Change slug detection to work with permalinks, but dates currenly ign…

…ored

src/djpress/models/post.py

@@ -11,7 +11,7 @@
 
 from djpress.conf import settings
 from djpress.models import Category
-from djpress.utils import render_markdown
+from djpress.utils import extract_parts_from_path, render_markdown
 
 logger = logging.getLogger(__name__)
 
@@ -65,6 +65,8 @@ def get_published_page_by_path(
         """Return a single published post from a path.
 
         For now, we'll only allow a top level path.
+
+        This will raise a ValueError if the path is invalid.
         """
         # Check for a single item in the path
         if path.count("/") > 0:
@@ -127,9 +129,7 @@ def _get_cached_recent_published_posts(self: "PostsManager") -> models.QuerySet:
 
             timeout = self._get_cache_timeout(queryset)
 
-            queryset = queryset.filter(date__lte=timezone.now())[
-                : settings.RECENT_PUBLISHED_POSTS_COUNT
-            ]
+            queryset = queryset.filter(date__lte=timezone.now())[: settings.RECENT_PUBLISHED_POSTS_COUNT]
             cache.set(
                 PUBLISHED_POSTS_CACHE_KEY,
                 queryset,
@@ -189,9 +189,16 @@ def get_published_post_by_path(
     ) -> "Post":
         """Return a single published post from a path.
 
-        This is a complex piece of logic...
+        This is a complex piece of logic that needs to extract the slug from the following path structures:
+
+        `/{POST_PREFIX}/{POST_PERMALINK}/{slug}`
+
+        - POST_PREFIX is optional and could be an empty string or a custom string.
+        - POST_PERMALINK is optional and could be an empty string or a custom date format string.
+        - slug is everything after the slash
 
         Here are all the different valid paths that we need to check:
+
         - /blog/2021/01/01/post-slug
         - /blog/2021/01/post-slug
         - /blog/2021/post-slug
@@ -210,22 +217,17 @@ def get_published_post_by_path(
 
         First, we need to know the following:
         - Is there a POST_PREFIX defined?
+        - Is there a POST_PERMALINK defined?
         - Are DATE_ARCHIVES enabled?
-        - If DATE_ARCHIVES are enabled, what is the POST_PERMALINK set to?
 
         I don't think I can avoid regex matching here...
 
         For now, we'll just look at the POST_PREFIX.
         """
-        if settings.POST_PREFIX and path.startswith(settings.POST_PREFIX):
-            slug = path.split(settings.POST_PREFIX + "/")[1]
-            return self.get_published_post_by_slug(slug)
+        # This will raise a SlugNotFoundError exception if the path is invalid
+        path_parts = extract_parts_from_path(path)
 
-        if settings.POST_PREFIX and not path.startswith(settings.POST_PREFIX):
-            msg = "Invalid path"
-            raise ValueError(msg)
-
-        return self.get_published_post_by_slug(path)
+        return self.get_published_post_by_slug(path_parts.slug)
 
     def get_published_posts_by_category(
         self: "PostsManager",
@@ -236,11 +238,7 @@ def get_published_posts_by_category(
         Must have a date less than or equal to the current date/time for a specific
         category, ordered by date in descending order.
         """
-        return (
-            self.get_published_posts()
-            .filter(categories=category)
-            .prefetch_related("categories", "author")
-        )
+        return self.get_published_posts().filter(categories=category).prefetch_related("categories", "author")
 
     def get_published_posts_by_author(
         self: "PostsManager",
@@ -251,11 +249,7 @@ def get_published_posts_by_author(
         Must have a date less than or equal to the current date/time for a specific
         author, ordered by date in descending order.
         """
-        return (
-            self.get_published_posts()
-            .filter(author=author)
-            .prefetch_related("categories", "author")
-        )
+        return self.get_published_posts().filter(author=author).prefetch_related("categories", "author")
 
 
 class Post(models.Model):
@@ -312,10 +306,7 @@ def content_markdown(self: "Post") -> str:
     def truncated_content_markdown(self: "Post") -> str:
         """Return the truncated content as HTML converted from Markdown."""
         read_more_index = self.content.find(settings.TRUNCATE_TAG)
-        if read_more_index != -1:
-            truncated_content = self.content[:read_more_index]
-        else:
-            truncated_content = self.content
+        truncated_content = self.content[:read_more_index] if read_more_index != -1 else self.content
         return render_markdown(truncated_content)
 
     @property

src/djpress/utils.py

@@ -1,11 +1,15 @@
 """Utility functions that are used in the project."""
 
