improve Billa parsing

StefanBratanov · Oct 25, 2023 · d7db630 · d7db630
1 parent 451c35d
commit d7db630
Show file tree

Hide file tree

Showing 3 changed files with 89 additions and 3 deletions.
diff --git a/src/main/kotlin/com/stefanbratanov/sofiasupermarketsapi/extractors/BillaProductsExtractor.kt b/src/main/kotlin/com/stefanbratanov/sofiasupermarketsapi/extractors/BillaProductsExtractor.kt
@@ -37,9 +37,11 @@ class BillaProductsExtractor : UrlProductsExtractor {
       "\\*+".toRegex(IGNORE_CASE),
       "Специфика\\s*:.*$".toRegex(IGNORE_CASE),
       "Супер цена".toRegex(IGNORE_CASE),
-      "МУЛТИ ПАК \\d+\\+\\d+".toRegex(IGNORE_CASE),
-      "Цена\\s+за\\s+\\d+\\s*бр\\.\\s*((без|с)\\s+отстъпка)?\\s*(\\d|\\.)+\\s*лв\\."
-        .toRegex(IGNORE_CASE)
+      "(?<=МУЛТИ ПАК.+)\\d+(х|x)".toRegex(IGNORE_CASE),
+      "МУЛТИ ПАК (\\d+\\+\\d+)?".toRegex(IGNORE_CASE),
+      "(\\d+(х|x)\\s*)?Цена\\s+за\\s+\\d+\\s*бр\\.\\s*((без|с)\\s+отстъпка)?\\s*(\\d|\\.|,)+\\s*лв\\.".toRegex(
+        IGNORE_CASE
+      )
     )
 
   private val regexesToDeleteBilla =

diff --git a/src/test/resources/extractors/billa/expected.json b/src/test/resources/extractors/billa/expected.json
@@ -1948,5 +1948,45 @@
     "picUrl": null,
     "validFrom": "2021-03-18",
     "validUntil": "2021-03-24"
+  },
+  {
+    "name": "Бира Tuborg кен",
+    "quantity": "0.5л",
+    "price": 1.33,
+    "oldPrice": 2.19,
+    "category": null,
+    "picUrl": null,
+    "validFrom": "2021-03-18",
+    "validUntil": "2021-03-24"
+  },
+  {
+    "name": "Вино Montado",
+    "quantity": "0.75л",
+    "price": 7.99,
+    "oldPrice": 10.99,
+    "category": null,
+    "picUrl": null,
+    "validFrom": "2021-03-18",
+    "validUntil": "2021-03-24"
+  },
+  {
+    "name": "Уиски Johnnie Walker Red Label",
+    "quantity": "0.7л",
+    "price": 18.99,
+    "oldPrice": 28.99,
+    "category": null,
+    "picUrl": null,
+    "validFrom": "2021-03-18",
+    "validUntil": "2021-03-24"
+  },
+  {
+    "name": "Бира Starobrno кен",
+    "quantity": "0.5л",
+    "price": 1.225,
+    "oldPrice": 2.45,
+    "category": null,
+    "picUrl": null,
+    "validFrom": "2021-03-18",
+    "validUntil": "2021-03-24"
   }
 ]
diff --git a/src/test/resources/extractors/billa/input.html b/src/test/resources/extractors/billa/input.html
@@ -3695,6 +3695,50 @@
       <div style="float:left; width:15%">
         <span class="price">13.99</span><span class="currency">лв.</span>	</div>
     </div>
+    <div class="product">
+      <div class="actualProduct" style="float:left; width:25%; margin-right:2%;">МУЛТИ ПАК Бира Tuborg кен 12х0.5л                                                                                                                  Цена за 1 бр. 2,19 лв.
+        Цена за 12 бр. без отстъпка 26,28 лв.</div>
+      <div class="priceText" style="float:left; width:10%;">СТАРА<br>ЦЕНА</div>
+      <div style="float:left; width:22%">
+        <span class="price">2.19</span> <span class="currency">лв.</span>	</div>
+      <div class="priceText" style="float:left; width:10%;">НОВА<br>ЦЕНА</div>
+      <div class="discount" style="float:left; width:10%;"> - -39%</div>
+      <div style="float:left; width:15%">
+        <span class="price">1.33</span><span class="currency">лв.</span>	</div>
+    </div>
+    <div class="product">
+      <div class="actualProduct" style="float:left; width:25%; margin-right:2%;">МУЛТИ ПАК Вино Montado 2x0.75л                                                                                                                              Цена за 1 бр. 10,99 лв.
+        Цена за 2 бр. без отстъпка 21,98 лв.</div>
+      <div class="priceText" style="float:left; width:10%;">СТАРА<br>ЦЕНА</div>
+      <div style="float:left; width:22%">
+        <span class="price">10.99</span> <span class="currency">лв.</span>	</div>
+      <div class="priceText" style="float:left; width:10%;">НОВА<br>ЦЕНА</div>
+      <div class="discount" style="float:left; width:10%;"> - -27%</div>
+      <div style="float:left; width:15%">
+        <span class="price">7.99</span><span class="currency">лв.</span>	</div>
+    </div>
+    <div class="product">
+      <div class="actualProduct" style="float:left; width:25%; margin-right:2%;">МУЛТИ ПАК 1+1 Уиски Johnnie Walker Red Label 2x0.7л                                                                                     Цена за 1 бр. 28,99 лв.
+        Цена за 2 бр. без отстъпка 57,98 лв.</div>
+      <div class="priceText" style="float:left; width:10%;">СТАРА<br>ЦЕНА</div>
+      <div style="float:left; width:22%">
+        <span class="price">28.99</span> <span class="currency">лв.</span>	</div>
+      <div class="priceText" style="float:left; width:10%;">НОВА<br>ЦЕНА</div>
+      <div class="discount" style="float:left; width:10%;"> - -34%</div>
+      <div style="float:left; width:15%">
+        <span class="price">18.99</span><span class="currency">лв.</span>	</div>
+    </div>
+    <div class="product">
+      <div class="actualProduct" style="float:left; width:25%; margin-right:2%;">МУЛТИ ПАК 1+1 Бира Starobrno кен 0.5л                                                                                                               Цена за 1 бр. 2,45 лв.
+        Цена за 2 бр. без отстъпка 4,90 лв.</div>
+      <div class="priceText" style="float:left; width:10%;">СТАРА<br>ЦЕНА</div>
+      <div style="float:left; width:22%">
+        <span class="price">2.45</span> <span class="currency">лв.</span>	</div>
+      <div class="priceText" style="float:left; width:10%;">НОВА<br>ЦЕНА</div>
+      <div class="discount" style="float:left; width:10%;"> - -50%</div>
+      <div style="float:left; width:15%">
+        <span class="price">1.225</span><span class="currency">лв.</span>	</div>
+    </div>
   </div>
 </div>