zifeng/ctgov

.gitattributes

@@ -5697,3 +5697,5 @@ documents/10.1056_nejmoa1710030/protocol.pdf filter=lfs diff=lfs merge=lfs -text
 documents/10.1056_nejmoa1900906/sap.pdf filter=lfs diff=lfs merge=lfs -text
 documents/10.1056_nejmoa1900906/protocol.pdf filter=lfs diff=lfs merge=lfs -text
 documents/10.1001_jama.2018.6496/sap.pdf filter=lfs diff=lfs merge=lfs -text
+*.pdf filter=lfs diff=lfs merge=lfs -text
+raw_data/**/*.txt filter=lfs diff=lfs merge=lfs -text

README.md

@@ -11,3 +11,19 @@ configs:
 # TrialDesignBench Source
 
 CSV index of clinical trials with linked protocol and SAP PDFs under `documents/<row_id>/`.
+
+## Bulk download from AACT
+
+`download_ctgov_prot_sap.py` downloads protocol and SAP PDFs in bulk from an [AACT](https://aact.ctti-clinicaltrials.org/) static dataset dump. It reads `<aact_dir>/provided_documents.txt` and saves each PDF to `documents/<nct_id>/<original-filename>.pdf`.
+
+```bash
+pip install pandas requests tqdm
+python download_ctgov_prot_sap.py /path/to/aact_pipe_delimited_dump
+```
+
+Useful flags: `--output-dir`, `--limit N`, `--full` (re-download existing), `--include-icf` (also fetch ICF-only rows).
+
+Example output (already in `documents/`):
+
+- `documents/NCT03254108/Prot_000.pdf`, `documents/NCT03254108/SAP_001.pdf` — separate protocol and SAP
+- `documents/NCT03521102/Prot_SAP_001.pdf` — combined protocol + SAP

documents/NCT02267317/Prot_SAP_000.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:349d4d0886372ee6119efafb136d2da59a1ab80a9a95838c826722f5dcf13df5
+size 1772156

documents/NCT03254108/Prot_000.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ad137c6bd5a4adf9704ed1c85533e717e8abbf6cfc346c462dd53c5a96cd990
+size 1800675

documents/NCT03254108/SAP_001.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72dd6b64c16937aa39ca95717cc5852481168c3a985b34727f2887f92d177923
+size 383474

documents/NCT03345004/Prot_000.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b5b652bc823b10f8379ba932fc73e25b2fe95e5f2084c45f6f3854e913f91b8
+size 1952713

documents/NCT03442868/Prot_SAP_000.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:803988d6e31c805accbabe47c36b3974b08fa3dd1be34a98c3050802a9b0362d
+size 725875

documents/NCT03521102/Prot_SAP_001.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73bf099f89e3ff654d4b20311399e611012be96aa234ec672ec623519ef5f519
+size 366996

documents/NCT03531632/Prot_000.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e50087b762eb91cfe91c38c807d9b229b683a1ea63ea48e1c2b7ff16d27304f
+size 4507356

documents/NCT03531632/SAP_001.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7987bc2610dafd2eb32940fb5664b17668cf8bc0a742d279b77700b4a67e7918
+size 961049

documents/NCT03834168/Prot_SAP_000.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1682af36e92c0301e4d962ef74eacd05438fc81721bb9d5aead49273048a07a
+size 192900

documents/NCT04493892/Prot_SAP_000.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51ad4a48016cfd435de6d1d4f7a07b743dcb9c25e43e0591a26396d0b9d0cd68
+size 376349

