Add importer for Statiegeld Nederland

import/__main__.py

@@ -41,6 +41,7 @@ if __name__ == "__main__":
     importers = [
         rabobank.Importer("Activa:Betaalrekening", "EUR"),
         pdf.MollieInvoiceImporter(),
+        pdf.StatiegeldImporter(),
     ]
     hooks = [classify_contra(rabobank.guess_contra, "Activa:Betaalrekening")]
     main = Ingest(importers, hooks)

import/pdf.py

@@ -84,3 +84,48 @@ class MollieInvoiceImporter(Importer):
             ],
         )
         return [tx, doc]
+
+
+class StatiegeldImporter(Importer):
+    def identify(self, filepath):
+        mimetype, encoding = mimetypes.guess_type(filepath)
+        if mimetype != "application/pdf":
+            return False
+
+        lines = pdf_to_text(filepath).split("\n")
+        return any(line.startswith("Statiegeld Nederland") for line in lines)
+
+    def account(self, filepath):
+        return "Inkomsten:Statiegeld"
+
+    def tx_ref(self, filepath):
+        lines = pdf_to_text(filepath).split("\n")
+        for line in lines:
+            if m := re.search(r"^Factuurnr : (RP\d+)$", line):
+                return f"SNL-{m[1]}"
+        raise Exception("Mollie invoice reference not found")
+
+    def filename(self, filepath):
+        return f"{self.tx_ref(filepath)}.pdf"
+
+    def date(self, filepath):
+        lines = pdf_to_text(filepath).split("\n")
+        for line in lines:
+            if m := re.search(r"^Datum : (\d{2})/(\d{2})/(\d{4})$", line):
+                return date(int(m[3]), int(m[2]), int(m[1]))
+        raise Exception("Date not found")
+
+    def extract(self, filepath, existing):
+        name = self.filename(filepath)
+        date = self.date(filepath)
+        link = self.tx_ref(filepath)
+
+        doc = Document(
+            meta=data.new_metadata(filepath, 0),
+            date=date,
+            account=self.account(filepath),
+            filename=f"docs/Inkomsten/Statiegeld/{date}.{name}",
+            tags=set(),
+            links={link},
+        )
+        return [doc]