Zaious commited on
Commit
c7b621c
·
verified ·
1 Parent(s): 7e4da7f

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +11 -2
app.py CHANGED
@@ -29,9 +29,18 @@ def generate_answer(brand_name,question, files):
29
 
30
  # 函数:解析PDF文件
31
  def extract_text_from_pdf(file_path):
32
- doc = fitz.open(file_path)
 
 
 
 
 
 
 
 
 
33
  text = ""
34
- for page_num in range(len(doc)):
35
  page = doc.load_page(page_num)
36
  text += page.get_text()
37
  return text
 
29
 
30
  # 函数:解析PDF文件
31
  def extract_text_from_pdf(file_path):
32
+ pdf_document = fitz.open(file_path)
33
+
34
+ total_pages = pdf_document.page_count
35
+ print(f"总页数: {total_pages}")
36
+
37
+ # 读取文本
38
+ page = pdf_document.load_page(0) # 读取第一页
39
+ text = page.get_text("text")
40
+ print(f"第一页文本:\n{text}")
41
+
42
  text = ""
43
+ for page_num in range(len(pdf_document)):
44
  page = doc.load_page(page_num)
45
  text += page.get_text()
46
  return text