From 571b47badb3e33633ae7d2ab709920b03c8058ca Mon Sep 17 00:00:00 2001
From: elsapet <elizabeth@bearer.sh>
Date: Mon, 3 Jun 2024 09:56:01 +0200
Subject: [PATCH] feat(python): add Google BigQuery third party rule (CWE-201)

---
 rules/python/third_parties/bigquery.yml       | 63 +++++++++++++++++++
 tests/python/third_parties/bigquery/test.js   | 20 ++++++
 .../third_parties/bigquery/testdata/main.py   | 20 ++++++
 3 files changed, 103 insertions(+)
 create mode 100644 rules/python/third_parties/bigquery.yml
 create mode 100644 tests/python/third_parties/bigquery/test.js
 create mode 100644 tests/python/third_parties/bigquery/testdata/main.py

diff --git a/rules/python/third_parties/bigquery.yml b/rules/python/third_parties/bigquery.yml
new file mode 100644
index 00000000..0d669ad7
--- /dev/null
+++ b/rules/python/third_parties/bigquery.yml
@@ -0,0 +1,63 @@
+imports:
+  - python_shared_lang_datatype
+  - python_shared_lang_instance
+  - python_shared_lang_import3
+patterns:
+  - pattern: |
+      $<CLIENT>.$<METHOD>($<...>$<DATA_TYPE>$<...>)
+    filters:
+      - variable: CLIENT
+        detection: python_third_parties_bigquery_client
+        scope: cursor
+      - variable: METHOD
+        values:
+          - insert_rows
+          - insert_rows_json
+      - variable: DATA_TYPE
+        detection: python_shared_lang_datatype
+        scope: result
+auxiliary:
+  - id: python_third_parties_bigquery_client
+    patterns:
+      - pattern: $<CLIENT>
+        filters:
+          - variable: CLIENT
+            detection: python_shared_lang_instance
+            scope: cursor
+            filters:
+              - variable: CLASS
+                detection: python_shared_lang_import3
+                scope: cursor
+                filters:
+                  - variable: MODULE1
+                    values: [google]
+                  - variable: MODULE2
+                    values: [cloud]
+                  - variable: MODULE3
+                    values: [bigquery]
+                  - variable: NAME
+                    values: [Client]
+languages:
+  - python
+severity: medium
+skip_data_types:
+  - Unique Identifier
+metadata:
+  description: Leakage of sensitive data to BigQuery
+  remediation_message: |
+    ## Description
+
+    Leaking sensitive data to third-party data tools like BigQuery is a common cause of data leaks and can lead to data breaches.
+
+    ## Remediations
+
+    - **Do** ensure all sensitive data is removed when sending data to third-party services like BigQuery.
+
+    ## References
+    - [Python Client for Google BigQuery](https://github.com/googleapis/python-bigquery)
+    - [BigQuery docs](https://cloud.google.com/python/docs/reference/bigquery/latest)
+  cwe_id:
+    - 201
+  associated_recipe: Google Cloud BigQuery
+  id: python_third_parties_bigquery
+  documentation_url: https://docs.bearer.com/reference/rules/python_third_parties_bigquery
diff --git a/tests/python/third_parties/bigquery/test.js b/tests/python/third_parties/bigquery/test.js
new file mode 100644
index 00000000..84168897
--- /dev/null
+++ b/tests/python/third_parties/bigquery/test.js
@@ -0,0 +1,20 @@
+const {
+  createNewInvoker,
+  getEnvironment,
+} = require("../../../helper.js")
+const { ruleId, ruleFile, testBase } = getEnvironment(__dirname)
+
+describe(ruleId, () => {
+  const invoke = createNewInvoker(ruleId, ruleFile, testBase)
+
+  test("bigquery", () => {
+    const testCase = "main.py"
+
+    const results = invoke(testCase)
+
+    expect(results).toEqual({
+      Missing: [],
+      Extra: []
+    })
+  })
+})
\ No newline at end of file
diff --git a/tests/python/third_parties/bigquery/testdata/main.py b/tests/python/third_parties/bigquery/testdata/main.py
new file mode 100644
index 00000000..eaa14926
--- /dev/null
+++ b/tests/python/third_parties/bigquery/testdata/main.py
@@ -0,0 +1,20 @@
+from google.cloud import bigquery
+
+def insert_user():
+  client = bigquery.Client()
+  dataset_ref = client.dataset("my_dataset")
+  table_ref = dataset_ref.table("my_table")
+  
+  schema = [
+    bigquery.SchemaField("id", "INTEGER", mode="REQUIRED"),
+    bigquery.SchemaField("username", "INTEGER", mode="REQUIRED"),
+    bigquery.SchemaField("email", "INTEGER", mode="REQUIRED"),
+  ]
+  
+  rows = [
+    { "id": user.id, "username": user.username, "email": user.email}
+  ]
+  
+  # bearer:expected python_third_parties_bigquery
+  errors = client.insert_rows(table_ref, rows, selected_fields=schema)
+  print("Insert errors: ", errors)