add download.py

download.py

@@ -0,0 +1,286 @@
+"""
+Stage 2 — Download PDF dari metadata yang dikumpulkan collect.py.
+
+Fitur:
+  - Resumable: skip file yang sudah ada
+  - Parallel: ThreadPoolExecutor (configurable --workers)
+  - Retry: 3x dengan exponential backoff
+  - Validasi: cek magic bytes %PDF dan ukuran minimum
+  - Laporan: download_report.json
+
+Usage:
+  python3 download.py
+  python3 download.py --workers 8             # lebih banyak thread
+  python3 download.py --input custom.json     # JSON lain
+  python3 download.py --venues SIGIR ACL      # venue tertentu saja
+  python3 download.py --limit 5              # max 5 paper per venue
+"""
+
+import argparse
+import json
+import os
+import threading
+import time
+from concurrent.futures import ThreadPoolExecutor, as_completed
+from pathlib import Path
+
+import requests
+
+# ─── CONFIG ──────────────────────────────────────────────────────────────────
+
+DEFAULT_INPUT   = Path("papers_metadata.json")
+DEFAULT_OUT_DIR = Path("papers_2025")
+DEFAULT_WORKERS = 5
+DEFAULT_REPORT  = Path("download_report.json")
+MIN_PDF_BYTES   = 10_000
+
+HEADERS = {"User-Agent": "IR-course-downloader/1.0 (academic research)"}
+
+print_lock = threading.Lock()
+
+
+def log(msg: str):
+    with print_lock:
+        print(msg, flush=True)
+
+
+# ─── DOWNLOAD ────────────────────────────────────────────────────────────────
+
+def safe_filename(title: str, idx: int) -> str:
+    clean = "".join(c if c.isalnum() or c in " -_" else "" for c in title)
+    slug  = clean.strip().replace(" ", "_")[:60]
+    return f"{idx:03d}_{slug}.pdf"
+
+
+def download_pdf(url: str, dest: Path, retries: int = 3) -> tuple[bool, str]:
+    """
+    Download satu PDF.
+    Return (sukses: bool, pesan: str)
+    """
+    for attempt in range(retries):
+        try:
+            resp = requests.get(url, headers=HEADERS, timeout=45, stream=True)
+            resp.raise_for_status()
+
+            dest.parent.mkdir(parents=True, exist_ok=True)
+            with open(dest, "wb") as f:
+                for chunk in resp.iter_content(8192):
+                    f.write(chunk)
+
+            size = dest.stat().st_size
+            if size < MIN_PDF_BYTES:
+                dest.unlink(missing_ok=True)
+                return False, f"terlalu kecil ({size} bytes)"
+
+            # Validasi magic bytes %PDF
+            with open(dest, "rb") as f:
+                magic = f.read(4)
+            if magic != b"%PDF":
+                dest.unlink(missing_ok=True)
+                return False, f"bukan PDF (magic={magic!r})"
+
+            return True, f"{size // 1024} KB"
+
+        except requests.HTTPError as e:
+            code = e.response.status_code if e.response else 0
+            wait = 3 * (2 ** attempt)
+            if code == 429:
+                wait = max(wait, 30)
+            if attempt < retries - 1:
+                time.sleep(wait)
+            else:
+                dest.unlink(missing_ok=True)
+                return False, f"HTTP {code}"
+
+        except Exception as e:
+            dest.unlink(missing_ok=True)
+            if attempt < retries - 1:
+                time.sleep(3 * (attempt + 1))
+            else:
+                return False, f"{type(e).__name__}: {e}"
+
+    return False, "semua retry gagal"
+
+
+# ─── TASK RUNNER ─────────────────────────────────────────────────────────────
+
+def run_downloads(tasks: list[dict], workers: int) -> list[dict]:
+    """
+    Jalankan download secara paralel.
+    Setiap task: {idx, venue, title, pdf_url, dest}
+    Return list hasil: {... + ok, msg, size_kb}
+    """
+    results = []
+    total   = len(tasks)
+
+    with ThreadPoolExecutor(max_workers=workers) as ex:
+        future_map = {
+            ex.submit(download_pdf, t["pdf_url"], Path(t["dest"])): t
+            for t in tasks
+        }
+
+        done = 0
+        for future in as_completed(future_map):
+            task = future_map[future]
+            ok, msg = future.result()
+            done += 1
+
+            venue  = task["venue"]
+            title  = task["title"][:55]
+            marker = "✓" if ok else "✗"
+            log(f"  [{done:3d}/{total}] {marker} [{venue}] {title}")
+            if not ok:
+                log(f"           Gagal: {msg}")
+
+            results.append({**task, "ok": ok, "msg": msg})
+
+    return results
+
+
+# ─── MAIN ────────────────────────────────────────────────────────────────────
+
+def main():
+    parser = argparse.ArgumentParser(description="Stage 2: Download PDF paper IR 2025")
+    parser.add_argument("--input",   default=str(DEFAULT_INPUT),
+                        help=f"File metadata JSON (default: {DEFAULT_INPUT})")
+    parser.add_argument("--out-dir", default=str(DEFAULT_OUT_DIR),
+                        help=f"Folder output (default: {DEFAULT_OUT_DIR})")
+    parser.add_argument("--report",  default=str(DEFAULT_REPORT),
+                        help=f"File laporan JSON (default: {DEFAULT_REPORT})")
+    parser.add_argument("--workers", type=int, default=DEFAULT_WORKERS,
+                        help=f"Jumlah thread paralel (default: {DEFAULT_WORKERS})")
+    parser.add_argument("--venues",  nargs="+", default=None,
+                        help="Filter venue tertentu (default: semua)")
+    parser.add_argument("--limit",   type=int, default=None,
+                        help="Batas paper per venue (default: semua)")
+    args = parser.parse_args()
+
+    input_path = Path(args.input)
+    out_dir    = Path(args.out_dir)
+    report_path = Path(args.report)
+
+    # Baca metadata
+    if not input_path.exists():
+        print(f"[ERROR] File tidak ditemukan: {input_path}")
+        print("Jalankan collect.py terlebih dahulu.")
