small fantastico parsing improvement

StefanBratanov · Nov 4, 2023 · 67ecbe7 · 67ecbe7
1 parent caec36e
commit 67ecbe7
Show file tree

Hide file tree

Showing 5 changed files with 137 additions and 3 deletions.
diff --git a/src/main/kotlin/com/stefanbratanov/sofiasupermarketsapi/api/AlcoholController.kt b/src/main/kotlin/com/stefanbratanov/sofiasupermarketsapi/api/AlcoholController.kt
@@ -210,7 +210,7 @@ class AlcoholController(
               val cdnUrl = cdnUploader.upload(productKey, picUrl!!)
               product.copy(picUrl = cdnUrl)
             } catch (ex: Exception) {
-              log.error("Error while uploading to CDN. Will fallback to Google search result")
+              log.error("Error while uploading to CDN. Will fallback to Google search result", ex)
               product.copy(picUrl = picUrl)
             }
           } else {

diff --git a/.../kotlin/com/stefanbratanov/sofiasupermarketsapi/extractors/FantasticoProductsExtractor.kt b/.../kotlin/com/stefanbratanov/sofiasupermarketsapi/extractors/FantasticoProductsExtractor.kt
@@ -61,7 +61,8 @@ class FantasticoProductsExtractor : PdfProductsExtractor {
       ",\\s*\$".toRegex(RegexOption.IGNORE_CASE),
       "➥+".toRegex(RegexOption.IGNORE_CASE),
       "количествата са лимитирани".toRegex(RegexOption.IGNORE_CASE),
-      "предложението е валидно.*цени".toRegex(RegexOption.IGNORE_CASE)
+      "предложението е валидно(.*цени)?".toRegex(RegexOption.IGNORE_CASE),
+      "от.*до.*г\\.".toRegex(RegexOption.IGNORE_CASE)
     )
 
   private val productSectionResolver: Map<ProductSection, (TextWithCoordinates) -> Boolean> =

diff --git a/...lin/com/stefanbratanov/sofiasupermarketsapi/extractors/FantasticoProductsExtractorTest.kt b/...lin/com/stefanbratanov/sofiasupermarketsapi/extractors/FantasticoProductsExtractorTest.kt
@@ -33,7 +33,8 @@ internal class FantasticoProductsExtractorTest {
     "fantastico_test_10.pdf,expected_10.json",
     "fantastico_test_11.pdf,expected_11.json",
     "fantastico_test_12.pdf,expected_12.json",
-    "fantastico_test_13.pdf,expected_13.json"
+    "fantastico_test_13.pdf,expected_13.json",
+    "fantastico_test_14.pdf,expected_14.json"
   )
   fun `test extracting products`(inputFile: String, expectedFile: String) {
     val testPdf = getPath("/extractors/fantastico/$inputFile")

diff --git a/src/test/resources/extractors/fantastico/expected_14.json b/src/test/resources/extractors/fantastico/expected_14.json
@@ -0,0 +1,132 @@
+[
+  {
+    "name": "Чипс от царевица GRIVAS различни вкусове от един и същи вкус",
+    "quantity": "2 бр х 100 г",
+    "price": 2.28,
+    "oldPrice": 4.56,
+    "category": null,
+    "picUrl": null,
+    "validFrom": null,
+    "validUntil": null
+  },
+  {
+    "name": "Крекер CHIO Party Mix",
+    "quantity": "2 бр х 200 г",
+    "price": 3.98,
+    "oldPrice": 7.96,
+    "category": null,
+    "picUrl": null,
+    "validFrom": null,
+    "validUntil": null
+  },
+  {
+    "name": "Печен чипс LAY`S различни вкусове",
+    "quantity": "125 г",
+    "price": 2.59,
+    "oldPrice": 5.35,
+    "category": null,
+    "picUrl": null,
+    "validFrom": null,
+    "validUntil": null
+  },
+  {
+    "name": "Бира STELLA ARTOIS амбалаж",
+    "quantity": "500 мл 0.20 лв бутилка",
+    "price": 1.55,
+    "oldPrice": 2.59,
+    "category": null,
+    "picUrl": null,
+    "validFrom": null,
+    "validUntil": null
+  },
+  {
+    "name": "Бира CORONA Екстра",
+    "quantity": "355 мл",
+    "price": 1.89,
+    "oldPrice": 3.29,
+    "category": null,
+    "picUrl": null,
+    "validFrom": null,
+    "validUntil": null
+  },
+  {
+    "name": "Бира HEINEKEN Classic; Silver",
+    "quantity": "мултипак, кен 4 бр х 500 мл",
+    "price": 4.99,
+    "oldPrice": 7.47,
+    "category": null,
+    "picUrl": null,
+    "validFrom": null,
+    "validUntil": null
+  },
+  {
+    "name": "Водка КРЕСКОВА",
+    "quantity": "1 л",
+    "price": 0.1,
+    "oldPrice": 21.99,
+    "category": null,
+    "picUrl": null,
+    "validFrom": null,
+    "validUntil": null
+  },
+  {
+    "name": "MARTINI Fiero; Bianco Torino",
+    "quantity": "750 мл",
+    "price": 11.99,
+    "oldPrice": 19.99,
+    "category": null,
+    "picUrl": null,
+    "validFrom": null,
+    "validUntil": null
+  },
+  {
+    "name": "Вермут Напитка ABSOLUT Passion Fruit Martini; Малинова Лимонада от един и същи вид",
+    "quantity": "кен, 2 бр х 250 мл",
+    "price": 3.78,
+    "oldPrice": 7.56,
+    "category": null,
+    "picUrl": null,
+    "validFrom": null,
+    "validUntil": null
+  },
+  {
+    "name": "Печено кашу ROIS",
+    "quantity": "2 бр х 140 г",
+    "price": 6.88,
+    "oldPrice": 13.76,
+    "category": null,
+    "picUrl": null,
+    "validFrom": null,
+    "validUntil": null
+  },
+  {
+    "name": "Печени бадеми KRONOS",
+    "quantity": "2 бр х 70 г",
+    "price": 4.28,
+    "oldPrice": 8.56,
+    "category": null,
+    "picUrl": null,
+    "validFrom": null,
+    "validUntil": null
+  },
+  {
+    "name": "Ядки AYA",
+    "quantity": "парти 500 г",
+    "price": 2.99,
+    "oldPrice": 5.99,
+    "category": null,
+    "picUrl": null,
+    "validFrom": null,
+    "validUntil": null
+  },
+  {
+    "name": "Тиквени семки DETELINA",
+    "quantity": "2 бр х 100 г",
+    "price": 2.58,
+    "oldPrice": 5.16,
+    "category": null,
+    "picUrl": null,
+    "validFrom": null,
+    "validUntil": null
+  }
+]
diff --git a/src/test/resources/extractors/fantastico/fantastico_test_14.pdf b/src/test/resources/extractors/fantastico/fantastico_test_14.pdf