docs(eval): Phase 2 canary retry GO — success 37/40 (92.5%) failed 2 skipped 1

옵션 C 실행 (2026-05-03 02:36-02:39 UTC):
- 5201 documents stuck processing → failed (conditional UPDATE 1 row)
- 3817 재 enqueue → success 35.8s
- 4059 재 enqueue → success 100.7s
- GPU contention 해소 확인 (free 8820 MiB)

최종 tally: success 37 / failed 2 (3810 corrupt PDF + 5201 scan-likely
timeout) / skipped 1 (5090 MAX_PAGES). Plan 3 게이트 모두 PASS.

다음 = 사용자 승인 게이트 (2-C 진입 + nightly 모드 선택). main 머지 +
parent pull + cron 추가 4단계 대기.

후속 백로그 (Phase 1B+, 별도 PR):
- B1 scan-likely auto-skip (5201 패턴)
- B2 OOM 503 transient (야간 contention 자동 복구)
- B3 queue exhausted → doc.md_status 동기화 (corner case 정리)

evals/markdown/README.md

@@ -117,3 +117,140 @@ docker compose exec fastapi python /app/scripts/phase1d_pilot.py select \
 ```
 
 **enqueue 의 `--yes` 또는 `--no-dry-run` 류 실행은 별도 사용자 승인 + 야간 단발 sweep 윈도우 (23:00~03:00 KST) 안에서만**. 30건 backfill = marker-service BATCH_SIZE=1 × 평균 5분/건 ≈ 2.5h.
+
+---
+
+# Phase 2 — Full Backfill (legacy pending PDFs)
+
+> Plan: `~/.claude/plans/iridescent-gathering-clover.md`
+> Script: `scripts/phase2_backfill.py` (subcommands: inventory / select-canary / enqueue / nightly-enqueue / post-report)
+
+## 목적
+
+1D pilot 결과 = engineering go signal. legacy pending PDF (1D 후 잔여 ~237건) 을 marker_worker 로 변환해 `md_status='success'` 누적. **신규 업로드 우선권 보존, 야간 저부하 sweep, DB state 기반 idempotent checkpoint**.
+
+진행 로드맵: **2-A dry-run inventory → 2-B canary 40건 → 2-C nightly sweep ~4-5 nights → 2-D post-report**.
+
+## 파일
+
+| 파일 | 역할 | 갱신 시점 |
+|---|---|---|
+| `phase2_inventory.csv` | pending PDFs dry-run inventory + skip forecast | 2-A 종료 (commit, 1회) |
+| `phase2_canary_sample.csv` | stratified 40건 canary sample (시드 `20260503`) | 2-B(a) (commit) |
+| `phase2_canary_result.md` | canary 결과 요약 + 1D 비교 + GO/HALT 결정 근거 | 2-B 종료 (commit) |
+| `phase2_nightly_log.tsv` | 야간 sweep 한 줄/일 (date / enqueued / active_queue_at_start / active_queue_oldest_age_min / pending_pool_remaining / abort_reason / marker_ready) | append 매 sweep, 주 1회 commit |
+| `phase2_post_report.csv` | Phase 2 sweep 처리된 doc 별 final state + quality | 2-D (commit) |
+| `phase2_post_report.md` | 처리 분포 + 1D baseline 비교 + skip/failed/outlier 목록 | 2-D (commit) |
+
+## Subcommand 사용법
+
+### inventory (read-only, dry-run)
+```bash
+docker exec hyungi_document_server-fastapi-1 python /app/scripts/phase2_backfill.py inventory \
+  --output /app/evals/markdown/phase2_inventory.csv
+```
+- pending PDFs 전체에 대해 doc_id / file_size / text_density / doc_type / forecast_skip_reason 적재.
+- forecast_skip_reason ∈ {unsupported_extension / doctype_skip / handwritten_hint / over_max_pages_estimated / none}. 'none' = 변환 시도 후보.
+- handwritten_hint = title/path 에 `필기|손글씨|handwritten|handwriting` 매칭 (marker_worker 의 7d0fca2 룰 미러).
+- over_max_pages_estimated = file_size > 25MB proxy. 실 page_count 는 marker_worker 가 PyMuPDF 로 결정.
+
+### select-canary (재현성 시드)
+```bash
+docker exec hyungi_document_server-fastapi-1 python /app/scripts/phase2_backfill.py select-canary \
+  --inventory /app/evals/markdown/phase2_inventory.csv \
+  --output /app/evals/markdown/phase2_canary_sample.csv \
+  --seed 20260503
+```
+- 40건 buckets: large 6 / scan_likely 2 / study_note 10 / Academic_Paper 8 / Reference 6 / {Standard,Manual,Specification} 4 / {Note,Report,Memo,NULL} 4
+- inventory 의 `forecast_skip_reason='none'` 만 선정 후보.
+- 시드 고정 → 재실행 시 동일 sample.
+
+### 경로 정책 (2-B canary vs 2-C nightly)
+
+`/app/scripts/` 와 `/app/evals/` 는 parent repo (`~/Documents/code/hyungi_Document_Server`, branch=main) 의 read-only bind-mount. `docker cp ... :/app/scripts/...` 는 read-only 위반으로 실패. `docker compose --build` 은 검색 실험 soft lock 위반. → 단계별로 다른 경로 사용:
+
+| 단계 | script 경로 | sample CSV 경로 | 메커니즘 |
+|---|---|---|---|
+| **2-B canary (pre-merge)** | `/app/logs/phase2_backfill.py` | `/app/logs/phase2_canary_sample.csv` | `docker cp` worktree → /app/logs (rw bind-mount) |
+| **2-C nightly (post-merge canonical)** | `/app/scripts/phase2_backfill.py` | `/app/evals/markdown/phase2_*` | feat/phase2-backfill main 머지 + parent `git pull` 후 bind-mount 자동 활성 |
+
+**2-B 임시 sync** (canary 실행 직전 1회):
+```bash
+docker cp ~/Documents/code/hyungi_Document_Server_phase2/scripts/phase2_backfill.py \
+  hyungi_document_server-fastapi-1:/app/logs/phase2_backfill.py
+docker cp ~/Documents/code/hyungi_Document_Server_phase2/evals/markdown/phase2_canary_sample.csv \
+  hyungi_document_server-fastapi-1:/app/logs/phase2_canary_sample.csv
+```
+
+**2-C 진입 시점** (canary GO 결정 후):
+```bash
+cd ~/Documents/code/hyungi_Document_Server_phase2 && git push origin feat/phase2-backfill
+cd ~/Documents/code/hyungi_Document_Server && git fetch origin \
+  && git merge --ff-only origin/feat/phase2-backfill && git push origin main
+```
+
+이유: 미검증 코드를 main 에 미리 박지 않음 / canary 결과 따라 worktree 에서 hot-fix 가능 / nightly cron 은 canonical path 만 사용 (cp 단계 없음).
+
+### enqueue (one-shot, 사용자 승인 게이트)
+```bash
+# dry-run (default) — 2-B 단계 = /app/logs 임시 경로 (위 §"경로 정책" 참조)
+docker exec hyungi_document_server-fastapi-1 python /app/logs/phase2_backfill.py enqueue \
+  --csv /app/logs/phase2_canary_sample.csv
+
+# actual (사용자 승인 후)
+docker exec hyungi_document_server-fastapi-1 python /app/logs/phase2_backfill.py enqueue \
+  --csv /app/logs/phase2_canary_sample.csv --no-dry-run
+```
+- marker-service `/ready` 사전 검증.
+- `enqueue_stage` idempotent — 중복 호출 안전.
+
+### nightly-enqueue (cron / manual)
+```bash
+docker exec hyungi_document_server-fastapi-1 python /app/scripts/phase2_backfill.py nightly-enqueue \
+  --limit 50 --max-active-queue 5 \
+  --log-tsv /app/evals/markdown/phase2_nightly_log.tsv
+```
+- 가드 순서: disable flag (`/tmp/phase2_disable`) → marker /ready → active_queue ≤ threshold → DB pool 비어있지 않음 → enqueue.
+- 매 sweep log_tsv 한 줄. abort_reason ∈ {disable_flag / marker_unhealthy / active_queue_threshold / pool_empty / empty}.
+- pool_empty = Phase 2 자연 완료 신호 (cron 제거 hard gate trigger).
+
+### post-report
+```bash
+docker exec hyungi_document_server-fastapi-1 python /app/scripts/phase2_backfill.py post-report \
+  --output-csv /app/evals/markdown/phase2_post_report.csv \
+  --output-md /app/evals/markdown/phase2_post_report.md \
+  --phase2-start 2026-05-03T00:00:00Z
+```
+- `--phase2-start` ISO timestamp 이후 `md_generated_at` 만 집계 (Phase 2 코드 push 시점 권장).
+- 1D baseline (success 92% / elapsed_p50 34s / text_length_ratio_p50 1.15) 와 비교.
+- outlier 후보: elapsed_ms > 300s, text_length_ratio < 0.5 또는 > 10, 신규 warning 종류.
+
+## 의사결정 게이트
+
+### 2-B canary GO/HALT
+- success ≥ 36/40 (90%) AND failed ≤ 2 AND skipped ≤ 6 → **2-C 진입 GO**
+- 위 미충족 → HALT, 사용자 보고 후 재검토
+
+### 2-C nightly abort
+- 1 night 안에 failed > 5 → script 가 disable flag 자동 생성
+- marker-service `/ready` 실패 → 그 sweep 건너뜀, 다음 sweep 재시도
+- active_queue_oldest_age_min > 60 (stuck 임계) → log 에 [warn], 사용자 morning check 으로 판단
+
+### 2-D 종료 hard gate (Phase 2 closed 선언 직전)
+- (cron 모드) `crontab -l | grep phase2_backfill` 결과 비어 있어야 함
+- `~/.phase2_disable` 파일 정리 됨
+- pending PDF (`md_status='pending'`, `file_format='pdf'`) ≤ 5
+- processing_queue markdown active = 0
+
+## 1D 와의 차이
+
+| 항목 | 1D | Phase 2 |
+|---|---|---|
+| 목적 | failure mode 진단 | 풀 변환 |
+| 대상 | 30건 stratified | 237건 잔여 |
+| sample_source | existing_success + controlled_backfill | controlled_backfill only |
+| 처리 모드 | one-shot cron (1회) | nightly cron (~4-5 nights) |
+| 평가 | 사용자 5축 rubric | marker 자가 metrics + 1D baseline 비교 |
+| anchor 보존 | doc 4809 forced_include | (재처리 안 함) |
+| handwritten | over-sample (3건) | marker_worker 자동 skip 신뢰 |
+

evals/markdown/phase2_canary_result.md

@@ -0,0 +1,166 @@
+# Phase 2 Canary Result — 40건 stratified backfill
+
+> 실행: 2026-05-02 23:50 UTC enqueue → ~02:30 UTC 완료 (wall ~2h 40m, marker BATCH_SIZE=1)
+> sample CSV: `evals/markdown/phase2_canary_sample.csv` (seed 20260503)
+> plan 결정 게이트: §"2-B canary GO/HALT"
+
+## 결과 분포
+
+| md_status | count | rate |
+|---|---|---|
+| success | 35 | 87.5% |
+| failed | 3 | 7.5% |
+| skipped | 1 | 2.5% |
+| processing (stuck) | 1 | 2.5% |
+
+총 40건 + 1건 corner case (stuck `processing`, queue 측은 max_attempts 도달 후 failed). **실효 doc-level 실패 = 4 / 40 = 10%**.
+
+## Plan 게이트 판정
+
+| 기준 | 임계 | 실측 | 판정 |
+|---|---|---|---|
+| success rate | ≥ 36/40 (90%) | 35/40 (87.5%) | **FAIL** (-1) |
+| failed | ≤ 2 | 4 (실효) | **FAIL** (+2) |
+| skipped | ≤ 6 | 1 | PASS |
+
+**→ HALT (사용자 보고 후 재검토)**.
+
+## 실패 분석 — 4건 모두 root cause 확인
+
+| doc_id | 분류 | 표면 에러 | root cause | 재시도 가능? |
+|---|---|---|---|---|
+| 3817 | failed | `OutOfMemoryError: CUDA out of memory. Tried to allocate 74 MiB` | **GPU contention** (canary 진행 중 다른 프로세스 5.93+5.35 GiB 사용 → free 55 MiB) | ✅ Yes (GPU free 시) |
+| 4059 | failed | `OutOfMemoryError: CUDA out of memory. Tried to allocate 176 MiB` | **GPU contention** 동일 | ✅ Yes |
+| 3810 | failed | `PdfiumError: Failed to load page` | 진짜 PDF 파싱 오류 (corrupt/protected page 가능성) | ❌ No (Marker 책임 아님) |
+| 5201 | processing (corner) | `ReadTimeout('')` 3회 (queue) | scan-likely 추정 (705 KB / text_len=15) — OCR-bound 처리 시간 > 300s timeout. queue 측 attempts=3 도달 후 failed, 그러나 doc.md_status='processing' 으로 stuck (Phase 1B 알려진 corner case) | ⚠ (timeout 늘리거나 SKIP rule 확장 필요) |
+
+**핵심 인사이트**: **Marker quality 자체는 0건 fail**. 실패 4건의 분류:
+- 50% (2/4) = **GPU 일시 contention** — infra 이슈, 사용자 활동 따라 재발 가능
+- 25% (1/4) = **PDF 자체 corrupt** — 정상 비율, non-retryable
+- 25% (1/4) = **scan-likely + 작은 file_size + 빈 텍스트** — 새 SKIP 후보 패턴
+
+## 처리 시간 (success 35건, 1D baseline 비교)
+
+| 메트릭 | 1D | 2-B canary | delta |
+|---|---|---|---|
+| avg | - | 44.3s | - |
+| p50 | 34s | 33.2s | -2% (안정) |
+| p90 | 112s | 92.6s | -17% (개선) |
+| max | 219s | 297.1s | +36% (large bias 영향) |
+
+전반적으로 1D 와 동등 또는 약간 빠름. **마커 자체 throughput 안정**.
+
+## Quality 메트릭 (success 35건, 1D baseline 비교)
+
+| 메트릭 | 1D | 2-B canary | delta |
+|---|---|---|---|
+| text_length_ratio p50 | 1.15 | 1.00 | -13% (정상 범위) |
+| warnings: heading_hierarchy_jump | 86% (24/28) | 94% (33/35) | +8pp |
+| warnings: low_image_alt_text_ratio | 89% (25/28) | 86% (30/35) | -3pp |
+
+신규 warning 종류 없음. heading_hierarchy_jump 가 거의 전건이라는 패턴 재확인 (1D 와 동일) — Marker output 의 heading 구조가 원본과 정확히 일치하지 않는 일반적 현상, score 기준 정의 필요.
+
+## GPU contention 상세
+
+canary 처리 시점 (~00:00-02:00 UTC) GPU 점유 process (canary 진행 중 nvidia-smi 추정):
+- text-embeddings-router: 1.35 GiB
+- python (process 1200154): 1.85 GiB
+- python (process 1562575): 5.93 GiB ← **largest, suspect Ollama or marker self**
+- python (process 1570943): 5.35 GiB
+- python (process 1612684): 1.01 GiB
+- **합계 ~15.5 GiB / 16.4 GiB capacity → free ~50-100 MiB**
+
+canary 종료 후 (02:31 UTC) 측정:
+- text-embeddings-router: 1.38 GiB
+- python (1200154): 1.89 GiB
+- python (1562575): 3.83 GiB
+- **합계 7.1 GiB → free 8.8 GiB**
+
+→ 5.35 GiB + 5.93 GiB → 3.83 GiB 으로 줄어듦 (또는 둘 중 하나 종료). canary 시점에만 일시적으로 cumulative load 였음.
