From a29c3c6abe9d83efe3c59a795fe85e522a2b0aac Mon Sep 17 00:00:00 2001
From: Soxoj <31013580+soxoj@users.noreply.github.com>
Date: Sat, 26 Feb 2022 13:38:15 +0300
Subject: [PATCH 1/8] CI autoupdate (#359)

* CI autoupdate

* Updated site list and statistics
---
 .github/workflows/update-site-data.yml | 24 ++++++++++--
 sites.md                               | 51 +++++++++++++++++++++++++-
 2 files changed, 71 insertions(+), 4 deletions(-)

diff --git a/.github/workflows/update-site-data.yml b/.github/workflows/update-site-data.yml
index d4689da..f0321e1 100644
--- a/.github/workflows/update-site-data.yml
+++ b/.github/workflows/update-site-data.yml
@@ -1,18 +1,36 @@
-name: Update sites rating
+name: Update sites rating and statistics
 
 on:
   push:
     branches: [ main ]
   pull_request:
     branches: [ main ]
+    types: [opened, synchronize]
 
 jobs:
   build:
     runs-on: ubuntu-latest
     steps:
-    - name: checkout repo
-      uses: actions/checkout@v2
+    - name: Checkout repository
+      uses: actions/checkout@v2.3.2
+      with:
+        ref: ${{ github.event.pull_request.head.sha }}
+        fetch-depth: 0 # otherwise, there would be errors pushing refs to the destination repository.
+
     - name: build application
       run: |
         pip3 install .
         python3 ./utils/update_site_data.py --empty-only
+
+    - name: Commit and push changes
+      run: |
+        git config --global user.name "Maigret autoupdate"
+        git config --global user.email "soxoj@protonmail.com"
+        echo `git name-rev ${{ github.event.pull_request.head.sha }} --name-only`
+        export BRANCH=`git name-rev ${{ github.event.pull_request.head.sha }} --name-only | sed 's/remotes\/origin\///'`
+        echo $BRANCH
+        git remote -v
+        git checkout $BRANCH
+        git add sites.md
+        git commit -m "Updated site list and statistics"
+        git push origin $BRANCH
\ No newline at end of file
diff --git a/sites.md b/sites.md
index e2b7d20..6bd93d0 100644
--- a/sites.md
+++ b/sites.md
@@ -2599,4 +2599,53 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://www.hozpitality.com) [hozpitality (https://www.hozpitality.com)](https://www.hozpitality.com)*: top 100M*
 1. ![](https://www.google.com/s2/favicons?domain=https://kazanlashkigalab.com) [kazanlashkigalab.com (https://kazanlashkigalab.com)](https://kazanlashkigalab.com)*: top 100M, kz*
 
-Alexa.com rank data fetched at (2022-01-21 21:28:53.863014 UTC)
+Alexa.com rank data fetched at (2022-02-23 22:14:26.029891 UTC)
+## Statistics
+
+Enabled/total sites: 2449/2595
+
+Incomplete checks: 586/1978 (false positive risks)
+
+Top 20 profile URLs:
+- (796)	`{urlMain}/index/8-0-{username} (uCoz)`
+- (221)	`{urlMain}{urlSubpath}/members/?username={username} (XenForo)`
+- (221)	`/{username}`
+- (138)	`/user/{username}`
+- (134)	`{urlMain}{urlSubpath}/member.php?username={username} (vBulletin)`
+- (97)	`/profile/{username}`
+- (87)	`{urlMain}/u/{username}/summary (Discourse)`
+- (74)	`/users/{username}`
+- (44)	`{urlMain}{urlSubpath}/search.php?author={username} (phpBB/Search)`
+- (41)	`/members/?username={username}`
+- (39)	`SUBDOMAIN`
+- (36)	`/@{username}`
+- (28)	`/u/{username}`
+- (27)	`{urlMain}{urlSubpath}/memberlist.php?username={username} (phpBB)`
+- (24)	`/members/{username}`
+- (18)	`/forum/members/?username={username}`
+- (18)	`/forum/search.php?keywords=&terms=all&author={username}`
+- (17)	`/search.php?keywords=&terms=all&author={username}`
+- (15)	`/author/{username}`
+- (14)	`/profile.php?mode=viewprofile&u={username}`
+
+Top 20 tags:
+- (255)	`forum`
+- (50)	`gaming`
+- (40)	`NO_TAGS` (non-standard)
+- (24)	`coding`
+- (23)	`photo`
+- (19)	`news`
+- (18)	`blog`
+- (18)	`music`
+- (15)	`tech`
+- (13)	`freelance`
+- (12)	`sharing`
+- (12)	`finance`
+- (11)	`shopping`
+- (10)	`dating`
+- (10)	`art`
+- (9)	`hobby`
+- (8)	`movies`
+- (7)	`sport`
+- (7)	`hacking`
+- (5)	`stock`

From 8a865a1ce6dc85887fb29881672714a6cd3cdea3 Mon Sep 17 00:00:00 2001
From: Soxoj <31013580+soxoj@users.noreply.github.com>
Date: Sat, 26 Feb 2022 14:16:13 +0300
Subject: [PATCH 2/8] Op.gg fixes (#363)

* Fixed op.gg sites

* Added testing docs, fixed some error

* Updated site list and statistics
---
 docs/source/development.rst | 31 +++++++++++++++
 maigret/checking.py         |  2 +-
 maigret/resources/data.json | 76 +++++++++++--------------------------
 sites.md                    |  2 +-
 4 files changed, 55 insertions(+), 56 deletions(-)

diff --git a/docs/source/development.rst b/docs/source/development.rst
index 95ae37a..c17ed63 100644
--- a/docs/source/development.rst
+++ b/docs/source/development.rst
@@ -3,6 +3,37 @@
 Development
 ==============
 
+Testing
+-------
+
+It is recommended use Python 3.7/3.8 for test due to some conflicts in 3.9.
+
+Install test requirements:
+
+.. code-block:: console
+
+  pip install -r test-requirements.txt
+
+
+Use the following commands to check Maigret:
+
+.. code-block:: console
+
+  # run linter and typing checks
+  # order of checks%
+  # - critical syntax errors or undefined names
+  # - flake checks
+  # - mypy checks
+  make lint
+
+  # run testing with coverage html report
+  # current test coverage is 60%
+  make text
+
+  # open html report
+  open htmlcov/index.html
+
+
 How to publish new version of Maigret
 -------------------------------------
 
diff --git a/maigret/checking.py b/maigret/checking.py
index a196cfc..b281287 100644
--- a/maigret/checking.py
+++ b/maigret/checking.py
@@ -132,7 +132,7 @@ class SimpleAiohttpChecker(CheckerBase):
                 error = CheckError("Unexpected", str(e))
 
         if error == "Invalid proxy response":
-            self.logger.debug(e, exc_info=True)
+            self.logger.debug(error, exc_info=True)
 
