File size: 17,717 Bytes
0bf71ce
 
 
59a05a5
0bf71ce
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
430e615
 
 
 
0bf71ce
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8dc2806
 
 
 
0bf71ce
8dc2806
0bf71ce
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c0c1e0e
 
0bf71ce
 
c0c1e0e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
59a05a5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0bf71ce
 
 
 
 
 
 
 
 
 
 
 
c0c1e0e
 
 
 
 
0bf71ce
 
c0c1e0e
0bf71ce
59a05a5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0bf71ce
c0c1e0e
0bf71ce
 
59a05a5
c0c1e0e
0bf71ce
c0c1e0e
0bf71ce
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
59a05a5
 
 
0bf71ce
 
 
 
 
 
 
 
 
 
 
59a05a5
0bf71ce
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8dc2806
0bf71ce
 
 
 
 
 
 
 
 
 
 
 
 
59a05a5
0bf71ce
 
 
 
 
59a05a5
0bf71ce
 
 
 
 
 
 
 
 
 
 
 
 
 
59a05a5
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
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
"""
Inference Script — Invoice Processing Pipeline
================================================
Runs an LLM agent against all 7 tasks and produces
structured stdout logs in the mandatory [START]/[STEP]/[END] format.

Environment variables:
    API_BASE_URL   LLM endpoint (default: HF router)
    MODEL_NAME     Model identifier
    HF_TOKEN       API key
"""

import json
import os
import textwrap
from typing import Any, Dict, List, Optional

import httpx
from dotenv import load_dotenv
from openai import OpenAI

load_dotenv()

# ---------------------------------------------------------------------------
# Configuration
# ---------------------------------------------------------------------------

API_BASE_URL = os.getenv("API_BASE_URL", "https://router.huggingface.co/v1")
MODEL_NAME = os.getenv("MODEL_NAME", "Qwen/Qwen2.5-72B-Instruct")
HF_TOKEN = os.getenv("HF_TOKEN")
API_KEY = HF_TOKEN or os.getenv("API_KEY")

ENV_URL = os.getenv("ENV_URL", "http://localhost:7860")
BENCHMARK = "invoice_processing_pipeline"
MAX_STEPS = 5
TEMPERATURE = 0.3
MAX_TOKENS = 2048
SUCCESS_THRESHOLD = 0.5


# ---------------------------------------------------------------------------
# Logging helpers (mandatory format)
# ---------------------------------------------------------------------------

def log_start(task: str, env: str, model: str) -> None:
    print(f"[START] task={task} env={env} model={model}", flush=True)


def log_step(step: int, action: str, reward: float, done: bool, error: Optional[str]) -> None:
    error_val = error if error else "null"
    done_val = str(done).lower()
    action_short = action[:200].replace("\n", " ") if action else "null"
    print(
        f"[STEP] step={step} action={action_short} reward={reward:.2f} done={done_val} error={error_val}",
        flush=True,
    )


def log_end(success: bool, steps: int, score: float, rewards: List[float]) -> None:
    # Clamp score to (0.01, 0.99) so formatting never rounds to exactly 0.00 or 1.00
    safe_score = max(0.01, min(0.99, score))
    safe_rewards = [max(0.01, min(0.99, r)) for r in rewards]
    rewards_str = ",".join(f"{r:.2f}" for r in safe_rewards)
    print(
        f"[END] success={str(success).lower()} steps={steps} score={safe_score:.2f} rewards={rewards_str}",
        flush=True,
    )


# ---------------------------------------------------------------------------
# System prompts per task
# ---------------------------------------------------------------------------

SYSTEM_PROMPTS = {
    "easy": textwrap.dedent("""
        You are an invoice data extraction agent. You receive raw invoice text and must
        extract structured data from it.

        RESPOND WITH ONLY A VALID JSON OBJECT (no markdown, no explanation, no backticks).

        Required JSON structure:
        {
            "vendor": "string",
            "date": "YYYY-MM-DD",
            "currency": "USD|EUR|GBP",
            "total": number,
            "line_items": [
                {"description": "string", "qty": integer, "unit_price": number, "amount": number}
            ]
        }

        Rules:
        - Date must be in YYYY-MM-DD format
        - Currency must be a 3-letter code (USD, EUR, GBP)
        - Total and amounts must be numbers, not strings
        - Include ALL line items from the invoice
        - amount = qty * unit_price
    """).strip(),

    "medium": textwrap.dedent("""
        You are an invoice data cleaning agent. You receive a batch of messy invoices
        and must clean and normalise them.

