Fix output Metadata (#87)

- Fix URL naming, closes #84
- Add timezone to last_modified using local tz as default
- Fix ComicInfo pages

Author: Buried-In-Code

perdoo/__main__.py

@@ -1,6 +1,7 @@
 import logging
 from datetime import date
 from enum import Enum
+from io import BytesIO
 from pathlib import Path
 from platform import python_version
 from typing import Annotated
@@ -129,17 +130,37 @@ def get_search_details(
 
 
 def load_page_info(entry: Comic, comic_info: ComicInfo) -> list[Page]:
+    from PIL import Image  # noqa: PLC0415
+
+    from perdoo.metadata.comic_info import PageType  # noqa: PLC0415
+
     pages = set()
     image_files = [
         x
         for x in entry.archive.get_filename_list()
         if Path(x).suffix.lower() in SUPPORTED_IMAGE_EXTENSIONS
     ]
     for idx, file in enumerate(image_files):
-        img_file = Path(file)
-        is_final_page = idx == len(image_files) - 1
         page = next((x for x in comic_info.pages if x.image == idx), None)
-        pages.add(Page.from_path(file=img_file, index=idx, is_final_page=is_final_page, page=page))
+        if page:
+            page_type = page.type
+        elif idx == 0:
+            page_type = PageType.FRONT_COVER
+        elif idx == len(image_files) - 1:
+            page_type = PageType.BACK_COVER
+        else:
+            page_type = PageType.STORY
+        if not page:
+            page = Page(image=idx)
+        page.type = page_type
+        page_bytes = entry.archive.read_file(file)
+        page.image_size = len(page_bytes)
+        with Image.open(BytesIO(page_bytes)) as page_data:
+            width, height = page_data.size
+            page.double_page = width >= height
+            page.image_height = height
+            page.image_width = width
+        pages.add(page)
     return sorted(pages)
 
 

perdoo/metadata/comic_info.py

@@ -4,22 +4,14 @@
 from collections.abc import Callable
 from datetime import date
 from enum import Enum
-from pathlib import Path
 
 from natsort import humansorted, ns
-from PIL import Image
 from pydantic import HttpUrl, NonNegativeFloat
 from pydantic_xml import attr, computed_attr, element, wrapped
 
 from perdoo.metadata._base import PascalModel
 from perdoo.settings import Naming
 
-try:
-    from typing import Self  # Python >= 3.11
-except ImportError:
-    from typing_extensions import Self  # Python < 3.11
-
-
 LOGGER = logging.getLogger(__name__)
 
 
@@ -41,7 +33,7 @@ class YesNo(Enum):
     YES = "Yes"
 
     @staticmethod
-    def load(value: str) -> Self:
+    def load(value: str) -> "YesNo":
         for entry in YesNo:
             if entry.value.replace(" ", "").casefold() == value.replace(" ", "").casefold():
                 return entry
@@ -59,7 +51,7 @@ class Manga(Enum):
     YES_AND_RIGHT_TO_LEFT = "YesAndRightToLeft"
 
     @staticmethod
-    def load(value: str) -> Self:
+    def load(value: str) -> "Manga":
         for entry in Manga:
             if entry.value.replace(" ", "").casefold() == value.replace(" ", "").casefold():
                 return entry
@@ -88,7 +80,7 @@ class AgeRating(Enum):
     X18 = "X18+"
 
     @staticmethod
-    def load(value: str) -> Self:
+    def load(value: str) -> "AgeRating":
         for entry in AgeRating:
             if entry.value.replace(" ", "").casefold() == value.replace(" ", "").casefold():
                 return entry
@@ -113,7 +105,7 @@ class PageType(Enum):
     DELETED = "Deleted"
 
     @staticmethod
-    def load(value: str) -> Self:
+    def load(value: str) -> "PageType":
         for entry in PageType:
             if entry.value.replace(" ", "").casefold() == value.replace(" ", "").casefold():
                 return entry
@@ -147,27 +139,6 @@ def __eq__(self, other) -> bool:  # noqa: ANN001
     def __hash__(self) -> int:
         return hash((type(self), self.image))
 
-    @staticmethod
-    def from_path(file: Path, index: int, is_final_page: bool, page: Self | None) -> Self:
-        if page:
-            page_type = page.type
-        elif index == 0:
-            page_type = PageType.FRONT_COVER
-        elif is_final_page:
-            page_type = PageType.BACK_COVER
-        else:
-            page_type = PageType.STORY
-        with Image.open(file) as img:
-            width, height = img.size
-        return Page(
-            image=index,
-            type=page_type,
-            double_page=width >= height,
-            image_size=file.stat().st_size,
-            image_height=height,
-            image_width=width,
-        )
-
 
 class ComicInfo(PascalModel):
     age_rating: AgeRating = element(default=AgeRating.UNKNOWN)

perdoo/metadata/metron_info.py

@@ -24,17 +24,12 @@
 from enum import Enum
 from typing import Generic, TypeVar
 
-from pydantic import HttpUrl, NonNegativeInt, PositiveInt
+from pydantic import HttpUrl, NonNegativeInt, PositiveInt, field_validator
 from pydantic_xml import attr, computed_attr, element, wrapped
 
 from perdoo.metadata._base import PascalModel
 from perdoo.settings import Naming
 
-try:
-    from typing import Self  # Python >= 3.11
-except ImportError:
-    from typing_extensions import Self  # Python < 3.11
-
 LOGGER = logging.getLogger(__name__)
 T = TypeVar("T")
 
@@ -49,7 +44,7 @@ class AgeRating(Enum):
     ADULT = "Adult"
 
     @staticmethod
-    def load(value: str) -> Self:
+    def load(value: str) -> "AgeRating":
         for entry in AgeRating:
             if entry.value.replace(" ", "").casefold() == value.replace(" ", "").casefold():
                 return entry
@@ -142,7 +137,7 @@ class Role(Enum):
     OTHER = "Other"
 
     @staticmethod
-    def load(value: str) -> Self:
+    def load(value: str) -> "Role":
         for entry in Role:
             if entry.value.replace(" ", "").casefold() == value.replace(" ", "").casefold():
                 return entry
@@ -191,7 +186,7 @@ class InformationSource(Enum):
     LEAGUE_OF_COMIC_GEEKS = "League of Comic Geeks"
 
     @staticmethod
-    def load(value: str) -> Self:
+    def load(value: str) -> "InformationSource":
         for entry in InformationSource:
             if entry.value.replace(" ", "").casefold() == value.replace(" ", "").casefold():
                 return entry
@@ -366,7 +361,7 @@ class MetronInfo(PascalModel):
     universes: list[Universe] = wrapped(
         path="Universes", entity=element(tag="Universe", default_factory=list)
     )
-    urls: list[Url] = wrapped(path="URLS", entity=element(tag="URLs", default_factory=list))
+    urls: list[Url] = wrapped(path="URLs", entity=element(tag="URL", default_factory=list))
 
     @computed_attr(ns="xsi", name="noNamespaceSchemaLocation")
     def schema_location(self) -> str:
@@ -389,6 +384,13 @@ def get_filename(self, settings: Naming) -> str:
             seperator=settings.seperator,
         )
 
+    @field_validator("last_modified", mode="before")
+    def ensure_timezone(cls, v: str | datetime | None) -> str | datetime | None:
+        if isinstance(v, datetime) and v.tzinfo is None:
+            timezone = datetime.now().astimezone().tzinfo
+            return v.replace(tzinfo=timezone)
+        return v
+
 
 PATTERN_MAP: dict[str, Callable[[MetronInfo], str | int | None]] = {
     "cover-date": lambda x: x.cover_date,