Remove rank from all of the queries and update materialized view

8Knot/cache_manager/db_init.py

@@ -204,7 +204,6 @@ def _create_application_tables() -> None:
                 repo_id int,
                 login text,
                 action text,
-                rank int,
                 cntrb_company text,
                 email_list text
             )
@@ -220,8 +219,7 @@ def _create_application_tables() -> None:
                 cntrb_id text,
                 created_at text,
                 login text,
-                action text,
-                rank int
+                action text
             )
             """
         )

8Knot/pages/codebase/visualizations/cntrb_file_heatmap.py

@@ -432,7 +432,7 @@ def cntrb_to_last_activity(df_actions: pd.DataFrame, df_dynamic_directory: pd.Da
     # drop unneccessary columns not needed after preprocessing steps
     df_actions = df_actions.reset_index()
     df_actions.drop(
-        ["index", "repo_id", "repo_name", "login", "Action", "rank"],
+        ["index", "repo_id", "repo_name", "login", "Action"],
         axis=1,
         inplace=True,
     )

8Knot/pages/codebase/visualizations/reviewer_file_heatmap.py

@@ -431,7 +431,7 @@ def cntrb_to_last_activity(df_actions: pd.DataFrame, df_dynamic_directory: pd.Da
     # drop unneccessary columns not needed after preprocessing steps
     df_actions = df_actions.reset_index()
     df_actions.drop(
-        ["index", "repo_id", "repo_name", "login", "Action", "rank"],
+        ["index", "repo_id", "repo_name", "login", "Action"],
         axis=1,
         inplace=True,
     )

8Knot/queries/affiliation_query.py

@@ -40,7 +40,6 @@ def affiliation_query(self, repos):
                         c.repo_id,
                         c.login,
                         c.action,
-                        c.rank,
                         con.cntrb_company,
                         string_agg(ca.alias_email, ' , ' order by ca.alias_email) as email_list
                     FROM
@@ -53,7 +52,7 @@ def affiliation_query(self, repos):
                         c.repo_id in %s
                         and timezone('utc', c.created_at) < now() -- created_at is a timestamptz value
                         -- don't need to check non-null for created_at because it's non-null by definition.
-                    GROUP BY c.cntrb_id, c.created_at, c.repo_id, c.login, c.action, c.rank, con.cntrb_company
+                    GROUP BY c.cntrb_id, c.created_at, c.repo_id, c.login, c.action, con.cntrb_company
                     ORDER BY
                         c.created_at
                     """

8Knot/queries/contributors_query.py

@@ -44,8 +44,7 @@ def contributors_query(self, repos):
                         left(ca.cntrb_id::text, 15) as cntrb_id, -- first 15 characters of the uuid
                         timezone('utc', ca.created_at) AS created_at,
                         ca.login,
-                        ca.action,
-                        ca.rank
+                        ca.action
                     FROM
                         explorer_contributor_actions ca
                     WHERE

docs/materialized_views/explorer_contributor_actions.sql

@@ -1,99 +1,143 @@
 /* This is the SQL query that populates the explorer_contributor_actions materialized view*/
 