         return str(html_text), status_code, error
 
diff --git a/maigret/resources/data.json b/maigret/resources/data.json
index d2673a8..c9fb0a9 100644
--- a/maigret/resources/data.json
+++ b/maigret/resources/data.json
@@ -16574,12 +16574,7 @@
                 "br",
                 "us"
             ],
-            "checkType": "message",
-            "presenseStrs": "common.SummonerHistory.Favorite",
-            "absenceStrs": [
-                "SummonerNotFoundLayout"
-            ],
-            "alexaRank": 331,
+            "engine": "op.gg",
             "urlMain": "https://br.op.gg/",
             "url": "https://br.op.gg/summoner/userName={username}",
             "usernameClaimed": "adam",
@@ -17120,12 +17115,7 @@
                 "gaming",
                 "us"
             ],
-            "checkType": "message",
-            "presenseStrs": "common.SummonerHistory.Favorite",
-            "absenceStrs": [
-                "SummonerNotFoundLayout"
-            ],
-            "alexaRank": 331,
+            "engine": "op.gg",
             "urlMain": "https://eune.op.gg/",
             "url": "https://eune.op.gg/summoner/userName={username}",
             "usernameClaimed": "adam",
@@ -17136,12 +17126,7 @@
                 "gaming",
                 "us"
             ],
-            "checkType": "message",
-            "presenseStrs": "common.SummonerHistory.Favorite",
-            "absenceStrs": [
-                "SummonerNotFoundLayout"
-            ],
-            "alexaRank": 331,
+            "engine": "op.gg",
             "urlMain": "https://euw.op.gg/",
             "url": "https://euw.op.gg/summoner/userName={username}",
             "usernameClaimed": "blue",
@@ -18080,12 +18065,7 @@
             "tags": [
                 "us"
             ],
-            "checkType": "message",
-            "presenseStrs": "common.SummonerHistory.Favorite",
-            "absenceStrs": [
-                "SummonerNotFoundLayout"
-            ],
-            "alexaRank": 331,
+            "engine": "op.gg",
             "urlMain": "https://lan.op.gg/",
             "url": "https://lan.op.gg/summoner/userName={username}",
             "usernameClaimed": "adam",
@@ -18096,12 +18076,7 @@
                 "gaming",
                 "us"
             ],
-            "checkType": "message",
-            "presenseStrs": "common.SummonerHistory.Favorite",
-            "absenceStrs": [
-                "SummonerNotFoundLayout"
-            ],
-            "alexaRank": 331,
+            "engine": "op.gg",
             "urlMain": "https://las.op.gg/",
             "url": "https://las.op.gg/summoner/userName={username}",
             "usernameClaimed": "adam",
@@ -18485,12 +18460,7 @@
             "tags": [
                 "gaming"
             ],
-            "checkType": "message",
-            "presenseStrs": "common.SummonerHistory.Favorite",
-            "absenceStrs": [
-                "SummonerNotFoundLayout"
-            ],
-            "alexaRank": 331,
+            "engine": "op.gg",
             "urlMain": "https://na.op.gg/",
             "url": "https://na.op.gg/summoner/userName={username}",
             "usernameClaimed": "adam",
@@ -18670,12 +18640,7 @@
                 "gaming",
                 "us"
             ],
-            "checkType": "message",
-            "presenseStrs": "common.SummonerHistory.Favorite",
-            "absenceStrs": [
-                "SummonerNotFoundLayout"
-            ],
-            "alexaRank": 331,
+            "engine": "op.gg",
             "urlMain": "https://oce.op.gg/",
             "url": "https://oce.op.gg/summoner/userName={username}",
             "usernameClaimed": "adam",
@@ -19142,12 +19107,7 @@
                 "ru",
                 "us"
             ],
-            "checkType": "message",
-            "presenseStrs": "common.SummonerHistory.Favorite",
-            "absenceStrs": [
-                "SummonerNotFoundLayout"
-            ],
-            "alexaRank": 331,
+            "engine": "op.gg",
             "urlMain": "https://ru.op.gg/",
             "url": "https://ru.op.gg/summoner/userName={username}",
             "usernameClaimed": "adam",
@@ -19694,12 +19654,7 @@
                 "tr",
                 "us"
             ],
-            "checkType": "message",
-            "presenseStrs": "common.SummonerHistory.Favorite",
-            "absenceStrs": [
-                "SummonerNotFoundLayout"
-            ],
-            "alexaRank": 331,
+            "engine": "op.gg",
             "urlMain": "https://tr.op.gg/",
             "url": "https://tr.op.gg/summoner/userName={username}",
             "usernameClaimed": "adam",
@@ -29054,6 +29009,19 @@
                     "404"
                 ]
             }
+        },
+        "op.gg": {
+            "name": "op.gg",
+            "site": {
+                "checkType": "message",
+                "presenseStrs": [
+                    "<div class=\"profile-icon\">"
+                ],
+                "absenceStrs": [
+                    "<h2 class=\"header__title\">This summoner is not registered"
+                ],
+                "alexaRank": 331
+            }
         }
     },
     "tags": [
diff --git a/sites.md b/sites.md
index 6bd93d0..37523c4 100644
--- a/sites.md
+++ b/sites.md
@@ -2599,7 +2599,7 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://www.hozpitality.com) [hozpitality (https://www.hozpitality.com)](https://www.hozpitality.com)*: top 100M*
 1. ![](https://www.google.com/s2/favicons?domain=https://kazanlashkigalab.com) [kazanlashkigalab.com (https://kazanlashkigalab.com)](https://kazanlashkigalab.com)*: top 100M, kz*
 
-Alexa.com rank data fetched at (2022-02-23 22:14:26.029891 UTC)
+Alexa.com rank data fetched at (2022-02-26 11:12:43.279519 UTC)
 ## Statistics
 
 Enabled/total sites: 2449/2595

From be204ff11923de9a049137ece9c97cdeed249be4 Mon Sep 17 00:00:00 2001
From: Soxoj <31013580+soxoj@users.noreply.github.com>
Date: Sat, 26 Feb 2022 14:27:08 +0300
Subject: [PATCH 3/8] Wikipedia fix (#365)

* Fixed op.gg sites

* Added testing docs, fixed some error

* Fixed Wikipedia
---
 .github/workflows/update-site-data.yml | 1 -
 maigret/resources/data.json            | 3 ++-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/update-site-data.yml b/.github/workflows/update-site-data.yml
index f0321e1..2d2c81e 100644
--- a/.github/workflows/update-site-data.yml
+++ b/.github/workflows/update-site-data.yml
@@ -5,7 +5,6 @@ on:
     branches: [ main ]
   pull_request:
     branches: [ main ]
-    types: [opened, synchronize]
 
 jobs:
   build:
diff --git a/maigret/resources/data.json b/maigret/resources/data.json
index c9fb0a9..60dd729 100644
--- a/maigret/resources/data.json
+++ b/maigret/resources/data.json
@@ -15518,7 +15518,8 @@
             ],
             "checkType": "message",
             "absenceStrs": [
-                "is not registered"
+                "is not registered",
+                "Wikipedia does not have a"
             ],
             "alexaRank": 12,
             "urlMain": "https://www.wikipedia.org/",

From 61452d56d394789d32bc0f811b8787de45cc0a8b Mon Sep 17 00:00:00 2001
From: Soxoj <31013580+soxoj@users.noreply.github.com>
Date: Sat, 26 Feb 2022 14:49:43 +0300
Subject: [PATCH 4/8] Disabled Netvibes and LeetCode (#366)

* Disabled Netvibes and LeetCode

* Specified types of PR for tests in CI

* Updated site list and statistics
---
 .github/workflows/python-package.yml |  1 +
 maigret/resources/data.json          | 10 +++++++---
 sites.md                             |  8 ++++----
 3 files changed, 12 insertions(+), 7 deletions(-)

diff --git a/.github/workflows/python-package.yml b/.github/workflows/python-package.yml
index 295160b..132fe8e 100644
--- a/.github/workflows/python-package.yml
+++ b/.github/workflows/python-package.yml
@@ -5,6 +5,7 @@ on:
     branches: [ main ]
   pull_request:
     branches: [ main ]
+    types: [opened, synchronize]
 
 jobs:
   build:
diff --git a/maigret/resources/data.json b/maigret/resources/data.json
index 60dd729..6ce0b6a 100644
--- a/maigret/resources/data.json
+++ b/maigret/resources/data.json
@@ -7760,6 +7760,7 @@
             "tags": [
                 "coding"
             ],
+            "disabled": true,
             "checkType": "status_code",
             "alexaRank": 1859,
             "urlMain": "https://leetcode.com/",
@@ -9711,12 +9712,15 @@
         },
         "Netvibes": {
             "tags": [
-                "de",
-                "fr",
-                "jp"
+                "business",
+                "fr"
             ],
+            "disabled": true,
             "checkType": "status_code",
             "alexaRank": 5730,
+            "headers": {
+                "User-Agent": "curl/7.64.1"
+            },
             "urlMain": "https://www.netvibes.com",
             "url": "https://www.netvibes.com/{username}#General",
             "usernameClaimed": "blue",
diff --git a/sites.md b/sites.md
index 37523c4..f035818 100644
--- a/sites.md
+++ b/sites.md
@@ -197,7 +197,7 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://www.polygon.com/) [Polygon (https://www.polygon.com/)](https://www.polygon.com/)*: top 5K, us*
 1. ![](https://www.google.com/s2/favicons?domain=https://otzovik.com/) [Otzovik (https://otzovik.com/)](https://otzovik.com/)*: top 5K, ru*
 1. ![](https://www.google.com/s2/favicons?domain=https://www.liveinternet.ru) [LiveInternet (https://www.liveinternet.ru)](https://www.liveinternet.ru)*: top 5K, ru*
-1. ![](https://www.google.com/s2/favicons?domain=https://leetcode.com/) [LeetCode (https://leetcode.com/)](https://leetcode.com/)*: top 5K, coding*
+1. ![](https://www.google.com/s2/favicons?domain=https://leetcode.com/) [LeetCode (https://leetcode.com/)](https://leetcode.com/)*: top 5K, coding*, search is disabled
 1. ![](https://www.google.com/s2/favicons?domain=https://www.kaggle.com/) [Kaggle (https://www.kaggle.com/)](https://www.kaggle.com/)*: top 5K, tech*
 1. ![](https://www.google.com/s2/favicons?domain=https://codepen.io/) [Codepen (https://codepen.io/)](https://codepen.io/)*: top 5K, coding, in*
 1. ![](https://www.google.com/s2/favicons?domain=https://www.rajce.idnes.cz/) [Rajce.net (https://www.rajce.idnes.cz/)](https://www.rajce.idnes.cz/)*: top 5K, cz*
@@ -300,7 +300,7 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://forums.digitalspy.com/) [Digitalspy (https://forums.digitalspy.com/)](https://forums.digitalspy.com/)*: top 10K, forum, gb, us*
 1. ![](https://www.google.com/s2/favicons?domain=https://www.bibsonomy.org) [Bibsonomy (https://www.bibsonomy.org)](https://www.bibsonomy.org)*: top 10K, in*
 1. ![](https://www.google.com/s2/favicons?domain=https://slashdot.org) [Slashdot (https://slashdot.org)](https://slashdot.org)*: top 10K, news*
-1. ![](https://www.google.com/s2/favicons?domain=https://www.netvibes.com) [Netvibes (https://www.netvibes.com)](https://www.netvibes.com)*: top 10K, de, fr, jp*
+1. ![](https://www.google.com/s2/favicons?domain=https://www.netvibes.com) [Netvibes (https://www.netvibes.com)](https://www.netvibes.com)*: top 10K, business, fr*, search is disabled
 1. ![](https://www.google.com/s2/favicons?domain=https://opensource.com/) [opensource (https://opensource.com/)](https://opensource.com/)*: top 10K, in, us*
 1. ![](https://www.google.com/s2/favicons?domain=https://discuss.elastic.co/) [Discuss.Elastic.co (https://discuss.elastic.co/)](https://discuss.elastic.co/)*: top 10K, forum, tech, us*
 1. ![](https://www.google.com/s2/favicons?domain=https://www.baby.ru/) [Baby.ru (https://www.baby.ru/)](https://www.baby.ru/)*: top 10K, ru*
@@ -2599,10 +2599,10 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://www.hozpitality.com) [hozpitality (https://www.hozpitality.com)](https://www.hozpitality.com)*: top 100M*
 1. ![](https://www.google.com/s2/favicons?domain=https://kazanlashkigalab.com) [kazanlashkigalab.com (https://kazanlashkigalab.com)](https://kazanlashkigalab.com)*: top 100M, kz*
 
-Alexa.com rank data fetched at (2022-02-26 11:12:43.279519 UTC)
+Alexa.com rank data fetched at (2022-02-26 11:41:48.847517 UTC)
 ## Statistics
 
-Enabled/total sites: 2449/2595
+Enabled/total sites: 2447/2595
 
 Incomplete checks: 586/1978 (false positive risks)
 

From dcf5181e2851e02e0cfb001520d3ef979fb9208e Mon Sep 17 00:00:00 2001
From: Soxoj <31013580+soxoj@users.noreply.github.com>
Date: Sat, 26 Feb 2022 15:31:15 +0300
Subject: [PATCH 5/8] Fixed several false positives, improved statistics info
 (#368)

* Fixed several false positives, improved statistics info

* Updated site list and statistics
---
 maigret/resources/data.json | 42 +++++++++++++++++++++++++------------
 maigret/sites.py            |  7 +++++--
 sites.md                    | 12 +++++------
 3 files changed, 40 insertions(+), 21 deletions(-)

diff --git a/maigret/resources/data.json b/maigret/resources/data.json
index 6ce0b6a..6d521b1 100644
--- a/maigret/resources/data.json
+++ b/maigret/resources/data.json
@@ -5203,11 +5203,14 @@
             ],
             "checkType": "message",
             "presenceStrs": [
-                "userStatsTitle"
+                "<meta content=\"name=Profile"
+            ],
+            "absenceStrs": [
+                "<title>Foursquare "
             ],
             "alexaRank": 3413,
-            "urlMain": "https://ru.foursquare.com/",
-            "url": "https://ru.foursquare.com/{username}",
+            "urlMain": "https://foursquare.com/",
+            "url": "https://foursquare.com/{username}",
             "usernameClaimed": "adam",
             "usernameUnclaimed": "noonewouldeverusethis7"
         },
@@ -6310,7 +6313,10 @@
             ],
             "checkType": "message",
             "absenceStrs": [
-                "Page not found."
+                "Page not found"
+            ],
+            "presenseStrs": [
+                "title=\"Gumroad\""
             ],
             "alexaRank": 4728,
             "urlMain": "https://www.gumroad.com/",
@@ -8857,7 +8863,10 @@
             ],
             "checkType": "message",
             "absenceStrs": [
-                "\u0417\u0434\u0435\u0441\u044c \u043f\u043e\u043a\u0430 \u043d\u0438\u0447\u0435\u0433\u043e \u043d\u0435\u0442"
+                "\u041f\u043e \u0412\u0430\u0448\u0435\u043c\u0443 \u0437\u0430\u043f\u0440\u043e\u0441\u0443 \u043d\u0438\u0447\u0435\u0433\u043e \u043d\u0435 \u043d\u0430\u0439\u0434\u0435\u043d\u043e"
+            ],
+            "presenseStrs": [
+                "<span>\u041b\u044e\u0434\u0438</span>"
             ],
             "alexaRank": 6409,
             "urlMain": "https://mirtesen.ru",
@@ -10166,10 +10175,7 @@
             "tags": [
                 "ru"
             ],
-            "checkType": "message",
-            "absenceStrs": [
-                "404 - Not Found"
-            ],
+            "checkType": "status_code",
             "alexaRank": 25200,
             "urlMain": "https://overclockers.ru",
             "url": "https://overclockers.ru/cpubase/user/{username}",
@@ -10714,7 +10720,11 @@
             "checkType": "message",
             "absenceStrs": [
                 "Hmm, it seems that you've come across an invalid username",
-                "404 Not Found"
+                "404 Not Found",
+                "Member Not Found"
+            ],
+            "presenseStrs": [
+                "profile on Planet Minecraft to see their public Minecraft community activity"
             ],
             "alexaRank": 9050,
             "urlMain": "https://www.planetminecraft.com",
@@ -12851,7 +12861,13 @@
             "tags": [
                 "music"
             ],
-            "checkType": "status_code",
+            "checkType": "message",
+            "presenseStrs": [
+                "Profile: "
+            ],
+            "absenceStrs": [
+                "Smule | Page Not Found (404)"
+            ],
             "alexaRank": 11742,
             "urlMain": "https://www.smule.com/",
             "url": "https://www.smule.com/{username}",
@@ -13117,7 +13133,7 @@
                 "us"
             ],
             "headers": {
-                "authorization": "Bearer BQC-v69M-AcXsPLrSktz0Era-J2P1SXWB42HLKRHnCNpj00jLEbbbDFpIFo1UhBKrHrL7FqLQd-X4MIuhFo"
+                "authorization": "Bearer BQBFTijjpshGAhX7n9-sO46wb8zJIkhu6TT3Ss7b-0V1dw_jXZhcff1agUpqRgbhznOG8pSIRlHtJAtd2TU"
             },
             "errors": {
                 "Spotify is currently not available in your country.": "Access denied in your country, use proxy/vpn"
@@ -14973,7 +14989,7 @@
                 "video"
             ],
             "headers": {
-                "Authorization": "jwt eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJleHAiOjE2NDExNzg4NjAsInVzZXJfaWQiOm51bGwsImFwcF9pZCI6NTg0NzksInNjb3BlcyI6InB1YmxpYyIsInRlYW1fdXNlcl9pZCI6bnVsbH0.9rznMue0JmX9SAPuWQDIYR-mmsozFq5PoKUvlvElpkQ"
+                "Authorization": "jwt eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJleHAiOjE2NDU4Nzg1NDAsInVzZXJfaWQiOm51bGwsImFwcF9pZCI6NTg0NzksInNjb3BlcyI6InB1YmxpYyIsInRlYW1fdXNlcl9pZCI6bnVsbH0.Bs6VBcKPsl-5dqoThdAImBIex1mas1UcyG2pSnIYqYk"
             },
             "activation": {
                 "url": "https://vimeo.com/_rv/viewer",
diff --git a/maigret/sites.py b/maigret/sites.py
index 9ea540d..402eb29 100644
--- a/maigret/sites.py
+++ b/maigret/sites.py
@@ -450,8 +450,11 @@ class MaigretDatabase:
             for tag in filter(lambda x: not is_country_tag(x), site.tags):
                 tags[tag] = tags.get(tag, 0) + 1
 
-        output += f"Enabled/total sites: {total_count - disabled_count}/{total_count}\n\n"
-        output += f"Incomplete checks: {message_checks_one_factor}/{message_checks} (false positive risks)\n\n"
+        enabled_perc = round(100*(total_count-disabled_count)/total_count, 2)
+        output += f"Enabled/total sites: {total_count - disabled_count}/{total_count} = {enabled_perc}%\n\n"
+
+        checks_perc = round(100*message_checks_one_factor/message_checks, 2)
+        output += f"Incomplete checks: {message_checks_one_factor}/{message_checks} = {checks_perc}% (false positive risks)\n\n"
 
         top_urls_count = 20
         output += f"Top {top_urls_count} profile URLs:\n"
diff --git a/sites.md b/sites.md
index f035818..90f9c46 100644
--- a/sites.md
+++ b/sites.md
@@ -249,7 +249,7 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://forum.xda-developers.com) [XDA (https://forum.xda-developers.com)](https://forum.xda-developers.com)*: top 5K, apps, forum*, search is disabled
 1. ![](https://www.google.com/s2/favicons?domain=https://i.thechive.com/) [Thechive (https://i.thechive.com/)](https://i.thechive.com/)*: top 5K, us*
 1. ![](https://www.google.com/s2/favicons?domain=https://999.md) [999.md (https://999.md)](https://999.md)*: top 5K, freelance, md, shopping*
-1. ![](https://www.google.com/s2/favicons?domain=https://ru.foursquare.com/) [Foursquare (https://ru.foursquare.com/)](https://ru.foursquare.com/)*: top 5K, geosocial, in*
+1. ![](https://www.google.com/s2/favicons?domain=https://foursquare.com/) [Foursquare (https://foursquare.com/)](https://foursquare.com/)*: top 5K, geosocial, in*
 1. ![](https://www.google.com/s2/favicons?domain=https://4pda.ru/) [4pda (https://4pda.ru/)](https://4pda.ru/)*: top 5K, ru*
 1. ![](https://www.google.com/s2/favicons?domain=https://www.weforum.org) [Weforum (https://www.weforum.org)](https://www.weforum.org)*: top 5K, forum, us*
 1. ![](https://www.google.com/s2/favicons?domain=http://www.techspot.com/community/) [techspot.com (http://www.techspot.com/community/)](http://www.techspot.com/community/)*: top 5K, forum, us*
@@ -2599,12 +2599,12 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://www.hozpitality.com) [hozpitality (https://www.hozpitality.com)](https://www.hozpitality.com)*: top 100M*
 1. ![](https://www.google.com/s2/favicons?domain=https://kazanlashkigalab.com) [kazanlashkigalab.com (https://kazanlashkigalab.com)](https://kazanlashkigalab.com)*: top 100M, kz*
 
-Alexa.com rank data fetched at (2022-02-26 11:41:48.847517 UTC)
+Alexa.com rank data fetched at (2022-02-26 12:19:53.127789 UTC)
 ## Statistics
 
-Enabled/total sites: 2447/2595
+Enabled/total sites: 2447/2595 = 94.3%
 
-Incomplete checks: 586/1978 (false positive risks)
+Incomplete checks: 582/1978 = 29.42% (false positive risks)
 
 Top 20 profile URLs:
 - (796)	`{urlMain}/index/8-0-{username} (uCoz)`
@@ -2634,9 +2634,9 @@ Top 20 tags:
 - (40)	`NO_TAGS` (non-standard)
 - (24)	`coding`
 - (23)	`photo`
-- (19)	`news`
+- (18)	`news`
 - (18)	`blog`
-- (18)	`music`
+- (17)	`music`
 - (15)	`tech`
 - (13)	`freelance`
 - (12)	`sharing`

From bc787cdf519e4c4fa417c135f059fad2230948ac Mon Sep 17 00:00:00 2001
From: Soxoj <31013580+soxoj@users.noreply.github.com>
Date: Sat, 26 Feb 2022 16:01:22 +0300
Subject: [PATCH 6/8] Fix false positives  (#370)

* Fixed several false positives, improved statistics info

* Disabled some sites, fixed fp percent count method

* Updated site list and statistics
---
 maigret/resources/data.json | 32 +++++++++++++++++++-------------
 maigret/sites.py            |  2 +-
 sites.md                    | 30 +++++++++++++++---------------
 3 files changed, 35 insertions(+), 29 deletions(-)

