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
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679"""Job service for job status management."""
import json
import logging
import time
from datetime import UTC, datetime
from typing import Any
from redis.asyncio import Redis
from ..config import settings
from ..shared.constants.redis_keys import JOBS_BY_UPDATED
from ..shared.constants.statuses import STATUS_COMPLETED, STATUS_DENIED, STATUS_FAILED
from ..utils.token_generator import generate_secure_token
logger = logging.getLogger(__name__)
# ---------------------------------------------------------------------------
# Lua scripts – executed atomically by Redis to avoid race conditions between
# HSET, EXPIRE, and ZADD when the process could crash between calls.
# ---------------------------------------------------------------------------
# Atomically: HSET fields + EXPIRE ttl + ZADD to secondary index
# KEYS[1] = job hash key, KEYS[2] = jobs-by-updated sorted set key
# ARGV[1] = ttl, ARGV[2] = timestamp (score), ARGV[3] = job_id
# ARGV[4..N] = field/value pairs (must be even count)
_LUA_HSET_EXPIRE_ZADD = """
local key = KEYS[1]
local idx = KEYS[2]
local ttl = tonumber(ARGV[1])
local ts = tonumber(ARGV[2])
local jid = ARGV[3]
for i = 4, #ARGV, 2 do
redis.call('HSET', key, ARGV[i], ARGV[i+1])
end
redis.call('EXPIRE', key, ttl)
redis.call('ZADD', idx, ts, jid)
return 1
"""
# Atomically: HSET fields + ZADD to secondary index (no TTL change)
# KEYS[1] = job hash key, KEYS[2] = jobs-by-updated sorted set key
# ARGV[1] = timestamp (score), ARGV[2] = job_id
# ARGV[3..N] = field/value pairs
_LUA_HSET_ZADD = """
local key = KEYS[1]
local idx = KEYS[2]
local ts = tonumber(ARGV[1])
local jid = ARGV[2]
for i = 3, #ARGV, 2 do
redis.call('HSET', key, ARGV[i], ARGV[i+1])
end
redis.call('ZADD', idx, ts, jid)
return 1
"""
# Atomically: DEL job hash + ZREM from secondary index
# KEYS[1] = job hash key, KEYS[2] = jobs-by-updated sorted set key
# ARGV[1] = job_id
_LUA_DELETE_JOB = """
local deleted = redis.call('DEL', KEYS[1])
redis.call('ZREM', KEYS[2], ARGV[1])
return deleted
"""
class JobService:
"""Service for managing job status and metadata."""
def __init__(self, redis_client: Redis) -> None:
"""Initialize job service with Redis client.
Args:
redis_client: Redis async client instance
"""
self.redis = redis_client
self.status_prefix = settings.job_status_prefix
self.jobs_index_key = JOBS_BY_UPDATED
# TTL settings from config
self.job_ttl_active = settings.job_ttl_active
self.job_ttl_completed = settings.job_ttl_completed
self.job_ttl_failed = settings.job_ttl_failed
self.job_ttl_denied = settings.job_ttl_denied
# Register Lua scripts (sync – returns callable Script objects)
self._script_hset_expire_zadd = self.redis.register_script(_LUA_HSET_EXPIRE_ZADD)
self._script_hset_zadd = self.redis.register_script(_LUA_HSET_ZADD)
self._script_delete_job = self.redis.register_script(_LUA_DELETE_JOB)
def _get_ttl_for_status(self, status: str) -> int:
"""Get appropriate TTL (time-to-live) for job based on status.
Different statuses have different retention requirements:
- Completed jobs: 30 days (for result retrieval and audit)
- Failed jobs: 30 days (for debugging and retry decisions)
- Denied jobs: 7 days (shorter retention, decision recorded)
- Active jobs (processing, awaiting_approval, etc.): 7 days
Args:
status: Job status string
Returns:
TTL in seconds for the given status
Example:
>>> ttl = self._get_ttl_for_status("completed")
>>> print(f"Completed jobs expire after {ttl / 86400} days")
"""
if status == STATUS_COMPLETED:
return self.job_ttl_completed
elif status == STATUS_FAILED:
return self.job_ttl_failed
elif status == STATUS_DENIED:
return self.job_ttl_denied
else:
# Default for active states: pii_scanning, awaiting_approval, processing
return self.job_ttl_active
@staticmethod
def _flatten_mapping(mapping: dict) -> list[str]:
"""Flatten a dict to [field, value, field, value, ...] for Lua scripts."""
