queries: add completion-rate queries (pingcap#1664)

* queries: add completion-rate queries

* add self-merged-ratio query

configs/queries/orgs/issues/closed-ratio/template.sql

@@ -104,10 +104,10 @@ WITH repos AS (
 SELECT
     cpppt.type,
     cpppt.issues AS current_period_issues,
-    ROUND(cpppt.issues / cppt.issues_total * 100, 2) AS current_period_percentage,
+    ROUND(cpppt.issues / cppt.issues_total, 2) AS current_period_percentage,
     ppppt.issues AS past_period_issues,
-    ROUND(ppppt.issues / pppt.issues_total * 100, 2) AS past_period_percentage,
-    ROUND((cpppt.issues / cppt.issues_total - ppppt.issues / pppt.issues_total) * 100, 2) AS percentage_change
+    ROUND(ppppt.issues / pppt.issues_total, 2) AS past_period_percentage,
+    ROUND((cpppt.issues / cppt.issues_total - ppppt.issues / pppt.issues_total), 2) AS percentage_change
 FROM current_period_issues_per_type cpppt
 LEFT JOIN past_period_issues_per_type ppppt ON cpppt.type = ppppt.type
 LEFT JOIN current_period_issues_total cppt ON 1 = 1

configs/queries/orgs/issues/total/template.sql

@@ -39,7 +39,7 @@ WITH repos AS (
 SELECT
     IFNULL(cpp.issues, 0) AS current_period_total,
     IFNULL(ppp.issues, 0) AS past_period_total,
-    ROUND((cpp.issues - ppp.issues) / ppp.issues * 100, 2) AS growth_percentage
+    ROUND((cpp.issues - ppp.issues) / ppp.issues, 2) AS growth_percentage
 FROM current_period_issues cpp
 LEFT JOIN past_period_issues ppp ON 1 = 1
 ;

configs/queries/orgs/participants/organizations/filled-ratio/params.json → configs/queries/orgs/participants/locations/completion-rate/params.json

@@ -23,9 +23,15 @@
       "default": "past_28_days"
     },
     {
-      "name": "n",
-      "type": "integer",
-      "default": 10
+      "name": "role",
+      "type": "string",
+      "enums": ["all", "pr_creators", "pr_commenters", "pr_reviewers", "issue_creators", "issue_commenters", "commit_authors"],
+      "default": "all"
+    },
+    {
+      "name": "excludeBots",
+      "type": "boolean",
+      "default": true
     }
   ]
 }