diff --git a/maigret/resources/data.json b/maigret/resources/data.json
index 6d521b1..124c439 100644
--- a/maigret/resources/data.json
+++ b/maigret/resources/data.json
@@ -1880,7 +1880,7 @@
             ],
             "checkType": "message",
             "absenceStrs": [
-                "\u041f\u043e\u043b\u044c\u0437\u043e\u0432\u0430\u0442\u0435\u043b\u044f \u0441 \u0442\u0430\u043a\u0438\u043c \u0438\u043c\u0435\u043d\u0435\u043c \u043d\u0435 \u0441\u0443\u0449\u0435\u0441\u0442\u0432\u0443\u0435\u0442!"
+                "\u041e\u0448\u0438\u0431\u043a\u0430 / \u041f\u043e\u043b\u044c\u0437\u043e\u0432\u0430\u0442\u0435\u043b\u044f \u0441 \u0442\u0430\u043a\u0438\u043c \u0438\u043c\u0435\u043d\u0435\u043c \u043d\u0435 \u0441\u0443\u0449\u0435\u0441\u0442\u0432\u0443\u0435\u0442"
             ],
             "alexaRank": 2303903,
             "urlMain": "https://bgforum.ru",
@@ -3598,6 +3598,7 @@
             "tags": [
                 "ru"
             ],
+            "disabled": true,
             "checkType": "status_code",
             "urlMain": "https://dinsk.su",
             "url": "https://dinsk.su/user/{username}",
@@ -4571,6 +4572,7 @@
             "tags": [
                 "ru"
             ],
+            "disabled": true,
             "checkType": "status_code",
             "alexaRank": 1225740,
             "urlMain": "https://favera.ru",
@@ -8546,6 +8548,7 @@
             "tags": [
                 "forum"
             ],
+            "disabled": true,
             "checkType": "message",
             "absenceStrs": [
                 "The specified member cannot be found"
@@ -9082,12 +9085,9 @@
         },
         "Movescount": {
             "tags": [
-                "es",
-                "in",
-                "pk",
-                "ru",
-                "us"
+                "maps"
             ],
+            "disabled": true,
             "checkType": "message",
             "absenceStrs": [
                 "error=4&amp;"
@@ -12891,16 +12891,19 @@
         },
         "Snooth": {
             "tags": [
-                "in"
+                "news"
             ],
             "checkType": "message",
             "absenceStrs": [
-                "<title>Profiles on Snooth</title>"
+                "<title>Page not found"
+            ],
+            "presenseStrs": [
+                "content=\"https://www.snooth.com/author/"
             ],
             "alexaRank": 4088489,
             "urlMain": "https://www.snooth.com/",
-            "url": "https://www.snooth.com/profiles/{username}/",
-            "usernameClaimed": "GregT",
+            "url": "https://www.snooth.com/author/{username}/",
+            "usernameClaimed": "joshua",
             "usernameUnclaimed": "noonewouldeverusethis7"
         },
         "SocialLibremOne": {
@@ -14989,7 +14992,7 @@
                 "video"
             ],
             "headers": {
-                "Authorization": "jwt eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJleHAiOjE2NDU4Nzg1NDAsInVzZXJfaWQiOm51bGwsImFwcF9pZCI6NTg0NzksInNjb3BlcyI6InB1YmxpYyIsInRlYW1fdXNlcl9pZCI6bnVsbH0.Bs6VBcKPsl-5dqoThdAImBIex1mas1UcyG2pSnIYqYk"
+                "Authorization": "jwt eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJleHAiOjE2NDU4Nzk3NDAsInVzZXJfaWQiOm51bGwsImFwcF9pZCI6NTg0NzksInNjb3BlcyI6InB1YmxpYyIsInRlYW1fdXNlcl9pZCI6bnVsbH0.T8E8Vrx0sO-9WP4RdZGNQZw2EB1hYTIXbIguXIZbfNQ"
             },
             "activation": {
                 "url": "https://vimeo.com/_rv/viewer",
@@ -24833,7 +24836,7 @@
             "usernameClaimed": "alex",
             "usernameUnclaimed": "noonewouldeverusethis7"
         },
-        "hashnode.com": {
+        "hashnode": {
             "tags": [
                 "in"
             ],
@@ -24846,7 +24849,7 @@
                 " name="
             ],
             "absenceStrs": [
-                "></title><meta name="
+                "We can\u2019t find the page you\u2019re looking for!"
             ],
             "urlMain": "https://hashnode.com",
             "url": "https://hashnode.com/@{username}",
@@ -26947,6 +26950,9 @@
         },
         "forum.rastrnet.ru": {
             "urlMain": "http://forum.rastrnet.ru",
+            "errors": {
+                "\u0418\u0437\u0432\u0438\u043d\u0438\u0442\u0435, \u043f\u0440\u043e\u0432\u043e\u0434\u044f\u0442\u0441\u044f \u0442\u0435\u0445\u043d\u0438\u0447\u0435\u0441\u043a\u0438\u0435 \u0440\u0430\u0431\u043e\u0442\u044b.": "Site error"
+            },
             "engine": "vBulletin",
             "usernameClaimed": "alex",
             "usernameUnclaimed": "noonewouldeverusethis7",
diff --git a/maigret/sites.py b/maigret/sites.py
index 402eb29..c073fe7 100644
--- a/maigret/sites.py
+++ b/maigret/sites.py
@@ -438,7 +438,7 @@ class MaigretDatabase:
             url_type = site.get_url_template()
             urls[url_type] = urls.get(url_type, 0) + 1
 