+
+## 권장 조정안 (HALT 후)
+
+게이트 가지 못했지만 **failure 4건 모두 설명 가능**. Marker 본질 결함 0건. 다음 중 사용자 결정:
+
+### A. 그대로 2-C 진입 (수용)
+- 근거: Marker quality 0 fail, GPU OOM 은 야간 sweep (23-03 KST) 윈도우 안에 contention 적을 가능성 높음 (사용자 활동 종료, Ollama 호출 적음)
+- 보완: 2-C 가드에 GPU free memory 사전 체크 추가 검토 (script 1 줄 — 단, 본 plan scope 외)
+- 실패 4건 (3817/4059/3810/5201) 은 별도 retry 또는 backlog
+
+### B. Marker 코드 / 가드 보강 후 재시도
+- B1. **scan-likely + tiny text 자동 skip** (5201 패턴): marker_worker 에 `file_size < 1MB AND text_len < 100` 또는 `text_density < 1` 시 `md_status='skipped'`
+- B2. **OOM 을 transient 로 분류**: server.py 가 OutOfMemoryError 만 503 으로 raise → marker_worker 가 5xx → queue retry. (현재는 422 = non-retryable)
+- B3. **queue 영구 실패 시 doc.md_status='failed' 동기화**: Phase 1B 알려진 corner case 정리 (5201 패턴 제거)
+- 작업: marker_worker.py + server.py 변경 → 별도 PR/커밋 → canary 재실행
+
+### C. canary 일부 재시도 → 결과 갱신
+- 3817, 4059 만 재 enqueue (GPU 지금 free) → 성공 확인 → **실효 success 37/40 = 92.5%, failed 2 (3810+5201) = 5%, skipped 1 = 2.5% → GO 게이트 통과**
+- 가장 빠른 unblock 경로
+
+### D. HALT 유지 + 백로그 정리
+- canary 결과 그대로 인정. 4 failed/stuck doc 은 Phase 2 마지막 후속 정리.
+- 2-C 진입 보류.
+
+## 다음 단계 — 사용자 결정 필요
+
+위 A/B/C/D 중 선택. 추천 = **C (3817+4059 즉시 재 enqueue)** 또는 **A (그대로 진입, 야간 GPU contention 낮음 가정)**. B 는 plan scope 확장이라 별도 라운드 권장.
+
+5201 의 stuck 'processing' 은 어느 옵션 가도 **수동 정리 필요** (`UPDATE documents SET md_status='failed', md_extraction_error='ReadTimeout after 3 attempts (Phase 1B corner case)' WHERE id=5201`). production DB write 라 사용자 승인 후.
+
+---
+
+## Retry 결과 (2026-05-03 02:36-02:39 UTC) — GO 게이트 통과
+
+옵션 C 실행 결과:
+
+| 작업 | 결과 |
+|---|---|
+| 5201 documents 정리 (`md_status='processing' → 'failed'`, conditional UPDATE) | 1 row updated, error_message 새로 박음 |
+| 5201 queue 정리 | 불필요 (이미 status='failed' terminal) |
+| 3817 재 enqueue | success (35.8s, GPU free 8820 MiB) |
+| 4059 재 enqueue | success (100.7s, large bias 영향) |
+
+### 최종 분포
+
+| md_status | count | rate |
+|---|---|---|
+| success | **37** | 92.5% |
+| failed | **2** | 5.0% |
+| skipped | **1** | 2.5% |
+
+failed 2건 분류:
+- doc 3810: corrupt PDF (PdfiumError) — Marker 책임 아님, non-retryable
+- doc 5201: scan-likely + 빈 텍스트 + ReadTimeout — Phase 1B+ corner case backlog
+
+### Plan 게이트 재판정
+
+| 기준 | 임계 | 실측 | 판정 |
+|---|---|---|---|
+| success rate | ≥ 36/40 (90%) | 37/40 (92.5%) | ✅ PASS |
+| failed | ≤ 2 | 2 | ✅ PASS |
+| skipped | ≤ 6 | 1 | ✅ PASS |
+
+**→ 2-C nightly sweep 진입 GO.**
+
+### 다음 = 2-C 진입 게이트 (사용자 승인 필요)
+
+옵션:
+- **모드 A (cron)** — 추천. 23:00 KST nightly, 50건/limit, max-active-queue 5
+- 모드 B (manual nightly)
+- 모드 C (systemd timer)
+
+cron 추가 전 필수 작업:
+1. main 머지: feat/phase2-backfill (HEAD `79dc31b` 또는 retry 반영 후 새 commit) → main FF
+2. parent repo `git pull --ff-only` → `/app/scripts/phase2_backfill.py` canonical path 활성
+3. `~/logs/phase2/` 디렉토리 생성
+4. crontab entry 추가
+
+사용자 승인 + 모드 선택 시 위 4 단계 진행.
+
+### Phase 1B+ backlog (Phase 2 외 별도)
+
+retry 결과는 다음 후속 작업의 우선순위 데이터:
+
+- **B2 후보**: server.py 가 OutOfMemoryError 만 503 으로 분류 (현재 422 = non-retryable). 야간 GPU contention 발생 시 자동 retry 가능. → 1B+ small PR
+- **B1 후보**: marker_worker scan-likely + tiny text 자동 skip (5201 패턴). text_density < 1 OR text_len < 100 시 skipped. → 1B+ small PR  
+- **B3 후보**: queue 영구 실패 시 doc.md_status 동기화. processing 상태 stuck 방지. → 1B+ small PR

evals/markdown/phase2_canary_sample.csv

@@ -0,0 +1,41 @@
+doc_id,title,file_size,file_size_band,text_density,text_density_band,doc_type,bucket_label
+3817,"Safety and Health for Engineers_05_37 Safety and Health Data, Analysis and Management Information",11313981,L,11.264,mixed,study_note,large
+4059,공업역학 동역학(제13판)_Chapter 12 질점의 운동학,11140007,L,8.78,mixed,Academic_Paper,large
+5264,황현필의 진보를 위한 역사_2장 식민지 수탈과 학살의 진실을 말하라,14154100,L,2.14,scan-likely,Academic_Paper,large
+5271,황현필의 진보를 위한 역사_9장 박정희 신격화를 중단하라,16266788,L,1.878,scan-likely,Academic_Paper,large
+5267,황현필의 진보를 위한 역사_5장 해방정국을 감추지 말라,12320415,L,1.617,scan-likely,Note,large
+5090,일반기계기사_실전 모의고사,24976506,L,15.75,mixed,Academic_Paper,large
+8857,"TKP-26-0132_신양철강_74,290원 - 복사본",450026,S,4.407,scan-likely,Specification,scan_likely
+5116,표준기계설계(KS)_04_나사,21167224,L,2.759,scan-likely,Standard,scan_likely
+5238,근로자 노출평가제도 내 위험성평가 방법론의 적용,393908,S,89.488,born-digital,study_note,study_note
+5240,사업장 위험성평가에 관한 법제의 비교법적 고찰,437377,S,83.786,born-digital,study_note,study_note
+3771,Industrial Safety and Health Management(7-ED)_16 Welding,1239353,M,62.251,born-digital,study_note,study_note
+3762,Industrial Safety and Health Management(7-ED)_7 Buildings and Facilities,1123847,M,60.053,born-digital,study_note,study_note
+3768,Industrial Safety and Health Management(7-ED)_13 Fire Protection,725779,S,62.01,born-digital,study_note,study_note
+3764,Industrial Safety and Health Management(7-ED)_9 Health and Toxic Substances,1409470,M,66.395,born-digital,study_note,study_note
+3766,Industrial Safety and Health Management(7-ED)_11 Flammable and Explosive Materials,1185190,M,54.637,born-digital,study_note,study_note
+3756,Industrial Safety and Health Management(7-ED)_1 The Safety and Health Manager,550267,S,76.971,born-digital,study_note,study_note
+3761,Industrial Safety and Health Management(7-ED)_6 Process safety and Disaster Preparedness,773058,S,65.656,born-digital,study_note,study_note
+3767,Industrial Safety and Health Management(7-ED)_12 Personal Protection and First Aid,1553116,M,65.161,born-digital,study_note,study_note
+5156,스마트팩토리 구축 제조기업 조직원이 지각한 상사의 변혁적리더십이 셀프리더,449251,S,94.531,born-digital,Academic_Paper,Academic_Paper
+5257,Modifying Large Language Model Post-Training for Diverse Creative Writing,885908,S,106.149,born-digital,Academic_Paper,Academic_Paper
+5142,표준기계설계(KS)_10_키 및 스플라인,17185683,L,2.06,scan-likely,Standard,tech_doc
+4546,機械設計_7 不等速運動機構,1205358,M,16.38,mixed,Manual,tech_doc
+5205,2_ASME VIII_Impact Test,364827,S,8.791,mixed,Standard,tech_doc
+5112,표준기계설계(KS)_03_기하 공차의 도시방법,7742981,M,1.377,scan-likely,Standard,tech_doc
+5085,일반기계기사_실전 모의고사 2회,2357404,M,12.318,mixed,Note,minor_doc
+5201,A Endorsement Exam,705076,S,0.022,scan-likely,Memo,minor_doc
+5086,일반기계기사_실전 모의고사 3회,1643798,M,17.111,mixed,Note,minor_doc
+5199,2012_NB_A_March_homewrk_quiz,1937838,M,0.025,scan-likely,Memo,minor_doc
+3780,"Safety and Health for Engineers_02_0 Legal, Aspects of Safety and Health",5089329,M,0.415,scan-likely,study_note,filler
+5160,"스마트팩토리, 제조기업 분야에 새로운 경쟁우위 확보를 위한 견인차 역할을",1261747,M,29.268,mixed,Academic_Paper,filler
+5174,#8. 검사 및 시험 계획_테크니컬 코리아 R.A_현대로템 통합,156781,S,22.043,mixed,Note,filler
+3819,Safety and Health for Engineers_06_Alphabetical List of Organizations Cited in Text,9678536,M,13.504,mixed,study_note,filler
+3784,Safety and Health for Engineers_02_7 Products Liability,6618025,M,10.892,mixed,study_note,filler
+5056,기계진동 이론과 응용(제5판)_목차,835993,S,18.099,mixed,Reference,filler
+3810,Safety and Health for Engineers_04_31 Human Behavior and Performance in safety and Health,8384987,M,10.362,mixed,study_note,filler
+4547,機械設計_8 アクチュエータ,1242829,M,19.333,mixed,Reference,filler
+5093,재료역학_부록,3178000,M,28.455,mixed,Academic_Paper,filler
+5265,황현필의 진보를 위한 역사_3장 독립운동을 상처 내지 말라,10398784,M,2.302,scan-likely,Note,filler
+5161,스마트팩토리를 위한 운영빅데이터 분석 플랫폼,920193,S,22.36,mixed,Academic_Paper,filler
+5130,07_Mechanical Behavior of Materials_Yielding and Fracture under Combined Stresses,3953088,M,32.835,mixed,Academic_Paper,filler

evals/markdown/phase2_inventory.csv

@@ -0,0 +1,238 @@
+doc_id,title,file_path,file_size,file_size_band,text_len,text_density,text_density_band,doc_type,forecast_skip_reason,created_at
+3755,Industrial Safety and Health Management(7-ED)_0 Contents,PKM/Inbox/Industrial Safety and Health Management(7-ED)_0 Contents.pdf,476785,S,20537,44.108,mixed,study_note,none,2026-04-03T04:34:24.389026+00:00
+3756,Industrial Safety and Health Management(7-ED)_1 The Safety and Health Manager,PKM/Inbox/Industrial Safety and Health Management(7-ED)_1 The Safety and Health Manager.pdf,550267,S,41362,76.971,born-digital,study_note,none,2026-04-03T04:34:24.572086+00:00
+3761,Industrial Safety and Health Management(7-ED)_6 Process safety and Disaster Preparedness,PKM/Inbox/Industrial Safety and Health Management(7-ED)_6 Process safety and Disaster Preparedness.pdf,773058,S,49566,65.656,born-digital,study_note,none,2026-04-03T04:49:04.845585+00:00
+3762,Industrial Safety and Health Management(7-ED)_7 Buildings and Facilities,PKM/Inbox/Industrial Safety and Health Management(7-ED)_7 Buildings and Facilities.pdf,1123847,M,65909,60.053,born-digital,study_note,none,2026-04-03T04:49:05.295538+00:00
+3764,Industrial Safety and Health Management(7-ED)_9 Health and Toxic Substances,PKM/Inbox/Industrial Safety and Health Management(7-ED)_9 Health and Toxic Substances.pdf,1409470,M,91388,66.395,born-digital,study_note,none,2026-04-03T04:49:06.411247+00:00
+3766,Industrial Safety and Health Management(7-ED)_11 Flammable and Explosive Materials,PKM/Inbox/Industrial Safety and Health Management(7-ED)_11 Flammable and Explosive Materials.pdf,1185190,M,63237,54.637,born-digital,study_note,none,2026-04-03T04:49:07.380118+00:00
+3767,Industrial Safety and Health Management(7-ED)_12 Personal Protection and First Aid,PKM/Inbox/Industrial Safety and Health Management(7-ED)_12 Personal Protection and First Aid.pdf,1553116,M,98831,65.161,born-digital,study_note,none,2026-04-03T04:49:07.939877+00:00
+3768,Industrial Safety and Health Management(7-ED)_13 Fire Protection,PKM/Inbox/Industrial Safety and Health Management(7-ED)_13 Fire Protection.pdf,725779,S,43951,62.01,born-digital,study_note,none,2026-04-03T04:49:08.149678+00:00
+3770,Industrial Safety and Health Management(7-ED)_15 Machine Guarding,PKM/Inbox/Industrial Safety and Health Management(7-ED)_15 Machine Guarding.pdf,3315859,M,142588,44.034,mixed,study_note,none,2026-04-03T04:49:10.013144+00:00
+3771,Industrial Safety and Health Management(7-ED)_16 Welding,PKM/Inbox/Industrial Safety and Health Management(7-ED)_16 Welding.pdf,1239353,M,75343,62.251,born-digital,study_note,none,2026-04-03T04:49:10.470978+00:00
+3772,Industrial Safety and Health Management(7-ED)_17 Electrical Hazards,PKM/Inbox/Industrial Safety and Health Management(7-ED)_17 Electrical Hazards.pdf,2039508,M,72357,36.329,mixed,study_note,none,2026-04-03T04:49:11.157819+00:00
+3773,Industrial Safety and Health Management(7-ED)_18 Construction,PKM/Inbox/Industrial Safety and Health Management(7-ED)_18 Construction.pdf,1908991,M,102240,54.842,born-digital,study_note,none,2026-04-03T04:49:11.800261+00:00