configs/queries/orgs/participants/locations/completion-rate/template.sql

@@ -0,0 +1,71 @@
+WITH repos AS (
+    SELECT
+        gr.repo_id, gr.repo_name
+    FROM github_repos gr
+    WHERE
+        gr.owner_id = {{ownerId}}
+        {% if repoIds.size > 0 %}
+        AND gr.repo_id IN ({{ repoIds | join: ',' }})
+        {% endif %}
+), participants_summary AS (
+    SELECT
+        CASE WHEN gu.country_code NOT IN ('', 'N/A', 'UND') THEN 1 ELSE 0 END AS is_filled,
+        COUNT(DISTINCT actor_login) AS participants
+    FROM github_events ge
+    JOIN github_users gu ON ge.actor_login = gu.login
+    WHERE
+        ge.repo_id IN (SELECT repo_id FROM repos)
+
+        {% case role %}
+        {% when 'pr_creators' %}
+        AND ge.type = 'PullRequestEvent' AND ge.action = 'opened'
+        {% when 'pr_reviewers' %}
+        AND ge.type = 'PullRequestReviewEvent' AND ge.action = 'created'
+        {% when 'issue_creators' %}
+        AND ge.type = 'IssuesEvent' AND ge.action = 'opened'
+        {% when 'commit_authors' %}
+        AND ge.type = 'PushEvent' AND ge.action = ''
+        {% when 'pr_commenters' %}
+        AND ge.type = 'IssueCommentEvent' AND ge.action = 'created'
+        AND EXISTS (
+            SELECT 1
+            FROM mv_repo_pull_requests mrpr
+            WHERE mrpr.repo_id = ge.repo_id AND mrpr.number = ge.number
+        )
+        {% when 'issue_commenters' %}
+        AND ge.type = 'IssueCommentEvent' AND ge.action = 'created'
+        AND EXISTS (
+            SELECT 1
+            FROM mv_repo_issues mri
+            WHERE mri.repo_id = ge.repo_id AND mri.number = ge.number
+        )
+        {% else %}
+        -- Events considered as participation (Exclude `WatchEvent`, which means star a repo).
+        AND ge.type IN ('IssueCommentEvent',  'DeleteEvent',  'CommitCommentEvent',  'MemberEvent',  'PushEvent',  'PublicEvent',  'ForkEvent',  'ReleaseEvent',  'PullRequestReviewEvent',  'CreateEvent',  'GollumEvent',  'PullRequestEvent',  'IssuesEvent',  'PullRequestReviewCommentEvent')
+        AND ge.action IN ('added', 'published', 'reopened', 'closed', 'created', 'opened', '')
+        {% endcase %}
+
+        {% if excludeBots %}
+        -- Exclude bot users.
+        AND LOWER(ge.actor_login) NOT LIKE '%bot%'
+        AND ge.actor_login NOT IN (SELECT login FROM blacklist_users LIMIT 255)
+        {% endif %}
+
+        {% case period %}
+            {% when 'past_7_days' %} AND ge.created_at > (NOW() - INTERVAL 7 DAY)
+            {% when 'past_28_days' %} AND ge.created_at > (NOW() - INTERVAL 28 DAY)
+            {% when 'past_90_days' %} AND ge.created_at > (NOW() - INTERVAL 90 DAY)
+            {% when 'past_12_months' %} AND ge.created_at > (NOW() - INTERVAL 12 MONTH)
+        {% endcase %}
+    GROUP BY is_filled
+), participants_total AS (
+    SELECT SUM(participants) AS total FROM participants_summary
+)
+SELECT
+    ROUND(ps.participants / pt.total, 2) AS percentage
+FROM
+    participants_summary ps,
+    participants_total pt
+WHERE
+    is_filled = 1
+LIMIT 1

configs/queries/orgs/participants/locations/template.sql

@@ -92,7 +92,7 @@ participants_total AS (
 SELECT
     ppc.country_code,
     ppc.participants,
-    ROUND(ppc.participants / pt.total * 100, 2) AS percentage
+    ROUND(ppc.participants / pt.total, 2) AS percentage
 FROM
     participants_per_country ppc,
     participants_total pt

configs/queries/orgs/participants/organizations/completion-rate/params.json

@@ -0,0 +1,37 @@
+{
+  "cacheHours": 1,
+  "engine": "liquid",
+  "cacheProvider": "NORMAL_TABLE",
+  "params": [
+    {
+      "name": "ownerId",
+      "replaces": "11855343",
+      "type": "integer"
+    },
+    {
+      "name": "repoIds",
+      "replaces": "41986369",
+      "type": "array",
+      "default": [],
+      "itemType": "integer",
+      "maxArrayLength": 50
+    },
+    {
+      "name": "period",
+      "type": "string",
+      "enums": ["past_7_days", "past_28_days", "past_90_days", "past_12_months"],
+      "default": "past_28_days"
+    },
+    {
+      "name": "role",
+      "type": "string",
+      "enums": ["all", "pr_creators", "pr_commenters", "pr_reviewers", "issue_creators", "issue_commenters", "commit_authors"],
+      "default": "all"
+    },
+    {
+      "name": "excludeBots",
+      "type": "boolean",
+      "default": true
+    }
+  ]
+}

configs/queries/orgs/participants/organizations/filled-ratio/template.sql → configs/queries/orgs/participants/organizations/completion-rate/template.sql

