Coverage for tests/test_md.py: 100%

1import re 

2import os 

3import io 

4from app.models import User 

5from app import db 

6from werkzeug.datastructures import FileStorage 

7from unittest.mock import patch 

8# from app import app 

9 

10def login_user(client): 

11 test_user = User(username="testuser") 

12 test_user.set_password("password") 

13 db.session.add(test_user) 

14 db.session.commit() 

15 response = client.post( 

16 "/index", 

17 data={"username": "testuser", "password": "password"}, 

18 follow_redirects=True, 

19 ) 

20 return client 

21 

22@patch("app.doc_indexer.retrieve_document.query_database") 

23def test_md(mock_query_database, client, app): 

24 # When the database returns a similarity score, override it to avoid 'no doc found' 

25 # mock_query_database.return_value = [ 

26 # ("Mocked document content about DNA", 0.9), 

27 # ] 

28 # Add document to database for use in testing 

29 app.config["WTF_CSRF_ENABLED"] = ( 

30 False # In testing envionment, no need for CSRF_TOKEN 

31 ) 

32 with app.app_context(): 

33 with client: 

34 client = login_user(client) 

35 file_path = os.path.join(os.path.dirname(__file__), "test_data", "Dna.pdf") 

36 with open(file_path, "rb") as f: 

37 # Create a FileStorage object from the file content 

38 pdf_file = FileStorage( 

39 stream=io.BytesIO(f.read()), 

40 filename="Dna.pdf", 

41 content_type="application/pdf", 

42 ) 

43 data = { 

44 "pdf_file": (io.BytesIO(pdf_file.read()), "Dna.pdf"), 

45 } 

46 response = client.post( 

47 "/admin", data=data, content_type="multipart/form-data" 

48 ) 

49 

50 response = client.post("/chat", json={"message": "Tell me about DNA", "conversationHistory": []}) 

51 content = response.get_data(as_text=True) 

52 print(content) 

53 md_format = False 

54 patterns = [ 

55 r"^\s*#+\s", # Headers (allow leading spaces) 

56 r"(\*\*|__)(.*?)\1", # Bold text 

57 r"(\*|_)(.*?)\1", # Italic text 

58 r"!?\[(.*?)\]\((.*?)\)", # Images and links 

59 r"`(.*?)`", # Inline code 

60 r"^-{3,}$", # Horizontal rules 

61 r"^\s*[-*\+]\s", # Lists 

62 r">", # Blockquotes 

63 ] 

64 for pattern in patterns: 

65 if re.search(pattern, content, re.M): 

66 md_format = True 

67 break 

68 assert md_format == True 

69 # try: 

70 # html = markdown.markdown(content) 

71 # except Exception as e: 

72 # pytest.fail(f"Markdown conversion failed: {e}") 

73 # assert response.status_code == 200