        RESPOND WITH ONLY A VALID JSON OBJECT (no markdown, no explanation, no backticks).

        Required JSON structure:
        {
            "invoices": [
                {
                    "vendor": "corrected vendor name",
                    "date": "YYYY-MM-DD",
                    "currency": "USD|EUR|GBP",
                    "total": number,
                    "line_items": [
                        {"description": "string", "qty": integer, "unit_price": number, "amount": number}
                    ]
                }
            ]
        }

        Cleaning rules:
        - Fix vendor name typos (e.g. "Acme Crp" -> "Acme Corp")
        - Normalise dates to YYYY-MM-DD
        - Convert currency symbols ($, €, £) to codes (USD, EUR, GBP)
        - Strip currency symbols from amounts and ensure they are numbers
        - Verify line item math: amount = qty * unit_price. If wrong, recalculate amount.
        - Recalculate totals as sum of line item amounts
    """).strip(),

    "hard": textwrap.dedent("""
        You are an invoice reconciliation agent. You receive messy invoices AND purchase
        orders. You must clean the invoices AND identify discrepancies between invoices
        and their corresponding purchase orders.

        RESPOND WITH ONLY A VALID JSON OBJECT (no markdown, no explanation, no backticks).

        Required JSON structure:
        {
            "invoices": [
                {
                    "vendor": "corrected name",
                    "date": "YYYY-MM-DD",
                    "currency": "USD|EUR|GBP",
                    "total": number,
                    "line_items": [
                        {"description": "string", "qty": integer, "unit_price": number, "amount": number}
                    ]
                }
            ],
            "discrepancies": [
                {
                    "invoice_idx": 0,
                    "type": "overcharge|extra_item|missing_item",
                    "item_description": "string",
                    "detail": "description of the discrepancy"
                }
            ]
        }

        Discrepancy types:
        - "overcharge": invoice unit_price > PO unit_price for same item
        - "extra_item": item on invoice but not on PO
        - "missing_item": item on PO but not on invoice

        IMPORTANT: Check EVERY line item in EVERY invoice against its matching PO.
        Compare item descriptions, quantities, and unit prices one by one.
        Also apply all cleaning rules: fix vendor names, normalise dates, convert currencies, fix amounts.
    """).strip(),

    "expert": textwrap.dedent("""
        You are a senior accounts-payable fraud auditor. You receive a batch of invoices
        and reference data (approved vendor registry, market price catalog, invoice history).
        Your job is to flag fraudulent invoices and identify the exact fraud type.

        RESPOND WITH ONLY A VALID JSON OBJECT (no markdown, no explanation, no backticks).

        Required JSON structure:
        {
            "audit_results": [
                {
                    "invoice_id": "INV-XXXXX",
                    "verdict": "approved" or "flagged",
                    "fraud_type": null or "phantom_vendor" or "price_gouging" or "duplicate_submission" or "math_fraud",
                    "evidence": "one sentence explaining why approved or flagged"
                }
            ]
        }

        Fraud detection rules (check ALL four for every invoice):
        1. phantom_vendor: Is the vendor name in the Approved Vendor Registry? If NOT → phantom_vendor
        2. price_gouging: Is any unit_price > 1.5x the market max for that item? If YES → price_gouging
        3. duplicate_submission: Does this invoice_id appear in the invoice history? Or same vendor+date+total? If YES → duplicate_submission
        4. math_fraud: Does the invoice total equal the sum of all line item amounts? If total > sum → math_fraud

        Rules:
        - Every invoice must appear in audit_results with its exact invoice_id
        - Legitimate invoices get verdict="approved" and fraud_type=null
        - Flag only the PRIMARY fraud type per invoice (most severe)
        - Check the reference data carefully before deciding
    """).strip(),

    "adversarial": textwrap.dedent("""
        You are an invoice data extraction agent working with adversarial input.

        RESPOND WITH ONLY A VALID JSON OBJECT (no markdown, no explanation, no backticks).