+import re
+from typing import NamedTuple
+
 import markdown
 from django.contrib.auth.models import User
 from django.template.loader import TemplateDoesNotExist, select_template
 from django.utils import timezone
 
 from djpress.conf import settings
+from djpress.exceptions import SlugNotFoundError
 
 md = markdown.Markdown(
     extensions=settings.MARKDOWN_EXTENSIONS,
@@ -74,13 +78,13 @@ def validate_date(year: str, month: str, day: str) -> None:
 
     try:
         if int_month and int_day:
-            timezone.datetime(int_year, int_month, int_day)
+            timezone.make_aware(timezone.datetime(int_year, int_month, int_day))
 
         elif int_month:
-            timezone.datetime(int_year, int_month, 1)
+            timezone.make_aware(timezone.datetime(int_year, int_month, 1))
 
         else:
-            timezone.datetime(int_year, 1, 1)
+            timezone.make_aware(timezone.datetime(int_year, 1, 1))
 
     except ValueError as exc:
         msg = "Invalid date"
@@ -103,3 +107,84 @@ def get_template_name(templates: list[str]) -> str:
         raise TemplateDoesNotExist(msg) from exc
 
     return template
+
+
+class PathParts(NamedTuple):
+    """Named tuple for the path parts.
+
+    These are extracted by the extract_parts_from_path function.
+
+    Attributes:
+        year (int | None): The year.
+        month (int | None): The month.
+        day (int | None): The day.
+        slug (str): The slug.
+    """
+
+    year: int | None
+    month: int | None
+    day: int | None
+    slug: str
+
+
+def extract_parts_from_path(path: str) -> PathParts:
+    """Extract the parts from the path.
+
+    Args:
+        path (str): The path.
+
+    Returns:
+        PathParts: The parts extracted from the path.
+    """
+    # Remove leading and trailing slashes
+    path = path.strip("/")
+
+    # Build the regex pattern
+    pattern_parts = []
+
+    post_prefix = settings.POST_PREFIX
+    post_permalink = settings.POST_PERMALINK
+
+    # Add the post prefix to the pattern, if it exists
+    if post_prefix:
+        pattern_parts.append(re.escape(post_prefix))
+
+    # Add the date parts based on the permalink structure
+    if post_permalink:
+        if "%Y" in post_permalink:
+            post_permalink = post_permalink.replace("%Y", r"(?P<year>\d{4})")
+        if "%m" in post_permalink:
+            post_permalink = post_permalink.replace("%m", r"(?P<month>\d{2})")
+        if "%d" in post_permalink:
+            post_permalink = post_permalink.replace("%d", r"(?P<day>\d{2})")
+        pattern_parts.append(post_permalink)
+
+    # Add the slug capture group
+    pattern_parts.append(r"(?P<slug>[0-9A-Za-z_/-]+)")  # TODO: repeated code - urls.py
+
+    # Join patterns with optional slashes
+    pattern = "^" + "/".join(f"(?:{part})" for part in pattern_parts) + "$"
+
+    # Attempt to match the pattern
+    match = re.match(pattern, path)
+
+    if not match:
+        msg = "Slug could not be found in the provided path."
+        raise SlugNotFoundError(msg)
+
+    # Extract the date parts and slug
+    year = match.group("year") if "year" in match.groupdict() else None
+    month = match.group("month") if "month" in match.groupdict() else None
+    day = match.group("day") if "day" in match.groupdict() else None
+    slug = match.group("slug") if "slug" in match.groupdict() else None
+
+    if not slug:
+        msg = "Slug could not be found in the provided path."
+        raise SlugNotFoundError(msg)
+
+    # Convert year, month, day to integers (or None if not present)
+    year = int(year) if year else None
+    month = int(month) if month else None
+    day = int(day) if day else None
+
+    return PathParts(year=year, month=month, day=day, slug=slug)

src/djpress/views.py

@@ -12,6 +12,7 @@
 from django.shortcuts import render
 
 from djpress.conf import settings
+from djpress.exceptions import SlugNotFoundError
 from djpress.models import Category, Post
 from djpress.utils import get_template_name, validate_date
 
