Update lib/read_pdf.py

lib/read_pdf.py

@@ -54,7 +54,7 @@ def extract_and_format_paragraphs(pdf_path):
             if paragraph_lines and not is_end_of_sentence(paragraph_lines[-1]):
                 # This line is a continuation of the previous one
                 if paragraph_lines[-1][-1] == "-":
-                    paragraph_lines[-1] = paragraph_lines[-1][:-1]
+                    #paragraph_lines[-1] = paragraph_lines[-1][:-1]
                     paragraph_lines[-1] += line.strip()
                     
                 paragraph_lines[-1] += ' ' + line.strip()
@@ -104,7 +104,7 @@ def extract_and_format_paragraphs(pdf_path):
                     width = page.width
                     height = page.height
     
-                    header_height = height * 0.1  # Adjust this value based on your PDF
+                    header_height = height * 0.08  # Adjust this value based on your PDF
                     #footer_height = height * 0.1   # Adjust this value based on your PDF
     
                     left_bbox = (0, header_height, width / 2, height)  # Left column