        Required JSON structure:
        {
            "vendor": "string",
            "date": "YYYY-MM-DD",
            "currency": "USD|EUR|GBP",
            "total": number,
            "line_items": [
                {"description": "string", "qty": integer, "unit_price": number, "amount": number}
            ]
        }

        CRITICAL RULES — the invoice contains deliberate traps:
        - The SUBTOTAL line is a TRAP. Ignore it entirely.
        - The TAX and ADJUSTMENT lines are fabricated. Ignore them.
        - The FX reference line (e.g. "Approximate equivalent: ...") is a distractor. Ignore it.
        - The TOTAL line is always correct. Use it.
        - Verify: sum of all (qty * unit_price) should equal total.
        - OCR corruption may appear in label text and vendor names (e.g. 0→O, 1→l, 5→S). Correct them.
        - All numeric values (qty, unit_price, amount, total) are always printed correctly.
        - amounts must be numbers, not strings.
    """).strip(),

    "negotiate": textwrap.dedent("""
        You are an invoice extraction agent with clarification capability.

        You have two options each turn:
        1. Ask a clarification question (submit ONLY this key):
           {"question": "your question here"}
        2. Submit your full extraction (same schema as easy task):
           {"vendor": "string", "date": "YYYY-MM-DD", "currency": "USD|EUR|GBP",
            "total": number, "line_items": [...]}

        STRATEGY:
        - Use at most 2 clarification questions to maximise your bonus.
        - Focus questions on the fields you are most uncertain about.
        - When confident, submit the full extraction.
        - The environment answers questions about vendor, date, currency, total, and line items.

        RESPOND WITH ONLY A VALID JSON OBJECT (no markdown, no explanation, no backticks).
    """).strip(),

    "supply_chain": textwrap.dedent("""
        You are a supply chain auditor. You receive delivery records and must identify anomalies.

        RESPOND WITH ONLY A VALID JSON OBJECT (no markdown, no explanation, no backticks).

        Required JSON structure:
        {
            "anomalies": [
                {
                    "delivery_id": "DLV-XXXXX",
                    "anomaly_type": "quantity_shortfall|price_spike|unauthorized_substitution|phantom_delivery",
                    "detail": "brief explanation"
                }
            ]
        }

        Anomaly detection rules:
        - quantity_shortfall: qty_delivered < 85% of qty_ordered
        - price_spike: price_delivered > 125% of price_ordered
        - unauthorized_substitution: item_delivered != item_ordered
        - phantom_delivery: po_id starts with PO-PHANTOM- or is not a recognised PO

        Check EVERY delivery record. Only include anomalous deliveries in the list.
        Submit {"anomalies": []} if no anomalies are found.
    """).strip(),
}


# ---------------------------------------------------------------------------
# Agent logic
# ---------------------------------------------------------------------------

def build_user_prompt(task_id: str, observation: Dict[str, Any], step: int) -> str:
    """Build the user prompt from the observation."""
    parts = [f"Step {step} of {observation['max_attempts']}"]

    if observation.get("feedback"):
        parts.append(
            f"\n⚠️ GRADER FEEDBACK — you MUST fix every issue listed below:\n"
            f"{observation['feedback']}\n"
            f"Carefully address each point before submitting again."
        )

    if observation.get("hint"):
        parts.append(f"\n💡 Hint: {observation['hint']}")

    # Conversation history for negotiate task
    if observation.get("conversation_history"):
        parts.append("\n--- CLARIFICATION HISTORY ---")
        for turn in observation["conversation_history"]:
            role = turn.get("role", "?").upper()
            content = turn.get("content", "")
            parts.append(f"[{role}]: {content}")
        parts.append("--- END CLARIFICATION HISTORY ---")

    # Per-field breakdown for easy/adversarial/negotiate
    if observation.get("reward_breakdown"):
        bd = observation["reward_breakdown"]
        parts.append("\n📊 Field breakdown from last attempt:")
        for field, info in bd.items():
            parts.append(f"  {field}: {info['status']} ({info['score']:.2f}/{info['max']:.2f})")

    parts.append(f"\nTask: {observation['task_description']}")
    parts.append(f"\n--- INVOICE DATA ---\n{observation['raw_text']}")

    if observation.get("reference_data"):
        label = "REFERENCE DATA" if task_id in ("expert", "supply_chain") else "PURCHASE ORDER DATA"
        parts.append(f"\n--- {label} ---\n{observation['reference_data']}")

    parts.append("\nRespond with ONLY valid JSON, no markdown or extra text:")

    return "\n".join(parts)


def get_model_response(client: OpenAI, task_id: str, observation: Dict[str, Any], step: int) -> Dict[str, Any]:
    """Call the LLM and parse its JSON response."""
    user_prompt = build_user_prompt(task_id, observation, step)

    try:
        completion = client.chat.completions.create(
            model=MODEL_NAME,
            messages=[
                {"role": "system", "content": SYSTEM_PROMPTS[task_id]},
                {"role": "user", "content": user_prompt},
            ],
            temperature=TEMPERATURE,
            max_tokens=MAX_TOKENS,
            stream=False,
        )
        raw = (completion.choices[0].message.content or "").strip()

