")
    sender_email:   str = Field("", description="๋ฐœ์‹ ์ž ์ด๋ฉ”์ผ ์ฃผ์†Œ. ๋ชจ๋ฅด๋ฉด ๋นˆ ๋ฌธ์ž์—ด.")
    sender_company: str = Field("", description="๋ฐœ์‹ ์ž ์†Œ์† ํšŒ์‚ฌ๋ช…. ๋ชจ๋ฅด๋ฉด ๋นˆ ๋ฌธ์ž์—ด.")
    recipient:      str = Field("", description="์ˆ˜์‹ ์ž (๋‹ด๋‹น์ž/ํŒ€๋ช…). ๋ชจ๋ฅด๋ฉด ๋นˆ ๋ฌธ์ž์—ด.")
    subject:        str = Field("", description="์ œ๋ชฉ. ๋ณธ๋ฌธ์—์„œ ์ถ”์ •ํ•ด๋„ ๋จ.")
    language:       str = Field("en", description="์ด๋ฉ”์ผ ์–ธ์–ด (en/ko/ja/zh/...). ๊ธฐ๋ณธ en.")

    # ์ •์ œ๋œ ๋ณธ๋ฌธ โ€” ์ธ์‚ฌ๋ง/์„œ๋ช…/Disclaimer ์ œ์™ธ, ๋ณธ๋ฌธ ํ•ต์‹ฌ๋งŒ
    cleaned_body: str = Field(..., description="์ธ์‚ฌยท์„œ๋ช…ยท๋ฉด์ฑ…์กฐํ•ญ์„ ์ œ์™ธํ•œ ๋ณธ๋ฌธ ํ•ต์‹ฌ")

    # ์š”์ฒญ ์š”์•ฝ โ€” ํ•œ๋‘ ๋ฌธ์žฅ
    request_summary: str = Field(
        ...,
        description="์ด๋ฉ”์ผ์ด ๋ฌด์—‡์„ ์š”์ฒญํ•˜๋Š”์ง€ ํ•œ๋‘ ๋ฌธ์žฅ ์š”์•ฝ (์˜์–ด๋กœ). "
                    "์˜ˆ: 'Change quantity of item 10 on order 4500023456 to 100 units.'",
    )

    # ์ง€์‹ ์งˆ๋ฌธ๋งŒ ๋ถ„๋ฆฌยท์ •๊ทœํ™” โ€” RAG(worker_b) ๊ฒ€์ƒ‰ ์ฟผ๋ฆฌ ์ž…๋ ฅ์œผ๋กœ ์‚ฌ์šฉ
    question_summary: str = Field(
        "",
        description="์ด๋ฉ”์ผ์— SAP ์ง€์‹/๋ฐฉ๋ฒ•(how-to/์ •์ฑ…/๊ฐœ๋…) ์งˆ๋ฌธ์ด ์žˆ์œผ๋ฉด, ERP ์•ก์…˜ ๋งฅ๋ฝ์„ "
                    "๋ชจ๋‘ ์ œ๊ฑฐํ•˜๊ณ  ํ•˜๋‚˜์˜ ๋ช…ํ™•ํ•˜๊ณ  ์ž๊ธฐ์™„๊ฒฐ์ ์ธ ์˜์–ด ์งˆ๋ฌธ์œผ๋กœ ์žฌ์ž‘์„ฑ. ํŠน์ • ์ฃผ๋ฌธ/์•„์ดํ…œ "
                    "๋ฒˆํ˜ธ๋‚˜ ์ˆ˜๋Ÿ‰ ๊ฐ™์€ ํŠธ๋žœ์žญ์…˜ ๋””ํ…Œ์ผ์€ ๋นผ๊ณ  ์ผ๋ฐ˜ํ™”ํ•œ๋‹ค. ์ง€์‹ ์งˆ๋ฌธ์ด ์—†์œผ๋ฉด ๋นˆ ๋ฌธ์ž์—ด. "
                    "์˜ˆ: 'How do I view the incompletion log for a sales order before creating an outbound delivery?'",
    )

