From 77f8a9312bcc8f8fb532772bfdf1ef41d9984902 Mon Sep 17 00:00:00 2001
From: Lioman <dev@lioman.de>
Date: Wed, 12 Apr 2023 13:13:01 +0200
Subject: [PATCH] remove non printable chars from titles #31

---
 pelican/plugins/search/search.py        |  7 +++++--
 tests/test_search_settings_generator.py | 23 +++++++++++++++++++++++
 2 files changed, 28 insertions(+), 2 deletions(-)

diff --git a/pelican/plugins/search/search.py b/pelican/plugins/search/search.py
index 276b6c0..81d8268 100644
--- a/pelican/plugins/search/search.py
+++ b/pelican/plugins/search/search.py
@@ -115,12 +115,15 @@ def get_input_files(
                 page.save_as if self._index_output() else page.relative_source_path
             )
             # Escape double-quotation marks in the title
-            title = striptags(page.title).replace('"', '\\"')
+            title = "".join(
+                c if c.isprintable() else " "
+                for c in striptags(page.title).replace('"', '\\"')
+            )
             input_files.append(
                 {
                     "path": page_to_index,
                     "url": f"/{page.url}",
-                    "title": f"{title}",
+                    "title": title,
                 }
             )
 
diff --git a/tests/test_search_settings_generator.py b/tests/test_search_settings_generator.py
index f2193b0..a1a9f52 100644
--- a/tests/test_search_settings_generator.py
+++ b/tests/test_search_settings_generator.py
@@ -361,3 +361,26 @@ def test_template_pages_collected(
                     "title": "",
                 },
             ]
+
+        @pytest.mark.parametrize("non_printable", ["\u00a0", "\u200B", "\uFEFF"])
+        def test_unicode_non_printable_chars_stripped(self, non_printable):
+            """Non printable chars should be all converted to a standard space"""
+            generator = SearchSettingsGenerator(
+                context={
+                    "pages": [
+                        self.PageArticleMock(f"title with nbsp{non_printable}char")
+                    ],
+                    "articles": [],
+                },
+                settings={"TEMPLATE_PAGES": []},
+                path=None,
+                theme=None,
+                output_path="output",
+            )
+            assert generator.get_input_files() == [
+                {
+                    "path": "save_as",
+                    "url": "/url",
+                    "title": "title with nbsp char",
+                }
+            ]