+3775,Safety and Health for Engineers_00,PKM/Inbox/Safety and Health for Engineers_00.pdf,1427288,M,13113,9.408,mixed,study_note,none,2026-04-03T05:06:56.386163+00:00
+3776,Safety and Health for Engineers_01_0 Introduction,PKM/Inbox/Safety and Health for Engineers_01_0 Introduction.pdf,8934279,M,743,0.085,scan-likely,study_note,none,2026-04-03T05:07:00.165514+00:00
+3777,Safety and Health for Engineers_01_1 The Importance of Safety and Health,PKM/Inbox/Safety and Health for Engineers_01_1 The Importance of Safety and Health.pdf,8934119,M,91642,10.504,mixed,study_note,none,2026-04-03T05:07:03.292702+00:00
+3778,Safety and Health for Engineers_01_2 Safety and Health Professions,PKM/Inbox/Safety and Health for Engineers_01_2 Safety and Health Professions.pdf,5807403,M,59814,10.547,mixed,study_note,none,2026-04-03T05:07:07.898542+00:00
+3779,Safety and Health for Engineers_01_3 Fundamental Concepts and Terms,PKM/Inbox/Safety and Health for Engineers_01_3 Fundamental Concepts and Terms.pdf,5989787,M,60618,10.363,mixed,study_note,none,2026-04-03T05:07:12.664716+00:00
+3780,"Safety and Health for Engineers_02_0 Legal, Aspects of Safety and Health","PKM/Inbox/Safety and Health for Engineers_02_0 Legal, Aspects of Safety and Health.pdf",5089329,M,2065,0.415,scan-likely,study_note,none,2026-04-03T05:07:17.328796+00:00
+3781,Safety and Health for Engineers_02_4 Legal Aspects of Safety and Health,PKM/Inbox/Safety and Health for Engineers_02_4 Legal Aspects of Safety and Health.pdf,5089299,M,50272,10.115,mixed,study_note,none,2026-04-03T05:07:21.571656+00:00
+3783,Safety and Health for Engineers_02_6 Worker's Compensation,PKM/Inbox/Safety and Health for Engineers_02_6 Worker's Compensation.pdf,6079717,M,65499,11.032,mixed,study_note,none,2026-04-03T05:07:30.644769+00:00
+3784,Safety and Health for Engineers_02_7 Products Liability,PKM/Inbox/Safety and Health for Engineers_02_7 Products Liability.pdf,6618025,M,70394,10.892,mixed,study_note,none,2026-04-03T05:07:35.452588+00:00
+3785,Safety and Health for Engineers_02_8 Record Keeping and Reporting,PKM/Inbox/Safety and Health for Engineers_02_8 Record Keeping and Reporting.pdf,7100288,M,81011,11.683,mixed,study_note,none,2026-04-03T05:07:41.197634+00:00
+3786,Safety and Health for Engineers_03_0 Hazards and Their control,PKM/Inbox/Safety and Health for Engineers_03_0 Hazards and Their control.pdf,8562283,M,2214,0.265,scan-likely,study_note,none,2026-04-03T05:07:47.779152+00:00
+3787,Safety and Health for Engineers_03_9 General Principles of Hazard Control,PKM/Inbox/Safety and Health for Engineers_03_9 General Principles of Hazard Control.pdf,8562254,M,90283,10.797,mixed,study_note,none,2026-04-03T05:07:53.502349+00:00
+3788,Safety and Health for Engineers_03_10 Mechanics and Structures,PKM/Inbox/Safety and Health for Engineers_03_10 Mechanics and Structures.pdf,9868378,M,104028,10.795,mixed,study_note,none,2026-04-03T05:07:59.923610+00:00
+3789,Safety and Health for Engineers_03_11 Walking and Working Surfaces,PKM/Inbox/Safety and Health for Engineers_03_11 Walking and Working Surfaces.pdf,9542843,M,102275,10.975,mixed,study_note,none,2026-04-03T05:08:07.160154+00:00
+3790,Safety and Health for Engineers_03_12 Electrical Safety,PKM/Inbox/Safety and Health for Engineers_03_12 Electrical Safety.pdf,7811797,M,82384,10.799,mixed,study_note,none,2026-04-03T05:08:12.616433+00:00
+3791,Safety and Health for Engineers_03_13 Tools and Machines,PKM/Inbox/Safety and Health for Engineers_03_13 Tools and Machines.pdf,10631629,L,105491,10.161,mixed,study_note,none,2026-04-03T05:08:19.963028+00:00
+3792,Safety and Health for Engineers_03_14 Transportation,PKM/Inbox/Safety and Health for Engineers_03_14 Transportation.pdf,9457716,M,103492,11.205,mixed,study_note,none,2026-04-03T05:08:27.286422+00:00
+3793,Safety and Health for Engineers_03_15 Materials Handling,PKM/Inbox/Safety and Health for Engineers_03_15 Materials Handling.pdf,12096868,L,132105,11.183,mixed,study_note,none,2026-04-03T05:08:36.205673+00:00
+3794,Safety and Health for Engineers_03_16 Fire Protection and Prevention,PKM/Inbox/Safety and Health for Engineers_03_16 Fire Protection and Prevention.pdf,27524814,L,165727,6.166,mixed,study_note,over_max_pages_estimated,2026-04-03T05:08:55.100356+00:00
+3795,Safety and Health for Engineers_03_17 Explosions and Explosives,PKM/Inbox/Safety and Health for Engineers_03_17 Explosions and Explosives.pdf,4876363,M,53213,11.174,mixed,study_note,none,2026-04-03T05:08:59.606826+00:00
+3796,Safety and Health for Engineers_03_18 Heat and Cold,PKM/Inbox/Safety and Health for Engineers_03_18 Heat and Cold.pdf,8215266,M,87758,10.939,mixed,study_note,none,2026-04-03T05:09:05.490524+00:00
+3797,Safety and Health for Engineers_03_19 Pressure,PKM/Inbox/Safety and Health for Engineers_03_19 Pressure.pdf,5611782,M,59442,10.847,mixed,study_note,none,2026-04-03T05:09:10.679634+00:00
+3798,Safety and Health for Engineers_03_20 Visual Environment,PKM/Inbox/Safety and Health for Engineers_03_20 Visual Environment.pdf,5414637,M,58281,11.022,mixed,study_note,none,2026-04-03T05:09:14.984786+00:00
+3799,Safety and Health for Engineers_03_21 Nonionizing Radiation,PKM/Inbox/Safety and Health for Engineers_03_21 Nonionizing Radiation.pdf,5514954,M,58454,10.854,mixed,study_note,none,2026-04-03T05:09:18.999306+00:00
+3800,Safety and Health for Engineers_03_22 Ionizing Radiation,PKM/Inbox/Safety and Health for Engineers_03_22 Ionizing Radiation.pdf,5035759,M,52456,10.667,mixed,study_note,none,2026-04-03T05:09:22.940102+00:00
+3801,Safety and Health for Engineers_03_23 Noise and Vibration,PKM/Inbox/Safety and Health for Engineers_03_23 Noise and Vibration.pdf,8765522,M,87691,10.244,mixed,study_note,none,2026-04-03T05:09:29.219392+00:00
+3802,Safety and Health for Engineers_03_24 Chemicals,PKM/Inbox/Safety and Health for Engineers_03_24 Chemicals.pdf,11162176,L,120941,11.095,mixed,study_note,none,2026-04-03T05:09:37.851928+00:00
+3803,Safety and Health for Engineers_03_25 Ventilation,PKM/Inbox/Safety and Health for Engineers_03_25 Ventilation.pdf,6894217,M,69324,10.297,mixed,study_note,none,2026-04-03T05:09:44.185835+00:00
+3804,Safety and Health for Engineers_03_26 Biohazards,PKM/Inbox/Safety and Health for Engineers_03_26 Biohazards.pdf,6554931,M,76458,11.944,mixed,study_note,none,2026-04-03T05:09:49.536301+00:00
+3805,Safety and Health for Engineers_03_27 Hazardous Waste,PKM/Inbox/Safety and Health for Engineers_03_27 Hazardous Waste.pdf,6671999,M,68689,10.542,mixed,study_note,none,2026-04-03T05:09:54.300613+00:00
+3806,Safety and Health for Engineers_03_28 Personal Protective Equipment,PKM/Inbox/Safety and Health for Engineers_03_28 Personal Protective Equipment.pdf,7122464,M,73829,10.614,mixed,study_note,none,2026-04-03T05:10:00.083359+00:00
+3807,Safety and Health for Engineers_03_29 Emergencies and Security,PKM/Inbox/Safety and Health for Engineers_03_29 Emergencies and Security.pdf,7077502,M,73716,10.666,mixed,study_note,none,2026-04-03T05:10:05.256855+00:00
+3808,"Safety and Health for Engineers_03_30 Facility Planning, Design, and Maintenance","PKM/Inbox/Safety and Health for Engineers_03_30 Facility Planning, Design, and Maintenance.pdf",4677058,M,50098,10.969,mixed,study_note,none,2026-04-03T05:10:09.854386+00:00
+3809,Safety and Health for Engineers_04_0 The Human Element,PKM/Inbox/Safety and Health for Engineers_04_0 The Human Element.pdf,8384999,M,3429,0.419,scan-likely,study_note,none,2026-04-03T05:10:15.932112+00:00
+3810,Safety and Health for Engineers_04_31 Human Behavior and Performance in safety and Health,PKM/Inbox/Safety and Health for Engineers_04_31 Human Behavior and Performance in safety and Health.pdf,8384987,M,84847,10.362,mixed,study_note,none,2026-04-03T05:10:22.591531+00:00
+3811,"Safety and Health for Engineers_04_32 Procedures, Rules, and Training","PKM/Inbox/Safety and Health for Engineers_04_32 Procedures, Rules, and Training.pdf",7175749,M,78070,11.141,mixed,study_note,none,2026-04-03T05:10:27.685186+00:00
+3813,Safety and Health for Engineers_05_0 Managing Safety and Health,PKM/Inbox/Safety and Health for Engineers_05_0 Managing Safety and Health.pdf,9454413,M,1899,0.206,scan-likely,study_note,none,2026-04-03T05:10:43.948700+00:00
+3814,"Safety and Health for Engineers_05_34 Risk, Risk Assessment, and Risk Management","PKM/Inbox/Safety and Health for Engineers_05_34 Risk, Risk Assessment, and Risk Management.pdf",9454350,M,104290,11.296,mixed,study_note,none,2026-04-03T05:10:51.419167+00:00
+3815,Safety and Health for Engineers_05_35 Safety and Health Management,PKM/Inbox/Safety and Health for Engineers_05_35 Safety and Health Management.pdf,8340393,M,91707,11.259,mixed,study_note,none,2026-04-03T05:10:57.600214+00:00
+3816,Safety and Health for Engineers_05_36 System Safety,PKM/Inbox/Safety and Health for Engineers_05_36 System Safety.pdf,7165903,M,70595,10.088,mixed,study_note,none,2026-04-03T05:11:02.595452+00:00
+3817,"Safety and Health for Engineers_05_37 Safety and Health Data, Analysis and Management Information","PKM/Inbox/Safety and Health for Engineers_05_37 Safety and Health Data, Analysis and Management Information.pdf",11313981,L,124455,11.264,mixed,study_note,none,2026-04-03T05:11:10.229570+00:00
+3818,Safety and Health for Engineers_05_38 Safety and Health Plans and Programs,PKM/Inbox/Safety and Health for Engineers_05_38 Safety and Health Plans and Programs.pdf,5832841,M,62752,11.017,mixed,study_note,none,2026-04-03T05:11:14.894009+00:00
+3819,Safety and Health for Engineers_06_Alphabetical List of Organizations Cited in Text,PKM/Inbox/Safety and Health for Engineers_06_Alphabetical List of Organizations Cited in Text.pdf,9678536,M,127636,13.504,mixed,study_note,none,2026-04-03T05:11:21.915005+00:00
+4057,공업역학 동력학(제13판)_APPENDIX,PKM/Inbox/공업역학 동력학(제13판)_APPENDIX.pdf,8966645,M,152829,17.453,mixed,Note,none,2026-04-03T07:06:39.166342+00:00
+4058,공업역학 동역학(제13판)_00 목차,PKM/Inbox/공업역학 동역학(제13판)_00 목차.pdf,2013932,M,20665,10.507,mixed,Reference,none,2026-04-03T07:06:41.434400+00:00
+4059,공업역학 동역학(제13판)_Chapter 12 질점의 운동학,PKM/Inbox/공업역학 동역학(제13판)_Chapter 12 질점의 운동학.pdf,11140007,L,95518,8.78,mixed,Academic_Paper,none,2026-04-03T07:06:50.001782+00:00
+4060,공업역학 동역학(제13판)_Chapter 13 질점의 운동역학_힘과 가속도,PKM/Inbox/공업역학 동역학(제13판)_Chapter 13 질점의 운동역학_힘과 가속도.pdf,7011688,M,56816,8.298,mixed,Manual,none,2026-04-03T07:06:56.438865+00:00
