Add importer for Mollie invoices

.gitignore

@@ -0,0 +1,2 @@
+__pycache__
+.mypy_cache

Uitgaven.beancount

@@ -0,0 +1,18 @@
+option "name_assets" "Activa"
+option "name_equity" "Vermogen"
+option "name_expenses" "Uitgaven"
+option "name_income" "Inkomsten"
+option "name_liabilities" "Passiva"
+option "account_previous_balances" "Openingsbalans"
+
+2025-04-30 * "Mollie B.V." "iDeal transactiekosten" ^MOL-NL-R2025.0000581421
+  Uitgaven:Bankkosten   7.74 EUR
+  Passiva:RevBank      -7.74 EUR
+
+2025-04-30 document Uitgaven:Bankkosten "docs/Uitgaven/Bankkosten/2025-04-30.MOL-NL-R2025.0000581421.pdf" ^MOL-NL-R2025.0000581421
+
+2025-05-31 * "Mollie B.V." "iDeal transactiekosten" ^MOL-NL-R2025.0000653802
+  Uitgaven:Bankkosten   17.04 EUR
+  Passiva:RevBank      -17.04 EUR
+
+2025-05-31 document Uitgaven:Bankkosten "docs/Uitgaven/Bankkosten/2025-05-31.MOL-NL-R2025.0000653802.pdf" ^MOL-NL-R2025.0000653802

bitlair.beancount

@@ -50,5 +50,6 @@ plugin "beancount_periodic.recur" "{'generate_until':'2025-05-31'}"
 include "Activa/Betaalrekening.beancount"
 include "Activa/Debiteuren/Huurders.beancount"
 include "Activa/Debiteuren/Deelnemers.beancount"
+include "Uitgaven.beancount"
 
 include "reimburse.beancount"

import/__main__.py

@@ -0,0 +1,47 @@
+#!/usr/bin/env python3
+
+from beangulp import Ingest
+from beangulp.testing import main
+from beancount.core.data import Transaction, Posting
+
+import rabobank
+import pdf
+
+
+def classify_contra(guess_contra, primary_account):
+    def _inner(extracted_entries_list, ledger_entries):
+        def _classify(entry):
+            if not isinstance(entry, Transaction):
+                return entry
+            posting = entry.postings[0]
+            if posting.account != primary_account:
+                return entry
+            contra = guess_contra(entry)
+            entry.postings.append(
+                Posting(
+                    contra or "Inkomsten:TODO",
+                    -posting.units,
+                    posting.cost,
+                    posting.price,
+                    None if contra else "!",
+                    None,
+                )
+            )
+            return entry
+
+        return [
+            (filename, [_classify(entry) for entry in entries], account, importer)
+            for filename, entries, account, importer in extracted_entries_list
+        ]
+
+    return _inner
+
+
+if __name__ == "__main__":
+    importers = [
+        rabobank.Importer("Activa:Betaalrekening", "EUR"),
+        pdf.MollieInvoiceImporter(),
+    ]
+    hooks = [classify_contra(rabobank.guess_contra, "Activa:Betaalrekening")]
+    main = Ingest(importers, hooks)
+    main()

import/pdf.py

@@ -0,0 +1,86 @@
+from pypdf import PdfReader
+from beangulp import mimetypes, Importer
+from beangulp.cache import cache
+import re
+from datetime import date
+from beancount.core.data import Transaction, Posting, Document
+from beancount.core import flags
+from beancount.core import data
+from beancount.core import amount
+from beancount.core.number import D
+
+
+@cache
+def pdf_to_text(filename):
+    reader = PdfReader(filename)
+    page1 = reader.pages[0]
+    return page1.extract_text()
+
+
+class MollieInvoiceImporter(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("Mollie B.V.") for line in lines)
+
+    def account(self, filepath):
+        return "Uitgaven:Bankkosten"
+
+    def tx_ref(self, filepath):
+        lines = pdf_to_text(filepath).split("\n")
+        for line in lines:
+            if m := re.search(r"^Invoice reference (MOL-.+)$", line):
+                return m[1]
+        raise Exception("Mollie invoice reference not found")
+
+    def filename(self, filepath):
+        return f"{self.tx_ref(filepath)}.pdf"
+
+    def date(self, filepath):
+        lines = pdf_to_text(filepath).split("\n")
+        for line in lines:
+            if m := re.search(r"^Invoice date (\d{2})-(\d{2})-(\d{4})$", line):
+                return date(int(m[3]), int(m[2]), int(m[1]))
+        raise Exception("Mollie invoice date not found")
+
+    def extract(self, filepath, existing):
+        lines = pdf_to_text(filepath).split("\n")
+        for line in lines:
+            if m := re.search(r"^Total including VAT ([\d.]+) EUR$", line):
+                total = D(m[1])
+                break
+        else:
+            raise Exception("Mollie invoice total not found")
+
+        contra = "Passiva:RevBank"
+
+        name = self.filename(filepath)
+        date = self.date(filepath)
+        link = self.tx_ref(filepath)
+
+        units = amount.Amount(total, "EUR")
+        doc = Document(
+            meta=data.new_metadata(filepath, 0),
+            date=date,
+            account=self.account(filepath),
+            filename=f"docs/Uitgaven/Bankkosten/{date}.{name}",
+            tags=set(),
+            links={link},
+        )
+        tx = Transaction(
+            meta=data.new_metadata(filepath, 0),
+            date=date,
+            flag=flags.FLAG_OKAY,
+            payee="Mollie B.V.",
+            narration="iDeal transactiekosten",
+            tags=set(),
+            links={link},
+            postings=[
+                Posting(self.account(filepath), units, None, None, None, None),
+                Posting(contra, -units, None, None, None, None),
+            ],
+        )
+        return [tx, doc]

import/rabobank.py

@@ -1,11 +1,8 @@
 #!/usr/bin/env python3
 
 from os.path import basename
-from beangulp import Ingest
 from beangulp import mimetypes
 from beangulp.importers import csvbase
-from beangulp.testing import main
-from beancount.core.data import Transaction, Posting
 from deelnemers import deelnemers
 
 
@@ -121,34 +118,3 @@ def guess_contra(entry):
             return f"Activa:Debiteuren:Deelnemers:{p.nickname.title().replace('_', '')}"
 
     return None
-
-
-def classify_hook(extracted_entries_list, ledger_entries):
-    def _classify(entry):
-        if not isinstance(entry, Transaction):
-            return entry
-        contra = guess_contra(entry)
-        posting = entry.postings[0]
-        entry.postings.append(
-            Posting(
-                contra or "Inkomsten:TODO",
-                -posting.units,
-                posting.cost,
-                posting.price,
-                None if contra else "!",
-                None,
-            )
-        )
-        return entry
-
-    return [
-        (filename, [_classify(entry) for entry in entries], account, importer)
-        for filename, entries, account, importer in extracted_entries_list
-    ]
-
-
-if __name__ == "__main__":
-    importers = [Importer("Activa:Betaalrekening", "EUR")]
-    hooks = [classify_hook]
-    main = Ingest(importers, hooks)
-    main()