Test

invoices.py

@@ -0,0 +1,54 @@
+from bs4 import BeautifulSoup
+from weasyprint import HTML, CSS
+import os
+
+# ---- Configuration ----
+input_file = r"C:\Users\conne\Downloads\Print Invoice Set - Canadian Avalanche Association.html"
+css_folder = r"C:\Users\conne\Downloads\Print Invoice Set - Canadian Avalanche Association_files"  # 👈 Update this to the actual CSS folder path
+base_output_folder = os.path.join("data", "invoices")
+# ------------------------
+
+os.makedirs(base_output_folder, exist_ok=True)
+
+# Load HTML file
+with open(input_file, "r", encoding="utf-8") as f:
+    html = f.read()
+
+soup = BeautifulSoup(html, "lxml")
+
+# Find all invoices
+invoices = soup.find_all("form", id=lambda x: x and x.startswith("frmInvoice-"))
+print(f"Found {len(invoices)} invoices")
+
+# Load all CSS files from folder
+css_files = [CSS(filename=os.path.join(css_folder, f)) for f in os.listdir(css_folder) if f.endswith(".css")]
+
+for form in invoices:
+    bill_to_cell = form.find("th", string=lambda x: x and "Bill to" in x)
+    if not bill_to_cell:
+        continue
+
+    bill_to_text = bill_to_cell.find_next("td").get_text(separator=" ", strip=True)
+    name = bill_to_text.split("\n")[0].strip()
+    safe_name = "".join(c for c in name if c.isalnum() or c in " _-").strip()
+
+    person_folder = os.path.join(base_output_folder, safe_name)
+    os.makedirs(person_folder, exist_ok=True)
+
+    invoice_number = form.find("th", string=lambda x: x and "Invoice #" in x)
+    invoice_number = (
+        invoice_number.find_next("td").get_text(strip=True)
+        if invoice_number
+        else "unknown"
+    )
+
+    output_pdf = os.path.join(person_folder, f"Invoice_{invoice_number}.pdf")
+
+    # Convert to PDF with CSS and base URL for relative paths
+    HTML(string=str(form), base_url=os.path.dirname(input_file)).write_pdf(
+        output_pdf, stylesheets=css_files
+    )
+
+    print(f"Saved PDF → {output_pdf}")
+
+print("✅ Done! All invoices saved as PDFs under data/invoices/")