@@ -7,14 +7,15 @@ WITH repos AS (
         {% if repoIds.size > 0 %}
         AND gr.repo_id IN ({{ repoIds | join: ',' }})
         {% endif %}
-), stars_overview AS (
+), participants_summary AS (
     SELECT
-        SUM(IF(gu.organization_formatted IS NOT NULL AND LENGTH(gu.organization_formatted) != 0, 1, 0)) AS stars_with_org,
-        COUNT(*) AS stars_total
+        IF(gu.organization_formatted IS NOT NULL AND LENGTH(gu.organization_formatted) != 0, 1, 0) AS is_filled,
+        COUNT(DISTINCT actor_login) AS participants
     FROM github_events ge
     JOIN github_users gu ON ge.actor_login = gu.login
     WHERE
         ge.repo_id IN (SELECT repo_id FROM repos)
+
         {% case role %}
         {% when 'pr_creators' %}
         AND ge.type = 'PullRequestEvent' AND ge.action = 'opened'
@@ -28,47 +29,42 @@ WITH repos AS (
         AND ge.type = 'IssueCommentEvent' AND ge.action = 'created'
         AND EXISTS (
             SELECT 1
-            FROM github_events ge2
-            WHERE
-                ge2.type = 'PullRequestEvent'
-                AND ge2.action = 'opened'
-                AND ge2.created_at < ge.created_at
-                AND ge2.repo_id = ge.repo_id
-                AND ge2.number = ge.number
+            FROM mv_repo_pull_requests mrpr
+            WHERE mrpr.repo_id = ge.repo_id AND mrpr.number = ge.number
         )
         {% when 'issue_commenters' %}
         AND ge.type = 'IssueCommentEvent' AND ge.action = 'created'
         AND EXISTS (
             SELECT 1
-            FROM github_events ge2
-            WHERE
-                ge2.type = 'IssuesEvent'
-                AND ge2.action = 'opened'
-                AND ge2.created_at < ge.created_at
-                AND ge2.repo_id = ge.repo_id
-                AND ge2.number = ge.number
+            FROM mv_repo_issues mri
+            WHERE mri.repo_id = ge.repo_id AND mri.number = ge.number
         )
         {% else %}
         -- Events considered as participation (Exclude `WatchEvent`, which means star a repo).
         AND ge.type IN ('IssueCommentEvent',  'DeleteEvent',  'CommitCommentEvent',  'MemberEvent',  'PushEvent',  'PublicEvent',  'ForkEvent',  'ReleaseEvent',  'PullRequestReviewEvent',  'CreateEvent',  'GollumEvent',  'PullRequestEvent',  'IssuesEvent',  'PullRequestReviewCommentEvent')
         AND ge.action IN ('added', 'published', 'reopened', 'closed', 'created', 'opened', '')
         {% endcase %}
+
         {% if excludeBots %}
         -- Exclude bot users.
         AND ge.actor_login NOT LIKE '%bot%'
         {% endif %}
+
         {% case period %}
-        {% when 'past_7_days' %}
-        AND ge.created_at > (NOW() - INTERVAL 7 DAY)
-        {% when 'past_28_days' %}
-        AND ge.created_at > (NOW() - INTERVAL 28 DAY)
-        {% when 'past_90_days' %}
-        AND ge.created_at > (NOW() - INTERVAL 90 DAY)
-        {% when 'past_12_months' %}
-        AND ge.created_at > (NOW() - INTERVAL 12 MONTH)
+            {% when 'past_7_days' %} AND ge.created_at > (NOW() - INTERVAL 7 DAY)
+            {% when 'past_28_days' %} AND ge.created_at > (NOW() - INTERVAL 28 DAY)
+            {% when 'past_90_days' %} AND ge.created_at > (NOW() - INTERVAL 90 DAY)
+            {% when 'past_12_months' %} AND ge.created_at > (NOW() - INTERVAL 12 MONTH)
         {% endcase %}
+    GROUP BY is_filled
+), participants_total AS (
+    SELECT SUM(participants) AS total FROM participants_summary
 )
 SELECT
-    ROUND(so.stars_with_org / so.stars_total * 100, 2) AS percentage
+    ROUND(ps.participants / pt.total, 2) AS percentage
 FROM
-    stars_overview so
+    participants_summary ps,
+    participants_total pt
+WHERE
+    is_filled = 1
+LIMIT 1

configs/queries/orgs/participants/organizations/template.sql

@@ -97,7 +97,7 @@ participants_total AS (
 SELECT
     ppo.organization_name,
     ppo.participants,
-    ROUND(ppo.participants / pt.participants_total * 100, 2) AS percentage
+    ROUND(ppo.participants / pt.participants_total, 2) AS percentage
 FROM
     participants_per_org ppo,
     participants_total pt