-            if site.check_type == 'message':
+            if site.check_type == 'message' and not site.disabled:
                 message_checks += 1
                 if site.absence_strs and site.presense_strs:
                     continue
diff --git a/sites.md b/sites.md
index 90f9c46..4e470c1 100644
--- a/sites.md
+++ b/sites.md
@@ -812,7 +812,7 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://bbs.huami.com) [bbs.huami.com (https://bbs.huami.com)](https://bbs.huami.com)*: top 10M, cn, in, ir, ru, us*, search is disabled
 1. ![](https://www.google.com/s2/favicons?domain=https://amazfitwatchfaces.com) [AmazfitWatchFaces (https://amazfitwatchfaces.com)](https://amazfitwatchfaces.com)*: top 10M, ae, es, forum, gr, id, ir, ru*
 1. ![](https://www.google.com/s2/favicons?domain=https://moikrug.ru/) [MoiKrug (https://moikrug.ru/)](https://moikrug.ru/)*: top 10M, career, us*
-1. ![](https://www.google.com/s2/favicons?domain=http://www.movescount.com) [Movescount (http://www.movescount.com)](http://www.movescount.com)*: top 10M, es, in, pk, ru, us*
+1. ![](https://www.google.com/s2/favicons?domain=http://www.movescount.com) [Movescount (http://www.movescount.com)](http://www.movescount.com)*: top 10M, maps*, search is disabled
 1. ![](https://www.google.com/s2/favicons?domain=https://tamtam.chat/) [TamTam (https://tamtam.chat/)](https://tamtam.chat/)*: top 10M, ru*
 1. ![](https://www.google.com/s2/favicons?domain=https://forum.velomania.ru/) [Velomania (https://forum.velomania.ru/)](https://forum.velomania.ru/)*: top 10M, forum, ru*
 1. ![](https://www.google.com/s2/favicons?domain=https://forum.itvdn.com) [ITVDN Forum (https://forum.itvdn.com)](https://forum.itvdn.com)*: top 10M, forum, ru, ua*
@@ -1199,7 +1199,7 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://mastodon.xyz/) [mastodon.xyz (https://mastodon.xyz/)](https://mastodon.xyz/)*: top 10M, th*
 1. ![](https://www.google.com/s2/favicons?domain=https://www.gays.com) [Gays (https://www.gays.com)](https://www.gays.com)*: top 10M, in*, search is disabled
 1. ![](https://www.google.com/s2/favicons?domain=http://transit-club.com) [transit-club.com (http://transit-club.com)](http://transit-club.com)*: top 10M, ru*
-1. ![](https://www.google.com/s2/favicons?domain=https://favera.ru) [Favera (https://favera.ru)](https://favera.ru)*: top 10M, ru*
+1. ![](https://www.google.com/s2/favicons?domain=https://favera.ru) [Favera (https://favera.ru)](https://favera.ru)*: top 10M, ru*, search is disabled
 1. ![](https://www.google.com/s2/favicons?domain=https://soylentnews.org) [soylentnews (https://soylentnews.org)](https://soylentnews.org)*: top 10M, us*
 1. ![](https://www.google.com/s2/favicons?domain=http://chan4chan.com/) [Chan4chan (http://chan4chan.com/)](http://chan4chan.com/)*: top 10M, hu*
 1. ![](https://www.google.com/s2/favicons?domain=http://the-mainboard.com/index.php) [the-mainboard.com (http://the-mainboard.com/index.php)](http://the-mainboard.com/index.php)*: top 10M, forum, us*
@@ -1390,7 +1390,7 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://uaksu.forum24.ru/) [Uaksu (https://uaksu.forum24.ru/)](https://uaksu.forum24.ru/)*: top 10M, forum, ru, ua*
 1. ![](https://www.google.com/s2/favicons?domain=http://0-3.ru) [0-3.RU (http://0-3.ru)](http://0-3.ru)*: top 10M, forum, ru*
 1. ![](https://www.google.com/s2/favicons?domain=http://www.forumsi.org) [Forumsi (http://www.forumsi.org)](http://www.forumsi.org)*: top 10M, forum, ru*
-1. ![](https://www.google.com/s2/favicons?domain=https://www.snooth.com/) [Snooth (https://www.snooth.com/)](https://www.snooth.com/)*: top 10M, in*
+1. ![](https://www.google.com/s2/favicons?domain=https://www.snooth.com/) [Snooth (https://www.snooth.com/)](https://www.snooth.com/)*: top 10M, news*
 1. ![](https://www.google.com/s2/favicons?domain=http://soft-deniz.ucoz.ru) [soft-deniz.ucoz.ru (http://soft-deniz.ucoz.ru)](http://soft-deniz.ucoz.ru)*: top 10M*
 1. ![](https://www.google.com/s2/favicons?domain=http://oih.at.ua) [oih.at.ua (http://oih.at.ua)](http://oih.at.ua)*: top 10M, ua*
 1. ![](https://www.google.com/s2/favicons?domain=http://gorodanapa.ru/) [Gorodanapa (http://gorodanapa.ru/)](http://gorodanapa.ru/)*: top 10M, ru*, search is disabled
@@ -1559,7 +1559,7 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://cyberclock.cc) [Cyberclock (https://cyberclock.cc)](https://cyberclock.cc)*: top 100M, ru*
 1. ![](https://www.google.com/s2/favicons?domain=http://www.cydak.ru) [Cydak (http://www.cydak.ru)](http://www.cydak.ru)*: top 100M, ru*
 1. ![](https://www.google.com/s2/favicons?domain=https://www.designspiration.net/) [Designspiration (https://www.designspiration.net/)](https://www.designspiration.net/)*: top 100M*
-1. ![](https://www.google.com/s2/favicons?domain=https://dinsk.su) [Dinsk (https://dinsk.su)](https://dinsk.su)*: top 100M, ru*
+1. ![](https://www.google.com/s2/favicons?domain=https://dinsk.su) [Dinsk (https://dinsk.su)](https://dinsk.su)*: top 100M, ru*, search is disabled
 1. ![](https://www.google.com/s2/favicons?domain=https://forum.djangoproject.co) [Djangoproject.co (https://forum.djangoproject.co)](https://forum.djangoproject.co)*: top 100M, coding, forum*
 1. ![](https://www.google.com/s2/favicons?domain=https://www.dublikat.shop) [Dublikat (https://www.dublikat.shop)](https://www.dublikat.shop)*: top 100M, ru*
 1. ![](https://www.google.com/s2/favicons?domain=http://eightbit.me/) [Eightbit (http://eightbit.me/)](http://eightbit.me/)*: top 100M*
@@ -1586,7 +1586,7 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://macqa.ru) [Macqa (https://macqa.ru)](https://macqa.ru)*: top 100M, ru*
 1. ![](https://www.google.com/s2/favicons?domain=https://magiimir.com) [Magiimir (https://magiimir.com)](https://magiimir.com)*: top 100M, ru*
 1. ![](https://www.google.com/s2/favicons?domain=https://mamochki.by/) [Mamochki (https://mamochki.by/)](https://mamochki.by/)*: top 100M, by, ru*
-1. ![](https://www.google.com/s2/favicons?domain=https://mastersofcrypto.com) [Mastersofcrypto (https://mastersofcrypto.com)](https://mastersofcrypto.com)*: top 100M, forum*
+1. ![](https://www.google.com/s2/favicons?domain=https://mastersofcrypto.com) [Mastersofcrypto (https://mastersofcrypto.com)](https://mastersofcrypto.com)*: top 100M, forum*, search is disabled
 1. ![](https://www.google.com/s2/favicons?domain=http:/mixlr.com/) [Mixlr (http:/mixlr.com/)](http:/mixlr.com/)*: top 100M, gb*
 1. ![](https://www.google.com/s2/favicons?domain=https://www.munzee.com/) [Munzee (https://www.munzee.com/)](https://www.munzee.com/)*: top 100M, gb*
 1. ![](https://www.google.com/s2/favicons?domain=http://murmansk-life.ru) [MurmanskLife (http://murmansk-life.ru)](http://murmansk-life.ru)*: top 100M, ru*
@@ -2313,7 +2313,7 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=http://mednolit.ru) [mednolit.ru (http://mednolit.ru)](http://mednolit.ru)*: top 100M, ru*
 1. ![](https://www.google.com/s2/favicons?domain=http://mikele-loconte.ru) [mikele-loconte.ru (http://mikele-loconte.ru)](http://mikele-loconte.ru)*: top 100M*
 1. ![](https://www.google.com/s2/favicons?domain=http://mkuniverse.ru) [mkuniverse.ru (http://mkuniverse.ru)](http://mkuniverse.ru)*: top 100M*
-1. ![](https://www.google.com/s2/favicons?domain=https://hashnode.com) [hashnode.com (https://hashnode.com)](https://hashnode.com)*: top 100M, in*
+1. ![](https://www.google.com/s2/favicons?domain=https://hashnode.com) [hashnode (https://hashnode.com)](https://hashnode.com)*: top 100M, in*
 1. ![](https://www.google.com/s2/favicons?domain=https://www.change.org) [www.change.org (https://www.change.org)](https://www.change.org)*: top 100M, us*
 1. ![](https://www.google.com/s2/favicons?domain=https://www.ifunny.co) [www.ifunny.co (https://www.ifunny.co)](https://www.ifunny.co)*: top 100M, us*
 1. ![](https://www.google.com/s2/favicons?domain=https://localcryptosapi.com) [LocalCryptos (https://localcryptosapi.com)](https://localcryptosapi.com)*: top 100M*
@@ -2599,12 +2599,12 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://www.hozpitality.com) [hozpitality (https://www.hozpitality.com)](https://www.hozpitality.com)*: top 100M*
 1. ![](https://www.google.com/s2/favicons?domain=https://kazanlashkigalab.com) [kazanlashkigalab.com (https://kazanlashkigalab.com)](https://kazanlashkigalab.com)*: top 100M, kz*
 
-Alexa.com rank data fetched at (2022-02-26 12:19:53.127789 UTC)
+Alexa.com rank data fetched at (2022-02-26 12:55:54.605333 UTC)
 ## Statistics
 
-Enabled/total sites: 2447/2595 = 94.3%
+Enabled/total sites: 2443/2595 = 94.14%
 
-Incomplete checks: 582/1978 = 29.42% (false positive risks)
+Incomplete checks: 525/1853 = 28.33% (false positive risks)
 
 Top 20 profile URLs:
 - (796)	`{urlMain}/index/8-0-{username} (uCoz)`
@@ -2625,18 +2625,18 @@ Top 20 profile URLs:
 - (18)	`/forum/members/?username={username}`
 - (18)	`/forum/search.php?keywords=&terms=all&author={username}`
 - (17)	`/search.php?keywords=&terms=all&author={username}`
-- (15)	`/author/{username}`
+- (16)	`/author/{username}`
 - (14)	`/profile.php?mode=viewprofile&u={username}`
 
 Top 20 tags:
-- (255)	`forum`
+- (271)	`forum`
+- (79)	`NO_TAGS` (non-standard)
 - (50)	`gaming`
-- (40)	`NO_TAGS` (non-standard)
+- (24)	`photo`
 - (24)	`coding`
-- (23)	`photo`
 - (18)	`news`
 - (18)	`blog`
-- (17)	`music`
+- (18)	`music`
 - (15)	`tech`
 - (13)	`freelance`
 - (12)	`sharing`
@@ -2645,7 +2645,7 @@ Top 20 tags:
 - (10)	`dating`
 - (10)	`art`
 - (9)	`hobby`
-- (8)	`movies`
+- (9)	`movies`
 - (7)	`sport`
 - (7)	`hacking`
 - (5)	`stock`

From 8a53a3854304a3ea097fc4d5d9cad10502c6b1ad Mon Sep 17 00:00:00 2001
From: Soxoj <31013580+soxoj@users.noreply.github.com>
Date: Sat, 26 Feb 2022 16:43:40 +0300
Subject: [PATCH 7/8] Fixed the rest of false positives for now (#371)

