fix #17 Unicode characters in slugs (#80)

Co-authored-by: Loic Teixeira <hello@lta.me>

CHANGELOG.md

@@ -11,6 +11,10 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
 
 - Compatibility with Python 3.11 and Wagtail 4.2. [#79](https://github.com/wagtail/wagtail-bakery/pull/79)
 
+### Fixed
+
+- Allow unicode characters in slug [#80](https://github.com/wagtail-nest/wagtail-bakery/pull/80)
+
 ### Removed
 
 - Drop support for Wagtail before 4.1 [#79](https://github.com/wagtail/wagtail-bakery/pull/79)

src/wagtailbakery/views.py

@@ -1,6 +1,6 @@
 import logging
 import os
-from urllib.parse import urlparse
+from urllib.parse import unquote, urlparse
 
 from bakery.views import BuildableDetailView
 from django.conf import settings
@@ -55,17 +55,18 @@ def get_build_path(self, obj):
         if url.startswith('http'):
             # Multisite has absolute urls
             url_parsed = urlparse(url)
-            path = url_parsed.path
+            path = unquote(url_parsed.path[1:])
             hostname = url_parsed.hostname
 
             if getattr(settings, 'BAKERY_MULTISITE', False):
                 build_path = os.path.join(
-                    settings.BUILD_DIR, hostname, path[1:])
+                    settings.BUILD_DIR, hostname, path)
             else:
-                build_path = os.path.join(settings.BUILD_DIR, path[1:])
+                build_path = os.path.join(settings.BUILD_DIR, path)
         else:
             # Single site has relative urls
-            build_path = os.path.join(settings.BUILD_DIR, url[1:])
+            path = unquote(url[1:])
+            build_path = os.path.join(settings.BUILD_DIR, path)
 
         # Make sure the (deeply) directories are created
         os.path.exists(build_path) or os.makedirs(build_path)

tests/fixtures.py

@@ -19,14 +19,30 @@ def site():
 def page_tree(page):
     from tests.factories.page import PageFactory
 
-    PageFactory(depth=3, path='000100010001', slug='first', numchild=2)
+    # /first
+    first_page = PageFactory(depth=3, path='000100010001', slug='first', numchild=2)
+
+    # /first/first
+    PageFactory(depth=4, path=f'{first_page.path}0001', slug='first')
+    # /first/second
+    PageFactory(depth=4, path=f'{first_page.path}0002', slug='second')
+
+    # /second
     PageFactory(depth=3, path='000100010002', slug='second')
+
+    # /third
     PageFactory(depth=3, path='000100010003', slug='third')
 
-    PageFactory(depth=4, path='0001000100010001', slug='first')
-    PageFactory(depth=4, path='0001000100010002', slug='second')
+    # /unicode-children
+    unicode_page = PageFactory(depth=3, path='000100010004', slug='unicode-children', numchild=3)
+    # /unicode-children/latin-capital-letter-i-with-diaeresis-Ï
+    PageFactory(depth=4, path=f'{unicode_page.path}0001', slug='latin-capital-letter-i-with-diaeresis-Ï')
+    # /unicode-children/cyrillic-capital-letter-ya-Я
+    PageFactory(depth=4, path=f'{unicode_page.path}0002', slug='cyrillic-capital-letter-ya-Я')
+    # /unicode-children/cjk-fire-火
+    PageFactory(depth=4, path=f'{unicode_page.path}0003', slug='cjk-fire-火')
 
-    page.numchild = 3
+    page.numchild = 4
     page.save()
 
     return page

tests/integration/test_api_views.py

@@ -34,10 +34,10 @@ def test_wagtail_bakery_pages_api_detail_view(page_tree):
     build_path = view.get_build_path(child_page)
     assert build_path == 'api/pages/detail/3.json'
 
-    # Check child build path of the first child page
+    # Check child build path of the first grandchild page
     child_page = child_page.get_descendants().first()
     build_path = view.get_build_path(child_page)
-    assert build_path == 'api/pages/detail/6.json'
+    assert build_path == 'api/pages/detail/4.json'
 
 
 @pytest.mark.django_db
@@ -75,7 +75,7 @@ def test_wagtail_bakery_pages_api_detail_view_content(page_tree):
     content = json.loads(view.get_content(grandchild_page).decode('UTF-8'))
     assert set(content.keys()) == DEFAULT_PAGE_FIELDS
     assert set(content['meta'].keys()) == DEFAULT_PAGE_META_FIELDS.union({'parent'})
-    assert content['id'] == 6
+    assert content['id'] == 4
     assert content['title'] == "Page"
 