@@ -223,9 +224,14 @@ def post_detail(request: HttpRequest, path: str) -> HttpResponse:
         # If the page is found, use the page template
         template_names.insert(0, "djpress/page.html")
     except ValueError:
+        # A ValueError means we were able to parse the path, but the page was not found
         try:
             post = Post.post_objects.get_published_post_by_path(path)
             context: dict = {"post": post}
+        except SlugNotFoundError as exc:
+            # A SlugNotFoundError means we were not able to parse the path
+            msg = "Post not found"
+            raise Http404(msg) from exc
         except ValueError as exc:
             msg = "Post not found"
             raise Http404(msg) from exc

tests/test_models_post.py

@@ -8,6 +8,7 @@
 from unittest.mock import patch
 
 from djpress.models.post import PUBLISHED_POSTS_CACHE_KEY
+from djpress.exceptions import SlugNotFoundError
 
 
 @pytest.fixture
@@ -94,9 +95,7 @@ def test_post_methods(test_post1, test_post2, category1, category2):
     test_post1.categories.add(category1)
 
     assert Post.post_objects.all().count() == 2
-    assert (
-        Post.post_objects.get_published_post_by_slug("test-post1").title == "Test Post1"
-    )
+    assert Post.post_objects.get_published_post_by_slug("test-post1").title == "Test Post1"
     assert Post.post_objects.get_published_posts_by_category(category1).count() == 1
     assert Post.post_objects.get_published_posts_by_category(category2).count() == 0
 
@@ -319,7 +318,7 @@ def test_post_permalink(user):
         slug="test-post",
         content="This is a test post.",
         author=user,
-        date=timezone.datetime(2024, 1, 1),
+        date=timezone.make_aware(timezone.datetime(2024, 1, 1)),
         status="published",
         post_type="post",
     )
@@ -451,27 +450,40 @@ def test_get_published_post_by_path(user):
 
     # Confirm settings are set according to settings_testing.py
     assert settings.POST_PREFIX == "test-posts"
+    assert settings.POST_PERMALINK == ""
 
     # Create a post
-    post = Post.objects.create(title="Test Post", status="published", author=user)
+    post = Post.objects.create(
+        title="Test Post",
+        status="published",
+        author=user,
+        date=timezone.make_aware(timezone.datetime(2024, 1, 1)),
+    )
 
-    # Test case 1: POST_PREFIX is set and path starts with POST_PREFIX
+    # Test case 1: POST_PREFIX is set and no POST_PERMALINK
     post_path = f"test-posts/{post.slug}"
     assert post == Post.post_objects.get_published_post_by_path(post_path)
 
     # Test case 2: POST_PREFIX is set but path does not start with POST_PREFIX
     post_path = f"/incorrect-path/{post.slug}"
-    # Should raise a ValueError
-    with pytest.raises(ValueError):
+    # Should raise a SlugNotFoundError since we can't parse the path to get the slug
+    with pytest.raises(SlugNotFoundError):
         Post.post_objects.get_published_post_by_path(post_path)
 
     # Test case 3: POST_PREFIX is not set but path starts with POST_PREFIX
     settings.set("POST_PREFIX", "")
     post_path = f"test-posts/non-existent-slug"
-    # Should raise a ValueError
+    # Should raise a ValueError since we can parse the path but the slug doesn't exist
     with pytest.raises(ValueError):
         Post.post_objects.get_published_post_by_path(post_path)
 
+    # # Test case 4: POST_PREFIX is set and POST_PERMALINK is set
+    # settings.set("POST_PERMALINK", "%Y/%m/%d")
+    # assert settings.POST_PREFIX == ""
+    # assert settings.POST_PERMALINK == "%Y/%m/%d"
+    # post_path = f"2024/01/01/{post.slug}"
+    # # assert post == Post.post_objects.get_published_post_by_path(post_path)
+
     # Set back to default
     settings.set("POST_PREFIX", "test-posts")
 
@@ -561,9 +573,7 @@ def test_get_cached_recent_published_posts(user, mock_timezone_now, monkeypatch)
 
     # Check if the timeout is correct (should be close to 2 hours)
     expected_timeout = 7200  # 2 hours in seconds
-    actual_timeout = (
-        kwargs.get("timeout") or args[2]
-    )  # timeout might be a kwarg or the third positional arg
+    actual_timeout = kwargs.get("timeout") or args[2]  # timeout might be a kwarg or the third positional arg
     assert abs(actual_timeout - expected_timeout) < 5  # Allow a small margin of error
 
     settings.set("CACHE_RECENT_PUBLISHED_POSTS", False)