parser logic added

2026-08-30 12:24:29 +02:00
parent 35d5c90d81
commit 3b698a3667
28 changed files with 3018 additions and 346 deletions
@@ -0,0 +1,186 @@
+{
+  "_comment": "Keyword/pattern config for the receipt parser. See docs/receipt-parsing.md to add a new shop. Match rules are implemented in receipt_parse_config.dart (short tokens use word boundaries; longer tokens use substring + compact OCR form).",
+
+  "known_stores": {
+    "_comment": "Used to detect store name from the header block (first 1-5 lines). Match against the STORE_HEADER lines.",
+    "list": [
+      "JUMBO",
+      "LIDL",
+      "KRUIDVAT",
+      "ALBERT HEIJN",
+      "AH",
+      "ALDI",
+      "PLUS",
+      "DIRK",
+      "COOP",
+      "SPAR",
+      "VOMAR",
+      "EKOPLAZA",
+      "HOOGVLIET",
+      "ACTION",
+      "ETOS",
+      "WORLD TOKO"
+    ],
+    "fallback_rule": "If no known store name matches, take the first non-address, non-phone-number line in the header block as a best-effort store name (free text)."
+  },
+
+  "column_header_keywords": {
+    "_comment": "Lines matching these are table headers, not data — ignore entirely.",
+    "list": [
+      "OMSCHRIJVING",
+      "BEDRAG IN",
+      "ARTIKEL",
+      "PRIJS",
+      "AANTAL ARTIKEL",
+      "BESCHRIJVING"
+    ],
+    "separator_pattern": "^[=\\-#]{3,}$"
+  },
+
+  "quantity_modifier_pattern": {
+    "_comment": "Regex to detect an 'N X price' line, optionally followed by a unit word. Group 1 = quantity, group 2 = unit price.",
+    "regex": "^\\s*(\\d+)\\s*[xX]\\s*(\\d+[,.]\\d{2})\\s*(PER\\s+STUK|PER\\s+KG|ST\\.?|STUKS?)?\\s*$",
+    "inline_regex": "(\\d+)\\s*[xX]\\s*(\\d+[,.]\\d{2})\\s*(PER\\s+STUK|PER\\s+KG)?",
+    "note": "Use 'regex' for standalone quantity lines (Jumbo-style). Use 'inline_regex' to detect the same pattern embedded within a product-name line (Kruidvat-style)."
+  },
+
+  "discount_keywords": {
+    "_comment": "Lines matching ANY of these AND carrying a negative amount are discounts, not products.",
+    "list": [
+      "ACTIE",
+      "KORTING",
+      "IN PRIJS VERLAAGD",
+      "BONUSKORTING",
+      "KASSAKORTING",
+      "ZEGELKORTING",
+      "REDUCTIE",
+      "AANBIEDING"
+    ],
+    "negative_amount_pattern": "-\\d+[,.]\\d{2}",
+    "receipt_level_total_discount_keywords": [
+      "TOTAAL KORTING",
+      "TOTALE KORTING",
+      "TOTAAL BESPARING"
+    ]
+  },
+
+  "total_keywords": {
+    "_comment": "Order matters for disambiguation — check subtotal_keywords before grand_total_keywords, since 'SUBTOTAAL' also contains 'TOTAAL'.",
+    "subtotal_keywords": [
+      "SUBTOTAAL",
+      "SUB TOTAAL"
+    ],
+    "grand_total_keywords": [
+      "TOTAAL",
+      "TOTAL"
+    ],
+    "tax_inclusive_total_keywords": [
+      "TOTAAL INCL. BTW",
+      "TOTAAL INCL BTW",
+      "TOTAAL EXCL. BTW",
+      "TOTAAL EXCL BTW"
+    ],
+    "rule": "The FIRST grand_total match encountered after the last item line is the authoritative receipt total. Any match containing a discount keyword (see discount_keywords.receipt_level_total_discount_keywords) is the total discount amount, not the grand total — check discount keywords first."
+  },
+
+  "payment_keywords": {
+    "_comment": "Lines matching these mark the start of the payment block. Everything from the first match onward (until noise/footer) should be excluded from item parsing, though some fields can be stored as receipt metadata.",
+    "list": [
+      "BETAALD",
+      "BETALING",
+      "VISA",
+      "MASTERCARD",
+      "MAESTRO",
+      "V PAY",
+      "PIN",
+      "CONTACTLOZE BETALING",
+      "AKKOORD",
+      "KAARTNR",
+      "KAART:",
+      "TRANSACTIE",
+      "TERMINAL",
+      "MERCHANT",
+      "AUTH. CODE",
+      "AUTH CODE",
+      "KAARTHOUDER",
+      "CVM",
+      "CHIP",
+      "PAY",
+      "IDEAL"
+    ],
+    "metadata_fields_to_extract": [
+      "payment_method (e.g. VISA, MASTERCARD, PIN, iDEAL)",
+      "transaction_id",
+      "auth_code",
+      "terminal_id"
+    ]
+  },
+
+  "tax_breakdown_keywords": {
+    "_comment": "Lines describing BTW/tax categories, not products. Also used to detect the trailing tax-category letter appended to some item lines.",
+    "list": [
+      "BTW",
+      "BTW-CODE",
+      "BEDR.EXCL",
+      "BEDR.INCL",
+      "BEDRAG EXCL",
+      "BEDRAG INCL"
+    ],
+    "category_row_pattern": "^[A-Z]\\s+\\d+%",
+    "trailing_category_letter_pattern": "\\s[A-Z]$",
+    "note": "trailing_category_letter_pattern strips a single trailing tax-category letter (e.g. ' B' or ' C') from the end of an item line if present, before extracting the price. Do not treat it as part of the price."
+  },
+
+  "noise_footer_keywords": {
+    "_comment": "Lines matching these (or appearing after the payment block) are noise — ignore entirely.",
+    "list": [
+      "BEWAAR",
+      "KASSABON",
+      "RETOURNEREN",
+      "RETOUR",
+      "DAGEN MET JE BON",
+      "WWW.",
+      ".COM",
+      ".NL",
+      "OPENINGSTIJDEN",
+      "MAANDAG",
+      "DINSDAG",
+      "WOENSDAG",
+      "DONDERDAG",
+      "VRIJDAG",
+      "ZATERDAG",
+      "ZONDAG",
+      "FACEBOOK",
+      "INSTAGRAM",
+      "BEDANKT",
+      "TOT ZIENS",
+      "MEDEWERKER",
+      "KASSA",
+      "TRANS",
+      "STORE",
+      "POS",
+      "REGISTREER",
+      "PLUS-APP",
+      "BESPAAR MEER",
+      "ZELF SERVICE",
+      "PRIVACY STATEMENT",
+      "KOOPZEGELS"
+    ]
+  },
+
+  "quantity_and_price_general": {
+    "price_pattern": "\\d+[,.]\\d{2}",
+    "decimal_separator": ",",
+    "note": "Some receipts (e.g. World Toko) show weight-based items like '900g', '500ml', '400g' as part of the product name — these are NOT quantities/units for the parser's purposes, just part of the descriptive name. Only treat a leading integer directly adjacent to 'X' or 'x' as a quantity."
+  },
+
+  "line_item_count_keywords": {
+    "_comment": "Some receipts show an item count line near the payment block, useful as a validation cross-check (does parsed item count match this?).",
+    "list": [
+      "AANTAL ARTIKEL",
+      "AANTAL ARTIKELEN",
+      "ART.",
+      "ARTIKELEN"
+    ]
+  }
+}