From 4eada16b949fa097d1389ca08f0ee4b9774910f2 Mon Sep 17 00:00:00 2001
From: Soxoj <31013580+soxoj@users.noreply.github.com>
Date: Sun, 8 Dec 2024 13:35:27 +0100
Subject: [PATCH] Added a test for submitter (#1944)

---
 maigret/checking.py      |  4 +++-
 maigret/errors.py        |  2 +-
 maigret/maigret.py       |  4 +++-
 maigret/submit.py        | 14 ++++++++----
 tests/db.json            | 20 ++++++++++++++++-
 tests/test_errors.py     | 44 ++++++++++++++++++++++++++++++++------
 tests/test_maigret.py    |  1 +
 tests/test_permutator.py | 46 +++++++++++++++++++++++++++-------------
 8 files changed, 105 insertions(+), 30 deletions(-)

diff --git a/maigret/checking.py b/maigret/checking.py
index f622483..e5ef6dd 100644
--- a/maigret/checking.py
+++ b/maigret/checking.py
@@ -381,7 +381,9 @@ def process_site_result(
         extracted_ids_data = extract_ids_data(html_text, logger, site)
         if extracted_ids_data:
             new_usernames = parse_usernames(extracted_ids_data, logger)
-            results_info = update_results_info(results_info, extracted_ids_data, new_usernames)
+            results_info = update_results_info(
+                results_info, extracted_ids_data, new_usernames
+            )
             result.ids_data = extracted_ids_data
 
     # Save status of request
diff --git a/maigret/errors.py b/maigret/errors.py
index c9e1b07..573511e 100644
--- a/maigret/errors.py
+++ b/maigret/errors.py
@@ -174,4 +174,4 @@ def notify_about_errors(
             ('You can see detailed site check errors with a flag `--print-errors`', '-')
         )
 
-    return results
\ No newline at end of file
+    return results
diff --git a/maigret/maigret.py b/maigret/maigret.py
index 1b639da..82c9577 100755
--- a/maigret/maigret.py
+++ b/maigret/maigret.py
@@ -665,7 +665,9 @@ async def main():
             check_domains=args.with_domains,
         )
 
-        errs = errors.notify_about_errors(results, query_notify, show_statistics=args.verbose)
+        errs = errors.notify_about_errors(
+            results, query_notify, show_statistics=args.verbose
+        )
         for e in errs:
             query_notify.warning(*e)
 
diff --git a/maigret/submit.py b/maigret/submit.py
index 3092e35..fbe86ae 100644
--- a/maigret/submit.py
+++ b/maigret/submit.py
@@ -1,6 +1,7 @@
 import asyncio
 import json
 import re
+import os
 from typing import Any, Dict, List, Optional
 
 from aiohttp import ClientSession, TCPConnector
@@ -61,7 +62,10 @@ class Submitter:
         proxy = self.args.proxy
         cookie_jar = None
         if args.cookie_file:
-            cookie_jar = import_aiohttp_cookies(args.cookie_file)
+            if not os.path.exists(args.cookie_file):
+                logger.error(f"Cookie file {args.cookie_file} does not exist!")
+            else:
+                cookie_jar = import_aiohttp_cookies(args.cookie_file)
 
         connector = ProxyConnector.from_url(proxy) if proxy else TCPConnector(ssl=False)
         connector.verify_ssl = False
@@ -123,7 +127,9 @@ class Submitter:
     async def detect_known_engine(
         self, url_exists, url_mainpage
     ) -> [List[MaigretSite], str]:
+
         resp_text = ''
+
         try:
             r = await self.session.get(url_mainpage)
             content = await r.content.read()
@@ -131,8 +137,8 @@ class Submitter:
             resp_text = content.decode(charset, "ignore")
             self.logger.debug(resp_text)
         except Exception as e:
-            self.logger.warning(e)
-            print("Some error while checking main page")
+            self.logger.warning(e, exc_info=True)
+            print(f"Some error while checking main page: {e}")
             return [], resp_text
 
         for engine in self.db.engines:
@@ -160,7 +166,7 @@ class Submitter:
                     for u in usernames_to_check:
                         site_data = {
                             "urlMain": url_mainpage,
-                            "name": url_mainpage.split("//")[1],
+                            "name": url_mainpage.split("//")[1].split("/")[0],
                             "engine": engine_name,
                             "usernameClaimed": u,
                             "usernameUnclaimed": "noonewouldeverusethis7",
diff --git a/tests/db.json b/tests/db.json
index 46cbdf4..93e5969 100644
--- a/tests/db.json
+++ b/tests/db.json
@@ -1,5 +1,23 @@
 {
-    "engines": {},
+    "engines": {
+        "Discourse": {
+            "name": "Discourse",
+            "site": {
+                "presenseStrs": [
+                    "<meta name=\"generator\" content=\"Discourse"
+                ],
+                "absenceStrs": [
+                    "Oops! That page doesn\u2019t exist or is private.",
+                    "wrap not-found-container"
+                ],
+                "checkType": "message",
+                "url": "{urlMain}/u/{username}/summary"
+            },
+            "presenseStrs": [
+                "<meta name=\"generator\" content=\"Discourse"
+            ]
+        }
+    },
     "sites": {
         "ValidActive": {
             "tags": ["global", "us"],
diff --git a/tests/test_errors.py b/tests/test_errors.py
index d2067cc..e44bf7d 100644
--- a/tests/test_errors.py
+++ b/tests/test_errors.py
@@ -6,18 +6,48 @@ from maigret.result import MaigretCheckResult, MaigretCheckStatus
 
