notebook

Medidata/download_edc_datalistings.py

@@ -72,6 +72,15 @@ def wait_load(page, extra_ms=1000):
 
 def dbg(page, label):
     print(f"[{label}] URL: {page.url}")
+    try:
+        from pathlib import Path
+        shots = Path(__file__).parent / "debug_shots"
+        shots.mkdir(exist_ok=True)
+        path = shots / f"{label}.png"
+        page.screenshot(path=str(path), full_page=True)
+        print(f"[{label}] screenshot: {path}")
+    except Exception as e:
+        print(f"[{label}] screenshot failed: {e}")
 
 
 def extract_study_label(study_search: str) -> str:
@@ -178,16 +187,32 @@ def select_role(page):
                 print(f"  Vybráno: '{txt}'")
                 break
 
+    clicked = False
     for btn_sel in ['input[value="Continue"]', 'input[type="submit"]',
                     'button:has-text("Continue")', 'button[type="submit"]']:
         try:
             btn = page.query_selector(btn_sel)
         except Exception:
-            break
+            continue
         if btn:
-            btn.click()
-            wait_load(page, 2000)
-            break
+            try:
+                with page.expect_navigation(timeout=15_000):
+                    btn.click()
+                clicked = True
+                break
+            except PWTimeout:
+                print(f"  Click on {btn_sel} nezpůsobil navigaci, zkouším další...")
+                continue
+
+    if not clicked:
+        print("  Fallback: submituji formulář přes JS...")
+        try:
+            with page.expect_navigation(timeout=15_000):
+                page.evaluate("document.forms[0] && document.forms[0].submit()")
+        except PWTimeout:
+            print("  JS submit fallback také neprošel.")
+
+    wait_load(page, 1500)
     dbg(page, "after-role")
 
 
@@ -404,8 +429,12 @@ def download_datalisting(study: str, forms: list[str], country: str | None = Non
     results = []
 
     with sync_playwright() as p:
-        browser = p.chromium.launch(headless=False, slow_mo=200)
-        ctx_kwargs = {"accept_downloads": True}
+        browser = p.chromium.launch(
+            headless=False,
+            slow_mo=200,
+            args=["--start-maximized"],
+        )
+        ctx_kwargs = {"accept_downloads": True, "no_viewport": True}
 
         use_saved = auth_valid()
         if use_saved: