queries: add org stars API (#1600)

pingcap · Sep 6, 2023 · b38c2b1 · b38c2b1
1 parent 4387190
commit b38c2b1
Show file tree

Hide file tree

Showing 10 changed files with 379 additions and 0 deletions.
diff --git a/configs/queries/orgs/stars/locations/params.json b/configs/queries/orgs/stars/locations/params.json
@@ -0,0 +1,30 @@
+{
+  "cacheHours": 1,
+  "engine": "liquid",
+  "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": "n",
+      "type": "integer",
+      "default": 10
+    }
+  ]
+}
diff --git a/configs/queries/orgs/stars/locations/template.sql b/configs/queries/orgs/stars/locations/template.sql
@@ -0,0 +1,28 @@
+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 %}
+)
+SELECT
+    IF(gu.country_code IN ('', 'N/A', 'UND'), 'UND', gu.country_code) AS country_code,
+    COUNT(*) AS stars
+FROM github_events ge
+JOIN github_users gu ON ge.actor_login = gu.login
+WHERE
+    ge.repo_id IN (SELECT repo_id FROM repos)
+    AND ge.type = 'WatchEvent'
+    AND ge.action = 'started'
+    {% 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 gu.country_code
+ORDER BY stars DESC
+LIMIT {{ n }}
diff --git a/configs/queries/orgs/stars/organizations/params.json b/configs/queries/orgs/stars/organizations/params.json
@@ -0,0 +1,30 @@
+{
+  "cacheHours": 1,
+  "engine": "liquid",
+  "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": "n",
+      "type": "integer",
+      "default": 10
+    }
+  ]
+}
diff --git a/configs/queries/orgs/stars/organizations/template.sql b/configs/queries/orgs/stars/organizations/template.sql
@@ -0,0 +1,31 @@
+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 %}
+)
+SELECT
+    TRIM(REPLACE(REPLACE(REPLACE(REPLACE(REPLACE(REPLACE(REPLACE(REPLACE(gu.organization, ',', ''), '-', ''), '@', ''), 'www.', ''), 'inc', ''), '.com', ''), '.cn', ''), '.', '')) AS organization_name,
+    CASE WHEN
+    COUNT(DISTINCT actor_login) AS stars
+FROM github_events ge
+JOIN github_users gu ON ge.actor_login = gu.login
+WHERE
+    ge.repo_id IN (SELECT repo_id FROM repos)
+    AND ge.type = 'WatchEvent'
+    AND ge.action = 'started'
+    {% 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 %}
+    AND LENGTH(gu.organization) != 0
+    AND gu.organization NOT IN ('', '-', 'none', 'no', 'home', 'n/a', 'null', 'unknown')
+GROUP BY organization_name
+ORDER BY stars DESC
+LIMIT {{ n }}
diff --git a/configs/queries/orgs/stars/top-repos/params.json b/configs/queries/orgs/stars/top-repos/params.json
@@ -0,0 +1,25 @@
+{
+  "cacheHours": 1,
+  "engine": "liquid",
+  "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"
+    }
+  ]
+}
diff --git a/configs/queries/orgs/stars/top-repos/template.sql b/configs/queries/orgs/stars/top-repos/template.sql
@@ -0,0 +1,40 @@
+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 %}
+), repos_with_stars AS (
+    SELECT
+        repo_id,
+        COUNT(*) AS stars
+    FROM github_events ge
+    WHERE
+        ge.repo_id IN (SELECT repo_id FROM repos)
+        AND ge.type = 'WatchEvent'
+        AND ge.action = 'started'
+        {% case period %}
+            {% when 'past_7_days' %} AND created_at > (NOW() - INTERVAL 7 DAY)
+            {% when 'past_28_days' %} AND created_at > (NOW() - INTERVAL 28 DAY)
+            {% when 'past_90_days' %} AND created_at > (NOW() - INTERVAL 90 DAY)
+            {% when 'past_12_months' %} AND created_at > (NOW() - INTERVAL 12 MONTH)
+        {% endcase %}
+    GROUP BY repo_id
+    ORDER BY stars DESC
+    LIMIT 10
+)
+SELECT
+    gr.repo_id,
+    gr.repo_name,
+    rws.stars
+FROM repos_with_stars rws
+JOIN github_repos gr USING (repo_id)
+ORDER BY stars DESC
+LIMIT 10
+
+
+
+
diff --git a/configs/queries/orgs/stars/total/params.json b/configs/queries/orgs/stars/total/params.json
@@ -0,0 +1,25 @@
+{
+  "cacheHours": 1,
+  "engine": "liquid",
+  "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"
+    }
+  ]
+}
diff --git a/configs/queries/orgs/stars/total/template.sql b/configs/queries/orgs/stars/total/template.sql
@@ -0,0 +1,43 @@
+WITH repos AS (
+    SELECT gr.repo_id
+    FROM github_repos gr
+    WHERE
+        gr.owner_id = {{ownerId}}
+        {% if repoIds.size > 0 %}
+        AND gr.repo_id IN ({{ repoIds | join: ',' }})
+        {% endif %}
+), stars_per_period AS (
+    SELECT
+        -- Divide periods.
