Service order adjustments (#15)

- Set service order, fixes #14
- Only process necessary archives

Author: Buried-In-Code

perdoo/__init__.py

@@ -64,6 +64,7 @@ def setup_logging(debug: bool = False) -> None:
     console_handler.setLevel(logging.DEBUG if debug else logging.INFO)
     console_handler.setFormatter(logging.Formatter("%(message)s"))
     file_handler = logging.FileHandler(filename=log_folder / "perdoo.log")
+    file_handler.setLevel(logging.DEBUG)
     logging.basicConfig(
         format="[%(asctime)s] [%(levelname)-8s] {%(name)s} | %(message)s",
         datefmt="%Y-%m-%d %H:%M:%S",

perdoo/__main__.py

@@ -19,7 +19,7 @@
 from perdoo.models.metadata import Format, Meta, Source, Tool
 from perdoo.models.metron_info import InformationSource
 from perdoo.services import Comicvine, League, Marvel, Metron
-from perdoo.settings import OutputFormat, Settings
+from perdoo.settings import OutputFormat, Service, Settings
 from perdoo.utils import Details, Identifications, get_metadata_id, list_files, sanitize
 
 LOGGER = logging.getLogger("perdoo")
@@ -47,14 +47,15 @@ def convert_collection(path: Path, output: OutputFormat) -> None:
         archive_type.convert(old_archive=archive)
 
 
-def read_meta(archive: BaseArchive) -> tuple[Meta, Details]:
+def read_meta(archive: BaseArchive) -> tuple[Meta | None, Details | None]:
     filenames = archive.list_filenames()
 
     def read_meta_file(cls: type[InfoModel], filename: str) -> InfoModel | None:
         if filename in filenames:
             return cls.from_bytes(content=archive.read_file(filename=filename))
         return None
 
+    # region Read Metadata
     try:
         metadata = read_meta_file(cls=Metadata, filename="/Metadata.xml") or read_meta_file(
             cls=Metadata, filename="Metadata.xml"
@@ -97,6 +98,9 @@ def read_meta_file(cls: type[InfoModel], filename: str) -> InfoModel | None:
             return meta, details
     except ValidationError:
         LOGGER.error("%s contains an invalid Metadata file", archive.path.name)  # noqa: TRY400
+    # endregion
+
+    # region Read MetronInfo
     try:
         metron_info = read_meta_file(cls=MetronInfo, filename="/MetronInfo.xml") or read_meta_file(
             cls=MetronInfo, filename="MetronInfo.xml"
@@ -139,6 +143,9 @@ def read_meta_file(cls: type[InfoModel], filename: str) -> InfoModel | None:
             return Meta(date_=date.today(), tool=Tool(value="MetronInfo")), details
     except ValidationError:
         LOGGER.error("%s contains an invalid MetronInfo file", archive.path.name)  # noqa: TRY400
+    # endregion
+
+    # region Read ComicInfo
     try:
         comic_info = read_meta_file(cls=ComicInfo, filename="/ComicInfo.xml") or read_meta_file(
             cls=ComicInfo, filename="ComicInfo.xml"
@@ -151,11 +158,28 @@ def read_meta_file(cls: type[InfoModel], filename: str) -> InfoModel | None:
             return Meta(date_=date.today(), tool=Tool(value="ComicInfo")), details
     except ValidationError:
         LOGGER.error("%s contains an invalid ComicInfo file", archive.path.name)  # noqa: TRY400
-
-    return Meta(date_=date.today(), tool=Tool(value="Manual")), Details(
-        series=Identifications(search=Prompt.ask("Series title", console=CONSOLE)),
-        issue=Identifications(),
-    )
+    # endregion
+
+    return None, None
+
+
+def load_archives(
+    path: Path, output: OutputFormat, force: bool = False
+) -> list[tuple[Path, BaseArchive, Details | None]]:
+    archives = []
+    with CONSOLE.status(f"Searching for {output} files"):
+        for file in list_files(path, f".{output}"):
+            archive = get_archive(path=file)
+            LOGGER.debug("Reading %s", file.stem)
+            meta, details = read_meta(archive=archive)
+            if not meta or not details:
+                archives.append((file, archive, details))
+                continue
+            difference = abs(date.today() - meta.date_)
+            if force or meta.tool != Tool() or difference.days >= 28:
+                archives.append((file, archive, details))
+                continue
+    return archives
 
 
 def fetch_from_services(
@@ -181,7 +205,15 @@ def fetch_from_services(
         LOGGER.warning("No external services configured")
         return None, None, None
 
