115 lines
4.2 KiB
Python
115 lines
4.2 KiB
Python
from playwright.sync_api import sync_playwright
|
|
import os
|
|
import glob
|
|
import datetime
|
|
|
|
import pandas as pd
|
|
|
|
# ── CONFIG ──────────────────────────────────────────────────────────────────
|
|
BASE_URL = "https://janssen.4gclinical.com"
|
|
EMAIL = "vbuzalka@its.jnj.com"
|
|
PASSWORD = "Vlado123++-+"
|
|
|
|
STUDIES = ["77242113UCO3001", "42847922MDD3003"]
|
|
|
|
BASE_DIR = os.path.dirname(os.path.abspath(__file__))
|
|
INCOMING_DIR = os.path.join(BASE_DIR, "IncomingSourceReports")
|
|
DETAILS_DIR = os.path.join(BASE_DIR, "IncomingSourceReportsDetails")
|
|
# ────────────────────────────────────────────────────────────────────────────
|
|
|
|
|
|
def get_subjects(study):
|
|
pattern = os.path.join(INCOMING_DIR, f"* {study} Subject Summary Report.xlsx")
|
|
files = sorted(
|
|
[f for f in glob.glob(pattern) if not os.path.basename(f).startswith("~$")],
|
|
key=os.path.getmtime,
|
|
reverse=True,
|
|
)
|
|
if not files:
|
|
raise FileNotFoundError(f"Nenalezen Subject Summary Report pro {study}")
|
|
today = datetime.date.today().strftime("%Y-%m-%d")
|
|
if not os.path.basename(files[0]).startswith(today):
|
|
raise FileNotFoundError(f"Dnešní Subject Summary Report pro {study} neexistuje — spusť nejdříve download_subject_summary.py")
|
|
path = files[0]
|
|
print(f" Čtu subjekty z: {os.path.basename(path)}")
|
|
|
|
raw = pd.read_excel(path, header=None)
|
|
header_row = None
|
|
for i, row in raw.iterrows():
|
|
if "Subject" in [str(v).strip() for v in row]:
|
|
header_row = i
|
|
break
|
|
if header_row is None:
|
|
raise ValueError("Hlavičkový řádek nenalezen")
|
|
|
|
df = pd.read_excel(path, header=header_row)
|
|
subjects = df["Subject"].dropna().astype(str).str.strip().tolist()
|
|
return subjects
|
|
|
|
|
|
def run(page, study):
|
|
out_dir = os.path.join(DETAILS_DIR, study)
|
|
os.makedirs(out_dir, exist_ok=True)
|
|
|
|
subjects = get_subjects(study)
|
|
print(f" Nalezeno {len(subjects)} subjektů")
|
|
today = datetime.date.today().strftime("%Y-%m-%d")
|
|
|
|
page.goto(f"{BASE_URL}/report/patient_detail_report")
|
|
page.wait_for_load_state("networkidle", timeout=120000)
|
|
|
|
for subject in subjects:
|
|
filename = os.path.join(out_dir, f"{today} {study} {subject} Subject Detail.xlsx")
|
|
print(f" [{subject}] Stahuji...")
|
|
input_field = page.locator('input[placeholder="search"], input[type="text"]').first
|
|
input_field.click()
|
|
input_field.fill(subject)
|
|
page.wait_for_timeout(500)
|
|
page.locator("mat-option").first.dispatch_event("click")
|
|
page.wait_for_load_state("networkidle", timeout=120000)
|
|
|
|
with page.expect_download(timeout=120000) as dl:
|
|
page.get_by_role("button", name="Download XLS").click()
|
|
dl.value.save_as(filename)
|
|
print(f" [{subject}] OK")
|
|
|
|
page.get_by_role("button", name="Clear").click()
|
|
page.wait_for_load_state("networkidle", timeout=120000)
|
|
|
|
print(f" [{study}] Subject details hotovo.")
|
|
|
|
|
|
def main():
|
|
os.makedirs(DETAILS_DIR, exist_ok=True)
|
|
|
|
with sync_playwright() as p:
|
|
for study in STUDIES:
|
|
print(f"\n[{study}] Přihlášení...")
|
|
browser = p.chromium.launch(headless=False)
|
|
context = browser.new_context(accept_downloads=True)
|
|
page = context.new_page()
|
|
|
|
page.goto(BASE_URL)
|
|
page.wait_for_load_state("networkidle")
|
|
page.get_by_label("Email *").fill(EMAIL)
|
|
page.get_by_label("Password *").fill(PASSWORD)
|
|
page.locator("#login__submit").click()
|
|
page.wait_for_load_state("networkidle")
|
|
|
|
page.get_by_label("Study *").click()
|
|
page.get_by_role("option", name=study).click()
|
|
page.get_by_role("button", name="SELECT").click()
|
|
page.wait_for_load_state("networkidle")
|
|
|
|
try:
|
|
run(page, study)
|
|
except Exception as e:
|
|
print(f" [{study}] CHYBA: {e}")
|
|
|
|
browser.close()
|
|
|
|
print("\nVše hotovo.")
|
|
|
|
|
|
main()
|