    # โ”€โ”€ ๋น ๋ฅธ ๋ผ์šฐํŒ… ํžŒํŠธ โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€
    mentions_action: bool = Field(
        False,
        description="ERP ์ˆ˜์ • ์š”์ฒญ(์ˆ˜๋Ÿ‰/๋‚ ์งœ/์ทจ์†Œ/์ฃผ์†Œ ๋ณ€๊ฒฝ ๋“ฑ)์ด ๋ช…์‹œ๋˜์–ด ์žˆ๋Š”๊ฐ€",
    )
    mentions_question: bool = Field(
        False,
        description="SAP/์ •์ฑ…/๋งค๋‰ด์–ผ์— ๋Œ€ํ•œ ์ง€์‹/๋ฌธ์˜ ์งˆ๋ฌธ์ด ๋ช…์‹œ๋˜์–ด ์žˆ๋Š”๊ฐ€",
    )

    # โ”€โ”€ ์‚ฌ์ „ ์ถ”์ถœ๋œ ํ•ต์‹ฌ ์—”ํ‹ฐํ‹ฐ (๋‹ค์šด์ŠคํŠธ๋ฆผ์ด hint๋กœ ํ™œ์šฉ) โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€
    order_ids: list[str] = Field(
        default_factory=list,
        description="์ด๋ฉ”์ผ์— ๋ช…์‹œ๋œ ๋ชจ๋“  ์˜์—… ์˜ค๋” ๋ฒˆํ˜ธ(VBELN) ํ›„๋ณด โ€” ์›๋ณธ ์ˆซ์ž ๊ทธ๋Œ€๋กœ, ํŒจ๋”ฉ ์—†์ด",
    )
    item_nos: list[str] = Field(
        default_factory=list,
        description="์ด๋ฉ”์ผ์— ๋ช…์‹œ๋œ ๋ชจ๋“  ์•„์ดํ…œ ๋ฒˆํ˜ธ(POSNR) ํ›„๋ณด โ€” ์›๋ณธ ์ˆซ์ž ๊ทธ๋Œ€๋กœ",
    )

    # โ”€โ”€ ๋ฉ”ํƒ€ โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€
    preprocess_ok: bool = Field(True, description="์ „์ฒ˜๋ฆฌ ์„ฑ๊ณต ์—ฌ๋ถ€ (LLM ์‹คํŒจ ์‹œ False)")
    error: str = Field("", description="์‹คํŒจ ์‚ฌ์œ  (preprocess_ok=False์ผ ๋•Œ๋งŒ)")


# ---------------------------------------------------------------------------
# Prompt
# ---------------------------------------------------------------------------

_SYSTEM_PROMPT = """\
You are a B2B email preprocessor for an SAP ERP support pipeline. Your job is to
read a single customer email and produce a structured summary that downstream
agents (router, ERP action extractor, RAG question extractor, reply composer)
will consume.

Extract the following fields from the email and output ONLY valid JSON matching
the required schema.

Field guidance:
  โ€ข sender_name / sender_email / sender_company:
      Look in the signature block ("Best regards, John Smith / john@acme.com /
      Acme Corp"), or in "From:" headers if present. If a field is not
      recoverable from the email, return an empty string. Do NOT invent.
  โ€ข recipient:
      The addressed party ("Hi Support Team", "Dear SAP Desk"). Empty if unclear.
  โ€ข subject:
      The Subject: line if present, otherwise a short topic phrase you infer
      from the body (under 80 chars).
  โ€ข language:
      Two-letter code: "en", "ko", "ja", "zh", "de", "es", "fr". Default "en".
  โ€ข cleaned_body:
      The substantive body โ€” strip greetings ("Dear ..."), sign-offs ("Best
      regards, ..."), legal disclaimers, and "Sent from my iPhone"-style
      footers. Keep the actual request and any context the customer provides.
      Preserve order numbers, item numbers, quantities, dates exactly as
      written.
  โ€ข request_summary:
      ONE or TWO sentences in ENGLISH that describe what the customer wants.
      Be specific โ€” include order number, item number, action, and quantities
      or dates if mentioned. Examples:
        "Change quantity of item 10 on order 4500023456 to 100 units."
        "Reduce quantity of item 20 on order 6105 by 50, and ask about the
         late-delivery penalty policy."
  โ€ข question_summary:
      If (and only if) the email contains an SAP knowledge / how-to / policy /
      concept question, restate THAT question as ONE clear, self-contained
      question in ENGLISH โ€” optimized as a documentation search query.
      Remove ONLY the transaction instance data (order/item numbers, quantities,
      dates) and the action request itself. Otherwise stay close to the
      customer's own wording: PRESERVE, verbatim, every SAP term, feature /
      screen / field name, and concept keyword they used โ€” these are the search
      keys the retriever matches on. Do NOT paraphrase a specific SAP term into a
      generic one, and do NOT swap in a related but different concept for the one
      asked about. If the question has two distinct parts, keep both. If there is
      no knowledge question, return an empty string.
  โ€ข mentions_action:
      true if the email asks to MODIFY anything in the ERP โ€” change quantity,
      change delivery date, cancel an item, change shipping address, unblock a
      delivery, change carrier/batch/payment terms. false if it is a pure
      knowledge question.
  โ€ข mentions_question:
      true if the email asks about HOW SAP works / what a policy says / how to
      configure something / definitions โ€” i.e. knowledge-base questions answered
      from documentation. false if it is purely a transaction request.
      mentions_action and mentions_question are NOT mutually exclusive โ€” an
      email can have both.
  โ€ข order_ids:
      EVERY distinct order number (VBELN) mentioned. Copy digits EXACTLY as
      written โ€” do NOT zero-pad, do NOT add or drop digits. Numbers may
      appear after "order", "order #", "sales order", "PO". Return as strings,
      no commas/dashes. Empty list if none.
  โ€ข item_nos:
      EVERY distinct item / line / position number (POSNR) mentioned. Copy
      digits EXACTLY. Numbers may appear after "item", "line", "position".
      Empty list if none.

