Fixed the rest of false positives for now (#371)

* Fixed the rest of false positives for now * Fixed tag * Updated site list and statistics
soxoj · Feb 26, 2022 · 8a53a38 · 8a53a38
1 parent bc787cd
commit 8a53a38
Show file tree

Hide file tree

Showing 4 changed files with 35 additions and 22 deletions.
diff --git 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
@@ -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
@@ -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*
@@ -2599,20 +2599,20 @@ 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)`
 - (221)	`{urlMain}{urlSubpath}/members/?username={username} (XenForo)`
 - (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
@@ -137,7 +137,7 @@ def get_readable_rank(r):
             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)