adding more rules

Re #2039

changedetectionio/processors/restock_diff.py

@@ -61,4 +61,4 @@ class perform_site_check(difference_detection_processor):
         # Always record the new checksum
         update_obj["previous_md5"] = fetched_md5
 
-        return changed_detected, update_obj, self.fetcher.instock_data.encode('utf-8')
+        return changed_detected, update_obj, self.fetcher.instock_data.encode('utf-8').strip()

changedetectionio/res/stock-not-in-stock.js

@@ -1,117 +1,131 @@
 function isItemInStock() {
-  // @todo Pass these in so the same list can be used in non-JS fetchers
+    // @todo Pass these in so the same list can be used in non-JS fetchers
-  const outOfStockTexts = [
+    const outOfStockTexts = [
-    ' أخبرني عندما يتوفر',
+        ' أخبرني عندما يتوفر',
-    '0 in stock',
+        '0 in stock',
-    'agotado',
+        'agotado',
-    'article épuisé',
+        'article épuisé',
-    'artikel zurzeit vergriffen',
+        'artikel zurzeit vergriffen',
-    'as soon as stock is available',
+        'as soon as stock is available',
-    'ausverkauft', // sold out
+        'ausverkauft', // sold out
-    'available for back order',
+        'available for back order',
-    'back-order or out of stock',
+        'back-order or out of stock',
-    'backordered',
+        'backordered',
-    'benachrichtigt mich', // notify me
+        'benachrichtigt mich', // notify me
-    'brak na stanie',
+        'brak na stanie',
-    'brak w magazynie',
+        'brak w magazynie',
-    'coming soon',
+        'coming soon',
-    'currently have any tickets for this',
+        'currently have any tickets for this',
-    'currently unavailable',
+        'currently unavailable',
-    'dostępne wkrótce',
+        'dostępne wkrótce',
-    'en rupture de stock',
+        'en rupture de stock',
-    'ist derzeit nicht auf lager',
+        'ist derzeit nicht auf lager',
-    'item is no longer available',
+        'item is no longer available',
-    'let me know when it\'s available',
+        'let me know when it\'s available',
-    'message if back in stock',
+        'message if back in stock',
-    'nachricht bei',
+        'nachricht bei',
-    'nicht auf lager',
+        'nicht auf lager',
-    'nicht lieferbar',
+        'nicht lieferbar',
-    'nicht zur verfügung',
+        'nicht zur verfügung',
-    'niet beschikbaar',
+        'niet beschikbaar',
-    'niet leverbaar',
+        'niet leverbaar',
-    'no disponible temporalmente',
+        'no disponible temporalmente',
-    'no longer in stock',
+        'no longer in stock',
-    'no tickets available',
+        'no tickets available',
-    'not available',
+        'not available',
-    'not currently available',
+        'not currently available',
-    'not in stock',
+        'not in stock',
-    'notify me when available',
+        'notify me when available',
-    'não estamos a aceitar encomendas',
+        'não estamos a aceitar encomendas',
-    'out of stock',
+        'out of stock',
-    'out-of-stock',
+        'out-of-stock',
-    'produkt niedostępny',
+        'produkt niedostępny',
-    'sold out',
+        'sold out',
-    'sold-out',
+        'sold-out',
-    'temporarily out of stock',
+        'temporarily out of stock',
-    'temporarily unavailable',
+        'temporarily unavailable',
-    'tickets unavailable',
+        'tickets unavailable',
-    'tijdelijk uitverkocht',
+        'tijdelijk uitverkocht',
-    'unavailable tickets',
+        'unavailable tickets',
-    'we do not currently have an estimate of when this product will be back in stock.',
+        'we do not currently have an estimate of when this product will be back in stock.',
-    'zur zeit nicht an lager',
+        'we don\'t know when or if this item will be back in stock.',
-    '品切れ',
+        'zur zeit nicht an lager',
-    '已售完',
+        '品切れ',
-    '품절'
+        '已售完',
-  ];
+        '품절'
+    ];
 