CRITICAL โ€” number handling:
  A single email often contains the order number, the item number, AND a
  quantity. They are DIFFERENT things. Map each to the correct field. Never
  put a quantity ("100 units", "50 pcs") into order_ids or item_nos.

Output ONLY valid JSON matching the schema. Set raw_text to the EXACT input
email (unchanged). Set preprocess_ok=true and error="" โ€” these are only set to
false by the calling code on failure.
"""

_HUMAN_TEMPLATE = "Email to preprocess:\n\n{email}"

_PROMPT = ChatPromptTemplate.from_messages([
    ("system", _SYSTEM_PROMPT),
    ("human",  _HUMAN_TEMPLATE),
])


# ---------------------------------------------------------------------------
# LLM factory (OpenRouter)
# ---------------------------------------------------------------------------

def _build_llm() -> ChatOpenAI:
    """์ „์ฒ˜๋ฆฌ LLM. configs.yaml์˜ models.preprocessor๋ฅผ ์šฐ์„  ์‚ฌ์šฉํ•˜๊ณ ,
    ์„ค์ •์ด ์—†์œผ๋ฉด worker_a(์ด๋ฏธ ๊ตฌ์กฐํ™” ์ถ”์ถœ์— ์ž˜ ๋™์ž‘)์™€ ๋™์ผํ•œ ๋ชจ๋ธ๋กœ ํด๋ฐฑ."""
    cfg = get_config()

    # ๋™์  lookup โ€” ์ƒˆ ์„ค์ • ํ‚ค๊ฐ€ ์—†์–ด๋„ ๋™์ž‘ํ•˜๋„๋ก
    preproc_cfg = getattr(cfg.models, "preprocessor", None) or cfg.models.worker_a

    api_key = os.getenv("OPENROUTER_API_KEY", "")
    if not api_key:
        raise EnvironmentError("OPENROUTER_API_KEY is not set. Check your .env file.")

    return ChatOpenAI(
        model=preproc_cfg.name,
        temperature=preproc_cfg.temperature,
        openai_api_key=api_key,
        openai_api_base=cfg.openrouter.base_url,
        default_headers={
            "HTTP-Referer": "https://github.com/daisysooyeon/SAP-ERP-AI-Agent",
            "X-Title": "SAP ERP AI Agent - Preprocessor",
        },
    )


_llm: Optional[ChatOpenAI] = None
_chain = None


def _get_chain():
    """LLM ์ฒด์ธ์„ lazyํ•˜๊ฒŒ ์ƒ์„ฑ โ€” import ์‹œ์ ์— API ํ‚ค ๊ฒ€์ฆ์„ ๋ฏธ๋ฃจ๊ธฐ ์œ„ํ•จ."""
    global _llm, _chain
    if _chain is None:
        _llm = _build_llm()
        _chain = _PROMPT | _llm.with_structured_output(EmailContext)
    return _chain


# ---------------------------------------------------------------------------
# Regex ๊ธฐ๋ฐ˜ backstop entity ์ถ”์ถœ (LLM ์‹คํŒจ/๋ˆ„๋ฝ ๋ณด์กฐ)
# ---------------------------------------------------------------------------