 

tests/integration/test_models.py

@@ -16,19 +16,31 @@ def test_site(site):
 @pytest.mark.django_db
 def test_page_tree(page_tree):
     children = page_tree.get_children()
-    assert children.count() == 3
+    assert children.count() == 4
 
     # Check if children of homepage return correct urls
     assert children[0].url == '/first/'
     assert children[1].url == '/second/'
     assert children[2].url == '/third/'
-
-    children = children[0].get_children()
-    assert children.count() == 2
+    assert children[3].url == '/unicode-children/'
 
     # Check if children of first child return correct urls
-    assert children[0].url == '/first/first/'
-    assert children[1].url == '/first/second/'
+    grandchildren = children[0].get_children()
+    assert grandchildren.count() == 2
+    assert grandchildren[0].url == '/first/first/'
+    assert grandchildren[1].url == '/first/second/'
+
+    # Check if children of unicode-children return correct urls
+    grandchildren = children[3].get_children()
+    assert grandchildren.count() == 3
+
+    # Unicode characters are URL-encoded with `url` but not `url_path`.
+    # However manual testing shows that the page is indeed accessible with the non-encoded string.
+    assert grandchildren[0].url_path == (
+        '/home/unicode-children/latin-capital-letter-i-with-diaeresis-Ï/'
+    )
+    assert grandchildren[1].url_path == '/home/unicode-children/cyrillic-capital-letter-ya-Я/'
+    assert grandchildren[2].url_path == '/home/unicode-children/cjk-fire-火/'
 
 
 @pytest.mark.django_db

tests/integration/test_views.py

@@ -17,7 +17,7 @@ def test_wagtail_bakery_view_get_url(page_tree):
     url = view.get_url(child_page)
     assert url == '/first/'
 
-    # Check child url of the first child page
+    # Check child url of the first grandchild page
     child_page = child_page.get_descendants().first()
     url = view.get_url(child_page)
     assert url == '/first/first/'
@@ -31,16 +31,37 @@ def test_wagtail_bakery_view_build_path(page_tree):
     build_path = view.get_build_path(page_tree)
     assert build_path == settings.BUILD_DIR + '/index.html'
 
-    # Check child build path for first child page
-    child_page = page_tree.get_descendants().first()
+    # Check build path for first child page
+    child_page = page_tree.get_children().first()
     build_path = view.get_build_path(child_page)
     assert build_path == settings.BUILD_DIR + '/first/index.html'
 
-    # Check child build path of the first child page
-    child_page = child_page.get_descendants().first()
-    build_path = view.get_build_path(child_page)
+    # Check build path of the first grandchild page
+    grandchild_page = child_page.get_children().first()
+    build_path = view.get_build_path(grandchild_page)
     assert build_path == settings.BUILD_DIR + '/first/first/index.html'
 
+    # Check build path for unicode-children child page
+    child_page = page_tree.get_children().last()
+    build_path = view.get_build_path(child_page)
+    assert build_path == settings.BUILD_DIR + '/unicode-children/index.html'
+
+    # Check build path for unicode-children grandchild pages
+    grandchild_pages = child_page.get_children().all()
+
+    build_path = view.get_build_path(grandchild_pages[0])
+    assert build_path == (
+        settings.BUILD_DIR + '/unicode-children/latin-capital-letter-i-with-diaeresis-Ï/index.html'
+    )
+
+    build_path = view.get_build_path(grandchild_pages[1])
+    assert build_path == (
+        settings.BUILD_DIR + '/unicode-children/cyrillic-capital-letter-ya-Я/index.html'
+    )
+
+    build_path = view.get_build_path(grandchild_pages[2])
+    assert build_path == settings.BUILD_DIR + '/unicode-children/cjk-fire-火/index.html'
+
 
 @pytest.mark.django_db
 def test_wagtail_bakery_view_build_path_for_multisite(multisite):
@@ -75,7 +96,7 @@ def test_all_published_pages_for_multiple_pages(page_tree):
     qs = view.get_queryset()
 
     # Check if all pages in page tree are returned
-    assert qs.count() == 6
+    assert qs.count() == 10
 
 
 @pytest.mark.django_db