+        {% case period %}
+            {% when 'past_7_days' %} TIMESTAMPDIFF(DAY, created_at, NOW()) DIV 7
+            {% when 'past_28_days' %} TIMESTAMPDIFF(DAY, created_at, NOW()) DIV 28
+            {% when 'past_90_days' %} TIMESTAMPDIFF(DAY, created_at, NOW()) DIV 90
+            {% when 'past_12_months' %}  TIMESTAMPDIFF(MONTH, created_at, NOW()) DIV 12
+        {% endcase %} AS period,
+        COUNT(*) AS stars_total
+    FROM github_events ge
+    WHERE
+        ge.repo_id IN (SELECT repo_id FROM repos)
+        AND type = 'WatchEvent'
+        AND action = 'started'
+        {% case period %}
+            {% when 'past_7_days' %} AND created_at > (NOW() - INTERVAL 14 DAY)
+            {% when 'past_28_days' %} AND created_at > (NOW() - INTERVAL 56 DAY)
+            {% when 'past_90_days' %} AND created_at > (NOW() - INTERVAL 180 DAY)
+            {% when 'past_12_months' %} AND created_at > (NOW() - INTERVAL 24 MONTH)
+        {% endcase %}
+    GROUP BY period
+), current_period_stars AS (
+    SELECT stars_total FROM stars_per_period WHERE period = 0
+), past_period_stars AS (
+    SELECT stars_total FROM stars_per_period WHERE period = 1
+)
+SELECT
+    cps.stars_total AS current_period_total,
+    pps.stars_total AS past_period_total,
+    (cps.stars_total - pps.stars_total) / pps.stars_total AS growth_percentage
+FROM
+    current_period_stars cps,
+    past_period_stars pps
+;
diff --git a/configs/queries/orgs/stars/trends/params.json b/configs/queries/orgs/stars/trends/params.json
@@ -0,0 +1,25 @@
+{
+  "cacheHours": 1,
+  "engine": "liquid",
+  "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"
+    }
+  ]
+}
diff --git a/configs/queries/orgs/stars/trends/template.sql b/configs/queries/orgs/stars/trends/template.sql
@@ -0,0 +1,102 @@
+WITH RECURSIVE seq(idx, current_period_day, past_period_day) AS (
+    SELECT
+        1 AS idx,
+        {% case period %}
+            {% when 'past_7_days', 'past_28_days', 'past_90_days' %} DATE_FORMAT(CURRENT_DATE(), '%Y-%m-%d')
+            {% when 'past_12_months' %} DATE_FORMAT(CURRENT_DATE(), '%Y-%m-01')
+        {% endcase %} AS current_period_day,
+        {% case period %}
+            {% when 'past_7_days' %} DATE_FORMAT(DATE_SUB(CURRENT_DATE(), INTERVAL 7 DAY), '%Y-%m-%d')
+            {% when 'past_28_days' %} DATE_FORMAT(DATE_SUB(CURRENT_DATE(), INTERVAL 28 DAY), '%Y-%m-%d')
+            {% when 'past_90_days' %} DATE_FORMAT(DATE_SUB(CURRENT_DATE(), INTERVAL 90 DAY), '%Y-%m-%d')
+            {% when 'past_12_months' %} DATE_FORMAT(DATE_SUB(CURRENT_DATE(), INTERVAL 12 MONTH), '%Y-%m-01')
+        {% endcase %} AS past_period_day
+    UNION ALL
+    SELECT
+        idx + 1 AS idx,
+        {% case period %}
+            {% when 'past_7_days', 'past_28_days', 'past_90_days' %} DATE_FORMAT(DATE_SUB(CURRENT_DATE(), INTERVAL idx DAY), '%Y-%m-%d')
+            {% when 'past_12_months' %} DATE_FORMAT(DATE_SUB(CURRENT_DATE(), INTERVAL idx MONTH), '%Y-%m-01')
+        {% endcase %} AS current_period_day,
+        {% case period %}
+            {% when 'past_7_days' %} DATE_FORMAT(DATE_SUB(CURRENT_DATE(), INTERVAL idx + 7 DAY), '%Y-%m-%d')
+            {% when 'past_28_days' %} DATE_FORMAT(DATE_SUB(CURRENT_DATE(), INTERVAL idx + 28 DAY), '%Y-%m-%d')
