From af88452bb1ab0fb56e8f785bb66e17d43e8e672f Mon Sep 17 00:00:00 2001
From: harshsoni2024 <64592571+harshsoni2024@users.noreply.github.com>
Date: Tue, 20 May 2025 17:40:44 +0530
Subject: [PATCH] MINOR: presidio sample data lib fix (#21295)

---
 ingestion/setup.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/ingestion/setup.py b/ingestion/setup.py
index 96b3c41e9f3..39deb91a07f 100644
--- a/ingestion/setup.py
+++ b/ingestion/setup.py
@@ -66,6 +66,7 @@ VERSIONS = {
     "google-cloud-bigtable": "google-cloud-bigtable>=2.0.0",
     "pyathena": "pyathena~=3.0",
     "sqlalchemy-bigquery": "sqlalchemy-bigquery>=1.2.2",
+    "presidio-analyzer": "presidio-analyzer==2.2.355",
 }
 
 COMMONS = {
@@ -335,6 +336,7 @@ plugins: Dict[str, Set[str]] = {
         VERSIONS["avro"],
         VERSIONS["grpc-tools"],
         VERSIONS["sqlalchemy-bigquery"],
+        VERSIONS["presidio-analyzer"],
     },
     "sap-hana": {"hdbcli", "sqlalchemy-hana"},
     "sas": {},
@@ -350,8 +352,9 @@ plugins: Dict[str, Set[str]] = {
         VERSIONS["spacy"],
         VERSIONS["pandas"],
         VERSIONS["numpy"],
-        "presidio-analyzer==2.2.355",
+        VERSIONS["presidio-analyzer"],
     },
+    "presidio-analyzer": {VERSIONS["presidio-analyzer"]},
 }
 
 dev = {
@@ -457,6 +460,7 @@ playwright_dependencies = {
     *plugins["airflow"],
     *plugins["datalake-s3"],
     *plugins["dbt"],
+    *plugins["presidio-analyzer"],
     *e2e_test
     # Add other plugins as needed for Playwright tests
 }