* Fixed the rest of false positives for now

* Fixed tag

* Updated site list and statistics
---
 maigret/maigret.py          |  2 +-
 maigret/resources/data.json | 35 ++++++++++++++++++++++++-----------
 sites.md                    | 18 +++++++++---------
 utils/update_site_data.py   |  2 +-
 4 files changed, 35 insertions(+), 22 deletions(-)

diff --git a/maigret/maigret.py b/maigret/maigret.py
index 224b33c..17700d4 100755
--- a/maigret/maigret.py
+++ b/maigret/maigret.py
@@ -48,7 +48,7 @@ def notify_about_errors(search_results: QueryResultWrapper, query_notify):
     for e in errs:
         if not errors.is_important(e):
             continue
-        text = f'Too many errors of type "{e["err"]}" ({e["perc"]}%)'
+        text = f'Too many errors of type "{e["err"]}" ({round(e["perc"],2)}%)'
         solution = errors.solution_of(e['err'])
         if solution:
             text = '. '.join([text, solution.capitalize()])
diff --git a/maigret/resources/data.json b/maigret/resources/data.json
index 124c439..dd58ebf 100644
--- a/maigret/resources/data.json
+++ b/maigret/resources/data.json
@@ -1462,6 +1462,7 @@
                 "forum",
                 "ru"
             ],
+            "disabled": true,
             "checkType": "message",
             "absenceStrs": [
                 "\u041f\u043e\u043b\u044c\u0437\u043e\u0432\u0430\u0442\u0435\u043b\u044c \u043d\u0435 \u043d\u0430\u0439\u0434\u0435\u043d"
@@ -18145,6 +18146,7 @@
             "tags": [
                 "ru"
             ],
+            "disabled": true,
             "checkType": "status_code",
             "alexaRank": 6054365,
             "urlMain": "http://linuxmint.info",
@@ -24487,23 +24489,35 @@
         },
         "thoughts.com": {
             "tags": [
-                "in"
+                "blog"
+            ],
+            "checkType": "message",
+            "absenceStrs": [
+                "<title>Start a Blog"
+            ],
+            "presenseStrs": [
+                "&#8211; Thoughts.com</title>"
             ],
-            "engine": "engine404get",
             "urlMain": "http://thoughts.com",
-            "url": "http://thoughts.com/profile/{username}",
+            "url": "http://thoughts.com/members/{username}",
             "usernameUnclaimed": "noonewouldeverusethis7",
-            "usernameClaimed": "red"
+            "usernameClaimed": "alicia12"
         },
         "hackernoon.com": {
             "tags": [
-                "in",
-                "us"
+                "us",
+                "news"
+            ],
+            "checkType": "message",
+            "absenceStrs": [
+                "<title>HackerNoon"
+            ],
+            "presenseStrs": [
+                " | HackerNoon</title>"
             ],
-            "engine": "engine404message",
             "urlMain": "https://hackernoon.com",
             "url": "https://hackernoon.com/u/{username}",
-            "usernameUnclaimed": "noonewouldeverusethis7",
+            "usernameUnclaimed": "noonewouldeverusethis71",
             "usernameClaimed": "god"
         },
         "Intigriti": {
@@ -28479,10 +28493,9 @@
         },
         "photoshop-kopona.com": {
             "absenceStrs": [
-                "<title>noonewouldeverusethis7 &raquo; \u0420\u0435\u0441\u0443\u0440\u0441\u044b \u0434\u043b\u044f \u0424\u043e\u0442\u043e\u0448\u043e\u043f\u0430</title>"
+                "<div id='dle-content'></div></div></main></div></div><footer class=\"footer\">"
             ],
             "presenseStrs": [
-                "offline",
                 "uspusertitle"
             ],
             "url": "https://photoshop-kopona.com/ru/user/{username}/",
@@ -28645,7 +28658,7 @@
         },
         "coder.social": {
             "absenceStrs": [
-                "Not Found in Githubhelp"
+                "<title>Coder Social Home</title>"
             ],
             "presenseStrs": [
                 "nofollow"
diff --git a/sites.md b/sites.md
index 4e470c1..720f489 100644
--- a/sites.md
+++ b/sites.md
@@ -1442,7 +1442,7 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=http://rodobozhie.ru) [rodobozhie.ru (http://rodobozhie.ru)](http://rodobozhie.ru)*: top 10M, ru*
 1. ![](https://www.google.com/s2/favicons?domain=http://satwarez.ru) [satwarez.ru (http://satwarez.ru)](http://satwarez.ru)*: top 10M*
 1. ![](https://www.google.com/s2/favicons?domain=http://nf-club.ru) [nf-club.ru (http://nf-club.ru)](http://nf-club.ru)*: top 10M*
-1. ![](https://www.google.com/s2/favicons?domain=http://linuxmint.info) [linuxmint.info (http://linuxmint.info)](http://linuxmint.info)*: top 10M, ru*
+1. ![](https://www.google.com/s2/favicons?domain=http://linuxmint.info) [linuxmint.info (http://linuxmint.info)](http://linuxmint.info)*: top 10M, ru*, search is disabled
 1. ![](https://www.google.com/s2/favicons?domain=http://kiabongo.info) [kiabongo.info (http://kiabongo.info)](http://kiabongo.info)*: top 10M*
 1. ![](https://www.google.com/s2/favicons?domain=http://koshtoris.at.ua) [koshtoris.at.ua (http://koshtoris.at.ua)](http://koshtoris.at.ua)*: top 10M*
 1. ![](https://www.google.com/s2/favicons?domain=http://xn--90anbhklk.xn--p1ai) [xn--90anbhklk.xn--p1ai (http://xn--90anbhklk.xn--p1ai)](http://xn--90anbhklk.xn--p1ai)*: top 10M*
@@ -1469,7 +1469,7 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=http://wolga24.at.ua) [wolga24.at.ua (http://wolga24.at.ua)](http://wolga24.at.ua)*: top 10M*
 1. ![](https://www.google.com/s2/favicons?domain=http://millerovo161.ru) [millerovo161.ru (http://millerovo161.ru)](http://millerovo161.ru)*: top 10M*
 1. ![](https://www.google.com/s2/favicons?domain=http://videomuzon.ucoz.ru) [videomuzon.ucoz.ru (http://videomuzon.ucoz.ru)](http://videomuzon.ucoz.ru)*: top 10M*
-1. ![](https://www.google.com/s2/favicons?domain=https://community.autolenta.ru) [Autolenta (https://community.autolenta.ru)](https://community.autolenta.ru)*: top 10M, auto, forum, ru*
+1. ![](https://www.google.com/s2/favicons?domain=https://community.autolenta.ru) [Autolenta (https://community.autolenta.ru)](https://community.autolenta.ru)*: top 10M, auto, forum, ru*, search is disabled
 1. ![](https://www.google.com/s2/favicons?domain=http://amax-sb.ru) [amax-sb.ru (http://amax-sb.ru)](http://amax-sb.ru)*: top 10M*
 1. ![](https://www.google.com/s2/favicons?domain=http://angelgothics.ru) [Angelgothics (http://angelgothics.ru)](http://angelgothics.ru)*: top 10M, ru*
 1. ![](https://www.google.com/s2/favicons?domain=http://help-baby.org) [help-baby.org (http://help-baby.org)](http://help-baby.org)*: top 10M*
@@ -2283,8 +2283,8 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://freelance.ru) [freelance.ru (https://freelance.ru)](https://freelance.ru)*: top 100M, ru*
 1. ![](https://www.google.com/s2/favicons?domain=https://freelansim.ru) [freelansim.ru (https://freelansim.ru)](https://freelansim.ru)*: top 100M*
 1. ![](https://www.google.com/s2/favicons?domain=http://fotolog.com) [fotolog.com (http://fotolog.com)](http://fotolog.com)*: top 100M, in*
-1. ![](https://www.google.com/s2/favicons?domain=http://thoughts.com) [thoughts.com (http://thoughts.com)](http://thoughts.com)*: top 100M, in*
-1. ![](https://www.google.com/s2/favicons?domain=https://hackernoon.com) [hackernoon.com (https://hackernoon.com)](https://hackernoon.com)*: top 100M, in, us*
+1. ![](https://www.google.com/s2/favicons?domain=http://thoughts.com) [thoughts.com (http://thoughts.com)](http://thoughts.com)*: top 100M, blog*
+1. ![](https://www.google.com/s2/favicons?domain=https://hackernoon.com) [hackernoon.com (https://hackernoon.com)](https://hackernoon.com)*: top 100M, news, us*
 1. ![](https://www.google.com/s2/favicons?domain=https://intigriti.com) [Intigriti (https://intigriti.com)](https://intigriti.com)*: top 100M, hacking, in*
 1. ![](https://www.google.com/s2/favicons?domain=https://yamaya.ru) [yamaya.ru (https://yamaya.ru)](https://yamaya.ru)*: top 100M, ru*
 1. ![](https://www.google.com/s2/favicons?domain=https://www.tinkoff.ru/invest/) [Tinkoff Invest (https://www.tinkoff.ru/invest/)](https://www.tinkoff.ru/invest/)*: top 100M, ru*
@@ -2599,12 +2599,12 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://www.hozpitality.com) [hozpitality (https://www.hozpitality.com)](https://www.hozpitality.com)*: top 100M*
 1. ![](https://www.google.com/s2/favicons?domain=https://kazanlashkigalab.com) [kazanlashkigalab.com (https://kazanlashkigalab.com)](https://kazanlashkigalab.com)*: top 100M, kz*
 
