From 5a930c7a2d1fb397cf1ba9f3e210b50f8688fabf Mon Sep 17 00:00:00 2001
From: Andrey Rakhmatullin <wrar@wrar.name>
Date: Mon, 13 Jan 2025 18:46:45 +0400
Subject: [PATCH] Add a non-whitespace pattern to SearchQueriesParam. (#74)

* Add a non-whitespace pattern to SearchQueriesParam.

* Fix the regex.

* Improve the regex.

* Add a validation test for serp.
---
 tests/test_serp.py                    | 28 +++++++++++++++++++++++++++
 zyte_spider_templates/spiders/serp.py |  1 +
 2 files changed, 29 insertions(+)

diff --git a/tests/test_serp.py b/tests/test_serp.py
index 4dde596..9b80cd7 100644
--- a/tests/test_serp.py
+++ b/tests/test_serp.py
@@ -1,6 +1,7 @@
 from urllib.parse import quote_plus
 
 import pytest
+from pydantic import ValidationError
 from scrapy import Request
 from scrapy_spider_metadata import get_spider_metadata
 from scrapy_zyte_api.responses import ZyteAPITextResponse
@@ -288,6 +289,7 @@ def test_metadata():
                         {"type": "null"},
                     ],
                     "description": "Input 1 search query per line (e.g. foo bar).",
+                    "pattern": r"(.|\r?\n)*\S+(.|\r?\n)*",
                     "title": "Search Queries",
                     "widget": "textarea",
                 },
@@ -764,3 +766,29 @@ def test_item_type_mappings():
 
     # Also ensure that no dict value is repeated.
     assert len(actual_keys) == len(set(ITEM_TYPE_CLASSES.values()))
+
+
+@pytest.mark.parametrize(
+    "input_data,raises",
+    [
+        ({"search_queries": "foo"}, False),
+        ({"search_queries": "foo "}, False),
+        ({"search_queries": " foo "}, False),
+        ({"search_queries": " fo o "}, False),
+        ({"search_queries": "fo o"}, False),
+        ({"search_queries": "fo\n o "}, False),
+        ({"search_queries": ["fo", " o "]}, False),
+        ({"search_queries": ["fo", "  "]}, False),
+        ({"search_queries": " "}, True),
+        ({"search_queries": ""}, True),
+        ({"search_queries": "   "}, True),
+        ({"search_queries": " \n  "}, True),
+        ({"search_queries": [" ", "  "]}, True),
+    ],
+)
+def test_query_validation(input_data, raises):
+    if raises:
+        with pytest.raises(ValidationError):
+            GoogleSearchSpider(**input_data)
+    else:
+        GoogleSearchSpider(**input_data)
diff --git a/zyte_spider_templates/spiders/serp.py b/zyte_spider_templates/spiders/serp.py
index 6ee001e..b6162a6 100644
--- a/zyte_spider_templates/spiders/serp.py
+++ b/zyte_spider_templates/spiders/serp.py
@@ -93,6 +93,7 @@ class SearchQueriesParam(BaseModel):
         description="Input 1 search query per line (e.g. foo bar).",
         json_schema_extra={
             "widget": "textarea",
+            "pattern": r"(.|\r?\n)*\S+(.|\r?\n)*",
         },
     )