Merge pull request #482 from tutorcruncher/extra-details-spam-email-sentry-warnings

Author: tomhamiltonstubber

src/spam/email_checker.py

@@ -1,12 +1,27 @@
 import logging
-from html import escape
+import re
+from html import unescape
+from typing import Optional, Union
 
 from src.render.main import MessageDef, render_email
 from src.schemas.messages import EmailSendModel
 from src.spam.services import OpenAISpamEmailService, SpamCacheService
 
 logger = logging.getLogger('spam.email_checker')
 
+_html_tag_re = re.compile(r'<[^>]+>')
+_white_space_re = re.compile(r'\s+')
+
+
+def _clean_html_body(
+    html_body: Optional[str] = None,
+) -> Union[str, None]:
+    '''
+    Cleans the html body - removes HTML tags and whitespaces
+    '''
+    if isinstance(html_body, str) and html_body.strip():
+        return _white_space_re.sub(' ', unescape(_html_tag_re.sub('', html_body)))
+
 
 class EmailSpamChecker:
     def __init__(self, spam_service: OpenAISpamEmailService, cache_service: SpamCacheService):
@@ -40,9 +55,7 @@ async def check_spam(self, m: EmailSendModel):
         )
         email_info = render_email(message_def)
         company_name = m.context.get("company_name", "no_company")
-        escaped_html = escape(email_info.html_body)
         subject = email_info.subject
-        recipients = [recipient.address for recipient in m.recipients]
 
         spam_result = await self.spam_service.is_spam_email(email_info, company_name)
 
@@ -55,10 +68,10 @@ async def check_spam(self, m: EmailSendModel):
                 extra={
                     "reason": spam_result.reason,
                     "number of recipients": len(m.recipients),
-                    "to": recipients,
                     "subject": subject,
                     "company": company_name,
-                    "html_escaped": escaped_html,
+                    "company_code": m.company_code,
+                    "email_main_body": _clean_html_body(context.get('main_message__render')) or 'no main body',
                 },
             )
         return spam_result

tests/test_email.py

@@ -2,6 +2,7 @@
 import hashlib
 import hmac
 import json
+import logging
 import pytest
 import re
 from arq import Retry
@@ -1168,3 +1169,47 @@ def test_get_cache_key_with_emojis_and_special_chars():
         assert (
             cache_key != cache_key3
         ), f"Different company codes should produce different cache keys for {test_case['name']}"
+
+
+@pytest.mark.spam
+def test_spam_logging_includes_body(cli: TestClient, sync_db: SyncDb, worker, caplog):
+    caplog.set_level(logging.ERROR, logger='spam.email_checker')
+
+    recipients = []
+    for i in range(21):
+        recipients.append(
+            {
+                'first_name': f'User{i}',
+                'last_name': f'Last{i}',
+                'address': f'user{i}@example.org',
+                'tags': ['test'],
+            }
+        )
+
+    data = {
+        'uid': str(uuid4()),
+        'company_code': 'foobar',
+        'from_address': 'Spammer <[email protected]>',
+        'method': 'email-test',
+        'subject_template': 'Urgent: {{ company_name }} Alert!',
+        'main_template': '{{{ main_message }}}',
+        'context': {
+            'main_message__render': 'Hi {{ recipient_first_name }},\n\nDont miss out on <b>FREE MONEY</b>! '
+            'Click [here]({{ login_link }}) now!\n\nRegards,\n{{ company_name }}',
+            'company_name': 'TestCorp',
+            'login_link': 'https://spam.example.com/click',
+        },
+        'recipients': recipients,
+    }
+
+    r = cli.post('/send/email/', json=data, headers={'Authorization': 'testing-key'})
+    assert r.status_code == 201, r.text
+    assert worker.test_run() == len(recipients)
+
+    records = [r for r in caplog.records if r.name == 'spam.email_checker' and r.levelno == logging.ERROR]
+    assert len(records) == 1
+    body = getattr(records[0], 'email_main_body')
+    assert (
+        body == 'Hi {{ recipient_first_name }}, Dont miss out on FREE MONEY! '
+        'Click [here]({{ login_link }}) now! Regards, {{ company_name }}'
+    )