flat: list[str] = []
for k, v in mapping.items():
flat.append(str(k))
flat.append(str(v))
return flat
async def create_job(
self,
job_id: str,
s3_key: str,
status: str = "pii_scanning",
original_filename: str | None = None,
pii_skipped: bool = False,
pii_skip_reason: str | None = None,
debug_bundle_requested: bool = False,
review_mode: str | None = None,
max_rounds: int = 1,
ocr_languages: list[str] | None = None,
) -> None:
"""
Create a new job in Redis with automatic TTL.
Atomically sets hash fields, TTL, and secondary index via Lua script
to prevent memory leaks if the process crashes between operations.
Args:
job_id: Unique job identifier
s3_key: S3 key where document is stored
status: Initial job status (default: "pii_scanning")
original_filename: Original filename from upload
pii_skipped: Whether PII scan was skipped (for audit trail)
pii_skip_reason: Reason PII scan was skipped
debug_bundle_requested: Whether to generate debug bundle artifacts
review_mode: Review mode ('auto' | 'human') for agentic pipeline
max_rounds: Maximum number of iterative refinement rounds (1-5, default: 1)
ocr_languages: Tesseract OCR language codes for scanned documents
"""
created_at = datetime.now(UTC).isoformat()
mapping: dict[str, str] = {
"job_id": job_id,
"s3_key": s3_key,
"status": status,
"created_at": created_at,
"updated_at": created_at,
"max_rounds": str(max_rounds),
}
if original_filename:
mapping["original_filename"] = original_filename
# Add PII skip audit fields if applicable
if pii_skipped:
mapping["pii_skipped"] = "true"
if pii_skip_reason:
mapping["pii_skip_reason"] = pii_skip_reason
# Add debug bundle flag
if debug_bundle_requested:
mapping["debug_bundle_requested"] = "true"
# Add review mode for agentic pipeline
if review_mode:
mapping["review_mode"] = review_mode
# Add OCR languages for scanned document processing
if ocr_languages:
mapping["ocr_languages"] = ",".join(ocr_languages)
ttl = self._get_ttl_for_status(status)
key = f"{self.status_prefix}{job_id}"
now_ts = time.time()
await self._script_hset_expire_zadd(
keys=[key, self.jobs_index_key],
args=[str(ttl), str(now_ts), job_id] + self._flatten_mapping(mapping),
)
async def get_job(self, job_id: str) -> dict[str, Any] | None:
"""
Get job status and metadata.
Args:
job_id: Job identifier
Returns:
Job data dictionary or None if not found
"""
job_data_raw = await self.redis.hgetall(f"{self.status_prefix}{job_id}")
if not job_data_raw:
return None
# Convert to proper dict type
job_data: dict[str, Any] = dict(job_data_raw)
# Parse JSON array/object fields only
# These fields are stored as JSON strings and need to be parsed
json_fields = [
"pii_findings", # Array of PII findings
"correction_results", # Array of correction results per page
"page_image_keys", # Array of S3 keys for page images
"verification_summary", # Object with verification phase results
]
for field in json_fields:
if field in job_data and job_data[field]:
try:
job_data[field] = json.loads(job_data[field])
except json.JSONDecodeError:
logger.warning(f"Failed to parse {field} for job {job_id}")
return job_data
async def update_job_status(self, job_id: str, status: str, **additional_fields: Any) -> None:
"""
Update job status and additional fields with automatic TTL adjustment.
Atomically sets hash fields, TTL, and secondary index via Lua script.