+        return
+
+    with open(input_path, encoding="utf-8") as f:
+        all_papers = json.load(f)
+
+    # Filter venue
+    if args.venues:
+        all_papers = [p for p in all_papers if p["venue"] in args.venues]
+
+    # Filter yang punya PDF URL
+    all_papers = [p for p in all_papers if p.get("pdf_url")]
+
+    # Grup per venue, terapkan limit
+    by_venue: dict[str, list] = {}
+    for p in all_papers:
+        by_venue.setdefault(p["venue"], []).append(p)
+
+    if args.limit:
+        for vk in by_venue:
+            by_venue[vk] = by_venue[vk][: args.limit]
+
+    # Susun task — skip yang sudah ada
+    tasks    = []
+    skipped  = 0
+    counters: dict[str, int] = {}
+
+    for venue_key, papers in by_venue.items():
+        counters[venue_key] = 0
+        for paper in papers:
+            counters[venue_key] += 1
+            idx      = counters[venue_key]
+            filename = safe_filename(paper["title"], idx)
+            dest     = out_dir / venue_key / filename
+
+            if dest.exists() and dest.stat().st_size >= MIN_PDF_BYTES:
+                skipped += 1
+                continue
+
+            tasks.append({
+                "idx":     idx,
+                "venue":   venue_key,
+                "title":   paper["title"],
+                "pdf_url": paper["pdf_url"],
+                "dest":    str(dest),
+                "doi":     paper.get("doi", ""),
+                "source":  paper.get("source", ""),
+            })
+
+    total_planned = sum(len(v) for v in by_venue.values())
+    print(f"Metadata  : {len(all_papers)} paper")
+    print(f"Akan diunduh: {len(tasks)} | Sudah ada (skip): {skipped}")
+    print(f"Workers   : {args.workers}")
+    print(f"Output    : {out_dir.resolve()}\n")
+
+    if not tasks:
+        print("Tidak ada yang perlu diunduh.")
+        return
+
+    # Jalankan download
+    t0      = time.time()
+    results = run_downloads(tasks, workers=args.workers)
+    elapsed = time.time() - t0
+
+    # Hitung statistik
+    ok_list   = [r for r in results if r["ok"]]
+    fail_list = [r for r in results if not r["ok"]]
+
+    by_venue_ok: dict[str, int] = {}
+    for r in ok_list:
+        by_venue_ok[r["venue"]] = by_venue_ok.get(r["venue"], 0) + 1
+
+    # Simpan laporan
+    report = {
+        "total_attempted": len(tasks),
+        "success": len(ok_list),
+        "failed": len(fail_list),
+        "skipped": skipped,
+        "elapsed_seconds": round(elapsed, 1),
+        "by_venue": {
+            vk: {
+                "attempted": len(papers),
+                "success":   by_venue_ok.get(vk, 0),
+            }
+            for vk, papers in by_venue.items()
+        },
+        "failures": [
+            {"venue": r["venue"], "title": r["title"],
+             "pdf_url": r["pdf_url"], "error": r["msg"]}
+            for r in fail_list
+        ],
+    }
+    with open(report_path, "w", encoding="utf-8") as f:
+        json.dump(report, f, ensure_ascii=False, indent=2)
+
+    # Ringkasan akhir
+    print(f"\n{'='*60}")
+    print("RINGKASAN DOWNLOAD")
+    print(f"{'='*60}")
+    for vk, papers in by_venue.items():
+        ok_n  = by_venue_ok.get(vk, 0)
+        all_n = len(papers)
+        bar   = "█" * ok_n + "░" * (all_n - ok_n)
+        print(f"  {vk:8s}: {ok_n:3d}/{all_n} {bar}")
+
+    print(f"\nBerhasil  : {len(ok_list)}/{len(tasks)}")
+    print(f"Gagal     : {len(fail_list)}")
+    print(f"Skip      : {skipped} (sudah ada)")
+    print(f"Waktu     : {elapsed:.1f} detik ({elapsed/60:.1f} menit)")
+    print(f"Laporan   : {report_path}")
+    print(f"Folder PDF: {out_dir.resolve()}")
+
+    if fail_list:
+        print(f"\nGagal download ({len(fail_list)} paper):")
+        for r in fail_list[:10]:
+            print(f"  [{r['venue']}] {r['title'][:60]} — {r['msg']}")
+        if len(fail_list) > 10:
+            print(f"  ... (+{len(fail_list)-10} lagi, lihat {report_path})")
+
+
+if __name__ == "__main__":
+    main()