Metron Service (#8)

- Add Metron Implementation
- Change urls to Pydantic's HttpUrl
- Check for Info file that may have a leading "/"
- Add more Manual matches for MetronInfo.Format

Author: Buried-In-Code

.pre-commit-config.yaml

@@ -1,6 +1,6 @@
 repos:
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.3.2
+    rev: v0.3.3
     hooks:
       - id: ruff-format
       - id: ruff

perdoo/__main__.py

@@ -8,6 +8,8 @@
 from platform import python_version
 from tempfile import TemporaryDirectory
 
+from pydantic import ValidationError
+
 from perdoo import ARCHIVE_EXTENSIONS, IMAGE_EXTENSIONS, __version__, setup_logging
 from perdoo.archives import BaseArchive, CB7Archive, CBTArchive, CBZArchive, get_archive
 from perdoo.console import CONSOLE
@@ -53,14 +55,33 @@ def convert_collection(path: Path, output: OutputFormat) -> None:
 def read_archive(archive: BaseArchive) -> tuple[Metadata, MetronInfo, ComicInfo]:
     filenames = archive.list_filenames()
     metadata = None
-    if "Metadata.xml" in filenames:
-        metadata = Metadata.from_bytes(content=archive.read_file(filename="Metadata.xml"))
+    try:
+        if "/Metadata.xml" in filenames:
+            metadata = Metadata.from_bytes(content=archive.read_file(filename="/Metadata.xml"))
+        elif "Metadata.xml" in filenames:
+            metadata = Metadata.from_bytes(content=archive.read_file(filename="Metadata.xml"))
+    except ValidationError:
+        LOGGER.error("%s contains an invalid Metadata file", archive.path.name)  # noqa: TRY400
     metron_info = None
-    if "MetronInfo.xml" in filenames:
-        metron_info = MetronInfo.from_bytes(content=archive.read_file(filename="MetronInfo.xml"))
+    try:
+        if "/MetronInfo.xml" in filenames:
+            metron_info = MetronInfo.from_bytes(
+                content=archive.read_file(filename="/MetronInfo.xml")
+            )
+        elif "MetronInfo.xml" in filenames:
+            metron_info = MetronInfo.from_bytes(
+                content=archive.read_file(filename="MetronInfo.xml")
+            )
+    except ValidationError:
+        LOGGER.error("%s contains an invalid MetronInfo file", archive.path.name)  # noqa: TRY400
     comic_info = None
-    if "ComicInfo.xml" in filenames:
-        comic_info = ComicInfo.from_bytes(content=archive.read_file(filename="ComicInfo.xml"))
+    try:
+        if "/ComicInfo.xml" in filenames:
+            comic_info = ComicInfo.from_bytes(content=archive.read_file(filename="/ComicInfo.xml"))
+        elif "ComicInfo.xml" in filenames:
+            comic_info = ComicInfo.from_bytes(content=archive.read_file(filename="ComicInfo.xml"))
+    except ValidationError:
+        LOGGER.error("%s contains an invalid ComicInfo file", archive.path.name)  # noqa: TRY400
 
     if not metadata:
         if metron_info:
@@ -80,7 +101,7 @@ def fetch_from_services(
     settings: Settings, metainfo: tuple[Metadata, MetronInfo, ComicInfo]
 ) -> None:
     marvel = None
-    if settings.marvel and settings.marvel.public_key and settings.marve.private_key:
+    if settings.marvel and settings.marvel.public_key and settings.marvel.private_key:
         marvel = Marvel(settings=settings.marvel)
     metron = None
     if settings.metron and settings.metron.username and settings.metron.password:
@@ -115,7 +136,7 @@ def generate_filename(root: Path, extension: str, metadata: Metadata) -> Path:
     )
 
     number_str = (
-        f"_#{metadata.issue.number.zfill(3 if metadata.issue.format_ == Format.COMIC else 2)}"
+        f"_#{metadata.issue.number.zfill(3 if metadata.issue.format == Format.COMIC else 2)}"
         if metadata.issue.number
         else ""
     )
@@ -125,10 +146,10 @@ def generate_filename(root: Path, extension: str, metadata: Metadata) -> Path:
         Format.GRAPHIC_NOVEL: "_GN",
         Format.HARDCOVER: "_HC",
         Format.TRADE_PAPERBACK: "_TP",
-    }.get(metadata.issue.format_, "")
-    if metadata.issue.format_ in {Format.ANNUAL, Format.DIGITAL_CHAPTER}:
+    }.get(metadata.issue.format, "")
+    if metadata.issue.format in {Format.ANNUAL, Format.DIGITAL_CHAPTER}:
         issue_filename = sanitize(value=series_filename) + format_str + number_str
-    elif metadata.issue.format_ in {Format.GRAPHIC_NOVEL, Format.HARDCOVER, Format.TRADE_PAPERBACK}:
+    elif metadata.issue.format in {Format.GRAPHIC_NOVEL, Format.HARDCOVER, Format.TRADE_PAPERBACK}:
         issue_filename = sanitize(value=series_filename) + number_str + format_str
     else:
         issue_filename = sanitize(value=series_filename) + number_str
@@ -145,7 +166,7 @@ def rename_images(folder: Path, filename: str) -> None:
     image_list = list_files(folder, *IMAGE_EXTENSIONS)
     pad_count = len(str(len(image_list)))
     for index, img_file in enumerate(image_list):
-        new_filename = f"{filename}-{str(index).zfill(pad_count)}{img_file.suffix}"
+        new_filename = f"{filename}_{str(index).zfill(pad_count)}{img_file.suffix}"
         if img_file.name != new_filename:
             LOGGER.info("Renamed %s to %s", img_file.name, new_filename)
             img_file.rename(folder / f"{filename}-{str(index).zfill(pad_count)}{img_file.suffix}")