+4062,공업역학 동역학(제13판)_Chapter 15 질점의 운동역학_역적과 운동량,PKM/Inbox/공업역학 동역학(제13판)_Chapter 15 질점의 운동역학_역적과 운동량.pdf,8153718,M,76160,9.565,mixed,Academic_Paper,none,2026-04-03T07:07:08.670475+00:00
+4063,공업역학 동역학(제13판)_Chapter 16 강체 평면 운동학,PKM/Inbox/공업역학 동역학(제13판)_Chapter 16 강체 평면 운동학.pdf,10138853,M,70522,7.123,mixed,Academic_Paper,none,2026-04-03T07:07:17.049458+00:00
+4064,공업역학 동역학(제13판)_Chapter 17 강체의 평면 운동역학_힘과 가속도,PKM/Inbox/공업역학 동역학(제13판)_Chapter 17 강체의 평면 운동역학_힘과 가속도.pdf,7411889,M,52850,7.302,mixed,Academic_Paper,none,2026-04-03T07:07:23.207207+00:00
+4065,공업역학 동역학(제13판)_Chapter 18 강체의 평면 운동학,PKM/Inbox/공업역학 동역학(제13판)_Chapter 18 강체의 평면 운동학.pdf,5122916,M,32029,6.402,mixed,Academic_Paper,none,2026-04-03T07:07:27.893070+00:00
+4066,공업역학 동역학(제13판)_Chapter 19 강체의 평면 운동역학_역적과 운동량,PKM/Inbox/공업역학 동역학(제13판)_Chapter 19 강체의 평면 운동역학_역적과 운동량.pdf,4240748,M,33459,8.079,mixed,Academic_Paper,none,2026-04-03T07:07:31.640948+00:00
+4067,공업역학 동역학(제13판)_Chapter 20 3차원 강체 운동학,PKM/Inbox/공업역학 동역학(제13판)_Chapter 20 3차원 강체 운동학.pdf,3837746,M,28698,7.657,mixed,Academic_Paper,none,2026-04-03T07:07:35.050541+00:00
+4069,공업역학 동역학(제13판)_Chapter 22 진동,PKM/Inbox/공업역학 동역학(제13판)_Chapter 22 진동.pdf,3478353,M,31656,9.319,mixed,Academic_Paper,none,2026-04-03T07:07:42.935401+00:00
+4070,공업역학 동역학(제13판)_Review 1 질점의 운동학과 운동역학,PKM/Inbox/공업역학 동역학(제13판)_Review 1 질점의 운동학과 운동역학.pdf,1808025,M,12218,6.92,mixed,Note,none,2026-04-03T07:07:45.006654+00:00
+4071,공업역학 동역학(제13판)_Review 2 강체의 평면 운동학 및,PKM/Inbox/공업역학 동역학(제13판)_Review 2 강체의 평면 운동학 및.pdf,2070528,M,11892,5.881,mixed,Note,none,2026-04-03T07:07:47.379011+00:00
+4540,機械設計_1 機械設計の基礎,PKM/Inbox/機械設計_1 機械設計の基礎.pdf,1675175,M,27631,16.89,mixed,Reference,none,2026-04-07T02:59:21.853900+00:00
+4541,機械設計_2 機械材料と強度および安全率,PKM/Inbox/機械設計_2 機械材料と強度および安全率.pdf,1461925,M,29295,20.52,mixed,Reference,none,2026-04-07T02:59:22.999108+00:00
+4542,機械設計_3 締結•接合要素,PKM/Inbox/機械設計_3 締結•接合要素.pdf,2343158,M,50016,21.858,mixed,Standard,none,2026-04-07T02:59:26.325850+00:00
+4543,機械設計_4 軸系要素,PKM/Inbox/機械設計_4 軸系要素.pdf,2218086,M,40769,18.821,mixed,Reference,none,2026-04-07T02:59:27.312924+00:00
+4544,機械設計_5 軸受•案内要素,PKM/Inbox/機械設計_5 軸受•案内要素.pdf,3293674,M,70593,21.947,mixed,Manual,none,2026-04-07T02:59:28.849332+00:00
+4545,機械設計_6 動力伝達要素,PKM/Inbox/機械設計_6 動力伝達要素.pdf,3168370,M,62220,20.109,mixed,Manual,none,2026-04-07T02:59:30.485110+00:00
+4546,機械設計_7 不等速運動機構,PKM/Inbox/機械設計_7 不等速運動機構.pdf,1205358,M,19281,16.38,mixed,Manual,none,2026-04-07T02:59:32.619449+00:00
+4547,機械設計_8 アクチュエータ,PKM/Inbox/機械設計_8 アクチュエータ.pdf,1242829,M,23464,19.333,mixed,Reference,none,2026-04-07T02:59:33.985785+00:00
+4548,機械設計_9 機械システムの設計,PKM/Inbox/機械設計_9 機械システムの設計.pdf,1441721,M,30786,21.866,mixed,Manual,none,2026-04-07T02:59:35.258788+00:00
+4549,機械設計_10 練習問題解答,PKM/Inbox/機械設計_10 練習問題解答.pdf,1699097,M,22927,13.817,mixed,Manual,none,2026-04-07T02:59:36.108172+00:00
+5056,기계진동 이론과 응용(제5판)_목차,PKM/Inbox/기계진동 이론과 응용(제5판)_목차.pdf,835993,S,14776,18.099,mixed,Reference,none,2026-04-08T23:15:27.766489+00:00
+5057,기계진동 이론과 응용(제5판)_Appendix,PKM/Inbox/기계진동 이론과 응용(제5판)_Appendix.pdf,4017320,M,57998,14.783,mixed,Reference,none,2026-04-08T23:15:29.285679+00:00
+5058,기계진동 이론과 응용(제5판)_Chapter 1 진동운동,PKM/Inbox/기계진동 이론과 응용(제5판)_Chapter 1 진동운동.pdf,1518132,M,10701,7.218,mixed,Note,none,2026-04-08T23:15:30.317920+00:00
+5059,기계진동 이론과 응용(제5판)_Chapter 2 자유진동,PKM/Inbox/기계진동 이론과 응용(제5판)_Chapter 2 자유진동.pdf,2217720,M,23768,10.975,mixed,Academic_Paper,none,2026-04-08T23:15:31.704320+00:00
+5060,기계진동 이론과 응용(제5판)_Chapter 3 조화 가진진동,PKM/Inbox/기계진동 이론과 응용(제5판)_Chapter 3 조화 가진진동.pdf,2992028,M,33140,11.342,mixed,Academic_Paper,none,2026-04-08T23:15:32.983116+00:00
+5061,기계진동 이론과 응용(제5판)_Chapter 4 과도 진동,PKM/Inbox/기계진동 이론과 응용(제5판)_Chapter 4 과도 진동.pdf,2509875,M,30004,12.241,mixed,Academic_Paper,none,2026-04-08T23:15:34.301838+00:00
+5062,기계진동 이론과 응용(제5판)_Chapter 5 2자유도계와 다자유도계,PKM/Inbox/기계진동 이론과 응용(제5판)_Chapter 5 2자유도계와 다자유도계.pdf,2775517,M,26202,9.667,mixed,Academic_Paper,none,2026-04-08T23:15:35.605446+00:00
+5063,기계진동 이론과 응용(제5판)_Chapter 6 진동계의 특성,PKM/Inbox/기계진동 이론과 응용(제5판)_Chapter 6 진동계의 특성.pdf,2476898,M,27466,11.355,mixed,Reference,none,2026-04-08T23:15:36.927471+00:00
+5064,기계진동 이론과 응용(제5판)_Chapter 7 라그랑즈 방정식,PKM/Inbox/기계진동 이론과 응용(제5판)_Chapter 7 라그랑즈 방정식.pdf,2662848,M,20434,7.858,mixed,Academic_Paper,none,2026-04-08T23:15:38.281162+00:00
+5066,기계진동 이론과 응용(제5판)_Chapter 8 수치해석법,PKM/Inbox/기계진동 이론과 응용(제5판)_Chapter 8 수치해석법.pdf,1881462,M,31273,17.021,mixed,Academic_Paper,none,2026-04-08T23:15:39.385956+00:00
+5067,流体力学_01 完全流体の力学_0 Intro,PKM/Inbox/流体力学_01 完全流体の力学_0 Intro.pdf,319348,S,1049,3.364,scan-likely,Reference,none,2026-04-08T23:15:39.779551+00:00
+5068,기계진동 이론과 응용(제5판)_Chapter 9 연속계의 진동,PKM/Inbox/기계진동 이론과 응용(제5판)_Chapter 9 연속계의 진동.pdf,1904149,M,22813,12.268,mixed,Academic_Paper,none,2026-04-08T23:15:40.244821+00:00
+5069,流体力学_01 完全流体の力学_1 基礎方程式と基礎原理,PKM/Inbox/流体力学_01 完全流体の力学_1 基礎方程式と基礎原理.pdf,1590589,M,32132,20.686,mixed,Academic_Paper,none,2026-04-08T23:15:40.795292+00:00
+5070,기계진동 이론과 응용(제5판)_Chapter 10 유한 요소법 입문,PKM/Inbox/기계진동 이론과 응용(제5판)_Chapter 10 유한 요소법 입문.pdf,2918260,M,37484,13.153,mixed,Academic_Paper,none,2026-04-08T23:15:41.347817+00:00
+5071,流体力学_01 完全流体の力学_2 ベルヌーイの定理および運動量保存則とその応用,PKM/Inbox/流体力学_01 完全流体の力学_2 ベルヌーイの定理および運動量保存則とその応用.pdf,1327546,M,26457,20.408,mixed,Academic_Paper,none,2026-04-08T23:15:41.793483+00:00
+5072,기계진동 이론과 응용(제5판)_Chapter 11 연속계의 모드-합성 방법,PKM/Inbox/기계진동 이론과 응용(제5판)_Chapter 11 연속계의 모드-합성 방법.pdf,1719318,M,18079,10.768,mixed,Academic_Paper,none,2026-04-08T23:15:42.200052+00:00
+5073,流体力学_01 完全流体の力学_3 流関数と速度ポテンシャル,PKM/Inbox/流体力学_01 完全流体の力学_3 流関数と速度ポテンシャル.pdf,1053322,M,17580,17.091,mixed,Academic_Paper,none,2026-04-08T23:15:42.776633+00:00
+5074,기계진동 이론과 응용(제5판)_Chapter 12 고전적인 해석방법,PKM/Inbox/기계진동 이론과 응용(제5판)_Chapter 12 고전적인 해석방법.pdf,3361219,M,37053,11.288,mixed,Academic_Paper,none,2026-04-08T23:15:43.492238+00:00
+5075,流体力学_01 完全流体の力学_4 二次元ポテンシャル流れ,PKM/Inbox/流体力学_01 完全流体の力学_4 二次元ポテンシャル流れ.pdf,2564583,M,50432,20.137,mixed,Academic_Paper,none,2026-04-08T23:15:44.050776+00:00
+5076,기계진동 이론과 응용(제5판)_Chapter 13 불규칙 진동,PKM/Inbox/기계진동 이론과 응용(제5판)_Chapter 13 불규칙 진동.pdf,2470352,M,34766,14.411,mixed,Reference,none,2026-04-08T23:15:44.575733+00:00
+5077,流体力学_01 完全流体の力学_5 渦(うず),PKM/Inbox/流体力学_01 完全流体の力学_5 渦(うず).pdf,1394271,M,23084,16.954,mixed,Academic_Paper,none,2026-04-08T23:15:45.026236+00:00
+5078,기계진동 이론과 응용(제5판)_Chapter 14 비선형진동,PKM/Inbox/기계진동 이론과 응용(제5판)_Chapter 14 비선형진동.pdf,2265927,M,23347,10.551,mixed,Academic_Paper,none,2026-04-08T23:15:45.599500+00:00
+5079,流体力学_01 完全流体の力学_6 波,PKM/Inbox/流体力学_01 完全流体の力学_6 波.pdf,1467570,M,25309,17.659,mixed,Academic_Paper,none,2026-04-08T23:15:46.156458+00:00
+5080,流体力学_02 粘性流体の力学_0 Intro,PKM/Inbox/流体力学_02 粘性流体の力学_0 Intro.pdf,226546,S,3008,13.596,mixed,Reference,none,2026-04-08T23:15:46.584826+00:00
+5081,流体力学_02 粘性流体の力学_7 粘性流体の基礎方程式,PKM/Inbox/流体力学_02 粘性流体の力学_7 粘性流体の基礎方程式.pdf,890677,S,14019,16.117,mixed,Academic_Paper,none,2026-04-08T23:15:47.550174+00:00
+5082,流体力学_02 粘性流体の力学_8 ナビエーストークス方程式の厳密解,PKM/Inbox/流体力学_02 粘性流体の力学_8 ナビエーストークス方程式の厳密解.pdf,1019341,S,19638,19.728,mixed,Academic_Paper,none,2026-04-08T23:15:48.643897+00:00
+5083,00_일반기계기사_목차,PKM/Inbox/00_일반기계기사_목차.pdf,8471940,M,70301,8.497,mixed,Manual,none,2026-04-08T23:15:50.578723+00:00
+5084,일반기계기사_실전 모의고사 1회,PKM/Inbox/일반기계기사_실전 모의고사 1회.pdf,1809053,M,28814,16.31,mixed,Reference,none,2026-04-08T23:15:55.821375+00:00
+5085,일반기계기사_실전 모의고사 2회,PKM/Inbox/일반기계기사_실전 모의고사 2회.pdf,2357404,M,28359,12.318,mixed,Note,none,2026-04-08T23:15:57.064358+00:00
+5086,일반기계기사_실전 모의고사 3회,PKM/Inbox/일반기계기사_실전 모의고사 3회.pdf,1643798,M,27468,17.111,mixed,Note,none,2026-04-08T23:15:58.106360+00:00
+5087,일반기계기사_실전 모의고사 4회,PKM/Inbox/일반기계기사_실전 모의고사 4회.pdf,1732142,M,28874,17.07,mixed,Reference,none,2026-04-08T23:15:59.221928+00:00
+5088,일반기계기사_실전 모의고사 5회,PKM/Inbox/일반기계기사_실전 모의고사 5회.pdf,1793083,M,29438,16.812,mixed,Reference,none,2026-04-08T23:16:00.341841+00:00
+5089,재료과학_00_서문,PKM/Inbox/재료과학_00_서문.pdf,1529124,M,28276,18.935,mixed,Reference,none,2026-04-08T23:16:03.847623+00:00
+5090,일반기계기사_실전 모의고사,PKM/Inbox/일반기계기사_실전 모의고사.pdf,24976506,L,384157,15.75,mixed,Academic_Paper,none,2026-04-08T23:16:05.341794+00:00
+5091,재료과학_01_재료과학 소개,PKM/Inbox/재료과학_01_재료과학 소개.pdf,3005951,M,26499,9.027,mixed,Academic_Paper,none,2026-04-08T23:16:06.184639+00:00
+5092,재료과학_02_원자의 구조와 원자간 결합,PKM/Inbox/재료과학_02_원자의 구조와 원자간 결합.pdf,4064573,M,47817,12.047,mixed,Academic_Paper,none,2026-04-08T23:16:08.028960+00:00
+5093,재료역학_부록,PKM/Inbox/재료역학_부록.pdf,3178000,M,88312,28.455,mixed,Academic_Paper,none,2026-04-08T23:16:09.428509+00:00
+5095,정역학_01_정역학 서론,PKM/Inbox/정역학_01_정역학 서론.pdf,1935591,M,32230,17.051,mixed,Reference,none,2026-04-08T23:16:15.217698+00:00
+5096,정역학_02_힘계의 기본 연산,PKM/Inbox/정역학_02_힘계의 기본 연산.pdf,3493287,M,39573,11.6,mixed,Academic_Paper,none,2026-04-08T23:16:16.562088+00:00
+5097,정역학_03_힘계의 합성,PKM/Inbox/정역학_03_힘계의 합성.pdf,2571786,M,28667,11.414,mixed,Academic_Paper,none,2026-04-08T23:16:17.833253+00:00
+5098,정역학_04_공명 평형 해석,PKM/Inbox/정역학_04_공명 평형 해석.pdf,5655029,M,71376,12.925,mixed,Academic_Paper,none,2026-04-08T23:16:19.571857+00:00
+5099,정역학_05_3차원 평형,PKM/Inbox/정역학_05_3차원 평형.pdf,2614037,M,32234,12.627,mixed,Academic_Paper,none,2026-04-08T23:16:20.693129+00:00
+5100,정역학_06_보와 케이블,PKM/Inbox/정역학_06_보와 케이블.pdf,3066734,M,42430,14.168,mixed,Academic_Paper,none,2026-04-08T23:16:21.984222+00:00
+5101,정역학_07_건마찰,PKM/Inbox/정역학_07_건마찰.pdf,3232307,M,44035,13.95,mixed,Reference,none,2026-04-08T23:16:23.491935+00:00
+5102,정역학_08_도심과 분포 하중,PKM/Inbox/정역학_08_도심과 분포 하중.pdf,4144284,M,42923,10.606,mixed,Academic_Paper,none,2026-04-08T23:16:24.746403+00:00
+5103,표준기계설계(KS)_01_치수공차와 끼워맞춤,PKM/Inbox/표준기계설계(KS)_01_치수공차와 끼워맞춤.pdf,20210623,L,44633,2.261,scan-likely,Standard,none,2026-04-08T23:16:26.944309+00:00
+5104,정역학_09_면적 관성 모멘트와 과성 곱,PKM/Inbox/정역학_09_면적 관성 모멘트와 과성 곱.pdf,2632774,M,32837,12.772,mixed,Reference,none,2026-04-08T23:16:26.964918+00:00
+5105,정역학_10_가상 일과 퍼텐셜 에너지,PKM/Inbox/정역학_10_가상 일과 퍼텐셜 에너지.pdf,2594344,M,28313,11.175,mixed,Academic_Paper,none,2026-04-08T23:16:28.289238+00:00
+5106,정역학_부록,PKM/Inbox/정역학_부록.pdf,1962700,M,39631,20.677,mixed,Academic_Paper,none,2026-04-08T23:16:29.586996+00:00
+5107,JlSにもとづく機械設計製囲便覧_第13版_0 Intro,PKM/Inbox/JlSにもとづく機械設計製囲便覧_第13版_0 Intro.pdf,1198849,M,43577,37.221,mixed,Reference,none,2026-04-08T23:16:30.246865+00:00
+5108,정영학_00_목차,PKM/Inbox/정영학_00_목차.pdf,850729,S,13589,16.357,mixed,Reference,none,2026-04-08T23:16:30.707965+00:00
+5109,표준기계설계(KS)_02_일반공차 및 보통공차,PKM/Inbox/표준기계설계(KS)_02_일반공차 및 보통공차.pdf,9486493,M,20303,2.192,scan-likely,Standard,none,2026-04-08T23:16:30.963027+00:00
