ocr_api

Paused

Arafath10 commited on Jul 23, 2024

Commit

599b9d7

verified ·

1 Parent(s): a7db7e5

Update scraper.py

Files changed (1) hide show

scraper.py CHANGED Viewed

@@ -25,7 +25,7 @@ class Scraper:
             # Get page content (paragraphs, headers)
             page_content = await page.evaluate("""() => {
-            let elements = Array.from(document.querySelectorAll('p, h1, h2, h3, h4, h5, h6'));
             return elements.map(element => element.innerText).join('\\n');
             }""")

             # Get page content (paragraphs, headers)
             page_content = await page.evaluate("""() => {
+            let elements = Array.from(document.querySelectorAll('body *'));
             return elements.map(element => element.innerText).join('\\n');
             }""")