download_ctgov_prot_sap.py

@@ -0,0 +1,212 @@
+#!/usr/bin/env python3
+"""Download protocol/SAP PDFs listed in an AACT provided_documents.txt dump.
+
+Reads <aact_dir>/provided_documents.txt (pipe-delimited) and saves each PDF to
+<output_dir>/<nct_id>/<basename>.pdf, preserving the URL's filename.
+
+Usage:
+    python download_ctgov_prot_sap.py <aact_dir>
+    python download_ctgov_prot_sap.py <aact_dir> --output-dir ./documents
+    python download_ctgov_prot_sap.py <aact_dir> --limit 100
+    python download_ctgov_prot_sap.py <aact_dir> --include-icf
+    python download_ctgov_prot_sap.py <aact_dir> --full     # re-download existing
+"""
+
+from __future__ import annotations
+
+import argparse
+import logging
+import os
+import re
+import sys
+import time
+from pathlib import Path
+from urllib.parse import unquote, urlparse
+
+import pandas as pd
+import requests
+
+LOG = logging.getLogger("download_ctgov_prot_sap")
+
+SCRIPT_DIR = Path(__file__).resolve().parent
+DEFAULT_OUTPUT_DIR = SCRIPT_DIR / "documents"
+DEFAULT_DELAY = 0.5
+REQUEST_TIMEOUT = 120
+
+HEADERS = {
+    "User-Agent": (
+        "Mozilla/5.0 (compatible; ClinicalTrialCrawler/1.0; "
+        "+https://github.com/clinical-trial-crawler)"
+    ),
+    "Accept": "application/pdf,*/*",
+}
+
+
+def safe_basename(url: str) -> str:
+    """Filesystem-safe basename from a URL, defaulting to document.pdf."""
+    name = os.path.basename(unquote(urlparse(url).path)) or "document.pdf"
+    name = re.sub(r'[<>:"|?*\\\x00-\x1f]', "_", name).strip(".") or "document.pdf"
+    return name[:200]
+
+
+def _norm_url(url: str) -> str:
+    u = (url or "").strip()
+    if u.startswith("http://"):
+        u = "https://" + u[7:]
+    return u
+
+
+def load_rows(aact_dir: Path, include_icf: bool) -> pd.DataFrame:
+    """Load (nct_id, url) pairs for protocol/SAP rows from provided_documents.txt."""
+    path = aact_dir / "provided_documents.txt"
+    if not path.exists():
+        raise FileNotFoundError(f"provided_documents.txt not found in {aact_dir}")
+
+    df = pd.read_csv(path, sep="|", dtype=str, low_memory=False)
+    for col in ("nct_id", "url"):
+        if col not in df.columns:
+            raise ValueError(f"provided_documents.txt missing column: {col}")
+
+    df = df.dropna(subset=["nct_id", "url"]).copy()
+    df["nct_id"] = df["nct_id"].str.strip()
+    df["url"] = df["url"].map(_norm_url)
+    df = df[(df["nct_id"] != "") & (df["url"] != "")]
+
+    if not include_icf:
+        # Keep rows that have a protocol or SAP; drop ICF-only rows.
+        has_prot = df.get("has_protocol", "f").fillna("f").str.lower().eq("t")
+        has_sap = df.get("has_sap", "f").fillna("f").str.lower().eq("t")
+        df = df[has_prot | has_sap]
+
+    return df.drop_duplicates(subset=["nct_id", "url"]).reset_index(drop=True)
+
+
+def download_one(session: requests.Session, url: str, dest: Path) -> bool:
+    """Download url to dest. Returns True on success."""
+    try:
+        resp = session.get(url, timeout=REQUEST_TIMEOUT)
+        resp.raise_for_status()
+        dest.parent.mkdir(parents=True, exist_ok=True)
+        dest.write_bytes(resp.content)
+        return True
+    except requests.RequestException as e:
+        LOG.debug("Download failed %s: %s", url, e)
+        return False
+
+
+def run(
+    aact_dir: Path,
+    output_dir: Path,
+    delay: float = DEFAULT_DELAY,
+    limit: int = 0,
+    full: bool = False,
+    include_icf: bool = False,
+) -> dict[str, int]:
+    df = load_rows(aact_dir, include_icf=include_icf)
+    if limit > 0:
+        df = df.head(limit)
+    LOG.info(
+        "Loaded %d (nct_id, url) pairs from %s -> writing to %s",
+        len(df), aact_dir, output_dir,
+    )
+
+    session = requests.Session()
+    session.headers.update(HEADERS)
+
+    stats = {"downloaded": 0, "skipped": 0, "failed": 0}
+    try:
+        from tqdm import tqdm
+        it = tqdm(df.itertuples(index=False), total=len(df), desc="Download", unit="pdf")
+    except ImportError:
+        it = df.itertuples(index=False)
+
+    for row in it:
+        nct_id = row.nct_id
+        url = row.url
+        dest = output_dir / nct_id / safe_basename(url)
+
+        if not full and dest.exists() and dest.stat().st_size > 0:
+            stats["skipped"] += 1
+            continue
+
+        time.sleep(delay)
+        if download_one(session, url, dest):
+            stats["downloaded"] += 1
+        else:
+            stats["failed"] += 1
+
+    LOG.info(
+        "Done: %d downloaded, %d skipped, %d failed",
+        stats["downloaded"], stats["skipped"], stats["failed"],
+    )
+    return stats
+
+
+def main() -> int:
+    logging.basicConfig(
+        level=logging.INFO,
+        format="%(asctime)s - %(name)s - %(levelname)s - %(message)s",
+    )
+    parser = argparse.ArgumentParser(
+        description="Download CTGOV protocol/SAP PDFs from an AACT provided_documents.txt dump.",
+    )
+    parser.add_argument(
+        "aact_dir",
+        type=Path,
+        help="Path to AACT data directory containing provided_documents.txt",
+    )
+    parser.add_argument(
+        "--output-dir",
+        type=Path,
+        default=DEFAULT_OUTPUT_DIR,
+        help=f"Output base directory (default: {DEFAULT_OUTPUT_DIR})",
+    )
+    parser.add_argument(
+        "--delay",
+        type=float,
+        default=DEFAULT_DELAY,
+        help=f"Seconds between requests (default: {DEFAULT_DELAY})",
+    )
+    parser.add_argument(
+        "--limit",
+        type=int,
+        default=0,
+        help="Max rows to process (0 = all)",
+    )
+    parser.add_argument(
+        "--full",
+        action="store_true",
+        help="Re-download even if the destination file already exists",
+    )
+    parser.add_argument(
+        "--include-icf",
+        action="store_true",
+        help="Also download ICF-only rows (default: protocol/SAP only)",
+    )
+    args = parser.parse_args()
+
+    try:
+        stats = run(
+            aact_dir=args.aact_dir.resolve(),
+            output_dir=args.output_dir.resolve(),
+            delay=args.delay,
+            limit=args.limit,
+            full=args.full,
+            include_icf=args.include_icf,
+        )
+        print(
+            f"Downloaded: {stats['downloaded']} | "
+            f"Skipped: {stats['skipped']} | "
+            f"Failed: {stats['failed']}"
+        )
+        return 0
+    except (FileNotFoundError, ValueError) as e:
+        LOG.error("%s", e)
+        return 1
+    except Exception as e:
+        LOG.exception("%s", e)
+        return 1
+
+
+if __name__ == "__main__":
+    sys.exit(main())

raw_data/20260521_clinical_trials_ctgov/provided_documents.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:faffa0944014c5b8499cf042ce338b54c2779f65b146ed7b58ad6bfe5672c18d
+size 11189443