fix index entries in usage reports (#5)

* squash commits. fix version to 5, add signals / hooks

* make index entries visible to usage reports

* update tests

* improve error handling

* stop unpublished and draft pages from being indexed

.github/workflows/test.yml

@@ -20,7 +20,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python: [3.7, 3.8, 3.9]
+        python: [3.8, 3.9]
 
     steps:
       - uses: actions/checkout@v2

.gitignore

@@ -5,3 +5,8 @@ wagtail_streamfield_index.egg-info
 media/
 Pipfile
 Pipfile.lock
+venv/
+.venv/
+build/
+__pycache__
+.vscode/

setup.py

@@ -3,15 +3,15 @@
 with open("README.md", "r") as fh:
     long_description = fh.read()
 
-INSTALL_REQUIRES = ["django>=3.2", "Wagtail>=4.0"]
+INSTALL_REQUIRES = ["django>=3.2", "Wagtail>=5.0.0", " wagtail < 6.0"]
 
-TESTING_REQUIRES = ["pytest==5.2.1", "pytest-django==3.5.1", "pytest-pythonpath==0.7.3"]
+TESTING_REQUIRES = ["pytest==6.2.5", "pytest-django==3.5.1", "pytest-pythonpath==0.7.3", "factory-boy>=3.2"]
 
 LINTING_REQUIRES = ["black==20.8b1", "flake8==3.7.8", "flake8-black==0.1.1", "isort==5.7.0"]
 
 setup(
     name="wagtail-streamfield-index",
-    version="0.0.4",
+    version="1.0.0",
     description="Indexing for Wagtail streamfields",
     author="Mike Monteith",
     author_email="<[email protected]>",

streamfieldindex/indexer.py

@@ -23,19 +23,20 @@ def index_page(page):
     # Clear the index for this specific page
     IndexEntry.objects.filter(page__id=page.id).delete()
 
-    for field in page._meta.fields:
-        if not isinstance(field, StreamField):
-            # We are only interested in streamfields. Skip over non-streamfield fields
-            continue
+    if page.live:  # we dont want to index any draft/unpublished pages
+        for field in page._meta.fields:
+            if not isinstance(field, StreamField):
+                # We are only interested in streamfields. Skip over non-streamfield fields
+                continue
 
-        index_field(field, page)
+            index_field(field, page)
 
 
 def index_field(field, page):
 
     field_name = field.name
     streamvalue = getattr(page, field_name)
-    for (block, path) in flatten_streamfield(streamvalue):
+    for block, path in flatten_streamfield(streamvalue):
         field_name = field_name
         block_name = path[-1]
 

streamfieldindex/models.py

@@ -1,4 +1,5 @@
 from django.db import models
+from django.urls import reverse
 from wagtail.blocks import ListBlock, StreamBlock, StructBlock
 
 
@@ -21,6 +22,9 @@ class IndexEntry(models.Model):
 
     page = models.ForeignKey("wagtailcore.Page", on_delete=models.CASCADE)
 
+    def get_edit_url(self):
+        return reverse("wagtailadmin_pages:edit", args=[self.page.id])
+
     def get_bound_block(self):
         field_value = getattr(self.page.specific, self.field_name)
         path = self.block_path.split("/")
@@ -43,15 +47,15 @@ def get_sub_block(bound_block, path_list):
                 next_block = bound_block.value[index]
                 return get_sub_block(next_block, path_list)
             else:
-                raise Exception(f"We don't know how to iterate over block type {type(bound_block.block)}")
+                raise ValueError(f"We don't know how to iterate over block type {type(bound_block.block)}")
 
         first_index = path.pop(0)  # The first index must always be a number, since it is a streamfield
         path.pop(0)  # We can throw away the next path as it is just the block_type which we don't need
         block = field_value[int(first_index)]
         return get_sub_block(block, path)
 
     def __str__(self):
-        return f"<IndexEntry {self.page.title}::{self.field_name}::{self.block_path}>"
+        return f"Streamfield index: {self.page.title} {self.field_name} {self.block_path}" # noqa E231
 
     class Meta:
         verbose_name_plural = "Index Entries"

streamfieldindex/wagtail_hooks.py

@@ -1,7 +1,10 @@
+from django.urls import reverse
 from wagtail import hooks
-from wagtail.signals import page_published
+from wagtail.admin.admin_url_finder import ModelAdminURLFinder, register_admin_url_finder
+from wagtail.signals import page_published, page_unpublished, post_page_move
 
 from .indexer import index_page
+from .models import IndexEntry
 
 
 @hooks.register("after_create_page")
@@ -14,8 +17,35 @@ def index_after_edit_page(request, page):
     index_page(page)
 
 
+@hooks.register("after_copy_page")
+def index_after_copy_page(request, page):
+    index_page(page)
+
+
 def post_publish(sender, instance, **kwargs):
     index_page(instance)
 
 
+def index_post_page_move(sender, instance, **kwargs):
+    index_page(instance)
+
+
+def index_post_unpublished(sender, instance, **kwargs):
+    index_page(instance)
+
+
+class IndexEntryAdminURLFinder(ModelAdminURLFinder):
+    """
+    Custom URL finder for IndexEntry model
+    https://github.com/gasman/wagtail/blob/9174db40746514b6fa6d792b25507571381c9255/wagtail/admin/admin_url_finder.py#L28
+    """
+
+    def construct_edit_url(self, instance):
+        return reverse("wagtailadmin_pages:edit", args=[instance.page.id])
+
+
+register_admin_url_finder(IndexEntry, IndexEntryAdminURLFinder)
+
 page_published.connect(post_publish)
+page_unpublished.connect(index_post_unpublished)
+post_page_move.connect(index_post_page_move)

tests/test_wagtail_hooks.py

@@ -0,0 +1,21 @@
+import pytest
+from django.urls import reverse
+from wagtail.test.utils.wagtail_factories import PageFactory
+from wagtail.tests.utils import WagtailTestUtils
+
+import streamfieldindex
+from streamfieldindex.wagtail_hooks import IndexEntryAdminURLFinder
+
+
+@pytest.mark.django_db
+class TestIndexEntryAdminURLFinder(WagtailTestUtils):
+    def test_construct_edit_url(self):
+        finder = IndexEntryAdminURLFinder()
+        instance = streamfieldindex.models.IndexEntry(
+            page=PageFactory(),
+            block_type="text",
+            block_value="Hello, world!",
+            block_path="content",
+        )
+        expected_url = reverse("wagtailadmin_pages:edit", args=[instance.page.id])
+        assert finder.construct_edit_url(instance) == expected_url