add simple framework for runing data consistency checks

acros/management/commands/datacheck.py

@@ -0,0 +1,12 @@
+from django.core.management.base import BaseCommand
+
+from acros.utils.checks import registry
+
+
+class Command(BaseCommand):
+    help = 'check integrity of data'
+
+    def handle(self, *args, **kwargs):
+        errors = registry.run_checks()
+        for error in errors:
+            print(error)

acros/templates/acros/datacheck.html

@@ -0,0 +1,19 @@
+{% extends 'base.html' %}
+{% load static %}
+{% block heading %}
+  <h1 class="acronym">Data Checks</h1>
+{% endblock %}
+
+{% block content %}
+
+  {% for error in errors %}
+    <div class="alert alert-{{ error.level }}" role="alert">
+      <strong>{{ error.obj }}</strong>:
+      {{ error.msg }}
+      <br>
+      <a href="{{ error.edit_url }}" class="alert-link">Edit</a>
+      <a href="{{ error.admin_edit_url }}" class="alert-link">Admin-Edit</a>
+    </div>
+  {% endfor %}
+
+{% endblock %}

acros/urls.py

@@ -34,7 +34,8 @@ urlpatterns = [
     path('tags', views.TagListView.as_view(), name='tags'),
     path('tag', RedirectView.as_view(pattern_name="tags")),
     path('tag/<str:slug>', views.TagAcroView.as_view(), name='tag'),
-    path('integrations', views.IntegrationsView.as_view(), name="integrations")
+    path('integrations', views.IntegrationsView.as_view(), name="integrations"),
+    path('datachecks', views.DataCheckView.as_view(), name="datachecks")
 ]
 
 if settings.DEBUG:

acros/utils/checks/__init__.py

@@ -0,0 +1,5 @@
+from .basecheck import BaseCheck
+from .check_registry import CheckRegistry, registry
+from .messages import DEBUG, INFO, WARNING, ERROR, CRITICAL, CheckMessage, CheckWarning, CheckInfo
+from .checks.acronym import *
+from .checks.image import *

acros/utils/checks/basecheck.py

@@ -0,0 +1,5 @@
+class BaseCheck:
+    def run(self):
+        return []
+
+

acros/utils/checks/check_registry.py

@@ -0,0 +1,22 @@
+from typing import Set
+
+from . import BaseCheck
+
+
+class CheckRegistry:
+    def __init__(self):
+        self.checks: Set[BaseCheck] = set()
+
+    def register(self, check: BaseCheck):
+        self.checks.add(check)
+
+    def run_checks(self):
+        errors = []
+        for Check in self.checks:
+            inst = Check()
+            new_errors = list(inst.run())
+            errors.extend(new_errors)
+        return errors
+
+
+registry = CheckRegistry()

acros/utils/checks/checks/acronym.py

@@ -0,0 +1,23 @@
+from acros.models import Acronym
+from acros.utils.checks import BaseCheck, CheckWarning, registry, CheckInfo
+
+
+class LetterCheck(BaseCheck):
+    def run(self):
+        for acronym in Acronym.objects.all():
+            if acronym.acro_letters is None:
+                yield CheckInfo(
+                    "missing acronym letters",
+                    obj=acronym
+                )
+                continue
+            let_len = len(acronym.acro_letters)
+            acr_len = len(acronym.name)
+            if let_len != acr_len:
+                yield CheckWarning(
+                    f"number of letters selected ({let_len}) not equal to letters in acronym ({acr_len})",
+                    obj=acronym
+                )
+
+
+registry.register(LetterCheck)

acros/utils/checks/checks/image.py

@@ -0,0 +1,13 @@
+from acros.models import WikipediaImage
+from acros.utils.checks import BaseCheck, CheckWarning, registry
+
+
+class AuthorAttributionCheck(BaseCheck):
+    def run(self):
+        for image in WikipediaImage.objects.all():
+            if image.attribution_required and not image.artist:
+                yield CheckWarning("Image needs attribution, but is missing an artist", obj=image)
+
+
+registry.register(AuthorAttributionCheck)
+

acros/utils/checks/messages.py

@@ -0,0 +1,43 @@
+from django.urls import reverse
+
+from acros.models import Acronym
+
+DEBUG = 10
+INFO = "info"
+WARNING = "warning"
+ERROR = 40
+CRITICAL = 50
+
+
+class CheckMessage:
+    def __init__(self, level: str, msg: str, obj=None):
+        self.level = level
+        self.msg = msg
+        self.obj = obj
+
+    def __str__(self):
+        obj = str(self.obj)
+        print(self.obj._meta.label)
+        return f"{obj}: {self.msg}"
+
+    @property
+    def edit_url(self):
+        if isinstance(self.obj, Acronym):
+            return reverse("edit", args=[self.obj.slug])
+        return None
+
+    @property
+    def admin_edit_url(self):
+        if isinstance(self.obj, Acronym):
+            return reverse("admin:acros_acronym_change", args=[self.obj.id])
+        return None
+
+
+class CheckWarning(CheckMessage):
+    def __init__(self, msg: str, obj=None):
+        super(CheckWarning, self).__init__(WARNING, msg, obj)
+
+
+class CheckInfo(CheckMessage):
+    def __init__(self, msg: str, obj=None):
+        super(CheckInfo, self).__init__(INFO, msg, obj)

acros/views.py

@@ -11,6 +11,7 @@ from acros.forms import EditForm, AddForm, WikipediaForm, PaperForm, WeblinkForm
 from acros.models import Acronym, Tag, AcroOfTheDay, WikipediaLink, PaperReference, Weblink
 from acros.serializers import AcronymSerializer, AcronymListSerializer, TagSerializer
 from acros.utils.assets import get_css
+from acros.utils.checks import registry
 
 handler404 = 'acros.views.PageNotFoundView'
 
@@ -122,6 +123,16 @@ class TagAcroView(generic.ListView):
         return Acronym.objects.filter(tags__slug__exact=self.kwargs['slug'])
 
 
+class DataCheckView(generic.TemplateView, LoginRequiredMixin):
+    template_name = "acros/datacheck.html"
+
+    def get_context_data(self, **kwargs):
+        data = super().get_context_data(**kwargs)
+        errors = registry.run_checks()
+        data['errors'] = errors
+        return data
+
+
 #### API Views ####
 
 class AcronymViewSet(viewsets.ReadOnlyModelViewSet):