Cleanup exceptions (#8)

* chore: remove non existing sites from exceptions

* Support for pyszneeko.pl

* Remove not supported exceptions

* Add support for polskikoszyk

* Add suport for emaja and marketbio

Author: WezSieTato

README.rst

@@ -37,14 +37,15 @@ Wspierane sklepy
 * https://www.bee.pl/
 * https://e-prim.pl
 * https://lisek.app/
+* https://pyszneeko.pl/
+* https://polskikoszyk.pl/
+* https://www.emaja24.pl/
+* https://marketbio.pl/
 
 Pracujemy nad wsparciem dla
 ---------------------------
 * https://mamyito.pl
-* https://polskikoszyk.pl/
 * https://zakupy.biedronka.pl/
 * https://delio.com.pl/
 * https://topaz24.pl/
 * https://www.lidl.pl/
-* https://www.emaja24.pl/
-* https://marketbio.pl/

content_scripts/pola.js

@@ -2,25 +2,7 @@
     let domain = document.location.hostname || '';
     let result = null;
     let exceptions = {
-        'www.neo24.pl': ['meta[property="og:image"][content]', 'content', /^.*\/(\d{13}|\d{8})..\.[a-z]{3,4}$/i],
-        'www.neonet.pl': ['meta[property="og:image"][content]', 'content', /^.*\/(\d{13}|\d{8})..\.[a-z]{3,4}$/i],
-        'www.megamarket24.pl': ['meta[property="og:url"][content]', 'content', /^.*\-(\d{13}|\d{8})\.html(?:\?.*)?$/i],
-        'aptekasowa.pl': ['meta[property="og:url"][content]', 'content', /^.*\-(\d{13}|\d{8})\.html(?:\?.*)?$/i],
-        'yesorganic.pl': ['meta[property="og:url"][content]', 'content', /^.*\-(\d{13}|\d{8})\.html(?:\?.*)?$/i],
-        'biomarketcompany.pl': ['meta[property="og:url"][content]', 'content', /^.*\-(\d{13}|\d{8})\.html(?:\?.*)?$/i],
-        'www.tvokazje.pl': ['meta[property="og:url"][content]', 'content', /^.*\-(\d{13}|\d{8})\.html(?:\?.*)?$/i],
-        'sklepdietetyczny.pl': ['link[rel="canonical"][href]', 'href', /^.*\-(\d{13}|\d{8})\.html$/i],
         'lisek.app': ['link[rel="canonical"][href]', 'href', /^.*\/(\d{13}|\d{8})$/i],
-        'studencik.com.pl': ['meta[itemprop="url"][content]', 'content', /^.*\-(\d{13}|\d{8})\.html$/i],
-        'polskikoszyk.pl': ['img[class="product-main-thumb"][src]', 'src', /^.*\/(\d{13}|\d{8})\.[a-z]{3,4}$/i],
-        'www.leclerc24.pl': ['#productWindow .item-image img[src]', 'src', /^.*\/(\d{13}|\d{8}).*$/i],
-        'www.douglas.pl': ['img[id^="addProductToBasketLink_"]', 'id', /^addProductToBasketLink_(\d{13}|\d{8})$/i]
-    };
-    let exceptions2 = {
-        'ezakupy.tesco.pl': [/"gs1:gtin":"(\d{13}|\d{8})(?=")/, document.head],
-        'www.electro-outlet.pl': [/"code_extern":"(\d{13}|\d{8})(?=")/, document.body],
-        'pyszneeko.pl': [/"code_extern":"(\d{13}|\d{8})(?=")/, document.body],
-        'sliv.com.pl': [/"code_producer":"(\d{13}|\d{8})(?=")/, document.body]
     };
     if (exceptions.hasOwnProperty(domain)) {
         let e = exceptions[domain];
@@ -32,13 +14,7 @@
             }
         }
     }
-    if (exceptions2.hasOwnProperty(domain)) {
-        let e = exceptions2[domain];
-        let tmp = e[0].exec(e[1].innerHTML);
-        if (tmp !== null) {
-            result = [tmp[1]];
-        }
-    }
+
     if (result === null) {
         let text = document.body.innerHTML;
         text = text.replace(/<script\b[^<]*(?:(?!<\/script>)<[^<]*)*<\/script>/gi, ' ');
@@ -62,6 +38,17 @@
         for (let elem of flix) {
             result.push(elem.getAttribute('data-flix-ean'));
         }
+        // Kody w danych strukturalnych schema.org (JSON-LD) — generyczny skan
+        // ich nie widzi, bo usuwa wszystkie bloki <script>.
+        let ldjson = document.querySelectorAll('script[type="application/ld+json"]');
+        for (let elem of ldjson) {
+            for (let m of elem.textContent.matchAll(/"gtin(?:13|8)?"\s*:\s*"(\d{13}|\d{8})(?=")/g)) {
+                if (result === null) {
+                    result = [];
+                }
+                result.push(m[1]);
+            }
+        }
     }
     if (result !== null) {
         result = result.filter((v, i, a) => a.indexOf(v) === i);

test-e2e/ean-detection.spec.js

@@ -165,6 +165,30 @@ const productPages = [
         product: 'Kujawski — Olej Kujawski 2 l',
         ean: '5900012003608',
         url: 'https://e-prim.pl/pl/products/olej-kujawski-2l-11380',
+    },
+    {
+        store: 'polskikoszyk.pl',
+        product: 'Dega Sałatka pikantna z makrelą 130 g',
+        ean: '5907623267788',
+        url: 'https://polskikoszyk.pl/dega-salatka-pikantna-z-makrela-130g',
+    },
+    {
+        store: 'pyszneeko.pl',
+        product: 'Mąka pszenna tortowa typ 480 1 kg Bio Planet EKO',
+        ean: '5907814660121',
+        url: 'https://pyszneeko.pl/product-pol-2744-Maka-pszenna-tortowa-typ-480-1-kg-Bio-Planet-EKO.html?rec=102002101',
+    },
+    {
+        store: 'emaja24.pl',
+        product: 'Aga-Holtex — Daktyle suszone 120 g',
+        ean: '5905027000260',
+        url: 'https://www.emaja24.pl/pl/p/Aga-Holtex-Daktyle-suszone-120g-Bakalie/5470',
+    },
+    {
+        store: 'marketbio.pl',
+        product: 'Sol Kłodawska kam. nat. do przetworów niej. 1,1 kg',
+        ean: '5902587000952',
+        url: 'https://marketbio.pl/sol-klodawska-kam-nat-do-przetworow-niej-1-1-kg.html',
     }
 ];