-
+    function getElementBaseText(element) {
-  const negateOutOfStockRegexs = [
+        // .textContent can include text from children which may give the wrong results
-      '[0-9] in stock'
+        // scan only immediate TEXT_NODEs, which will be a child of the element
-  ]
+        var text = "";
-  var negateOutOfStockRegexs_r = [];
+        for (var i = 0; i < element.childNodes.length; ++i)
-  for (let i = 0; i < negateOutOfStockRegexs.length; i++) {
+            if (element.childNodes[i].nodeType === Node.TEXT_NODE)
-    negateOutOfStockRegexs_r.push(new RegExp(negateOutOfStockRegexs[0], 'g'));
+                text += element.childNodes[i].textContent;
-  }
+        return text.toLowerCase().trim();
-
-
-  const elementsWithZeroChildren = Array.from(document.getElementsByTagName('*')).filter(element => element.children.length === 0);
-
-  // REGEXS THAT REALLY MEAN IT'S IN STOCK
-  for (let i = elementsWithZeroChildren.length - 1; i >= 0; i--) {
-    const element = elementsWithZeroChildren[i];
-    if (element.offsetWidth > 0 || element.offsetHeight > 0 || element.getClientRects().length > 0) {
-      var elementText="";
-      if (element.tagName.toLowerCase() === "input") {
-        elementText = element.value.toLowerCase();
-      } else {
-        elementText = element.textContent.toLowerCase();
-      }
-
-      if (elementText.length) {
-        // try which ones could mean its in stock
-        for (let i = 0; i < negateOutOfStockRegexs.length; i++) {
-          if (negateOutOfStockRegexs_r[i].test(elementText)) {
-            return 'Possibly in stock';
-          }
-        }
-      }
     }
-  }
 
-  // OTHER STUFF THAT COULD BE THAT IT'S OUT OF STOCK
+    const negateOutOfStockRegexs = [
-  for (let i = elementsWithZeroChildren.length - 1; i >= 0; i--) {
+        '[0-9] in stock'
-    const element = elementsWithZeroChildren[i];
+    ]
-    if (element.offsetWidth > 0 || element.offsetHeight > 0 || element.getClientRects().length > 0) {
+    var negateOutOfStockRegexs_r = [];
-      var elementText="";
+    for (let i = 0; i < negateOutOfStockRegexs.length; i++) {
-      if (element.tagName.toLowerCase() === "input") {
+        negateOutOfStockRegexs_r.push(new RegExp(negateOutOfStockRegexs[0], 'g'));
-        elementText = element.value.toLowerCase();
-      } else {
-        elementText = element.textContent.toLowerCase();
-      }
-
-      if (elementText.length) {
-        // and these mean its out of stock
-        for (const outOfStockText of outOfStockTexts) {
-          if (elementText.includes(outOfStockText)) {
-            return elementText; // item is out of stock
-          }
-        }
-      }
     }
-  }
 
-  return 'Possibly in stock'; // possibly in stock, cant decide otherwise.
+    // The out-of-stock or in-stock-text is generally always above-the-fold
+    // and often below-the-fold is a list of related products that may or may not contain trigger text
+    // so it's good to filter to just the 'above the fold' elements
+    // and it should be atleast 100px from the top to ignore items in the toolbar, sometimes menu items like "Coming soon" exist
+    const elementsToScan = Array.from(document.getElementsByTagName('*')).filter(element => element.getBoundingClientRect().top + window.scrollY <= window.innerHeight && element.getBoundingClientRect().top + window.scrollY >= 100);
+
+    var elementText = "";
+
+    // REGEXS THAT REALLY MEAN IT'S IN STOCK
+    for (let i = elementsToScan.length - 1; i >= 0; i--) {
+        const element = elementsToScan[i];
+        elementText = "";
+        if (element.tagName.toLowerCase() === "input") {
+            elementText = element.value.toLowerCase();
+        } else {
+            elementText = getElementBaseText(element);
+        }
+
+        if (elementText.length) {
+            // try which ones could mean its in stock
+            for (let i = 0; i < negateOutOfStockRegexs.length; i++) {
+                if (negateOutOfStockRegexs_r[i].test(elementText)) {
+                    return 'Possibly in stock';
+                }
+            }
+        }
+    }
+
+    // OTHER STUFF THAT COULD BE THAT IT'S OUT OF STOCK
+    for (let i = elementsToScan.length - 1; i >= 0; i--) {
+        const element = elementsToScan[i];
+        if (element.offsetWidth > 0 || element.offsetHeight > 0 || element.getClientRects().length > 0) {
+            elementText = "";
+            if (element.tagName.toLowerCase() === "input") {
+                elementText = element.value.toLowerCase();
+            } else {
+                elementText = getElementBaseText(element);
+            }
+
+            if (elementText.length) {
+                // and these mean its out of stock
+                for (const outOfStockText of outOfStockTexts) {
+                    if (elementText.includes(outOfStockText)) {
+                        return outOfStockText; // item is out of stock
+                    }
+                }
+            }
+        }
+    }
+
+    return 'Possibly in stock'; // possibly in stock, cant decide otherwise.
 }
 
 // returns the element text that makes it think it's out of stock
-return isItemInStock();
+return isItemInStock().trim()
+