blaxx14 commited on
Commit
15a0b55
·
verified ·
1 Parent(s): 56295b7

Update backend/parser.py

Browse files
Files changed (1) hide show
  1. backend/parser.py +5 -5
backend/parser.py CHANGED
@@ -116,9 +116,9 @@ def parse_promotion_pdf(pdf_path):
116
  docx_path = pathname + ".docx"
117
 
118
  with open(pdf_path, 'rb') as f:
119
- convert_pdf_to_word(f, os.path.join('tmp', docx_path))
120
 
121
- tables = extract_tables_from_docx(os.path.join('tmp', docx_path))
122
  tables_result = parse_table_data(tables)
123
  del tables_result[0]
124
 
@@ -227,18 +227,18 @@ def parse_promotion_excel(excel_path, filename):
227
  data = df.to_dict(orient="records")
228
 
229
  # Buat folder temp jika belum ada
230
- os.makedirs('tmp', exist_ok=True)
231
 
232
  # Tambah .json jika belum ada
233
  if not filename.lower().endswith('.json'):
234
  filename += '.json'
235
 
236
  # Cegah overwrite file
237
- filepath = os.path.join('tmp', filename)
238
  base_name, ext = os.path.splitext(filename)
239
  copy_num = 1
240
  while os.path.exists(filepath):
241
- filepath = os.path.join('tmp', f"{base_name} ({copy_num}){ext}")
242
  copy_num += 1
243
 
244
  # Simpan file JSON
 
116
  docx_path = pathname + ".docx"
117
 
118
  with open(pdf_path, 'rb') as f:
119
+ convert_pdf_to_word(f, os.path.join('/tmp', docx_path))
120
 
121
+ tables = extract_tables_from_docx(os.path.join('/tmp', docx_path))
122
  tables_result = parse_table_data(tables)
123
  del tables_result[0]
124
 
 
227
  data = df.to_dict(orient="records")
228
 
229
  # Buat folder temp jika belum ada
230
+ os.makedirs('/tmp', exist_ok=True)
231
 
232
  # Tambah .json jika belum ada
233
  if not filename.lower().endswith('.json'):
234
  filename += '.json'
235
 
236
  # Cegah overwrite file
237
+ filepath = os.path.join('/tmp', filename)
238
  base_name, ext = os.path.splitext(filename)
239
  copy_num = 1
240
  while os.path.exists(filepath):
241
+ filepath = os.path.join('/tmp', f"{base_name} ({copy_num}){ext}")
242
  copy_num += 1
243
 
244
  # Simpan file JSON