-Alexa.com rank data fetched at (2022-02-26 12:55:54.605333 UTC)
+The list was updated at (2022-02-26 13:41:10.351473 UTC)
 ## Statistics
 
-Enabled/total sites: 2443/2595 = 94.14%
+Enabled/total sites: 2441/2595 = 94.07%
 
-Incomplete checks: 525/1853 = 28.33% (false positive risks)
+Incomplete checks: 523/1853 = 28.22% (false positive risks)
 
 Top 20 profile URLs:
 - (796)	`{urlMain}/index/8-0-{username} (uCoz)`
@@ -2612,7 +2612,7 @@ Top 20 profile URLs:
 - (221)	`/{username}`
 - (138)	`/user/{username}`
 - (134)	`{urlMain}{urlSubpath}/member.php?username={username} (vBulletin)`
-- (97)	`/profile/{username}`
+- (96)	`/profile/{username}`
 - (87)	`{urlMain}/u/{username}/summary (Discourse)`
 - (74)	`/users/{username}`
 - (44)	`{urlMain}{urlSubpath}/search.php?author={username} (phpBB/Search)`
@@ -2621,7 +2621,7 @@ Top 20 profile URLs:
 - (36)	`/@{username}`
 - (28)	`/u/{username}`
 - (27)	`{urlMain}{urlSubpath}/memberlist.php?username={username} (phpBB)`
-- (24)	`/members/{username}`
+- (25)	`/members/{username}`
 - (18)	`/forum/members/?username={username}`
 - (18)	`/forum/search.php?keywords=&terms=all&author={username}`
 - (17)	`/search.php?keywords=&terms=all&author={username}`
diff --git a/utils/update_site_data.py b/utils/update_site_data.py
index 106444a..05934e2 100755
--- a/utils/update_site_data.py
+++ b/utils/update_site_data.py
@@ -137,7 +137,7 @@ Rank data fetched from Alexa by domains.
             site_file.write(f'1. {favicon} [{site}]({url_main})*: top {valid_rank}{tags}*{note}\n')
             db.update_site(site)
 
-        site_file.write(f'\nAlexa.com rank data fetched at ({datetime.utcnow()} UTC)\n')
+        site_file.write(f'\nThe list was updated at ({datetime.utcnow()} UTC)\n')
         db.save_to_file(args.base_file)
 
         statistics_text = db.get_db_stats(is_markdown=True)

From 364187861da1dd45ea9d11ef3207740740c72bab Mon Sep 17 00:00:00 2001
From: Soxoj <31013580+soxoj@users.noreply.github.com>
Date: Sun, 27 Feb 2022 04:44:15 +0300
Subject: [PATCH 8/8] Fix false positive and CI (#372)

* Fix false positive and CI
* Updated site list and statistics
---
 .github/workflows/python-package.yml   | 3 +--
 .github/workflows/update-site-data.yml | 3 +--
 maigret/resources/data.json            | 1 +
 sites.md                               | 6 +++---
 4 files changed, 6 insertions(+), 7 deletions(-)

diff --git a/.github/workflows/python-package.yml b/.github/workflows/python-package.yml
index 132fe8e..382283f 100644
--- a/.github/workflows/python-package.yml
+++ b/.github/workflows/python-package.yml
@@ -2,10 +2,9 @@ name: Linting and testing
 
 on:
   push:
-    branches: [ main ]
   pull_request:
     branches: [ main ]
-    types: [opened, synchronize]
+    types: [opened, synchronize, reopened]
 
 jobs:
   build:
diff --git a/.github/workflows/update-site-data.yml b/.github/workflows/update-site-data.yml
index 2d2c81e..a7b3d0e 100644
--- a/.github/workflows/update-site-data.yml
+++ b/.github/workflows/update-site-data.yml
@@ -1,10 +1,9 @@
 name: Update sites rating and statistics
 
 on:
-  push:
-    branches: [ main ]
   pull_request:
     branches: [ main ]
+    types: [opened, synchronize]
 
 jobs:
   build:
diff --git a/maigret/resources/data.json b/maigret/resources/data.json
index dd58ebf..d0bc1d3 100644
--- a/maigret/resources/data.json
+++ b/maigret/resources/data.json
@@ -1867,6 +1867,7 @@
                 "pk",
                 "us"
             ],
+            "disabled": true,
             "checkType": "status_code",
             "alexaRank": 522588,
             "urlMain": "https://bethepro.com",
diff --git a/sites.md b/sites.md
index 720f489..68d3afc 100644
--- a/sites.md
+++ b/sites.md
@@ -1056,7 +1056,7 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://forum.swiftbook.ru) [swiftbook (https://forum.swiftbook.ru)](https://forum.swiftbook.ru)*: top 10M, forum, ru*
 1. ![](https://www.google.com/s2/favicons?domain=http://ladies.zp.ua) [Ladies (http://ladies.zp.ua)](http://ladies.zp.ua)*: top 10M, forum, ua*
 1. ![](https://www.google.com/s2/favicons?domain=https://archlinux.org.ru) [Archlinux (https://archlinux.org.ru)](https://archlinux.org.ru)*: top 10M, ru*
-1. ![](https://www.google.com/s2/favicons?domain=https://bethepro.com) [Bethepro (https://bethepro.com)](https://bethepro.com)*: top 10M, pk, us*
+1. ![](https://www.google.com/s2/favicons?domain=https://bethepro.com) [Bethepro (https://bethepro.com)](https://bethepro.com)*: top 10M, pk, us*, search is disabled
 1. ![](https://www.google.com/s2/favicons?domain=https://forum.mapillary.com) [Mapillary Forum (https://forum.mapillary.com)](https://forum.mapillary.com)*: top 10M, forum, us*
 1. ![](https://www.google.com/s2/favicons?domain=https://www.firearmstalk.com) [Firearmstalk (https://www.firearmstalk.com)](https://www.firearmstalk.com)*: top 10M, us*
 1. ![](https://www.google.com/s2/favicons?domain=https://www.houserepairtalk.com) [Houserepairtalk (https://www.houserepairtalk.com)](https://www.houserepairtalk.com)*: top 10M, forum, in, us*
@@ -2599,10 +2599,10 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=https://www.hozpitality.com) [hozpitality (https://www.hozpitality.com)](https://www.hozpitality.com)*: top 100M*
 1. ![](https://www.google.com/s2/favicons?domain=https://kazanlashkigalab.com) [kazanlashkigalab.com (https://kazanlashkigalab.com)](https://kazanlashkigalab.com)*: top 100M, kz*
 
-The list was updated at (2022-02-26 13:41:10.351473 UTC)
+The list was updated at (2022-02-27 01:43:04.619091 UTC)
 ## Statistics
 
-Enabled/total sites: 2441/2595 = 94.07%
+Enabled/total sites: 2440/2595 = 94.03%
 
 Incomplete checks: 523/1853 = 28.22% (false positive risks)