simplify b2 url

downloader.py

@@ -137,39 +137,11 @@ def get_invoice_links(page: Page) -> list[dict]:
         entry.update(meta)
         links.append(entry)
 
-    rows = page.locator("table.billing-table tbody tr, table#billingTable tbody tr, table tbody tr").all()
-    for row in rows:
-        anchors = row.locator("a[href*='billing_invoice'], a[href*='invoice']").all()
-        for anchor in anchors:
-            href = anchor.get_attribute("href") or ""
-            text = anchor.inner_text().strip()
-            if href:
-                if not href.startswith("http"):
-                    href = f"{BASE_URL}/{href.lstrip('/')}"
-                _add_link(href, text)
-
-    if not links:
-        all_anchors = page.locator("a[href*='invoice']").all()
-        for anchor in all_anchors:
-            href = anchor.get_attribute("href") or ""
-            text = anchor.inner_text().strip()
-            if href and "invoice" in href.lower():
-                if not href.startswith("http"):
-                    href = f"{BASE_URL}/{href.lstrip('/')}"
-                _add_link(href, text)
-
     print_links = page.locator("a.no-print[data-reference-object-id], a[data-reference-object-id]").all()
     for anchor in print_links:
-        href = anchor.get_attribute("href") or ""
         ref_id = anchor.get_attribute("data-reference-object-id") or ""
         label = anchor.inner_text().strip() or ref_id
 
-        if href:
-            if not href.startswith("http"):
-                href = f"{BASE_URL}/{href.lstrip('/')}"
-            _add_link(href, label)
-            continue
-
         if ref_id:
             _add_link(
                 f"{BASE_URL}/billing_print_invoice.htm",
@@ -179,7 +151,7 @@ def get_invoice_links(page: Page) -> list[dict]:
             )
             continue
 
-        logger.warning("Invoice link missing href and reference id")
+        logger.warning("Invoice link missing reference id")
 
     return links
 
@@ -266,32 +238,24 @@ def export_invoice_pdf(
     billing_url: str = BILLING_URL,
 ) -> tuple[Page, Path]:
     invoice_page = None
-    if invoice.get("reference_id"):
-        page = _goto_with_retry(context, page, billing_url)
-        selector = f"a[data-reference-object-id='{invoice['reference_id']}']"
-        anchor = page.locator(selector).first
-        if anchor.count() > 0:
-            anchor.scroll_into_view_if_needed()
-            with page.expect_popup() as popup_info:
-                anchor.click()
-            invoice_page = popup_info.value
-        else:
-            with page.expect_popup() as popup_info:
-                page.evaluate("printPayment", invoice["reference_id"])
-            invoice_page = popup_info.value
+    if not invoice.get("reference_id"):
+        raise RuntimeError("Invoice reference id is required to open the invoice popup")
 
-        invoice_page.wait_for_load_state("domcontentloaded")
-        invoice_page.wait_for_url("**/billing_print_invoice.htm**", timeout=30000)
+    page = _goto_with_retry(context, page, billing_url)
+    selector = f"a[data-reference-object-id='{invoice['reference_id']}']"
+    anchor = page.locator(selector).first
+    if anchor.count() > 0:
+        anchor.scroll_into_view_if_needed()
+        with page.expect_popup() as popup_info:
+            anchor.click()
+        invoice_page = popup_info.value
     else:
-        page = _goto_with_retry(context, page, invoice["url"])
-        invoice_page = page
-        invoice_page.wait_for_load_state("domcontentloaded")
-        if "billing_print_invoice.htm" not in invoice_page.url:
-            with page.expect_popup() as popup_info:
-                page.evaluate("printPayment", invoice.get("reference_id"))
-            invoice_page = popup_info.value
-            invoice_page.wait_for_load_state("domcontentloaded")
-            invoice_page.wait_for_url("**/billing_print_invoice.htm**", timeout=30000)
+        with page.expect_popup() as popup_info:
+            page.evaluate("printPayment", invoice["reference_id"])
+        invoice_page = popup_info.value
+
+    invoice_page.wait_for_load_state("domcontentloaded")
+    invoice_page.wait_for_url("**/billing_print_invoice.htm**", timeout=30000)
 
     time.sleep(1)
 
@@ -337,7 +301,10 @@ def download_all_invoices() -> list[Path]:
         billing_pages = ["b2", "groups"]
 
         for billing_page in billing_pages:
-            billing_url = f"{BILLING_URL}?billing_page={billing_page}"
+            if billing_page == "b2":
+                billing_url = f"{BILLING_URL}"
+            else:
+                billing_url = f"{BILLING_URL}?billing_page={billing_page}"
             page = _goto_with_retry(context, page, billing_url)
 
             page, groups = get_group_options(context, page, billing_url=billing_url)