HOTFIX: linksearchtotal_collect query performance

reworked query template to go from this:
```
SELECT COUNT(*) FROM externallinks
WHERE el_to_domain_index LIKE '%com.example%'
AND el_to_path LIKE '%%'
```

to this:
```
SELECT COUNT(*) FROM externallinks
WHERE el_to_domain_index LIKE 'http://com.example%'

SELECT COUNT(*) FROM externallinks
WHERE el_to_domain_index LIKE 'https://com.example%'
```

and from this:
```
SELECT COUNT(*) FROM externallinks
WHERE el_to_domain_index LIKE '%com.example%'
AND el_to_path LIKE '%/examples/%'
```

to this:
```
SELECT COUNT(*) FROM externallinks
WHERE el_to_domain_index LIKE 'http://com.example%'
AND el_to_path LIKE '/examples/%'

SELECT COUNT(*) FROM externallinks
WHERE el_to_domain_index LIKE 'https://com.example%'
AND el_to_path LIKE '/examples/%'
```

which is significantly faster since it avoids wildcards at the start of
the condition.

See: https://phabricator.wikimedia.org/T403209
Bug: T403209

Author: jsnshrmn

extlinks/links/management/commands/linksearchtotal_collect.py

@@ -16,7 +16,9 @@ class Command(BaseCommand):
     help = "Updates link totals from externallinks table"
 
     def _handle(self, *args, **options):
+        protocols = ["http", "https"]
 
+        print("reading wiki-list")
         with open(os.path.join(BASE_DIR, "wiki-list.csv"), "r") as wiki_list:
             csv_reader = csv.reader(wiki_list)
             wiki_list_data = []
@@ -27,6 +29,7 @@ def _handle(self, *args, **options):
 
         total_links_dictionary = {}
         for i, language in enumerate(wiki_list_data):
+            print("connecting to db {}".format(language))
             db = MySQLdb.connect(
                 host="{lang}wiki.analytics.db.svc.wikimedia.cloud".format(
                     lang=language
@@ -39,6 +42,7 @@ def _handle(self, *args, **options):
             cur = db.cursor()
 
             for urlpattern in all_urlpatterns:
+                print("searching url pattern {}".format(urlpattern))
                 # For the first language, initialise tracking
                 if i == 0:
                     total_links_dictionary[urlpattern.pk] = 0
@@ -50,27 +54,30 @@ def _handle(self, *args, **options):
                     url = urlpattern.url
 
                 url_parsed = urlparse(url)
-                url_path = url_parsed.path
                 url_host = url_parsed.hostname
-
-                query = f"""
-                   SELECT COUNT(*) FROM externallinks
-                    WHERE el_to_domain_index LIKE '%{reverse_host(url_host)}%'
-                    AND el_to_path LIKE '%{url_path}%'
-                    """
-
-                cur.execute(
-                    query
-                )
-
-                this_num_urls = cur.fetchone()[0]
-
-                total_links_dictionary[urlpattern.pk] += this_num_urls
+                url_path = url_parsed.path
+                for protocol in protocols:
+                    query = f"""
+                        SELECT COUNT(*) FROM externallinks
+                        WHERE el_to_domain_index LIKE '{protocol}://{reverse_host(url_host)}%'
+                        """
+                    if len(url_path) > 0:
+                        cond = f"""AND el_to_path LIKE '{url_path}%'
+                        """
+                        query += cond
+                    print("executing query {}".format(query))
+                    cur.execute(query)
+
+                    this_num_urls = cur.fetchone()[0]
+
+                    print("found {}".format(this_num_urls))
+                    total_links_dictionary[urlpattern.pk] += this_num_urls
 
         for urlpattern_pk, total_count in total_links_dictionary.items():
             linksearch_object = LinkSearchTotal(
                 url=URLPattern.objects.get(pk=urlpattern_pk), total=total_count
             )
+            print("saving linksearch_object {}".format(linksearch_object))
             linksearch_object.save()
 
         close_old_connections()