Args:
job_id: Job identifier
status: New status
**additional_fields: Additional fields to update (auto-serialized if dict/list)
"""
update_data: dict[str, str] = {
"status": status,
"updated_at": datetime.now(UTC).isoformat(),
}
# Serialize complex fields as JSON
for field_key, value in additional_fields.items():
if isinstance(value, (dict, list)):
update_data[field_key] = json.dumps(value)
else:
update_data[field_key] = str(value)
ttl = self._get_ttl_for_status(status)
key = f"{self.status_prefix}{job_id}"
now_ts = time.time()
await self._script_hset_expire_zadd(
keys=[key, self.jobs_index_key],
args=[str(ttl), str(now_ts), job_id] + self._flatten_mapping(update_data),
)
async def add_pii_findings(self, job_id: str, findings: list[dict[str, Any]]) -> None:
"""
Store PII scan results for a job.
Updates job with PII findings without changing status or TTL.
Keeps the secondary index timestamp fresh.
Args:
job_id: Job identifier
findings: List of PII finding dictionaries
"""
mapping = {
"pii_findings": json.dumps(findings),
"updated_at": datetime.now(UTC).isoformat(),
}
key = f"{self.status_prefix}{job_id}"
now_ts = time.time()
await self._script_hset_zadd(
keys=[key, self.jobs_index_key],
args=[str(now_ts), job_id] + self._flatten_mapping(mapping),
)
async def add_processing_result(self, job_id: str, result_url: str, confidence: float) -> None:
"""
Store processing completion data.
Updates job with processing results without changing status or TTL.
Keeps the secondary index timestamp fresh.
Args:
job_id: Job identifier
result_url: URL to the processed result
confidence: Processing confidence score (0.0 to 1.0)
"""
now = datetime.now(UTC).isoformat()
mapping = {
"result_url": result_url,
"confidence_score": str(confidence),
"completed_at": now,
"updated_at": now,
}
key = f"{self.status_prefix}{job_id}"
now_ts = time.time()
await self._script_hset_zadd(
keys=[key, self.jobs_index_key],
args=[str(now_ts), job_id] + self._flatten_mapping(mapping),
)
async def job_exists(self, job_id: str) -> bool:
"""
Check if job exists in Redis.
Args:
job_id: Job identifier
Returns:
True if job exists, False otherwise
"""
try:
exists = await self.redis.exists(f"{self.status_prefix}{job_id}")
return bool(exists)
except Exception:
return False
async def delete_job(self, job_id: str) -> None:
"""
Delete job hash and remove from secondary index atomically.
Args:
job_id: Job identifier
"""
try:
key = f"{self.status_prefix}{job_id}"
await self._script_delete_job(
keys=[key, self.jobs_index_key],
args=[job_id],
)
except Exception as e:
raise Exception(f"Failed to delete job {job_id}: {str(e)}")
async def set_expiration(self, job_id: str, ttl_seconds: int) -> None:
"""
Set TTL for automatic job cleanup.
Args:
job_id: Job identifier
ttl_seconds: Time-to-live in seconds
"""
try:
await self.redis.expire(f"{self.status_prefix}{job_id}", ttl_seconds)
except Exception as e:
raise Exception(f"Failed to set expiration for job {job_id}: {str(e)}")
async def list_all_jobs(self) -> list[str]:
"""List all job IDs in Redis.
This method scans for all job keys matching the status prefix
and extracts the job IDs. Used by orphan detection service.
Returns:
List of job IDs (without prefix)
Example:
>>> job_service = JobService(redis_client)
>>> job_ids = await job_service.list_all_jobs()
>>> print(f"Found {len(job_ids)} jobs")
"""
try:
# Use SCAN to find all job keys (safer than KEYS for production)
pattern = f"{self.status_prefix}*"
job_ids = []
cursor = 0
while True:
cursor, keys = await self.redis.scan(cursor=cursor, match=pattern, count=100)
# Extract job IDs from keys (remove prefix)
for key in keys:
# Remove prefix to get job ID
job_id = key.replace(self.status_prefix, "")
job_ids.append(job_id)
if cursor == 0:
break
logger.debug(f"Found {len(job_ids)} jobs in Redis")
return job_ids
except Exception as e:
logger.error(f"Error listing jobs: {str(e)}", exc_info=True)
return []
async def list_jobs_updated_before(self, cutoff_timestamp: float) -> list[str]:
"""List job IDs updated before the given Unix timestamp.
Uses the ZRANGEBYSCORE on the jobs-by-updated sorted set for
efficient lookup instead of scanning all keys.
Args:
cutoff_timestamp: Unix timestamp; jobs updated before this are returned.