 def test_notify_about_errors():
     results = {
-        'site1': {'status': MaigretCheckResult('', '', '', MaigretCheckStatus.UNKNOWN, error=CheckError('Captcha'))},
-        'site2': {'status': MaigretCheckResult('', '', '', MaigretCheckStatus.UNKNOWN, error=CheckError('Bot protection'))},
-        'site3': {'status': MaigretCheckResult('', '', '', MaigretCheckStatus.UNKNOWN, error=CheckError('Access denied'))},
-        'site4': {'status': MaigretCheckResult('', '', '', MaigretCheckStatus.CLAIMED, error=None)},
+        'site1': {
+            'status': MaigretCheckResult(
+                '', '', '', MaigretCheckStatus.UNKNOWN, error=CheckError('Captcha')
+            )
+        },
+        'site2': {
+            'status': MaigretCheckResult(
+                '',
+                '',
+                '',
+                MaigretCheckStatus.UNKNOWN,
+                error=CheckError('Bot protection'),
+            )
+        },
+        'site3': {
+            'status': MaigretCheckResult(
+                '',
+                '',
+                '',
+                MaigretCheckStatus.UNKNOWN,
+                error=CheckError('Access denied'),
+            )
+        },
+        'site4': {
+            'status': MaigretCheckResult(
+                '', '', '', MaigretCheckStatus.CLAIMED, error=None
+            )
+        },
     }
 
     results = notify_about_errors(results, query_notify=None, show_statistics=True)
 
     # Check the output
     expected_output = [
-        ('Too many errors of type "Captcha" (25.0%). Try to switch to another ip address or to use service cookies', '!'),
-        ('Too many errors of type "Bot protection" (25.0%). Try to switch to another ip address', '!'),
+        (
+            'Too many errors of type "Captcha" (25.0%). Try to switch to another ip address or to use service cookies',
+            '!',
+        ),
+        (
+            'Too many errors of type "Bot protection" (25.0%). Try to switch to another ip address',
+            '!',
+        ),
         ('Too many errors of type "Access denied" (25.0%)', '!'),
         ('Verbose error statistics:', '-'),
         ('Captcha: 25.0%', '!'),
@@ -25,4 +55,4 @@ def test_notify_about_errors():
         ('Access denied: 25.0%', '!'),
         ('You can see detailed site check errors with a flag `--print-errors`', '-'),
     ]
-    assert results == expected_output
\ No newline at end of file
+    assert results == expected_output
diff --git a/tests/test_maigret.py b/tests/test_maigret.py
index d9d9fc9..4c2f8ac 100644
--- a/tests/test_maigret.py
+++ b/tests/test_maigret.py
@@ -15,6 +15,7 @@ from maigret.sites import MaigretSite
 from maigret.result import MaigretCheckResult, MaigretCheckStatus
 from tests.conftest import RESULTS_EXAMPLE
 
+
 @pytest.mark.slow
 @pytest.mark.asyncio
 async def test_self_check_db(test_db):
diff --git a/tests/test_permutator.py b/tests/test_permutator.py
index 21a20a8..8b466cb 100644
--- a/tests/test_permutator.py
+++ b/tests/test_permutator.py
@@ -7,12 +7,18 @@ def test_gather_strict():
     permute = Permute(elements)
     result = permute.gather(method="strict")
     expected = {
-        'a_b': 1, 'b_a': 2,
-        'a-b': 1, 'b-a': 2,
-        'a.b': 1, 'b.a': 2,
-        'ab': 1, 'ba': 2,
-        '_ab': 1, 'ab_': 1,
-        '_ba': 2, 'ba_': 2
+        'a_b': 1,
+        'b_a': 2,
+        'a-b': 1,
+        'b-a': 2,
+        'a.b': 1,
+        'b.a': 2,
+        'ab': 1,
+        'ba': 2,
+        '_ab': 1,
+        'ab_': 1,
+        '_ba': 2,
+        'ba_': 2,
     }
     assert result == expected
 
@@ -22,13 +28,23 @@ def test_gather_all():
     permute = Permute(elements)
     result = permute.gather(method="all")
     expected = {
-        'a': 1, '_a': 1, 'a_': 1,
-        'b': 2, '_b': 2, 'b_': 2,
-        'a_b': 1, 'b_a': 2,
-        'a-b': 1, 'b-a': 2,
-        'a.b': 1, 'b.a': 2,
-        'ab': 1, 'ba': 2,
-        '_ab': 1, 'ab_': 1,
-        '_ba': 2, 'ba_': 2
+        'a': 1,
+        '_a': 1,
+        'a_': 1,
+        'b': 2,
+        '_b': 2,
+        'b_': 2,
+        'a_b': 1,
+        'b_a': 2,
+        'a-b': 1,
+        'b-a': 2,
+        'a.b': 1,
+        'b.a': 2,
+        'ab': 1,
+        'ba': 2,
+        '_ab': 1,
+        'ab_': 1,
+        '_ba': 2,
+        'ba_': 2,
     }
-    assert result == expected 
\ No newline at end of file
+    assert result == expected