-SELECT a.id AS cntrb_id,
+WITH commits_deduped AS (
+    SELECT DISTINCT ON (c.cmt_commit_hash)
+        c.cmt_ght_author_id        AS cntrb_id,
+        c.cmt_author_timestamp     AS created_at,
+        c.repo_id,
+        'commit'::text             AS action,
+        r.repo_name
+    FROM augur_data.commits c
+    JOIN augur_data.repo r ON r.repo_id = c.repo_id
+    ORDER BY c.cmt_commit_hash, c.cmt_author_timestamp DESC
+)
+SELECT
+    a.cntrb_id,
     a.created_at,
     a.repo_id,
     a.action,
-    repo.repo_name,
-    a.login,
-    row_number() OVER (PARTITION BY a.id, a.repo_id ORDER BY a.created_at DESC) AS rank
-    FROM ( SELECT commits.cmt_ght_author_id AS id,
-            commits.cmt_author_timestamp AS created_at,
-            commits.repo_id,
-            'commit'::text AS action,
-            contributors.cntrb_login AS login
-           FROM (augur_data.commits
-             LEFT JOIN augur_data.contributors ON (((contributors.cntrb_id)::text = (commits.cmt_ght_author_id)::text)))
-          GROUP BY commits.cmt_commit_hash, commits.cmt_ght_author_id, commits.repo_id, commits.cmt_author_timestamp, 'commit'::text, contributors.cntrb_login
-        UNION ALL
-         SELECT issues.reporter_id AS id,
-            issues.created_at,
-            issues.repo_id,
-            'issue_opened'::text AS action,
-            contributors.cntrb_login AS login
-           FROM (augur_data.issues
-             LEFT JOIN augur_data.contributors ON ((contributors.cntrb_id = issues.reporter_id)))
-          WHERE (issues.pull_request IS NULL)
-        UNION ALL
-         SELECT pull_request_events.cntrb_id AS id,
-            pull_request_events.created_at,
-            pull_requests.repo_id,
-            'pull_request_closed'::text AS action,
-            contributors.cntrb_login AS login
-           FROM augur_data.pull_requests,
-            (augur_data.pull_request_events
-             LEFT JOIN augur_data.contributors ON ((contributors.cntrb_id = pull_request_events.cntrb_id)))
-          WHERE ((pull_requests.pull_request_id = pull_request_events.pull_request_id) AND (pull_requests.pr_merged_at IS NULL) AND ((pull_request_events.action)::text = 'closed'::text))
-        UNION ALL
-         SELECT pull_request_events.cntrb_id AS id,
-            pull_request_events.created_at,
-            pull_requests.repo_id,
-            'pull_request_merged'::text AS action,
-            contributors.cntrb_login AS login
-           FROM augur_data.pull_requests,
-            (augur_data.pull_request_events
-             LEFT JOIN augur_data.contributors ON ((contributors.cntrb_id = pull_request_events.cntrb_id)))
-          WHERE ((pull_requests.pull_request_id = pull_request_events.pull_request_id) AND ((pull_request_events.action)::text = 'merged'::text))
-        UNION ALL
-         SELECT issue_events.cntrb_id AS id,
-            issue_events.created_at,
-            issues.repo_id,
-            'issue_closed'::text AS action,
-            contributors.cntrb_login AS login
-           FROM augur_data.issues,
-            (augur_data.issue_events
-             LEFT JOIN augur_data.contributors ON ((contributors.cntrb_id = issue_events.cntrb_id)))
-          WHERE ((issues.issue_id = issue_events.issue_id) AND (issues.pull_request IS NULL) AND ((issue_events.action)::text = 'closed'::text))
-        UNION ALL
-         SELECT pull_request_reviews.cntrb_id AS id,
-            pull_request_reviews.pr_review_submitted_at AS created_at,
-            pull_requests.repo_id,
-            ('pull_request_review_'::text || (pull_request_reviews.pr_review_state)::text) AS action,
-            contributors.cntrb_login AS login
-           FROM augur_data.pull_requests,
-            (augur_data.pull_request_reviews
-             LEFT JOIN augur_data.contributors ON ((contributors.cntrb_id = pull_request_reviews.cntrb_id)))
-          WHERE (pull_requests.pull_request_id = pull_request_reviews.pull_request_id)
-        UNION ALL
-         SELECT pull_requests.pr_augur_contributor_id AS id,
-            pull_requests.pr_created_at AS created_at,
-            pull_requests.repo_id,
-            'pull_request_open'::text AS action,
-            contributors.cntrb_login AS login
-           FROM (augur_data.pull_requests
-             LEFT JOIN augur_data.contributors ON ((pull_requests.pr_augur_contributor_id = contributors.cntrb_id)))
-        UNION ALL
-         SELECT message.cntrb_id AS id,
-            message.msg_timestamp AS created_at,
-            pull_requests.repo_id,
-            'pull_request_comment'::text AS action,
-            contributors.cntrb_login AS login
-           FROM augur_data.pull_requests,