+            {% when 'past_90_days' %} DATE_FORMAT(DATE_SUB(CURRENT_DATE(), INTERVAL idx + 90 DAY), '%Y-%m-%d')
+            {% when 'past_12_months' %} DATE_FORMAT(DATE_SUB(CURRENT_DATE(), INTERVAL idx + 12 MONTH), '%Y-%m-01')
+        {% endcase %} AS past_period_day
+    FROM seq
+    WHERE
+        1 = 1
+        {% case period %}
+            {% when 'past_7_days' %} AND idx < 7
+            {% when 'past_28_days' %} AND idx < 28
+            {% when 'past_90_days' %} AND idx < 90
+            {% when 'past_12_months' %} AND idx < 12
+        {% endcase %}
+), repos AS (
+    SELECT gr.repo_id
+    FROM github_repos gr
+    WHERE
+        gr.owner_id = {{ownerId}}
+        {% if repoIds.size > 0 %}
+        AND gr.repo_id IN ({{ repoIds | join: ',' }})
+        {% endif %}
+), group_by_day AS (
+    SELECT
+        {% case period %}
+            {% when 'past_7_days' %} TIMESTAMPDIFF(DAY, day, CURRENT_DATE()) % 7 + 1
+            {% when 'past_28_days' %} TIMESTAMPDIFF(DAY, day, CURRENT_DATE()) % 28 + 1
+            {% when 'past_90_days' %} TIMESTAMPDIFF(DAY, day, CURRENT_DATE()) % 90 + 1
+            {% when 'past_12_months' %} TIMESTAMPDIFF(MONTH, day, CURRENT_DATE()) % 12 + 1
+        {% endcase %} AS idx,
+        -- Divide periods.
+        {% case period %}
+            {% when 'past_7_days' %} TIMESTAMPDIFF(DAY, day, CURRENT_DATE()) DIV 7
+            {% when 'past_28_days' %} TIMESTAMPDIFF(DAY, day, CURRENT_DATE()) DIV 28
+            {% when 'past_90_days' %} TIMESTAMPDIFF(DAY, day, CURRENT_DATE()) DIV 90
+            {% when 'past_12_months' %} TIMESTAMPDIFF(MONTH, day, CURRENT_DATE()) DIV 12
+        {% endcase %} AS period,
+        day,
+        stars
+    FROM (
+        SELECT
+            {% case period %}
+                {% when 'past_7_days', 'past_28_days', 'past_90_days' %} DATE_FORMAT(created_at, '%Y-%m-%d')
+                {% when 'past_12_months' %} DATE_FORMAT(created_at, '%Y-%m-01')
+            {% endcase %} AS day,
+            COUNT(*) AS stars
+        FROM
+            github_events ge
+        WHERE
+            repo_id IN (SELECT repo_id FROM repos)
+            AND type = 'WatchEvent'
+            AND action = 'started'
+            {% case period %}
+                {% when 'past_7_days' %} AND created_at > (CURRENT_DATE() - INTERVAL 14 DAY)
+                {% when 'past_28_days' %} AND created_at > (CURRENT_DATE() - INTERVAL 56 DAY)
+                {% when 'past_90_days' %} AND created_at > (CURRENT_DATE() - INTERVAL 180 DAY)
+                {% when 'past_12_months' %} AND created_at > (CURRENT_DATE() - INTERVAL 24 MONTH)
+            {% endcase %}
+        GROUP BY day
+        ORDER BY day
+    ) sub
+), current_period AS (
+    SELECT idx, day, stars
+    FROM group_by_day
+    WHERE period = 0
+), past_period AS (
+    SELECT idx, day, stars
+    FROM group_by_day
+    WHERE period = 1
+)
+SELECT
+    s.idx AS idx,
+    s.current_period_day AS current_period_day,
+    IFNULL(cp.stars, 0) AS current_period_day_total,
+    s.past_period_day AS past_period_day,
+    IFNULL(pp.stars, 0) AS past_period_day_total
+FROM seq s
+LEFT JOIN current_period cp ON s.idx = cp.idx
+LEFT JOIN past_period pp ON s.idx = pp.idx
+ORDER BY idx
+;