+5110,JlSにもとづく機械設計製囲便覧_第13版_1章 諸単位,PKM/Inbox/JlSにもとづく機械設計製囲便覧_第13版_1章 諸単位.pdf,727694,S,22655,31.88,mixed,Standard,none,2026-04-08T23:16:31.668820+00:00
+5112,표준기계설계(KS)_03_기하 공차의 도시방법,PKM/Inbox/표준기계설계(KS)_03_기하 공차의 도시방법.pdf,7742981,M,10413,1.377,scan-likely,Standard,none,2026-04-08T23:16:32.995488+00:00
+5113,JlSにもとづく機械設計製囲便覧_第13版_3章 カ学,PKM/Inbox/JlSにもとづく機械設計製囲便覧_第13版_3章 カ学.pdf,833772,S,16544,20.319,mixed,Reference,none,2026-04-08T23:16:33.005005+00:00
+5114,JlSにもとづく機械設計製囲便覧_第13版_4章 材料カ学,PKM/Inbox/JlSにもとづく機械設計製囲便覧_第13版_4章 材料カ学.pdf,1172147,M,20208,17.654,mixed,Reference,none,2026-04-08T23:16:34.146033+00:00
+5116,표준기계설계(KS)_04_나사,PKM/Inbox/표준기계설계(KS)_04_나사.pdf,21167224,L,57023,2.759,scan-likely,Standard,none,2026-04-08T23:16:37.766254+00:00
+5117,01_Mechanical Behavior of Materials_Introduction,PKM/Inbox/01_Mechanical Behavior of Materials_Introduction.pdf,2893265,M,52850,18.705,mixed,Reference,none,2026-04-08T23:16:38.832286+00:00
+5118,"02_Mechanical Behavior of Materials_Structure, Defects, and Deformation in Materials","PKM/Inbox/02_Mechanical Behavior of Materials_Structure, Defects, and Deformation in Materials.pdf",4378386,M,56091,13.118,mixed,Note,none,2026-04-08T23:16:40.575592+00:00
+5119,"표준기계설계(KS)_05_볼트, 자리따기","PKM/Inbox/표준기계설계(KS)_05_볼트, 자리따기.pdf",14523342,L,31237,2.202,scan-likely,Standard,none,2026-04-08T23:16:42.196325+00:00
+5120,03_Mechanical Behavior of Materials_Mechanical Testing-Tension Test and Stress-Strain Mechanisms,PKM/Inbox/03_Mechanical Behavior of Materials_Mechanical Testing-Tension Test and Stress-Strain Mechanisms.pdf,3783032,M,138532,37.498,mixed,Reference,none,2026-04-08T23:16:42.870499+00:00
+5121,표준기계설계(KS)_06_너트,PKM/Inbox/표준기계설계(KS)_06_너트.pdf,4933016,M,13303,2.761,scan-likely,Standard,none,2026-04-08T23:16:44.448549+00:00
+5122,04_Mechanical Behavior of Materials_Mechanical Testing- Additional Basic Tests,PKM/Inbox/04_Mechanical Behavior of Materials_Mechanical Testing- Additional Basic Tests.pdf,2947668,M,62179,21.601,mixed,Reference,none,2026-04-08T23:16:44.738221+00:00
+5123,표준기계설계(KS)_07_와셔,PKM/Inbox/표준기계설계(KS)_07_와셔.pdf,7058048,M,15925,2.31,scan-likely,Standard,none,2026-04-08T23:16:47.338036+00:00
+5124,THERMODYNAMICS_An Engineering Approach 9-E_APPENDIX,PKM/Inbox/THERMODYNAMICS_An Engineering Approach 9-E_APPENDIX.pdf,7740918,M,159241,21.065,mixed,Reference,none,2026-04-08T23:16:48.026166+00:00
+5125,05_Mechanical Behavior of Materials_Stress-Strain Relationships and Behavior,PKM/Inbox/05_Mechanical Behavior of Materials_Stress-Strain Relationships and Behavior.pdf,2884730,M,99098,35.177,mixed,Academic_Paper,none,2026-04-08T23:16:49.247029+00:00
+5126,THERMODYNAMICS_An Engineering Approach 9-E_CHAPTER 01,PKM/Inbox/THERMODYNAMICS_An Engineering Approach 9-E_CHAPTER 01.pdf,5212258,M,166000,32.612,mixed,Academic_Paper,none,2026-04-08T23:16:50.848415+00:00
+5128,06_Mechanical Behavior of Materials_Review of Complex and Principal States of Stress and Strain,PKM/Inbox/06_Mechanical Behavior of Materials_Review of Complex and Principal States of Stress and Strain.pdf,3173000,M,81191,26.202,mixed,Reference,none,2026-04-08T23:16:51.964505+00:00
+5129,THERMODYNAMICS_An Engineering Approach 9-E_CHAPTER 02,PKM/Inbox/THERMODYNAMICS_An Engineering Approach 9-E_CHAPTER 02.pdf,4778125,M,213287,45.71,mixed,Reference,none,2026-04-08T23:16:53.289800+00:00
+5130,07_Mechanical Behavior of Materials_Yielding and Fracture under Combined Stresses,PKM/Inbox/07_Mechanical Behavior of Materials_Yielding and Fracture under Combined Stresses.pdf,3953088,M,126759,32.835,mixed,Academic_Paper,none,2026-04-08T23:16:53.797654+00:00
+5131,압력용기 핸드북_강 구조의 설계,PKM/Inbox/압력용기 핸드북_강 구조의 설계.pdf,1363261,M,14266,10.716,mixed,Reference,none,2026-04-08T23:16:54.838637+00:00
+5134,THERMODYNAMICS_An Engineering Approach 9-E_CHAPTER 04,PKM/Inbox/THERMODYNAMICS_An Engineering Approach 9-E_CHAPTER 04.pdf,5932814,M,159158,27.471,mixed,Reference,none,2026-04-08T23:16:58.605198+00:00
+5136,Pressure Vessel Design Manual_00,PKM/Inbox/Pressure Vessel Design Manual_00.pdf,372049,S,19436,53.494,born-digital,Manual,none,2026-04-08T23:17:00.194347+00:00
+5138,표준기계설계(KS)_09_축 관계,PKM/Inbox/표준기계설계(KS)_09_축 관계.pdf,26465006,L,52684,2.038,scan-likely,Standard,over_max_pages_estimated,2026-04-08T23:17:02.195454+00:00
+5139,압력용기 핸드북_압력용기 설계와 제작,PKM/Inbox/압력용기 핸드북_압력용기 설계와 제작.pdf,14093886,L,254595,18.498,mixed,Manual,none,2026-04-08T23:17:05.346798+00:00
+5140,Pressure Vessel Design Manual_02 General Design,PKM/Inbox/Pressure Vessel Design Manual_02 General Design.pdf,8285419,M,223799,27.659,mixed,Manual,none,2026-04-08T23:17:07.080394+00:00
+5142,표준기계설계(KS)_10_키 및 스플라인,PKM/Inbox/표준기계설계(KS)_10_키 및 스플라인.pdf,17185683,L,34581,2.06,scan-likely,Standard,none,2026-04-08T23:17:08.309320+00:00
+5143,Structural Analysiss and Design of Process Equipment_01_History and Organization of Codes,PKM/Inbox/Structural Analysiss and Design of Process Equipment_01_History and Organization of Codes.pdf,914921,S,16537,18.509,mixed,Reference,none,2026-04-08T23:17:08.926568+00:00
+5144,Pressure Vessel Design Manual_03 Flange Design,PKM/Inbox/Pressure Vessel Design Manual_03 Flange Design.pdf,3768131,M,112421,30.551,mixed,Manual,none,2026-04-08T23:17:09.602217+00:00
+5145,"Structural Analysiss and Design of Process Equipment_02_Selection of Vessel, Specifications, Reports and Allowable Stresses","PKM/Inbox/Structural Analysiss and Design of Process Equipment_02_Selection of Vessel, Specifications, Reports and Allowable Stresses.pdf",691082,S,33272,49.3,mixed,Reference,none,2026-04-08T23:17:09.766146+00:00
+5146,표준기계설계(KS)_11_멈춤링,PKM/Inbox/표준기계설계(KS)_11_멈춤링.pdf,5851521,M,15566,2.724,scan-likely,Standard,none,2026-04-08T23:17:10.529895+00:00
+5147,"Structural Analysiss and Design of Process Equipment_03_Strength Theories, Design Criteria and Design Equations","PKM/Inbox/Structural Analysiss and Design of Process Equipment_03_Strength Theories, Design Criteria and Design Equations.pdf",695405,S,20757,30.565,mixed,Reference,none,2026-04-08T23:17:10.657722+00:00
+5148,Structural Analysiss and Design of Process Equipment_04_Materials of Construction,PKM/Inbox/Structural Analysiss and Design of Process Equipment_04_Materials of Construction.pdf,3306699,M,70627,21.871,mixed,Reference,none,2026-04-08T23:17:12.650803+00:00
+5149,Pressure Vessel Design Manual_04 Design of Vessel Supports,PKM/Inbox/Pressure Vessel Design Manual_04 Design of Vessel Supports.pdf,8693390,M,234154,27.581,mixed,Manual,none,2026-04-08T23:17:13.743831+00:00
+5150,표준기계설계(KS)_12_벨트와 풀리,PKM/Inbox/표준기계설계(KS)_12_벨트와 풀리.pdf,18363580,L,38654,2.155,scan-likely,Standard,none,2026-04-08T23:17:17.375038+00:00
+5152,THE PIPE FITTERS BLUE BOOK,PKM/Inbox/THE PIPE FITTERS BLUE BOOK.pdf,42381135,L,238868,5.771,mixed,Manual,over_max_pages_estimated,2026-04-08T23:17:52.867389+00:00
+5153,로지스틱스 연구 조립제조산에서의 자재분류 요인분석 관한 사례연구,PKM/Inbox/로지스틱스 연구 조립제조산에서의 자재분류 요인분석 관한 사례연구.pdf,1137506,M,31149,28.041,mixed,Academic_Paper,none,2026-04-08T23:17:59.109278+00:00
+5156,스마트팩토리 구축 제조기업 조직원이 지각한 상사의 변혁적리더십이 셀프리더,PKM/Inbox/스마트팩토리 구축 제조기업 조직원이 지각한 상사의 변혁적리더십이 셀프리더.pdf,449251,S,41473,94.531,born-digital,Academic_Paper,none,2026-04-08T23:18:08.381775+00:00
+5157,스마트팩토리 운영전략이 경영성과와 지속사용의도 간의 구조적 관계분석,PKM/Inbox/스마트팩토리 운영전략이 경영성과와 지속사용의도 간의 구조적 관계분석.pdf,866978,S,35476,41.901,mixed,Academic_Paper,none,2026-04-08T23:18:09.256468+00:00
+5158,스마트팩토리 제조경쟁력 확보가 한국의 해외진출 제조기업 리쇼어링에 미치는,PKM/Inbox/스마트팩토리 제조경쟁력 확보가 한국의 해외진출 제조기업 리쇼어링에 미치는.pdf,1015118,S,37670,38.0,mixed,Academic_Paper,none,2026-04-08T23:18:10.193550+00:00
+5159,"스마트팩토리, 인공지능으로 날개를 달다","PKM/Inbox/스마트팩토리, 인공지능으로 날개를 달다.pdf",604946,S,9037,15.297,mixed,Report,none,2026-04-08T23:18:11.105141+00:00
+5160,"스마트팩토리, 제조기업 분야에 새로운 경쟁우위 확보를 위한 견인차 역할을","PKM/Inbox/스마트팩토리, 제조기업 분야에 새로운 경쟁우위 확보를 위한 견인차 역할을.pdf",1261747,M,36063,29.268,mixed,Academic_Paper,none,2026-04-08T23:18:12.225466+00:00
+5161,스마트팩토리를 위한 운영빅데이터 분석 플랫폼,PKM/Inbox/스마트팩토리를 위한 운영빅데이터 분석 플랫폼.pdf,920193,S,20093,22.36,mixed,Academic_Paper,none,2026-04-08T23:18:13.263588+00:00
+5162,철강 산업 기업의 스마트팩토리 도입 운영형태가 품질 안정화에 미치는 관계,PKM/Inbox/철강 산업 기업의 스마트팩토리 도입 운영형태가 품질 안정화에 미치는 관계.pdf,1139690,M,31606,28.398,mixed,Academic_Paper,none,2026-04-08T23:18:14.315451+00:00
+5165,금융 관련 공공기관의 ESG경영 현황,PKM/Inbox/금융 관련 공공기관의 ESG경영 현황.pdf,555648,S,33344,61.449,born-digital,Report,none,2026-04-08T23:18:37.418558+00:00
+5166,기업시민 포스코 ESG경영 사례,PKM/Inbox/기업시민 포스코 ESG경영 사례.pdf,662512,S,5868,9.07,mixed,Report,none,2026-04-08T23:18:38.237680+00:00
+5169,수소인프라 이슈 및 과제,PKM/Inbox/수소인프라 이슈 및 과제.pdf,1495185,M,26104,17.878,mixed,Academic_Paper,none,2026-04-08T23:19:16.572890+00:00
+5170,전주 완주 수소시범도시 주민수용성 분석,PKM/Inbox/전주 완주 수소시범도시 주민수용성 분석.pdf,976519,S,22013,23.083,mixed,study_note,none,2026-04-08T23:19:17.548849+00:00
+5172,탄소중립은 지속가능한 경제성장과 양립하는가,PKM/Inbox/탄소중립은 지속가능한 경제성장과 양립하는가.pdf,664629,S,24049,37.053,mixed,Academic_Paper,none,2026-04-08T23:19:19.371453+00:00
+5173,HyRAM 3.0을 이용한 수소충전소 정량적 위험성평가 분석에 관한 연구,PKM/Inbox/HyRAM 3.0을 이용한 수소충전소 정량적 위험성평가 분석에 관한 연구.pdf,1279610,M,26966,21.579,mixed,study_note,none,2026-04-08T23:19:20.401783+00:00
+5174,#8. 검사 및 시험 계획_테크니컬 코리아 R.A_현대로템 통합,PKM/Inbox/#8. 검사 및 시험 계획_테크니컬 코리아 R.A_현대로템 통합.pdf,156781,S,3375,22.043,mixed,Note,none,2026-04-08T23:19:22.531563+00:00
+5176,수소취성(최병학),PKM/Inbox/수소취성(최병학).pdf,7110851,M,11069,1.594,scan-likely,study_note,none,2026-04-08T23:19:30.535784+00:00
+5181,PTC46,PKM/Inbox/PTC46.PDF,8154549,M,0,0.0,scan-likely,,none,2026-04-08T23:19:47.730804+00:00
+5182,VIII-1_01,PKM/Inbox/VIII-1_01.pdf,1934884,M,85906,45.464,mixed,Standard,none,2026-04-08T23:19:48.887787+00:00
+5186,VIII-1_05-UG,PKM/Inbox/VIII-1_05-UG.pdf,6098647,M,531589,89.257,born-digital,Standard,none,2026-04-08T23:19:52.267197+00:00
+5187,VIII-1_06-UW,PKM/Inbox/VIII-1_06-UW.pdf,1520880,M,174445,117.453,born-digital,Standard,none,2026-04-08T23:19:53.320860+00:00
+5190,VIII-1_09-UCS,PKM/Inbox/VIII-1_09-UCS.pdf,737749,S,108861,151.1,born-digital,Standard,none,2026-04-08T23:19:55.892937+00:00
+5191,VIII-1_10-UNF,PKM/Inbox/VIII-1_10-UNF.pdf,359963,S,47261,134.445,born-digital,Standard,none,2026-04-08T23:19:56.741277+00:00
+5192,VIII-1_11-UHA,PKM/Inbox/VIII-1_11-UHA.pdf,379489,S,60782,164.012,born-digital,Standard,none,2026-04-08T23:19:57.550987+00:00
+5199,2012_NB_A_March_homewrk_quiz,PKM/Inbox/2012_NB_A_March_homewrk_quiz.pdf,1937838,M,47,0.025,scan-likely,Memo,none,2026-04-08T23:20:02.786018+00:00
