Update app.py

app.py

@@ -176,15 +176,41 @@ This translation app is useful for people who want to translate something or wan
             book = []  # PDF text data pool
             progressbar1 = st.empty()
             my_bar1 = progressbar1.progress(0)
-            #for index, page in enumerate(doc.pages):
+            
+            from bs4 import BeautifulSoup
+            xml_line_blocks = st.empty()
             for index, page in enumerate(doc):
                 #page_text = page.extract_text()
-                page_text = page.get_text(sort=True)
-                book.append((index, page_text))
+                #page_text = page.get_text(sort=True)
+                blocks = page.get_text("xml")
+                soup = BeautifulSoup(blocks,'lxml-xml')
+                page_text2 = ""
+
+                for tag0 in soup.find_all("block"):
+                    temp_y_posi = 0.0
+                    for tag1 in tag0.find_all("line"):
+                        for tag2 in tag1.find_all("font"):
+                            for tag3 in tag2.find_all("char"):
+                                y_posi = tag3.get("y")
+                                if y_posi != temp_y_posi:
+                                    page_text2 += "\n"
+                                    temp_y_posi = y_posi
+                                page_text2 += tag3.get("c")
+
+                    xml_line_blocks.write(page_text2)
+                    time.sleep(0.05)
+            #for index, page in enumerate(doc.pages):
+            #for index, page in enumerate(doc):
+            #    #page_text = page.extract_text()
+            #    page_text = page.get_text(sort=True)
+            #    book.append((index, page_text))
+                book.append((index, page_text2))
+
                 done = int(((index + 1) / page_count) * 100)
                 my_bar1.progress(done,
                                  text=f"Reading Page Number : {index + 1}")
             doc.close()
+            xml_line_blocks.empty()
             st.session_state.book = book
             my_bar1.empty()
             if os.path.isfile(