_ORDER_RE = re.compile(
    r"(?:order|order\s*#|sales\s+order|PO)\s*[:#]?\s*(\d{3,10})",
    re.IGNORECASE,
)
_ITEM_RE = re.compile(
    r"(?:item|line|position)\s*[:#]?\s*(\d{1,6})",
    re.IGNORECASE,
)


def _regex_entities(text: str) -> tuple[list[str], list[str]]:
    """LLM์ด ์‹คํŒจํ•˜๊ฑฐ๋‚˜ ๋น ๋œจ๋ฆด ๋•Œ๋ฅผ ๋Œ€๋น„ํ•œ ์ •๊ทœ์‹ backstop. ์ค‘๋ณต ์ œ๊ฑฐ."""
    order_ids = list(dict.fromkeys(_ORDER_RE.findall(text)))
    item_nos  = list(dict.fromkeys(_ITEM_RE.findall(text)))
    return order_ids, item_nos


# ---------------------------------------------------------------------------
# Public API
# ---------------------------------------------------------------------------

def preprocess_email(raw_email: str) -> EmailContext:
    """์ด๋ฉ”์ผ์„ LLM ํ•œ ๋ฒˆ ํ˜ธ์ถœ๋กœ ๊ตฌ์กฐํ™”๋œ EmailContext๋กœ ๋ณ€ํ™˜.

    LLM์ด ์‹คํŒจํ•ด๋„ EmailContext.preprocess_ok=False๋กœ ํ‘œ์‹œํ•˜๊ณ  ์ •๊ทœ์‹ backstop์œผ๋กœ
    ์ตœ์†Œํ•œ์˜ ์ •๋ณด๋ฅผ ์ฑ„์›Œ ๋ฐ˜ํ™˜ํ•œ๋‹ค. ํ˜ธ์ถœ์ž(graph entry point)๋Š” ๊ทธ๋ƒฅ ํ†ต๊ณผ์‹œํ‚ค๋ฉด
    ๋˜๊ณ , ๋‹ค์šด์ŠคํŠธ๋ฆผ ๋…ธ๋“œ๋Š” ๋นˆ ํ•„๋“œ๋ฅผ ๋ณด๋ฉด ๊ธฐ์กด user_input ํŒŒ์‹ฑ์œผ๋กœ ํด๋ฐฑํ•œ๋‹ค.
    """
    raw_email = raw_email or ""
    logger.info("[preprocess] Preprocessing email (%d chars)โ€ฆ", len(raw_email))

    try:
        chain = _get_chain()
        result: EmailContext = chain.invoke({"email": raw_email})

        # LLM์ด raw_text๋ฅผ ๋น„์šฐ๊ฑฐ๋‚˜ ๋ณ€ํ˜•ํ–ˆ์„ ์ˆ˜ ์žˆ์œผ๋ฏ€๋กœ ๊ฒฐ์ •๋ก ์ ์œผ๋กœ ๋ฎ์–ด์“ด๋‹ค
        result.raw_text = raw_email

        # LLM์ด ์—”ํ‹ฐํ‹ฐ๋ฅผ ๋น ๋œจ๋ ธ์œผ๋ฉด ์ •๊ทœ์‹์œผ๋กœ ๋ณด๊ฐ• (๋ฎ์–ด์“ฐ์ง€ ์•Š๊ณ  union)
        rx_orders, rx_items = _regex_entities(raw_email)
        if not result.order_ids and rx_orders:
            result.order_ids = rx_orders
        if not result.item_nos and rx_items:
            result.item_nos = rx_items

        logger.info(
            "[preprocess] OK | sender=%r action=%s question=%s "
            "orders=%s items=%s lang=%s",
            result.sender_name, result.mentions_action,
            result.mentions_question, result.order_ids, result.item_nos,
            result.language,
        )
        return result

    except Exception as e:
        # ์‹คํŒจํ•ด๋„ ํ๋ฆ„์€ ๋Š์ง€ ์•Š๋Š”๋‹ค โ€” backstop์œผ๋กœ ์ตœ์†Œ ์ •๋ณด ์ฑ„์›Œ ๋ฐ˜ํ™˜
        logger.error("[preprocess] FAILED: %s โ€” falling back to regex backstop only", e)
        rx_orders, rx_items = _regex_entities(raw_email)
        return EmailContext(
            raw_text=raw_email,
            cleaned_body=raw_email,         # ์›๋ณธ์„ ๊ทธ๋Œ€๋กœ ์œ ์ง€
            request_summary="",
            order_ids=rx_orders,
            item_nos=rx_items,
            preprocess_ok=False,
            error=f"{type(e).__name__}: {e}",
        )