From 8b195f8bd6372f0215ebc72d409e2c3e767bcdf0 Mon Sep 17 00:00:00 2001
From: Pierre-Yves Chibon <pingou@pingoured.fr>
Date: Wed, 16 May 2018 15:09:52 +0200
Subject: [PATCH] Add SAR script for pagure

Signed-off-by: Pierre-Yves Chibon <pingou@pingoured.fr>
---
 .../host_vars/pagure01.fedoraproject.org      |   5 +
 roles/pagure/frontend/files/pagure_sar.py     | 145 ++++++++++++++++++
 roles/pagure/frontend/tasks/main.yml          |   7 +
 3 files changed, 157 insertions(+)
 create mode 100644 roles/pagure/frontend/files/pagure_sar.py

diff --git a/inventory/host_vars/pagure01.fedoraproject.org b/inventory/host_vars/pagure01.fedoraproject.org
index 828de32abf..0e8efc3cba 100644
--- a/inventory/host_vars/pagure01.fedoraproject.org
+++ b/inventory/host_vars/pagure01.fedoraproject.org
@@ -26,3 +26,8 @@ effective_cache_size: "512MB"
 
 ssh_hostnames:
 - pagure.io
+
+# GDPR SAR variables - datanommer/datagrepper
+sar_script: /usr/local/bin/pagure_sar.py
+sar_script_user: git
+sar_output_file: pagure.json
diff --git a/roles/pagure/frontend/files/pagure_sar.py b/roles/pagure/frontend/files/pagure_sar.py
new file mode 100644
index 0000000000..f04c49f3f7
--- /dev/null
+++ b/roles/pagure/frontend/files/pagure_sar.py
@@ -0,0 +1,145 @@
+#!/usr/bin/python
+
+from __future__ import unicode_literals, print_function
+
+import json
+import os
+import sys
+
+import sqlalchemy
+
+import pagure.config
+import pagure.lib
+from pagure.lib import model
+
+
+if 'PAGURE_CONFIG' not in os.environ \
+        and os.path.exists('/etc/pagure/pagure.cfg'):
+    os.environ['PAGURE_CONFIG'] = '/etc/pagure/pagure.cfg'
+
+
+_config = pagure.config.reload_config()
+session = pagure.lib.create_session(_config['DB_URL'])
+
+
+def get_issue_users(session, user_id):
+    ''' Return all pagure.lib.model.Issue related to the usernames provided
+    '''
+    query1 = session.query(
+        model.Issue.uid
+    ).filter(
+        sqlalchemy.or_(
+            model.Issue.assignee_id == user_id,
+            model.Issue.user_id == user_id
+        )
+    )
+    query2 = session.query(
+        model.Issue.uid
+    ).filter(
+        model.Issue.uid == model.IssueComment.issue_uid
+    ).filter(
+        model.IssueComment.user_id == user_id
+    )
+
+    query = session.query(
+        model.Issue
+    ).filter(
+        sqlalchemy.or_(
+            model.Issue.uid.in_(query1.subquery()),
+            model.Issue.uid.in_(query2.subquery())
+        )
+    ).order_by(
+        model.Issue.date_created
+    )
+
+    return query.all()
+
+
+def get_pr_users(session, user_id):
+    ''' Return all pagure.lib.model.PullRequest related to the usernames provided
+    '''
+    query1 = session.query(
+        model.PullRequest.uid
+    ).filter(
+        sqlalchemy.or_(
+            model.PullRequest.assignee_id == user_id,
+            model.PullRequest.user_id == user_id
+        )
+    )
+    query2 = session.query(
+        model.PullRequest.uid
+    ).filter(
+        model.PullRequest.uid == model.PullRequestComment.pull_request_uid
+    ).filter(
+        model.PullRequestComment.user_id == user_id
+    )
+
+    query = session.query(
+        model.PullRequest
+    ).filter(
+        sqlalchemy.or_(
+            model.PullRequest.uid.in_(query1.subquery()),
+            model.PullRequest.uid.in_(query2.subquery())
+        )
+    ).order_by(
+        model.PullRequest.date_created
+    )
+
+    return query.all()
+
+
+def main():
+    ''' Prints out all the pagure project and comment related to the username
+    specified in the SAR_USERNAME environment variable or the email
+    specified in the SAR_EMAIL environment variable..
+    '''
+
+    username = os.getenv('SAR_USERNAME')
+    email = os.getenv('SAR_EMAIL')
+
+    users = []
+    if username:
+        users.append(pagure.lib.search_user(session, username=username))
+    if email:
+        user_email = pagure.lib.search_user(session, email=email)
+        if user_email not in users:
+            users.apennd(user_email)
+
+    output = {}
+
+    for user in users:
+        temp = {}
+        temp['user_info'] = user.to_json(public=False)
+
+        projects = pagure.lib.search_projects(session, user.username)
+        projects = [
+            project.to_json()
+            for project in projects
+        ]
+        temp['projects'] = projects
+
+        issues = get_issue_users(session, user.id)
+        issues = [
+            issue.to_json()
+            for issue in issues
+        ]
+        temp['issues'] = issues
+
+        prs = get_pr_users(session, user.id)
+        prs = [
+            pr.to_json()
+            for pr in prs
+        ]
+        temp['pull_requests'] = prs
+
+        output[user.username] = temp
+
+    session.remove()
+
+    print(json.dumps(
+        output, sort_keys=True, indent=4, separators=(',', ': ')
+    ).encode('utf-8'))
+
+
+if __name__ == '__main__':
+    sys.exit(main())
diff --git a/roles/pagure/frontend/tasks/main.yml b/roles/pagure/frontend/tasks/main.yml
index 787b7adb58..50d1892759 100644
--- a/roles/pagure/frontend/tasks/main.yml
+++ b/roles/pagure/frontend/tasks/main.yml
@@ -402,3 +402,10 @@
   tags:
   - config
   - apache
+
+- name: Add SAR script for pagure
+  copy: src=pagure_sar.py dest=/usr/local/bin/pagure_sar.py owner=git mode=0700
+  tags:
+  - SAR
+  - GDPR
+  - pagure