perdoo/models/comic_info.py

@@ -10,7 +10,7 @@
 import xmltodict
 from natsort import humansorted, ns
 from PIL import Image
-from pydantic import Field
+from pydantic import Field, HttpUrl
 
 from perdoo.models._base import InfoModel, PascalModel
 
@@ -134,7 +134,7 @@ def __str__(self: PageType) -> str:
 
 class Page(PascalModel):
     image: int = Field(alias="@Image")
-    type_: PageType = Field(alias="@Type", default=PageType.STORY)
+    type: PageType = Field(alias="@Type", default=PageType.STORY)
     double_page: bool = Field(alias="@DoublePage", default=False)
     image_size: int = Field(alias="@ImageSize", default=0)
     key: str | None = Field(alias="@Key", default=None)
@@ -158,7 +158,7 @@ def __hash__(self: Page) -> int:
     @staticmethod
     def from_path(file: Path, index: int, is_final_page: bool, page: Page | None) -> Page:
         if page:
-            page_type = page.type_
+            page_type = page.type
         elif index == 0:
             page_type = PageType.FRONT_COVER
         elif is_final_page:
@@ -169,7 +169,7 @@ def from_path(file: Path, index: int, is_final_page: bool, page: Page | None) ->
             width, height = img.size
         return Page(
             image=index,
-            type_=page_type,
+            type=page_type,
             double_page=width >= height,
             image_size=file.stat().st_size,
             image_height=height,
@@ -201,10 +201,10 @@ class ComicInfo(PascalModel, InfoModel):
     publisher: str | None = None
     imprint: str | None = None
     genre: str | None = None
-    web: str | None = None
+    web: HttpUrl | None = None
     page_count: int = 0
     language_iso: str | None = Field(alias="LanguageISO", default=None)
-    format_: str | None = Field(alias="Format", default=None)
+    format: str | None = None
     black_and_white: YesNo = YesNo.UNKNOWN
     manga: Manga = Manga.UNKNOWN
     characters: str | None = None

perdoo/models/metadata.py

@@ -184,7 +184,7 @@ class Issue(PascalModel):
     characters: list[TitledResource] = Field(default_factory=list)
     cover_date: date | None = None
     credits: list[Credit] = Field(default_factory=list)
-    format_: Format = Field(alias="Format", default=Format.COMIC)
+    format: Format = Format.COMIC
     language: str = Field(alias="@language", default="en")
     locations: list[TitledResource] = Field(default_factory=list)
     number: str | None = None
@@ -280,7 +280,7 @@ class Page(PascalModel):
     height: int = Field(alias="@height")
     index: int = Field(alias="@index")
     size: int = Field(alias="@size")
-    type_: PageType = Field(alias="@type", default=PageType.STORY)
+    type: PageType = Field(alias="@type", default=PageType.STORY)
     width: int = Field(alias="@width")
 
     def __lt__(self: Page, other) -> int:  # noqa: ANN001
@@ -299,7 +299,7 @@ def __hash__(self: Page) -> int:
     @staticmethod
     def from_path(file: Path, index: int, is_final_page: bool, page: Page | None) -> Page:
         if page:
-            page_type = page.type_
+            page_type = page.type
         elif index == 0:
             page_type = PageType.FRONT_COVER
         elif is_final_page:
@@ -314,7 +314,7 @@ def from_path(file: Path, index: int, is_final_page: bool, page: Page | None) ->
             height=height,
             index=index,
             size=file.stat().st_size,
-            type_=page_type,
+            type=page_type,
             width=width,
         )
 

perdoo/models/metron_info.py

@@ -27,7 +27,7 @@
 
 import xmltodict
 from PIL import Image
-from pydantic import Field
+from pydantic import Field, HttpUrl
 
