Updated logic of false positive risk estimating (#475)

maigret/resources/data.json

@@ -31965,8 +31965,12 @@
         },
         "wiki.tfes.org": {
             "checkType": "message",
-            "presenseStrs": ["History"],
-            "absenceStrs": ["is not registered."],
+            "presenseStrs": [
+                "History"
+            ],
+            "absenceStrs": [
+                "is not registered."
+            ],
             "usernameClaimed": "Tom_Bishop",
             "usernameUnclaimed": "noonewouldeverusethis7",
             "url": "https://wiki.tfes.org/User:{username}"

maigret/sites.py

@@ -431,6 +431,8 @@ class MaigretDatabase:
         message_checks = 0
         message_checks_one_factor = 0
 
+        status_checks = 0
+
         for _, site in sites_dict.items():
             if site.disabled:
                 disabled_count += 1
@@ -444,17 +446,26 @@ class MaigretDatabase:
                     continue
                 message_checks_one_factor += 1
 
+            if site.check_type == 'status_code':
+                status_checks += 1
+
             if not site.tags:
                 tags["NO_TAGS"] = tags.get("NO_TAGS", 0) + 1
 
             for tag in filter(lambda x: not is_country_tag(x), site.tags):
                 tags[tag] = tags.get(tag, 0) + 1
 
-        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"
+        enabled_count = total_count-disabled_count
+        enabled_perc = round(100*enabled_count/total_count, 2)
+        output += f"Enabled/total sites: {enabled_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"
+        checks_perc = round(100*message_checks_one_factor/enabled_count, 2)
+        output += f"Incomplete message checks: {message_checks_one_factor}/{enabled_count} = {checks_perc}% (false positive risks)\n\n"
+
+        status_checks_perc = round(100*status_checks/enabled_count, 2)
+        output += f"Status code checks: {status_checks}/{enabled_count} = {status_checks_perc}% (false positive risks)\n\n"
+
+        output += f"False positive risk (total): {checks_perc+status_checks_perc}%\n\n"
 
         top_urls_count = 20
         output += f"Top {top_urls_count} profile URLs:\n"

sites.md

@@ -2968,12 +2968,16 @@ Rank data fetched from Alexa by domains.
 1. ![](https://www.google.com/s2/favicons?domain=) [zeldadungeon.net ()]()*: top 100M*
 1. ![](https://www.google.com/s2/favicons?domain=) [zoig.com ()]()*: top 100M*
 
-The list was updated at (2022-05-10 07:37:30.589483 UTC)
+The list was updated at (2022-05-10 11:50:34.852022 UTC)
 ## Statistics
 
 Enabled/total sites: 2793/2964 = 94.23%
 
-Incomplete checks: 511/1995 = 25.61% (false positive risks)
+Incomplete message checks: 511/2793 = 18.3% (false positive risks)
+
+Status code checks: 733/2793 = 26.24% (false positive risks)
+
+False positive risk (total): 44.54%
 
 Top 20 profile URLs:
 - (796)	`{urlMain}/index/8-0-{username} (uCoz)`
@@ -2998,7 +3002,7 @@ Top 20 profile URLs:
 - (17)	`/search.php?keywords=&terms=all&author={username}`
 
 Top 20 tags:
-- (295)	`NO_TAGS` (non-standard)
+- (294)	`NO_TAGS` (non-standard)
 - (271)	`forum`
 - (50)	`gaming`
 - (24)	`photo`