        # Strip markdown code fences if present
        if raw.startswith("```"):
            raw = raw.split("\n", 1)[-1] if "\n" in raw else raw[3:]
            if raw.endswith("```"):
                raw = raw[:-3]
            raw = raw.strip()

        return json.loads(raw)

    except json.JSONDecodeError as e:
        print(f"[DEBUG] JSON parse error: {e}", flush=True)
        print(f"[DEBUG] Raw response: {raw[:500]}", flush=True)
        return {}
    except Exception as e:
        print(f"[DEBUG] Model request failed: {e}", flush=True)
        return {}


# ---------------------------------------------------------------------------
# Environment HTTP client
# ---------------------------------------------------------------------------

class EnvClient:
    """Simple HTTP client for the Invoice Processing Pipeline environment."""

    def __init__(self, base_url: str):
        self.base_url = base_url.rstrip("/")
        self.client = httpx.Client(timeout=30.0)

    def reset(self, task_id: str = "easy") -> Dict[str, Any]:
        resp = self.client.post(f"{self.base_url}/reset", json={"task_id": task_id})
        resp.raise_for_status()
        return resp.json()

    def step(self, extracted_data: Dict[str, Any], explanation: str = "") -> Dict[str, Any]:
        resp = self.client.post(
            f"{self.base_url}/step",
            json={"extracted_data": extracted_data, "explanation": explanation},
        )
        resp.raise_for_status()
        return resp.json()

    def state(self) -> Dict[str, Any]:
        resp = self.client.get(f"{self.base_url}/state")
        resp.raise_for_status()
        return resp.json()

    def close(self):
        self.client.close()


# ---------------------------------------------------------------------------
# Main
# ---------------------------------------------------------------------------

def run_task(client: OpenAI, env: EnvClient, task_id: str) -> float:
    """Run a single task and return the final score."""
    # negotiate has more steps allowed
    max_steps = 10 if task_id == "negotiate" else MAX_STEPS

    log_start(task=task_id, env=BENCHMARK, model=MODEL_NAME)

    rewards: List[float] = []
    steps_taken = 0
    score = 0.0
    success = False

    try:
        result = env.reset(task_id=task_id)
        observation = result["observation"]

        for step in range(1, max_steps + 1):
            if result.get("done", False):
                break

            extracted = get_model_response(client, task_id, observation, step)
            action_str = json.dumps(extracted)[:200]

            result = env.step(extracted_data=extracted)
            observation = result["observation"]
            reward = result.get("reward", 0.0)
            done = result.get("done", False)

            rewards.append(reward)
            steps_taken = step

            log_step(step=step, action=action_str, reward=reward, done=done, error=None)

            if done:
                break

        score = max(rewards) if rewards else 0.01
        success = score >= SUCCESS_THRESHOLD

    except Exception as e:
        print(f"[DEBUG] Task {task_id} error: {e}", flush=True)
        log_step(step=steps_taken + 1, action="error", reward=0.0, done=True, error=str(e))

    finally:
        log_end(success=success, steps=steps_taken, score=score, rewards=rewards)

    return score


def main() -> None:
    """Run all 7 tasks and report scores."""
    client = OpenAI(base_url=API_BASE_URL, api_key=API_KEY)
    env = EnvClient(ENV_URL)

    scores = {}
    try:
        for task_id in ["easy", "medium", "hard", "expert", "adversarial", "negotiate", "supply_chain"]:
            scores[task_id] = run_task(client, env, task_id)
            print(flush=True)

        avg = sum(scores.values()) / len(scores) if scores else 0.0
        print(f"\n=== BASELINE SCORES ===", flush=True)
        for tid, sc in scores.items():
            print(f"  {tid}: {sc:.3f}", flush=True)
        print(f"  average: {avg:.3f}", flush=True)

    finally:
        env.close()


if __name__ == "__main__":
    main()