-            augur_data.pull_request_message_ref,
-            (augur_data.message
-             LEFT JOIN augur_data.contributors ON ((contributors.cntrb_id = message.cntrb_id)))
-          WHERE ((pull_request_message_ref.pull_request_id = pull_requests.pull_request_id) AND (pull_request_message_ref.msg_id = message.msg_id))
-        UNION ALL
-         SELECT message.cntrb_id AS id,
-            message.msg_timestamp AS created_at,
-            issues.repo_id,
-            'issue_comment'::text AS action,
-            contributors.cntrb_login AS login
-           FROM augur_data.issues,
-            augur_data.issue_message_ref,
-            (augur_data.message
-             LEFT JOIN augur_data.contributors ON ((contributors.cntrb_id = message.cntrb_id)))
-          WHERE ((issue_message_ref.msg_id = message.msg_id) AND (issues.issue_id = issue_message_ref.issue_id) AND (issues.closed_at <> message.msg_timestamp))) a,
-    augur_data.repo
-  WHERE (a.repo_id = repo.repo_id)
-  ORDER BY a.created_at DESC
+    a.repo_name,
+    co.cntrb_login AS login
+FROM (
+    SELECT * FROM commits_deduped
+
+    UNION ALL
+
+    -- issues opened
+    SELECT
+        i.reporter_id                  AS cntrb_id,
+        i.created_at,
+        i.repo_id,
+        'issue_opened'::text           AS action,
+        r.repo_name
+    FROM augur_data.issues i
+    JOIN augur_data.repo r ON r.repo_id = i.repo_id
+    WHERE i.pull_request IS NULL
+
+    UNION ALL
+
+    -- pull requests closed (not merged)
+    SELECT
+        pre.cntrb_id,
+        pre.created_at,
+        pr.repo_id,
+        'pull_request_closed'::text    AS action,
+        r.repo_name
+    FROM augur_data.pull_request_events pre
+    JOIN augur_data.pull_requests pr
+        ON pr.pull_request_id = pre.pull_request_id
+        AND pr.pr_merged_at IS NULL
+    JOIN augur_data.repo r ON r.repo_id = pr.repo_id
+    WHERE pre.action = 'closed'
+
+    UNION ALL
+
+    -- pull requests merged
+    SELECT
+        pre.cntrb_id,
+        pre.created_at,
+        pr.repo_id,
+        'pull_request_merged'::text    AS action,
+        r.repo_name
+    FROM augur_data.pull_request_events pre
+    JOIN augur_data.pull_requests pr
+        ON pr.pull_request_id = pre.pull_request_id
+    JOIN augur_data.repo r ON r.repo_id = pr.repo_id
+    WHERE pre.action = 'merged'
+
+    UNION ALL
+
+    -- issues closed
+    SELECT
+        ie.cntrb_id,
+        ie.created_at,
+        i.repo_id,
+        'issue_closed'::text           AS action,
+        r.repo_name
+    FROM augur_data.issue_events ie
+    JOIN augur_data.issues i
+        ON i.issue_id = ie.issue_id
+        AND i.pull_request IS NULL
+    JOIN augur_data.repo r ON r.repo_id = i.repo_id
+    WHERE ie.action = 'closed'
+
+    UNION ALL
+
+    -- pull request reviews
+    SELECT
+        prr.cntrb_id,
+        prr.pr_review_submitted_at     AS created_at,
+        pr.repo_id,
+        ('pull_request_review_' || prr.pr_review_state::text) AS action,
+        r.repo_name
+    FROM augur_data.pull_request_reviews prr
+    JOIN augur_data.pull_requests pr
+        ON pr.pull_request_id = prr.pull_request_id
+    JOIN augur_data.repo r ON r.repo_id = pr.repo_id
+
+    UNION ALL
+
+    -- pull requests opened
+    SELECT
+        pr.pr_augur_contributor_id     AS cntrb_id,
+        pr.pr_created_at               AS created_at,
+        pr.repo_id,
+        'pull_request_open'::text      AS action,
+        r.repo_name
+    FROM augur_data.pull_requests pr
+    JOIN augur_data.repo r ON r.repo_id = pr.repo_id
+
+    UNION ALL
+
+    -- pull request comments
+    SELECT
+        m.cntrb_id,
+        m.msg_timestamp                AS created_at,
+        pr.repo_id,
+        'pull_request_comment'::text   AS action,
+        r.repo_name
+    FROM augur_data.pull_request_message_ref prmr
+    JOIN augur_data.pull_requests pr
+        ON pr.pull_request_id = prmr.pull_request_id
+    JOIN augur_data.repo r ON r.repo_id = pr.repo_id
+    JOIN augur_data.message m ON m.msg_id = prmr.msg_id
+
+    UNION ALL
+
+    -- issue comments
+    SELECT
+        m.cntrb_id,
+        m.msg_timestamp                AS created_at,
+        i.repo_id,
+        'issue_comment'::text          AS action,
+        r.repo_name
+    FROM augur_data.issue_message_ref imr
+    JOIN augur_data.message m ON m.msg_id = imr.msg_id
+    JOIN augur_data.issues i
+        ON i.issue_id = imr.issue_id
+        AND i.pull_request IS NULL
+        AND i.closed_at <> m.msg_timestamp
+    JOIN augur_data.repo r ON r.repo_id = i.repo_id
+) a
+LEFT JOIN augur_data.contributors co ON co.cntrb_id = a.cntrb_id
+ORDER BY a.created_at DESC