@govtechsg/oobee 0.10.76 → 0.10.78-alpha1

package/dist/crawlers/custom/extractText.js

@@ -0,0 +1,27 @@
+export function extractText() {
+    try {
+        // Extract text content from all specified elements (e.g., paragraphs)
+        const elements = document.querySelectorAll('p'); // Adjust selector as needed
+        const extractedSentences = [];
+        elements.forEach(element => {
+            const text = element.innerText.trim();
+            // Split the text into individual sentences
+            const sentencePattern = /[^.!?]*[.!?]+/g; // Match sentences ending with ., !, or ?
+            const matches = text.match(sentencePattern);
+            if (matches) {
+                // Add only sentences that end with punctuation
+                matches.forEach(sentence => {
+                    const trimmedSentence = sentence.trim(); // Trim whitespace from each sentence
+                    if (trimmedSentence.length > 0) {
+                        extractedSentences.push(trimmedSentence);
+                    }
+                });
+            }
+        });
+        return extractedSentences;
+    }
+    catch (error) {
+        console.error('Error extracting text:', error);
+        return []; // Return an empty string in case of an error
+    }
+}

package/dist/crawlers/custom/findElementByCssSelector.js

@@ -0,0 +1,36 @@
+import { framesCheck } from "./framesCheck.js";
+export function findElementByCssSelector(cssSelector) {
+    let doc = document;
+    // Check if the selector includes 'frame' or 'iframe' and update doc and selector
+    if (/\s*>\s*html\s*/.test(cssSelector)) {
+        const inFrames = framesCheck(cssSelector);
+        doc = inFrames.doc;
+        cssSelector = inFrames.remainingSelector;
+    }
+    // Query the element in the document (including inside frames)
+    let element = doc.querySelector(cssSelector);
+    // Handle Shadow DOM if the element is not found
+    if (!element) {
+        const shadowRoots = [];
+        const allElements = document.querySelectorAll('*');
+        // Look for elements with shadow roots
+        allElements.forEach(el => {
+            if (el.shadowRoot) {
+                shadowRoots.push(el.shadowRoot);
+            }
+        });
+        // Search inside each shadow root for the element
+        for (const shadowRoot of shadowRoots) {
+            const shadowElement = shadowRoot.querySelector(cssSelector);
+            if (shadowElement) {
+                element = shadowElement; // Found the element inside shadow DOM
+                break;
+            }
+        }
+    }
+    if (element) {
+        return element.outerHTML;
+    }
+    console.warn(`Unable to find element for css selector: ${cssSelector}`);
+    return null;
+}