Spaces:

bachpc
/

table-structure-recognition

Build error

pcback commited on Mar 2, 2023

Commit

76f6ada

1 Parent(s): 08161c3

Fix html escape

Files changed (1) hide show

app.py CHANGED Viewed

@@ -14,6 +14,7 @@ import pytesseract
 from pytesseract import Output
 from bs4 import BeautifulSoup as bs
 import sys, json
@@ -202,7 +203,7 @@ def cells_to_html(cells):
         for cell in r_cells:
             rowspan = cell['row_nums'][-1] - cell['row_nums'][0] + 1
             colspan = cell['column_nums'][-1] - cell['column_nums'][0] + 1
-            r_html += f'<td rowspan="{rowspan}" colspan="{colspan}">{cell["text"]}</td>'
         html_code += f'<tr>{r_html}</tr>'
     html_code = '''<html>
                    <head>

 from pytesseract import Output
 from bs4 import BeautifulSoup as bs
+from html import escape
 import sys, json
         for cell in r_cells:
             rowspan = cell['row_nums'][-1] - cell['row_nums'][0] + 1
             colspan = cell['column_nums'][-1] - cell['column_nums'][0] + 1
+            r_html += f'<td rowspan="{rowspan}" colspan="{colspan}">{escape(cell["text"])}</td>'
         html_code += f'<tr>{r_html}</tr>'
     html_code = '''<html>
                    <head>