+5201,A Endorsement Exam,PKM/Inbox/A Endorsement Exam.pdf,705076,S,15,0.022,scan-likely,Memo,none,2026-04-08T23:20:07.556782+00:00
+5204,1_ASME VIII_Introduction,PKM/Inbox/1_ASME VIII_Introduction.pdf,1206580,M,29525,25.057,mixed,Manual,none,2026-04-08T23:20:20.560415+00:00
+5205,2_ASME VIII_Impact Test,PKM/Inbox/2_ASME VIII_Impact Test.pdf,364827,S,3132,8.791,mixed,Standard,none,2026-04-08T23:20:21.422093+00:00
+5206,3_ASME VIII_Design_Basic_R1,PKM/Inbox/3_ASME VIII_Design_Basic_R1.pdf,3750689,M,40387,11.026,mixed,Manual,none,2026-04-08T23:20:22.835881+00:00
+5207,4_ASME VIII_Design_Practical_R1,PKM/Inbox/4_ASME VIII_Design_Practical_R1.pdf,1882509,M,24365,13.253,mixed,Manual,none,2026-04-08T23:20:23.956728+00:00
+5208,5_ASME VIII_Fabrication-Inspection_R1,PKM/Inbox/5_ASME VIII_Fabrication-Inspection_R1.pdf,2306232,M,26273,11.666,mixed,Manual,none,2026-04-08T23:20:26.704242+00:00
+5209,6_ASME IX_Welding,PKM/Inbox/6_ASME IX_Welding.pdf,1707491,M,38523,23.103,mixed,Manual,none,2026-04-08T23:20:27.826904+00:00
+5210,ASME B-PV CODE SECTION Ⅷ Div. 1압력용기 코드(2007),PKM/Inbox/ASME B-PV CODE SECTION Ⅷ Div. 1압력용기 코드(2007).pdf,35453616,L,1199923,34.657,mixed,Standard,over_max_pages_estimated,2026-04-08T23:20:36.746155+00:00
+5212,PTB-4-2013_01_GENERAL REQUIREMENTS,PKM/Inbox/PTB-4-2013_01_GENERAL REQUIREMENTS.pdf,256269,S,11370,45.432,mixed,Standard,none,2026-04-08T23:20:38.519481+00:00
+5213,PTB-4-2013_02_EXAMPLE PROBLEM DESCRIPTIONS,PKM/Inbox/PTB-4-2013_02_EXAMPLE PROBLEM DESCRIPTIONS.pdf,123771,S,1961,16.224,mixed,Standard,none,2026-04-08T23:20:39.242724+00:00
+5214,PTB-4-2013_03_MATERIALS REQUIREMENTS,PKM/Inbox/PTB-4-2013_03_MATERIALS REQUIREMENTS.pdf,808215,S,21380,27.088,mixed,Standard,none,2026-04-08T23:20:40.136240+00:00
+5225,code에_따른_wps_pqr_이해와_적용,PKM/Inbox/code에_따른_wps_pqr_이해와_적용.pdf,1471806,M,25363,17.646,mixed,Manual,none,2026-04-08T23:20:50.316232+00:00
+5226,ISO 45001(안전보건경영시스템)_요구사항 해설,PKM/Inbox/ISO 45001(안전보건경영시스템)_요구사항 해설.pdf,41321427,L,35770,0.886,scan-likely,study_note,over_max_pages_estimated,2026-04-08T23:21:07.053788+00:00
+5228,미국의 여름철 고온에 따른 산업안전 문제와 그 대응,PKM/Inbox/미국의 여름철 고온에 따른 산업안전 문제와 그 대응.pdf,980707,S,13012,13.586,mixed,study_note,none,2026-04-08T23:21:20.806314+00:00
+5229,사업체들의 산업안전 활동 최근 동향과 과제,PKM/Inbox/사업체들의 산업안전 활동 최근 동향과 과제.pdf,818746,S,30674,38.364,mixed,study_note,none,2026-04-08T23:21:21.773986+00:00
+5230,산업공학(産業工學)과 산업안전(産業安全),PKM/Inbox/산업공학(産業工學)과 산업안전(産業安全).pdf,1746354,M,13701,8.034,mixed,study_note,none,2026-04-08T23:21:22.870060+00:00
+5231,산업안전 수준 제고를 위한 노사관계의 과제,PKM/Inbox/산업안전 수준 제고를 위한 노사관계의 과제.pdf,1176219,M,19283,16.788,mixed,study_note,none,2026-04-08T23:21:23.918345+00:00
+5235,산업안전기사 2003년 3월,PKM/Inbox/산업안전기사 2003년 3월.pdf,1922633,M,38600,20.558,mixed,study_note,none,2026-04-08T23:21:30.549895+00:00
+5236,산업안전기사_건설공사 안전관리,PKM/Inbox/산업안전기사_건설공사 안전관리.pdf,35080540,L,59699,1.743,scan-likely,study_note,over_max_pages_estimated,2026-04-08T23:21:41.107632+00:00
+5237,4M 위험성평가 기법을 이용한 아차사고의 효과적인 발굴기법,PKM/Inbox/4M 위험성평가 기법을 이용한 아차사고의 효과적인 발굴기법.pdf,1428352,M,17312,12.411,mixed,study_note,none,2026-04-08T23:21:41.748592+00:00
+5238,근로자 노출평가제도 내 위험성평가 방법론의 적용,PKM/Inbox/근로자 노출평가제도 내 위험성평가 방법론의 적용.pdf,393908,S,34424,89.488,born-digital,study_note,none,2026-04-08T23:21:42.648168+00:00
+5239,기계안전의 위험성평가와 안전대책,PKM/Inbox/기계안전의 위험성평가와 안전대책.pdf,271735,S,6373,24.016,mixed,study_note,none,2026-04-08T23:21:43.516587+00:00
+5240,사업장 위험성평가에 관한 법제의 비교법적 고찰,PKM/Inbox/사업장 위험성평가에 관한 법제의 비교법적 고찰.pdf,437377,S,35787,83.786,born-digital,study_note,none,2026-04-08T23:21:44.434222+00:00
+5241,수소충전소 튜브트레일러 누출에 따른 위험성평가,PKM/Inbox/수소충전소 튜브트레일러 누출에 따른 위험성평가.pdf,1091248,M,12461,11.693,mixed,study_note,none,2026-04-08T23:21:45.145330+00:00
+5242,위험성평가 방법 A Literature Review,PKM/Inbox/위험성평가 방법 A Literature Review.pdf,210188,S,2602,12.676,mixed,study_note,none,2026-04-08T23:21:46.011697+00:00
+5243,위험성평가 사업장 구축 및 실행방안,PKM/Inbox/위험성평가 사업장 구축 및 실행방안.pdf,1029859,S,11524,11.458,mixed,study_note,none,2026-04-08T23:21:47.747234+00:00
+5244,산업안전기사_기계 기구 및 설비 안전관리,PKM/Inbox/산업안전기사_기계 기구 및 설비 안전관리.pdf,38948352,L,67629,1.778,scan-likely,study_note,over_max_pages_estimated,2026-04-08T23:21:48.974016+00:00
+5245,위험성평가 제도의 만족도 및 인식도 조사에 관한 연구,PKM/Inbox/위험성평가 제도의 만족도 및 인식도 조사에 관한 연구.pdf,208897,S,1840,9.02,mixed,study_note,none,2026-04-08T23:21:48.978130+00:00
+5246,위험성평가,PKM/Inbox/위험성평가.pdf,4051722,M,8267,2.089,scan-likely,study_note,none,2026-04-08T23:21:50.820505+00:00
+5249,산업안전기사_산업재해 예방 및 안전보건교육,PKM/Inbox/산업안전기사_산업재해 예방 및 안전보건교육.pdf,34442396,L,66969,1.991,scan-likely,study_note,over_max_pages_estimated,2026-04-08T23:21:56.802323+00:00
+5253,산업안전기사_인간공학 및 위험성평가,PKM/Inbox/산업안전기사_인간공학 및 위험성평가.pdf,29835234,L,59514,2.043,scan-likely,study_note,over_max_pages_estimated,2026-04-08T23:22:04.156790+00:00
+5257,Modifying Large Language Model Post-Training for Diverse Creative Writing,PKM/Inbox/Modifying Large Language Model Post-Training for Diverse Creative Writing.pdf,885908,S,91834,106.149,born-digital,Academic_Paper,none,2026-04-08T23:22:06.950818+00:00
+5260,산업안전기사_전기설비 안전관리,PKM/Inbox/산업안전기사_전기설비 안전관리.pdf,39221894,L,80499,2.102,scan-likely,study_note,over_max_pages_estimated,2026-04-08T23:22:12.862475+00:00
+5262,コード X AI_00_はじめに,PKM/Inbox/コード X AI_00_はじめに.pdf,4551317,M,18591,4.183,scan-likely,Reference,none,2026-04-08T23:22:17.418401+00:00
+5263,황현필의 진보를 위한 역사_1장 식민지근대화론 거짓말을 그만하라,PKM/Inbox/황현필의 진보를 위한 역사_1장 식민지근대화론 거짓말을 그만하라.pdf,16365941,L,26144,1.636,scan-likely,Academic_Paper,none,2026-04-08T23:22:41.986015+00:00
+5264,황현필의 진보를 위한 역사_2장 식민지 수탈과 학살의 진실을 말하라,PKM/Inbox/황현필의 진보를 위한 역사_2장 식민지 수탈과 학살의 진실을 말하라.pdf,14154100,L,29575,2.14,scan-likely,Academic_Paper,none,2026-04-08T23:22:43.326498+00:00
+5265,황현필의 진보를 위한 역사_3장 독립운동을 상처 내지 말라,PKM/Inbox/황현필의 진보를 위한 역사_3장 독립운동을 상처 내지 말라.pdf,10398784,M,23381,2.302,scan-likely,Note,none,2026-04-08T23:22:47.235541+00:00
+5266,황현필의 진보를 위한 역사_4장 김구를 모욕하지 말라,PKM/Inbox/황현필의 진보를 위한 역사_4장 김구를 모욕하지 말라.pdf,7617487,M,15170,2.039,scan-likely,Academic_Paper,none,2026-04-08T23:22:47.988096+00:00
+5267,황현필의 진보를 위한 역사_5장 해방정국을 감추지 말라,PKM/Inbox/황현필의 진보를 위한 역사_5장 해방정국을 감추지 말라.pdf,12320415,L,19457,1.617,scan-likely,Note,none,2026-04-08T23:22:49.154378+00:00
+5269,황현필의 진보를 위한 역사_7장 이승만 국부 만들기를 그만두라,PKM/Inbox/황현필의 진보를 위한 역사_7장 이승만 국부 만들기를 그만두라.pdf,14647972,L,34310,2.399,scan-likely,Note,none,2026-04-08T23:22:53.228481+00:00
+5270,황현필의 진보를 위한 역사_8장 6-25전쟁의 사실을은폐하지 말라,PKM/Inbox/황현필의 진보를 위한 역사_8장 6-25전쟁의 사실을은폐하지 말라.pdf,17737798,L,41579,2.4,scan-likely,Note,none,2026-04-08T23:22:59.687925+00:00
+5271,황현필의 진보를 위한 역사_9장 박정희 신격화를 중단하라,PKM/Inbox/황현필의 진보를 위한 역사_9장 박정희 신격화를 중단하라.pdf,16266788,L,29837,1.878,scan-likely,Academic_Paper,none,2026-04-08T23:23:04.499857+00:00
+5272,황현필의 진보를 위한 역사_10장 5-18 광주민주화운동을 폄훼하지 말라,PKM/Inbox/황현필의 진보를 위한 역사_10장 5-18 광주민주화운동을 폄훼하지 말라.pdf,13749286,L,24990,1.861,scan-likely,Note,none,2026-04-08T23:23:09.164908+00:00
+5273,황현필의 진보를 위한 역사_11장 통일을 반대하지 말라,PKM/Inbox/황현필의 진보를 위한 역사_11장 통일을 반대하지 말라.pdf,10701885,L,24856,2.378,scan-likely,Academic_Paper,none,2026-04-08T23:23:13.489086+00:00
+6675,TK-SUP-01_2026년 안전보건 경영목표 달성을 위한 세부 실시 계획서,PKM/Inbox/TK-SUP-01_2026년 안전보건 경영목표 달성을 위한 세부 실시 계획서.pdf,3852353,M,3988,1.06,scan-likely,study_note,none,2026-04-14T06:34:13.166804+00:00
+8852,"TKP-26-0112_신양철강_650,000원","PKM/Inbox/발주/2026-03/TKP-26-0112_신양철강/TKP-26-0112_신양철강_650,000원.pdf",448972,S,1845,4.208,scan-likely,Specification,none,2026-04-20T02:23:47.126481+00:00
+8854,"TKP-26-0114_대연기업_845,000원","PKM/Inbox/발주/2026-03/TKP-26-0114_대연기업/TKP-26-0114_대연기업_845,000원.pdf",447907,S,2006,4.586,scan-likely,Report,none,2026-04-20T02:23:47.132334+00:00
+8857,"TKP-26-0132_신양철강_74,290원 - 복사본","PKM/Inbox/발주/2026-03/TKP-26-0132_신양철강/TKP-26-0132_신양철강_74,290원 - 복사본.pdf",450026,S,1937,4.407,scan-likely,Specification,none,2026-04-20T02:23:47.140119+00:00
+8944,TrusBill print,PKM/Inbox/발주/2026-03/TKP-26-0112_신양철강/TrusBill print.pdf,209087,S,1800,8.815,mixed,Report,none,2026-04-20T05:51:45.513019+00:00

scripts/phase2_backfill.py

@@ -0,0 +1,790 @@
+"""Phase 2 — markdown canonical layer full backfill.
+
+* legacy pending PDF (Phase 1B/1C/1D 후 잔여) 을 야간 sweep + canary 단계로 변환.
+* 1D pilot (25건 controlled_backfill, 23 success / 2 skipped / 0 failed) 가 engineering go signal.
+* handwritten 자동 skip (commit 7d0fca2) 와 marker_worker SKIP_DOC_TYPES / MAX_PAGES guard 모두 운영 시점 가드. 본 스크립트는 inventory · sample · enqueue 만 담당.
+
+Subcommands:
+  inventory         pending PDFs 의 dry-run inventory CSV 작성 (skip forecast 포함)
+  select-canary     inventory 에서 stratified 40건 canary sample CSV 작성 (재현성 시드)
+  enqueue           sample CSV 의 doc_id 들을 markdown 큐에 enqueue (one-shot, --no-dry-run 필요)
+  nightly-enqueue   야간 sweep — disable flag / marker ready / active-queue threshold / DB pool 가드 후 limit 만큼 enqueue + log_tsv
+  post-report       최종 결과 CSV + markdown 요약 (1D baseline 비교 포함)
+
+실행 (GPU 서버):
+  docker exec hyungi_document_server-fastapi-1 python /app/scripts/phase2_backfill.py inventory \
+    --output /app/evals/markdown/phase2_inventory.csv
+
+  docker exec hyungi_document_server-fastapi-1 python /app/scripts/phase2_backfill.py select-canary \
+    --inventory /app/evals/markdown/phase2_inventory.csv \
+    --output /app/evals/markdown/phase2_canary_sample.csv \
+    --seed 20260503
+
+  docker exec hyungi_document_server-fastapi-1 python /app/scripts/phase2_backfill.py enqueue \
+    --csv /app/evals/markdown/phase2_canary_sample.csv --no-dry-run
+
+  # cron (nightly):
+  docker exec hyungi_document_server-fastapi-1 python /app/scripts/phase2_backfill.py nightly-enqueue \
+    --limit 50 --max-active-queue 5 \
+    --log-tsv /app/evals/markdown/phase2_nightly_log.tsv
+
+  docker exec hyungi_document_server-fastapi-1 python /app/scripts/phase2_backfill.py post-report \
+    --output-csv /app/evals/markdown/phase2_post_report.csv \
+    --output-md /app/evals/markdown/phase2_post_report.md
+"""
+
+import argparse
+import asyncio
+import csv
+import json
+import os
+import random
+import re
+import sys
+import urllib.error
+import urllib.request
+from collections import Counter, defaultdict
+from datetime import datetime, timezone
+from pathlib import Path
+
+# fastapi 컨테이너 WORKDIR=/app — `from models...` import 가능하게 path 추가.
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), ".."))
+
+from sqlalchemy import func, select, text
+from sqlalchemy.ext.asyncio import AsyncSession, async_sessionmaker, create_async_engine
+
+# marker_worker 룰 미러 (1D pilot 패턴 따름 — drift 회피 위해 한 곳).
