Better bounding boxes

index.js

@@ -36,18 +36,54 @@ async function main() {
     // This will contain all the collected information
     let info = [];
 
+    // Take a first screenshot
+    await page.screenshot({path: __dirname + '/' + 'screenshot1.png'});
+
+    // Edit the page to shrink elements in order to get better bounding boxes
     for (let selector of ["h1", "h2", "h3", "h4", "h5", "h6", "a", "img", "p", "ul", "ol", "li"]) {
 
         let query = selector;
-        if (query === "p") {
+
-            // Skip paragraphs that have children (which means that they are not text paragraphs)
+        let shouldCreateSpan = query !== "ul" && query !== "ol" && query != "img";
-            query += ":not(:has(*))";
+
+        if (shouldCreateSpan) {
+
+            await page.evaluate(([query]) => {
+                for (let e of document.querySelectorAll(query)) {
+                    if (e.children.length === 0) {
+                        e.innerHTML = '<span>' + e.innerHTML + '</span>';
+                    }
                 }
 
+            }, [query]);
+        }
+    }
+
+    // Take another screenshot and check the modification we made didn't change the layout of the page
+    await page.screenshot({path: __dirname + '/' + 'screenshot2.png'});
+
+    // Compare both screenshots
+    let file1 = await fs.readFile(__dirname + '/' + 'screenshot1.png');
+    let file2 = await fs.readFile(__dirname + '/' + 'screenshot2.png');
+    let filesAreSame = file1.map((x, i) => x === file2[i]).reduce((a, b) => a && b, true);
+
+    // Crash if they're different
+    if (!filesAreSame) {
+        throw new Error("Page edit changed the layout");
+    }
+
+    for (let selector of ["h1", "h2", "h3", "h4", "h5", "h6", "a", "img", "p", "ul", "ol", "li"]) {
+
+        let query = selector;
+
+        // Shrink the elements horizontally (to be able to get better bounding boxes)
+        let shouldCreateSpan = query !== "ul" && query !== "ol" && query != "img";
+
         // Query the considered element
-        let elements = await page.$$(query);
+        let elements = await page.$$(query + (shouldCreateSpan ? ' > *:first-child' : ''));
 
         for (let element of elements) {
+
             let box = await element.boundingBox();
 
             // Scale the bounding box