 from perdoo.models._base import InfoModel, PascalModel
 
@@ -91,6 +91,14 @@ def load(value: str) -> Format:
         for entry in Format:
             if entry.value.replace(" ", "").casefold() == value.replace(" ", "").casefold():
                 return entry
+        # region Manual matches
+        if value.casefold() == "Limited Series".casefold():
+            return Format.LIMITED
+        if value.casefold() == "Cancelled Series".casefold():
+            return Format.SERIES
+        if value.casefold() == "Hard Cover".casefold():
+            return Format.SERIES
+        # endregion
         raise ValueError(f"'{value}' isnt a valid metron_info.Format")
 
     def __lt__(self: Format, other) -> int:  # noqa: ANN001
@@ -108,7 +116,7 @@ class Series(PascalModel):
     name: str
     sort_name: str | None = None
     volume: int | None = None
-    format_: Format | None = Field(alias="Format", default=None)
+    format: Format | None = None
 
 
 class Price(PascalModel):
@@ -362,7 +370,7 @@ def __str__(self: PageType) -> str:
 
 class Page(PascalModel):
     image: int = Field(alias="@Image")
-    type_: PageType = Field(alias="@Type", default=PageType.STORY)
+    type: PageType = Field(alias="@Type", default=PageType.STORY)
     double_page: bool = Field(alias="@DoublePage", default=False)
     image_size: int = Field(alias="@ImageSize", default=0)
     key: str | None = Field(alias="@Key", default=None)
@@ -386,7 +394,7 @@ def __hash__(self: Page) -> int:
     @staticmethod
     def from_path(file: Path, index: int, is_final_page: bool, page: Page | None) -> Page:
         if page:
-            page_type = page.type_
+            page_type = page.type
         elif index == 0:
             page_type = PageType.FRONT_COVER
         elif is_final_page:
@@ -397,7 +405,7 @@ def from_path(file: Path, index: int, is_final_page: bool, page: Page | None) ->
             width, height = img.size
         return Page(
             image=index,
-            type_=page_type,
+            type=page_type,
             double_page=width >= height,
             image_size=file.stat().st_size,
             image_height=height,
@@ -428,7 +436,7 @@ class MetronInfo(PascalModel, InfoModel):
     gtin: GTIN | None = Field(alias="GTIN", default=None)
     age_rating: AgeRating = Field(default=AgeRating.UNKNOWN)
     reprints: list[Resource] = Field(default_factory=list)
-    url: str | None = Field(alias="URL", default=None)
+    url: HttpUrl | None = Field(alias="URL", default=None)
     credits: list[Credit] = Field(default_factory=list)
     pages: list[Page] = Field(default_factory=list)
 
@@ -481,7 +489,6 @@ def to_file(self: MetronInfo, file: Path) -> None:
         with file.open("wb") as stream:
             xmltodict.unparse(
                 {"MetronInfo": {k: content[k] for k in sorted(content)}},
-                # {"MetronInfo": content},
                 output=stream,
                 short_empty_elements=True,
                 pretty=True,

perdoo/services/_base.py

@@ -3,11 +3,63 @@
 __all__ = ["BaseService"]
 
 from abc import abstractmethod
+from typing import Generic, TypeVar
 
 from perdoo.models import ComicInfo, Metadata, MetronInfo
 
+P = TypeVar("P")
+S = TypeVar("S")
+C = TypeVar("C")
+
+
+class BaseService(Generic[P, S, C]):
+    @abstractmethod
+    def _search_publishers(self: BaseService, title: str | None) -> int | None: ...
+
+    @abstractmethod
+    def _get_publisher_id(
+        self: BaseService, metadata: Metadata, metron_info: MetronInfo
+    ) -> int | None: ...
+
+    @abstractmethod
+    def fetch_publisher(
+        self: BaseService, metadata: Metadata, metron_info: MetronInfo, comic_info: ComicInfo
+    ) -> P | None: ...
+
+    @abstractmethod
+    def _search_series(self: BaseService, publisher_id: int, title: str | None) -> int | None: ...
+
+    @abstractmethod
+    def _get_series_id(
+        self: BaseService, publisher_id: int, metadata: Metadata, metron_info: MetronInfo
+    ) -> int | None: ...
+
+    @abstractmethod
+    def fetch_series(
+        self: BaseService,
+        metadata: Metadata,
+        metron_info: MetronInfo,
+        comic_info: ComicInfo,
+        publisher_id: int,
+    ) -> S | None: ...
+
+    @abstractmethod
+    def _search_issues(self: BaseService, series_id: int, number: str | None) -> int | None: ...
+
+    @abstractmethod
+    def _get_issue_id(
+        self: BaseService, series_id: int, metadata: Metadata, metron_info: MetronInfo
+    ) -> int | None: ...
+
+    @abstractmethod
+    def fetch_issue(
+        self: BaseService,
+        metadata: Metadata,
+        metron_info: MetronInfo,
+        comic_info: ComicInfo,
+        series_id: int,
+    ) -> C | None: ...
 
-class BaseService:
     @abstractmethod
     def fetch(
         self: BaseService,