+SKIP_DOC_TYPES = {
+    "발주서", "세금계산서", "명세표",
+    "Invoice", "Purchase_Order", "Estimate", "Statement",
+}
+
+# handwritten 자동 skip 키워드 (commit 7d0fca2 의 marker_worker 정의와 동일).
+HANDWRITTEN_REGEX = re.compile(r"필기|손글씨|handwritten|handwriting", re.IGNORECASE)
+
+# Inventory 의 page_count proxy — file_size 가 25MB 초과면 MAX_PAGES=200 초과 가능성 높음.
+# 실제 marker_worker 가 PyMuPDF 로 page_count 확인 후 결정. 본 forecast 는 plan budget 추정용.
+MAX_PAGES_FILESIZE_PROXY = 25 * 1024 * 1024
+
+# stuck 판단 임계 — 평균 변환 시간 ~48s (1D 실측). 30분이면 의심, 60분이면 stuck likely.
+STUCK_THRESHOLD_MIN = 60.0
+
+# Disable flag — 사용자 또는 자동 abort 가드가 생성. cron 매 사이클 시작 시 체크.
+DISABLE_FLAG_PATH = Path("/tmp/phase2_disable")  # 컨테이너 내부 경로 (호스트 mount 별도)
+
+# marker-service ready endpoint (Phase 1B 명세).
+MARKER_READY_URL = "http://marker-service:3300/ready"
+
+
+# ─── helpers ────────────────────────────────────────────────────────────
+
+
+def _build_engine():
+    db_url = os.environ["DATABASE_URL"]
+    return create_async_engine(db_url, pool_pre_ping=True)
+
+
+def _file_size_band(file_size: int | None) -> str:
+    if file_size is None:
+        return "unknown"
+    if file_size < 1 * 1024 * 1024:
+        return "S"
+    if file_size < 10 * 1024 * 1024:
+        return "M"
+    return "L"
+
+
+def _text_density(text_len: int, file_size: int | None) -> float | None:
+    if not file_size or file_size <= 0:
+        return None
+    return text_len / (file_size / 1024.0)
+
+
+def _text_density_band(density: float | None) -> str:
+    if density is None:
+        return "unknown"
+    if density < 5.0:
+        return "scan-likely"
+    if density < 50.0:
+        return "mixed"
+    return "born-digital"
+
+
+def _forecast_skip_reason(file_format: str | None, doc_type: str | None,
+                          title: str | None, file_path: str | None,
+                          file_size: int | None) -> str:
+    if file_format and file_format.lower() != "pdf":
+        return "unsupported_extension"
+    if doc_type and doc_type in SKIP_DOC_TYPES:
+        return "doctype_skip"
+    blob = " ".join(filter(None, [title or "", file_path or ""]))
+    if HANDWRITTEN_REGEX.search(blob):
+        return "handwritten_hint"
+    if file_size and file_size > MAX_PAGES_FILESIZE_PROXY:
+        return "over_max_pages_estimated"
+    return "none"
+
+
+def _percentile(values: list[float], p: float) -> float | None:
+    if not values:
+        return None
+    s = sorted(values)
+    idx = max(0, min(len(s) - 1, int(len(s) * p)))
+    return s[idx]
+
+
+def _check_marker_ready() -> bool:
+    try:
+        with urllib.request.urlopen(MARKER_READY_URL, timeout=5) as resp:
+            data = json.load(resp)
+        return data.get("status") == "ready"
+    except (urllib.error.URLError, urllib.error.HTTPError, json.JSONDecodeError, OSError):
+        return False
+
+
+# ─── inventory ──────────────────────────────────────────────────────────
+
+INVENTORY_COLUMNS = [
+    "doc_id", "title", "file_path", "file_size", "file_size_band",
+    "text_len", "text_density", "text_density_band",
+    "doc_type", "forecast_skip_reason", "created_at",
+]
+
+
+async def cmd_inventory(output: Path) -> None:
+    engine = _build_engine()
+    Session = async_sessionmaker(engine, class_=AsyncSession, expire_on_commit=False)
+
+    from models.document import Document  # type: ignore
+
+    async with Session() as session:
+        rows = (
+            await session.execute(
+                select(
+                    Document.id, Document.title, Document.file_path, Document.file_size,
+                    Document.file_format, Document.document_type,
+                    Document.extracted_text, Document.created_at,
+                ).where(
+                    Document.deleted_at.is_(None),
+                    Document.file_format == "pdf",
+                    Document.md_status == "pending",
+                )
+                .order_by(Document.created_at.asc())
+            )
+        ).all()
+
+    enriched = []
+    for r in rows:
+        text_len = len(r.extracted_text or "")
+        density = _text_density(text_len, r.file_size)
+        enriched.append({
+            "doc_id": r.id,
+            "title": (r.title or ""),
+            "file_path": (r.file_path or ""),
+            "file_size": r.file_size or 0,
+            "file_size_band": _file_size_band(r.file_size),
+            "text_len": text_len,
+            "text_density": round(density, 3) if density is not None else "",
+            "text_density_band": _text_density_band(density),
+            "doc_type": r.document_type or "",
+            "forecast_skip_reason": _forecast_skip_reason(
+                r.file_format, r.document_type, r.title, r.file_path, r.file_size
+            ),
+            "created_at": r.created_at.isoformat() if r.created_at else "",
+        })
+
+    output.parent.mkdir(parents=True, exist_ok=True)
+    with output.open("w", newline="", encoding="utf-8") as f:
+        writer = csv.DictWriter(f, fieldnames=INVENTORY_COLUMNS, extrasaction="ignore")
+        writer.writeheader()
+        for row in enriched:
+            writer.writerow(row)
+
+    skip_dist = Counter(r["forecast_skip_reason"] for r in enriched)
+    band_dist = Counter(r["file_size_band"] for r in enriched)
+    density_dist = Counter(r["text_density_band"] for r in enriched)
+    type_dist = Counter(r["doc_type"] or "(NULL)" for r in enriched)
+
+    print(f"\n## Phase 2 inventory")
+    print(f"  total pending PDFs: {len(enriched)}")
+    print(f"  output: {output}")
+    print(f"\n  forecast_skip_reason:")
+    for k, v in skip_dist.most_common():
+        print(f"    {k:<30} {v:>4}")
+    print(f"\n  file_size_band: ", dict(band_dist))
+    print(f"  text_density_band: ", dict(density_dist))
+    print(f"\n  doc_type top 10:")
+    for k, v in type_dist.most_common(10):
+        print(f"    {k:<25} {v:>4}")
+
+    convert_target = sum(1 for r in enriched if r["forecast_skip_reason"] == "none")
+    print(f"\n  변환 시도 후보 (forecast_skip_reason='none'): {convert_target}건")
+    print(f"  즉시 skip 예상: {len(enriched) - convert_target}건")
+
+    await engine.dispose()
+
+
+# ─── select-canary ──────────────────────────────────────────────────────
+
+CANARY_COLUMNS = [
+    "doc_id", "title", "file_size", "file_size_band",
+    "text_density", "text_density_band", "doc_type", "bucket_label",
+]
+
+
+def _select_canary_buckets(candidates: list[dict], rng: random.Random) -> list[dict]:
+    """40건 stratified — plan §"Sample budget" 표 참조.
+
+    large 6 / scan_likely 2 / study_note 10 / Academic_Paper 8 / Reference 6 /
+    {Standard,Manual,Specification} 4 / {Note,Report,Memo,NULL} 4 = 40
+    """
+    selected: list[dict] = []
+    used: set[int] = set()
+
+    def take(pool_filter, n: int, label: str) -> None:
+        avail = [c for c in candidates if c["doc_id"] not in used and pool_filter(c)]
+        rng.shuffle(avail)
+        for c in avail[:n]:
+            picked = dict(c)
+            picked["bucket_label"] = label
+            selected.append(picked)
+            used.add(c["doc_id"])
+
+    take(lambda c: c["file_size_band"] == "L", 6, "large")
+    take(lambda c: c["text_density_band"] == "scan-likely", 2, "scan_likely")
+    take(lambda c: c["doc_type"] == "study_note" and c["text_density_band"] == "born-digital", 10, "study_note")
+    take(lambda c: c["doc_type"] == "Academic_Paper" and c["text_density_band"] == "born-digital", 8, "Academic_Paper")
+    take(lambda c: c["doc_type"] == "Reference" and c["text_density_band"] == "born-digital", 6, "Reference")
+    take(lambda c: c["doc_type"] in {"Standard", "Manual", "Specification"}, 4, "tech_doc")
+    take(lambda c: (c["doc_type"] in {"Note", "Report", "Memo"} or not c["doc_type"]), 4, "minor_doc")
+
+    if len(selected) < 40:
+        leftover = [c for c in candidates if c["doc_id"] not in used]
+        rng.shuffle(leftover)
+        for c in leftover[: 40 - len(selected)]:
+            picked = dict(c)
+            picked["bucket_label"] = "filler"
+            selected.append(picked)
+            used.add(c["doc_id"])
+
+    return selected[:40]
+
+
+def cmd_select_canary(inventory: Path, output: Path, seed: int) -> None:
+    if not inventory.is_file():
+        print(f"[error] inventory CSV 없음: {inventory}", file=sys.stderr)
+        sys.exit(1)
+
+    candidates: list[dict] = []
+    with inventory.open(encoding="utf-8") as f:
+        reader = csv.DictReader(f)
+        for row in reader:
+            if row["forecast_skip_reason"] != "none":
+                continue
+            candidates.append({
+                "doc_id": int(row["doc_id"]),
+                "title": row["title"],
+                "file_size": int(row["file_size"]) if row["file_size"] else 0,
+                "file_size_band": row["file_size_band"],
+                "text_density": row["text_density"],
+                "text_density_band": row["text_density_band"],
+                "doc_type": row["doc_type"],
+            })
+
+    rng = random.Random(seed)
+    selected = _select_canary_buckets(candidates, rng)
+
+    output.parent.mkdir(parents=True, exist_ok=True)
+    with output.open("w", newline="", encoding="utf-8") as f:
+        writer = csv.DictWriter(f, fieldnames=CANARY_COLUMNS, extrasaction="ignore")
+        writer.writeheader()
+        for s in selected:
+            writer.writerow({col: s.get(col, "") for col in CANARY_COLUMNS})
+
+    bucket_dist = Counter(s["bucket_label"] for s in selected)
+    print(f"\n## Canary sample (seed={seed})")
+    print(f"  selected: {len(selected)} / 40")
+    print(f"  output: {output}")
+    print(f"\n  bucket_label:")
+    for k, v in bucket_dist.most_common():
+        print(f"    {k:<20} {v:>3}")
+    print(f"\n  doc_id list:")
+    for s in sorted(selected, key=lambda x: (x["bucket_label"], x["doc_id"])):
+        print(f"    {s['doc_id']:>5}  [{s['bucket_label']:<14}]  {(s['title'] or '-')[:60]}")
+
+
+# ─── enqueue (one-shot from CSV) ─────────────────────────────────────────
+
+
+async def cmd_enqueue(csv_path: Path, dry_run: bool) -> None:
+    if not csv_path.is_file():
+        print(f"[error] sample CSV 없음: {csv_path}", file=sys.stderr)
+        sys.exit(1)
+
+    ids: list[int] = []
+    with csv_path.open(encoding="utf-8") as f:
+        reader = csv.DictReader(f)
+        for row in reader:
+            ids.append(int(row["doc_id"]))
+
+    if not ids:
+        print("[abort] enqueue 대상 없음.")
+        return
+
+    print(f"[targets] {len(ids)} doc_ids: {ids[:10]}{' …' if len(ids) > 10 else ''}")
+
+    if dry_run:
+        print("[dry-run] --no-dry-run 으로 다시 실행하면 실제 enqueue.")
+        return
+
+    if not _check_marker_ready():
+        print("[abort] marker-service /ready 가 ready 가 아님. enqueue 중단.")
+        sys.exit(2)
+
+    engine = _build_engine()
+    Session = async_sessionmaker(engine, class_=AsyncSession, expire_on_commit=False)
+    from models.queue import enqueue_stage  # type: ignore
+
+    enqueued, skipped = [], []
+    async with Session() as session:
+        for doc_id in ids:
+            ok = await enqueue_stage(session, doc_id, "markdown")
+            (enqueued if ok else skipped).append(doc_id)
+        await session.commit()
+
+    print(f"\nenqueued: {len(enqueued)}, skipped (이미 active): {len(skipped)}")
+    if skipped:
+        print(f"  skipped ids: {skipped[:20]}{' …' if len(skipped) > 20 else ''}")
+    await engine.dispose()
+
+
+# ─── nightly-enqueue (cron-friendly) ─────────────────────────────────────
+
+NIGHTLY_TSV_COLUMNS = [
+    "date", "enqueued", "active_queue_at_start", "active_queue_oldest_age_min",
+    "pending_pool_remaining", "abort_reason", "marker_ready",
+]
+
+
+def _append_tsv(log_tsv: Path, row: dict) -> None:
+    log_tsv.parent.mkdir(parents=True, exist_ok=True)
+    new_file = not log_tsv.exists() or log_tsv.stat().st_size == 0
+    with log_tsv.open("a", encoding="utf-8") as f:
+        if new_file:
+            f.write("\t".join(NIGHTLY_TSV_COLUMNS) + "\n")
+        f.write("\t".join(str(row.get(col, "")) for col in NIGHTLY_TSV_COLUMNS) + "\n")
+
+
+async def cmd_nightly_enqueue(limit: int, max_active_queue: int, log_tsv: Path,
+                              dry_run: bool) -> None:
+    today = datetime.now().strftime("%Y-%m-%d")
+    base_row = {
+        "date": today,
+        "enqueued": 0,
+        "active_queue_at_start": "",
+        "active_queue_oldest_age_min": "",
+        "pending_pool_remaining": "",
+        "abort_reason": "",
+        "marker_ready": "",
+    }
+
+    # 1. disable flag
+    if DISABLE_FLAG_PATH.exists():
+        base_row["abort_reason"] = "disable_flag"
+        _append_tsv(log_tsv, base_row)
+        print(f"[abort] disable flag 존재: {DISABLE_FLAG_PATH}")
+        return
+
+    # 2. marker ready
+    marker_ready = _check_marker_ready()
+    base_row["marker_ready"] = 1 if marker_ready else 0
+    if not marker_ready:
+        base_row["abort_reason"] = "marker_unhealthy"
+        _append_tsv(log_tsv, base_row)
+        print("[abort] marker-service /ready 가 ready 아님")
+        return
+
+    engine = _build_engine()
+    Session = async_sessionmaker(engine, class_=AsyncSession, expire_on_commit=False)
+
+    from models.document import Document  # type: ignore
+    from models.queue import ProcessingQueue, enqueue_stage  # type: ignore
+
+    # 3. active queue check + oldest age
+    async with Session() as session:
+        active_stats = (await session.execute(text("""
+            SELECT COUNT(*) AS active_count,
+                   COALESCE(EXTRACT(EPOCH FROM (NOW() - MIN(COALESCE(started_at, created_at))))/60, 0) AS oldest_age_min
+            FROM processing_queue
+            WHERE stage='markdown' AND status IN ('pending','processing')
+        """))).first()
+        active_count = int(active_stats[0])
+        oldest_age = float(active_stats[1])
+
+    base_row["active_queue_at_start"] = active_count
+    base_row["active_queue_oldest_age_min"] = round(oldest_age, 1)
+
+    if active_count > max_active_queue:
+        base_row["abort_reason"] = "active_queue_threshold"
+        _append_tsv(log_tsv, base_row)
+        print(f"[skip] active queue {active_count} > threshold {max_active_queue} (oldest {oldest_age:.1f}min)")
+        if oldest_age > STUCK_THRESHOLD_MIN:
+            print(f"[warn] oldest active row age {oldest_age:.1f}min > {STUCK_THRESHOLD_MIN}min — possible stuck")
+        await engine.dispose()
+        return
+
+    # 4. pending pool query
+    async with Session() as session:
+        pool_rows = (
+            await session.execute(text(f"""
+                SELECT id FROM documents
+                WHERE deleted_at IS NULL
+                  AND file_format='pdf'
+                  AND md_status='pending'
+                  AND id NOT IN (
+                    SELECT document_id FROM processing_queue
+                    WHERE stage='markdown' AND status IN ('pending','processing')
+                  )
+                ORDER BY created_at ASC
+                LIMIT {int(limit)}
+            """))
+        ).all()
+        pool_ids = [r[0] for r in pool_rows]
+
+    if not pool_ids:
+        base_row["abort_reason"] = "pool_empty"
+        base_row["pending_pool_remaining"] = 0
+        _append_tsv(log_tsv, base_row)
+        print("[done] pending pool empty — Phase 2 backfill 자연 완료 신호.")