Returns:
List of job IDs (without prefix)
"""
try:
members = await self.redis.zrangebyscore(
self.jobs_index_key, "-inf", str(cutoff_timestamp)
)
return [m if isinstance(m, str) else m.decode("utf-8") for m in members]
except Exception as e:
logger.error(f"Error querying jobs-by-updated index: {e}", exc_info=True)
return []
async def backfill_jobs_index(self) -> int:
"""One-time migration: populate the jobs-by-updated sorted set from existing keys.
Returns:
Number of jobs added to the index.
"""
job_ids = await self.list_all_jobs()
added = 0
for job_id in job_ids:
try:
job_data = await self.redis.hgetall(f"{self.status_prefix}{job_id}")
if not job_data:
continue
updated_at = job_data.get("updated_at") or job_data.get("created_at")
if updated_at:
dt = datetime.fromisoformat(str(updated_at).replace("Z", "+00:00"))
ts = dt.timestamp()
else:
ts = time.time()
await self.redis.zadd(self.jobs_index_key, {job_id: ts})
added += 1
except Exception as e:
logger.warning(f"Failed to backfill index for job {job_id}: {e}")
logger.info(f"Backfilled jobs-by-updated index with {added} jobs")
return added
async def get_job_status(self, job_id: str) -> dict[str, Any] | None:
"""Get job status and metadata (alias for get_job).
This method is an alias for get_job() to match the naming
convention used in other services (timeout_service, orphan_service).
Args:
job_id: Job identifier
Returns:
Job data dictionary or None if not found
"""
return await self.get_job(job_id)
async def cleanup_old_job(self, job_id: str) -> bool:
"""Delete old job hash and remove from secondary index atomically.
Args:
job_id: Job identifier (UUID)
Returns:
bool: True if job existed and was deleted, False if job didn't exist
"""
try:
key = f"{self.status_prefix}{job_id}"
deleted_count = await self._script_delete_job(
keys=[key, self.jobs_index_key],
args=[job_id],
)
if deleted_count and int(deleted_count) > 0:
logger.info(f"Cleaned up old job {job_id}")
return True
else:
logger.debug(f"Job {job_id} does not exist (already cleaned up)")
return False
except Exception as e:
logger.error(f"Error cleaning up job {job_id}: {str(e)}", exc_info=True)
return False
def _build_audit_record(self, job_data: dict[str, Any], reason: str) -> dict[str, Any]:
"""Build a structured audit record from job data.
Includes summary information only – no actual PII content is logged.
Args:
job_data: Full job hash data from Redis
reason: Why the job is being deleted/transitioned
Returns:
Structured dict suitable for JSON logging
"""
# Summarize PII findings without including actual PII content
pii_summary: dict[str, Any] | None = None
pii_raw = job_data.get("pii_findings")
if pii_raw:
try:
findings = json.loads(pii_raw) if isinstance(pii_raw, str) else pii_raw
if isinstance(findings, list):
entity_types = list({f.get("entity_type", "unknown") for f in findings})
pii_summary = {"count": len(findings), "entity_types": entity_types}
except (json.JSONDecodeError, TypeError):
pii_summary = {"count": 0, "entity_types": [], "parse_error": True}
return {
"event": "job_audit",
"job_id": job_data.get("job_id", "unknown"),
"status": job_data.get("status", "unknown"),
"created_at": job_data.get("created_at"),
"updated_at": job_data.get("updated_at"),
"completed_at": job_data.get("completed_at"),
"confidence_score": job_data.get("confidence_score"),
"error_message": job_data.get("error_message"),
"pii_summary": pii_summary,
"reason": reason,
}
async def emit_job_audit_log(self, job_id: str, reason: str) -> None:
"""Emit a structured JSON audit log for a job before deletion or forced transition.
Never blocks the caller on failure – catches all exceptions and logs a warning.
Args:
job_id: Job identifier
reason: Reason for the audit event (e.g., "retention_cleanup", "stuck_job_failed")
"""
try:
job_data = await self.redis.hgetall(f"{self.status_prefix}{job_id}")
if not job_data:
logger.warning(f"Cannot emit audit log for job {job_id}: job not found")
return
record = self._build_audit_record(dict(job_data), reason)
logger.info(json.dumps(record))
except Exception as e:
logger.warning(f"Failed to emit audit log for job {job_id}: {e}")
async def store_approval_token_mapping(self, approval_token: str, job_id: str, ttl_hours: int = 4) -> None:
"""Store approval token to job ID mapping for O(1) lookup.