+        await engine.dispose()
+        return
+
+    if dry_run:
+        print(f"[dry-run] would enqueue {len(pool_ids)} ids: {pool_ids[:10]}{' …' if len(pool_ids) > 10 else ''}")
+        await engine.dispose()
+        return
+
+    # 5. enqueue
+    enqueued, skipped = [], []
+    async with Session() as session:
+        for doc_id in pool_ids:
+            ok = await enqueue_stage(session, doc_id, "markdown")
+            (enqueued if ok else skipped).append(doc_id)
+        await session.commit()
+
+    # 6. pending pool remaining (post-enqueue)
+    async with Session() as session:
+        remaining_count = (
+            await session.execute(text("""
+                SELECT COUNT(*) FROM documents d
+                WHERE d.deleted_at IS NULL
+                  AND d.file_format='pdf'
+                  AND d.md_status='pending'
+                  AND d.id NOT IN (
+                    SELECT document_id FROM processing_queue
+                    WHERE stage='markdown' AND status IN ('pending','processing')
+                  )
+            """))
+        ).scalar()
+
+    base_row["enqueued"] = len(enqueued)
+    base_row["pending_pool_remaining"] = int(remaining_count or 0)
+    _append_tsv(log_tsv, base_row)
+
+    print(f"\n[ok] enqueued={len(enqueued)} skipped={len(skipped)} remaining_pool={remaining_count}")
+    await engine.dispose()
+
+
+# ─── post-report ────────────────────────────────────────────────────────
+
+POST_REPORT_COLUMNS = [
+    "doc_id", "title", "final_md_status", "md_extraction_engine", "md_extraction_engine_version",
+    "elapsed_ms_estimate", "text_length_ratio", "markdown_heading_count",
+    "markdown_table_row_count", "markdown_image_count", "warnings", "phase2_processed_at",
+]
+
+# 1D pilot baseline (project_markdown_canonical_layer.md, Phase 1D 결과 섹션).
+BASELINE_1D = {
+    "success_rate": 0.92,       # 23/25
+    "skipped_rate": 0.08,
+    "failed_rate": 0.0,
+    "elapsed_p50_ms": 34000,
+    "elapsed_p90_ms": 112000,
+    "text_length_ratio_p50": 1.15,
+    "warnings_heading_jump_pct": 0.86,    # 24/28
+    "warnings_low_image_alt_pct": 0.89,   # 25/28
+}
+
+
+async def cmd_post_report(output_csv: Path, output_md: Path,
+                          phase2_start: str | None) -> None:
+    """Phase 2 sweep 결과 집계 + 1D baseline 비교.
+
+    phase2_start: ISO timestamp (예: '2026-05-03T00:00:00Z'). 이후 처리된 doc 만 집계.
+                  None 이면 sample CSV 들에서 doc_id union 으로 한정 (별도 인자 미지원 — 기본 = NULL = 전 history).
+                  실용적으로는 phase2_start 사용 권장 (Phase 2 코드 push 후 시각).
+    """
+    engine = _build_engine()
+    Session = async_sessionmaker(engine, class_=AsyncSession, expire_on_commit=False)
+
+    from models.document import Document          # type: ignore
+    from models.queue import ProcessingQueue      # type: ignore
+
+    async with Session() as session:
+        cond = [Document.deleted_at.is_(None), Document.file_format == "pdf"]
+        if phase2_start:
+            cond.append(Document.md_generated_at >= phase2_start)
+        else:
+            cond.append(Document.md_generated_at.is_not(None))
+
+        rows = (
+            await session.execute(
+                select(
+                    Document.id, Document.title, Document.md_status,
+                    Document.md_extraction_engine, Document.md_extraction_engine_version,
+                    Document.md_extraction_quality, Document.md_extraction_error,
+                    Document.md_generated_at,
+                ).where(*cond)
+            )
+        ).all()
+
+        ids = [r.id for r in rows]
+        if ids:
+            q_rows = (
+                await session.execute(
+                    select(
+                        ProcessingQueue.document_id,
+                        ProcessingQueue.started_at,
+                        ProcessingQueue.completed_at,
+                    ).where(
+                        ProcessingQueue.document_id.in_(ids),
+                        ProcessingQueue.stage == "markdown",
+                    )
+                )
+            ).all()
+        else:
+            q_rows = []
+
+    elapsed: dict[int, float] = {}
+    for q in q_rows:
+        if q.started_at and q.completed_at:
+            elapsed[q.document_id] = (q.completed_at - q.started_at).total_seconds() * 1000
+
+    # CSV
+    enriched_rows = []
+    for r in rows:
+        q = r.md_extraction_quality or {}
+        m = q.get("metrics") if isinstance(q, dict) else {}
+        m = m or {}
+        warnings = q.get("warnings") if isinstance(q, dict) else []
+        enriched_rows.append({
+            "doc_id": r.id,
+            "title": (r.title or "")[:120],
+            "final_md_status": r.md_status,
+            "md_extraction_engine": r.md_extraction_engine or "",
+            "md_extraction_engine_version": r.md_extraction_engine_version or "",
+            "elapsed_ms_estimate": int(elapsed.get(r.id, 0)) if r.id in elapsed else "",
+            "text_length_ratio": m.get("text_length_ratio", "") if m else "",
+            "markdown_heading_count": m.get("markdown_heading_count", "") if m else "",
+            "markdown_table_row_count": m.get("markdown_table_row_count", "") if m else "",
+            "markdown_image_count": m.get("markdown_image_count", "") if m else "",
+            "warnings": ",".join(warnings) if isinstance(warnings, list) else "",
+            "phase2_processed_at": r.md_generated_at.isoformat() if r.md_generated_at else "",
+        })
+
+    output_csv.parent.mkdir(parents=True, exist_ok=True)
+    with output_csv.open("w", newline="", encoding="utf-8") as f:
+        writer = csv.DictWriter(f, fieldnames=POST_REPORT_COLUMNS, extrasaction="ignore")
+        writer.writeheader()
+        for row in enriched_rows:
+            writer.writerow(row)
+
+    # Markdown summary
+    total = len(rows)
+    by_status = Counter(r.md_status for r in rows)
+    success_n = by_status.get("success", 0)
+    skipped_n = by_status.get("skipped", 0)
+    failed_n = by_status.get("failed", 0)
+
+    success_rows = [r for r in rows if r.md_status == "success"]
+    text_ratios = []
+    heading_jump_count = 0
+    low_image_alt_count = 0
+    new_warnings: Counter = Counter()
+    KNOWN_WARNINGS = {"heading_hierarchy_jump", "low_image_alt_text_ratio"}
+    for r in success_rows:
+        q = r.md_extraction_quality or {}
+        m = q.get("metrics") if isinstance(q, dict) else {}
+        m = m or {}
+        if isinstance(m.get("text_length_ratio"), (int, float)):
+            text_ratios.append(float(m["text_length_ratio"]))
+        warnings = q.get("warnings") if isinstance(q, dict) else []
+        if isinstance(warnings, list):
+            if "heading_hierarchy_jump" in warnings:
+                heading_jump_count += 1
+            if "low_image_alt_text_ratio" in warnings:
+                low_image_alt_count += 1
+            for w in warnings:
+                if w not in KNOWN_WARNINGS:
+                    new_warnings[w] += 1
+
+    elapsed_succ = [elapsed[r.id] for r in success_rows if r.id in elapsed]
+    elapsed_p50 = _percentile(elapsed_succ, 0.5)
+    elapsed_p90 = _percentile(elapsed_succ, 0.9)
+    text_ratio_p50 = _percentile(text_ratios, 0.5)
+
+    skip_reasons = Counter()
+    for r in rows:
+        if r.md_status == "skipped":
+            reason = (r.md_extraction_error or "unknown").split(":", 1)[0]
+            skip_reasons[reason.strip()] += 1
+
+    failed_rows = [r for r in rows if r.md_status == "failed"]
+
+    outliers = []
+    for r in success_rows:
+        e = elapsed.get(r.id)
+        if e and e > 300_000:
+            outliers.append((r.id, f"elapsed_ms={int(e)}"))
+        q = r.md_extraction_quality or {}
+        m = q.get("metrics") if isinstance(q, dict) else {}
+        m = m or {}
+        ratio = m.get("text_length_ratio")
+        if isinstance(ratio, (int, float)) and (ratio < 0.5 or ratio > 10):
+            outliers.append((r.id, f"text_length_ratio={ratio:.2f}"))
+
+    def pct(n: int, d: int) -> str:
+        return f"{n/d*100:.0f}%" if d else "-"
+
+    def delta_pp(actual_n: int, actual_d: int, baseline: float) -> str:
+        if not actual_d:
+            return "-"
+        return f"{(actual_n/actual_d - baseline)*100:+.1f}pp"
+
+    def fmt_num(v) -> str:
+        if v is None:
+            return "-"
+        if isinstance(v, float):
+            return f"{v:.2f}"
+        return str(v)
+
+    scope_line = "Scope: file_format='pdf' AND deleted_at IS NULL"
+    if phase2_start:
+        scope_line += f" AND md_generated_at >= {phase2_start}"
+    else:
+        scope_line += " AND md_generated_at IS NOT NULL"
+
+    md_lines = [
+        "# Phase 2 Post-report",
+        "",
+        f"Generated: {datetime.now(timezone.utc).isoformat()}",
+        scope_line,
+        "",
+        f"## 처리 분포 (총 {total}건)",
+        "| status | count | rate |",
+        "|---|---|---|",
+        f"| success | {success_n} | {pct(success_n, total)} |",
+        f"| skipped | {skipped_n} | {pct(skipped_n, total)} |",
+        f"| failed | {failed_n} | {pct(failed_n, total)} |",
+        "",
+        "## vs 1D baseline 비교",
+        "| 메트릭 | 1D | Phase 2 | delta |",
+        "|---|---|---|---|",
+        f"| success rate | 92% | {pct(success_n, total)} | {delta_pp(success_n, total, BASELINE_1D['success_rate'])} |",
+        f"| skipped rate | 8% | {pct(skipped_n, total)} | {delta_pp(skipped_n, total, BASELINE_1D['skipped_rate'])} |",
+        f"| failed rate | 0% | {pct(failed_n, total)} | {delta_pp(failed_n, total, BASELINE_1D['failed_rate'])} |",
+        f"| elapsed_ms p50 | 34000 | {fmt_num(int(elapsed_p50)) if elapsed_p50 else '-'} | - |",
+        f"| elapsed_ms p90 | 112000 | {fmt_num(int(elapsed_p90)) if elapsed_p90 else '-'} | - |",
+        f"| text_length_ratio p50 | 1.15 | {fmt_num(text_ratio_p50)} | - |",
+        f"| warnings: heading_hierarchy_jump | 86% | {pct(heading_jump_count, success_n)} | - |",
+        f"| warnings: low_image_alt_text_ratio | 89% | {pct(low_image_alt_count, success_n)} | - |",
+        "",
+        "## skip reason 분포",
+    ]
+    for reason, c in skip_reasons.most_common():
+        md_lines.append(f"- `{reason}`: {c}건")
+    if not skip_reasons:
+        md_lines.append("- (no skipped)")
+    md_lines += [
+        f"",
+        f"## failed 케이스 ({len(failed_rows)}건)",
+    ]
+    for r in failed_rows:
+        md_lines.append(f"- doc {r.id} `{(r.title or '-')[:60]}`: {r.md_extraction_error or '-'}")
+    if not failed_rows:
+        md_lines.append("- (no failed)")
+    md_lines += [
+        f"",
+        f"## outlier ({len(outliers)}건)",
+    ]
+    for doc_id, note in outliers[:30]:
+        md_lines.append(f"- doc {doc_id}: {note}")
+    if not outliers:
+        md_lines.append("- (no outliers)")
+    if new_warnings:
+        md_lines += [f"", f"## 신규 warning 종류 (1D 미관측)"]
+        for w, c in new_warnings.most_common():
+            md_lines.append(f"- `{w}`: {c}건")
+
+    output_md.parent.mkdir(parents=True, exist_ok=True)
+    output_md.write_text("\n".join(md_lines) + "\n", encoding="utf-8")
+
+    print(f"\n## Phase 2 post-report")
+    print(f"  CSV: {output_csv}")
+    print(f"  MD:  {output_md}")
+    print(f"  scope: {total} rows (success {success_n} / skipped {skipped_n} / failed {failed_n})")
+
+    await engine.dispose()
+
+
+# ─── main ───────────────────────────────────────────────────────────────
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(
+        description="Phase 2 markdown canonical layer full backfill",
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+    )
+    sub = parser.add_subparsers(dest="cmd", required=True)
+
+    p_inv = sub.add_parser("inventory", help="pending PDFs dry-run inventory CSV")
+    p_inv.add_argument("--output", type=Path, required=True)
+
+    p_can = sub.add_parser("select-canary", help="stratified 40 canary sample CSV")
+    p_can.add_argument("--inventory", type=Path, required=True)
+    p_can.add_argument("--output", type=Path, required=True)
+    p_can.add_argument("--seed", type=int, default=20260503)
+
+    p_enq = sub.add_parser("enqueue", help="enqueue from sample CSV (one-shot)")
+    p_enq.add_argument("--csv", type=Path, required=True)
+    g = p_enq.add_mutually_exclusive_group()
+    g.add_argument("--dry-run", dest="dry_run", action="store_true", default=True)
+    g.add_argument("--no-dry-run", dest="dry_run", action="store_false")
+
+    p_nig = sub.add_parser("nightly-enqueue", help="nightly cron sweep")
+    p_nig.add_argument("--limit", type=int, default=50)
+    p_nig.add_argument("--max-active-queue", type=int, default=5)
+    p_nig.add_argument("--log-tsv", type=Path, required=True)
+    p_nig.add_argument("--dry-run", action="store_true", default=False)
+
+    p_rep = sub.add_parser("post-report", help="final results CSV + markdown summary")
+    p_rep.add_argument("--output-csv", type=Path, required=True)
+    p_rep.add_argument("--output-md", type=Path, required=True)
+    p_rep.add_argument("--phase2-start", type=str, default=None,
+                       help="ISO timestamp; only docs with md_generated_at >= this are counted")
+
+    args = parser.parse_args()
+    if args.cmd == "inventory":
+        asyncio.run(cmd_inventory(args.output))
+    elif args.cmd == "select-canary":
+        cmd_select_canary(args.inventory, args.output, args.seed)
+    elif args.cmd == "enqueue":
+        asyncio.run(cmd_enqueue(args.csv, args.dry_run))
+    elif args.cmd == "nightly-enqueue":
+        asyncio.run(cmd_nightly_enqueue(args.limit, args.max_active_queue, args.log_tsv, args.dry_run))
+    elif args.cmd == "post-report":
+        asyncio.run(cmd_post_report(args.output_csv, args.output_md, args.phase2_start))
+
+
+if __name__ == "__main__":
+    main()