Creates a Redis key: eq-pdf:approval-token:{token} → job_id
This enables direct token lookup without scanning all job hashes.
Args:
approval_token: Secure approval token
job_id: Job identifier
ttl_hours: Time-to-live in hours (matches approval expiration)
Example:
>>> job_service = JobService(redis_client)
>>> await job_service.store_approval_token_mapping(
... "abc123token",
... "job-uuid-123",
... ttl_hours=4
... )
"""
try:
token_key = f"eq-pdf:approval-token:{approval_token}"
ttl_seconds = ttl_hours * 3600
# Store mapping with expiration
await self.redis.set(token_key, job_id, ex=ttl_seconds)
logger.debug(f"Stored approval token mapping: {approval_token[:8]}... → {job_id}")
except Exception as e:
logger.error(f"Failed to store approval token mapping: {str(e)}", exc_info=True)
raise Exception(f"Failed to store token mapping: {str(e)}")
async def get_job_by_approval_token(self, token: str) -> dict[str, Any] | None:
"""Get job by approval token using O(1) Redis lookup.
Uses token mapping stored in Redis to directly fetch job ID,
then retrieves full job data. Much faster than scanning all jobs.
Args:
token: Approval token from URL
Returns:
Job data dictionary or None if token not found/expired
Example:
>>> job_service = JobService(redis_client)
>>> job = await job_service.get_job_by_approval_token("abc123token")
>>> if job:
... print(f"Found job: {job['job_id']}")
"""
try:
# O(1) lookup in Redis
token_key = f"eq-pdf:approval-token:{token}"
job_id = await self.redis.get(token_key)
if not job_id:
logger.debug("Approval token not found or expired")
return None
# Decode if bytes
if isinstance(job_id, bytes):
job_id = job_id.decode("utf-8")
# Fetch full job data
return await self.get_job(job_id)
except Exception as e:
logger.error(f"Error fetching job by token: {str(e)}", exc_info=True)
return None
async def create_stream_token(self, job_id: str) -> str:
"""Generate single-use stream token for SSE authentication.
Browser EventSource API cannot send custom headers, so this creates
a short-lived token that can be passed as a query parameter.
Token characteristics:
- 256-bit entropy (cryptographically secure)
- 5-minute TTL (connection should start immediately)
- Single-use (consumed on first validation via GETDEL)
- Job-scoped (validated against specific job_id)
Args:
job_id: Job identifier to associate with token
Returns:
Secure token string (43 characters, URL-safe)
Example:
>>> token = await job_service.create_stream_token("job-123")
>>> # Client uses: EventSource(f"/stream?token={token}")
"""
token = generate_secure_token()
token_key = f"eq-pdf:stream-token:{token}"
# Store token -> job_id mapping with short TTL (5 minutes)
await self.redis.set(token_key, job_id, ex=300)
logger.debug(f"Created stream token for job {job_id}: {token[:8]}...")
return token
async def validate_and_consume_stream_token(self, token: str) -> str | None:
"""Validate stream token and consume it atomically (single-use).
Uses Redis GETDEL for atomic get-and-delete operation, ensuring
the token can only be used once even under concurrent requests.
Args:
token: Stream token from query parameter
Returns:
job_id if token is valid, None if invalid/expired/already-consumed
Example:
>>> job_id = await job_service.validate_and_consume_stream_token("abc...")
>>> if job_id:
... # Token was valid and is now consumed
... # Verify job_id matches requested job for security
"""
token_key = f"eq-pdf:stream-token:{token}"
# GETDEL: Atomic get-and-delete ensures single-use
job_id = await self.redis.getdel(token_key)
if not job_id:
logger.debug("Stream token not found, expired, or already consumed")
return None
# Decode if bytes
if isinstance(job_id, bytes):
job_id = job_id.decode("utf-8")
logger.debug(f"Consumed stream token for job {job_id}")
return job_id