-    for service in (marvel, metron, comicvine, league):
+    services = {
+        Service.COMICVINE: comicvine,
+        Service.LEAGUE_OF_COMIC_GEEKS: league,
+        Service.MARVEL: marvel,
+        Service.METRON: metron,
+    }
+
+    for service_name in settings.service_order:
+        service = services[service_name]
         if not service:
             continue
         LOGGER.info("Fetching details from %s", type(service).__name__)
@@ -243,6 +275,7 @@ def process_pages(
     from perdoo.models.metadata import Page as MetadataPage
     from perdoo.models.metron_info import Page as MetronPage
 
+    LOGGER.info("Processing pages")
     rename_images(folder=folder, filename=filename)
     image_list = list_files(folder, *IMAGE_EXTENSIONS)
     metadata_pages = set()
@@ -272,16 +305,17 @@ def process_pages(
 def start(settings: Settings, force: bool = False) -> None:
     LOGGER.info("Starting Perdoo")
     convert_collection(path=settings.collection_folder, output=settings.output.format)
-    for file in list_files(settings.collection_folder, f".{settings.output.format}"):
-        archive = get_archive(path=file)
+    archives = load_archives(
+        path=settings.collection_folder, output=settings.output.format, force=force
+    )
+
+    for file, archive, details in archives:
         CONSOLE.rule(file.stem)
         LOGGER.info("Processing %s", file.stem)
-        meta, details = read_meta(archive=archive)
-
-        if not force:
-            difference = abs(date.today() - meta.date_)
-            if meta.tool == Tool() and difference.days < 28:
-                continue
+        details = details or Details(  # noqa: PLW2901
+            series=Identifications(search=Prompt.ask("Series title", console=CONSOLE)),
+            issue=Identifications(),
+        )
 
         metadata, metron_info, comic_info = fetch_from_services(settings=settings, details=details)
         new_file = generate_filename(
@@ -293,7 +327,6 @@ def start(settings: Settings, force: bool = False) -> None:
             temp_folder = Path(temp_str)
             if not archive.extract_files(destination=temp_folder):
                 return
-            LOGGER.info("Processing %s pages", file.stem)
             process_pages(
                 folder=temp_folder,
                 metadata=metadata,

perdoo/archives/cb7.py

@@ -78,7 +78,9 @@ def convert(old_archive: BaseArchive) -> CB7Archive | None:
             temp_folder = Path(temp_str)
             if not old_archive.extract_files(destination=temp_folder):
                 return None
-            archive_file = CB7Archive.archive_files(src=temp_folder, filename=old_archive.path.stem)
+            archive_file = CB7Archive.archive_files(
+                src=temp_folder, output_name=old_archive.path.stem
+            )
             if not archive_file:
                 return None
             new_filepath = old_archive.path.parent / f"{old_archive.path.stem}.cb7"

perdoo/archives/cbt.py

@@ -73,7 +73,9 @@ def convert(old_archive: BaseArchive) -> CBTArchive | None:
             temp_folder = Path(temp_str)
             if not old_archive.extract_files(destination=temp_folder):
                 return None
-            archive_file = CBTArchive.archive_files(src=temp_folder, filename=old_archive.path.stem)
+            archive_file = CBTArchive.archive_files(
+                src=temp_folder, output_name=old_archive.path.stem
+            )
             if not archive_file:
                 return None
             new_filepath = old_archive.path.parent / f"{old_archive.path.stem}.cbt"

perdoo/archives/cbz.py

@@ -61,7 +61,9 @@ def convert(old_archive: BaseArchive) -> CBZArchive | None:
             temp_folder = Path(temp_str)
             if not old_archive.extract_files(destination=temp_folder):
                 return None
-            archive_file = CBZArchive.archive_files(src=temp_folder, filename=old_archive.path.stem)
+            archive_file = CBZArchive.archive_files(
+                src=temp_folder, output_name=old_archive.path.stem
+            )
             if not archive_file:
                 return None
             new_filepath = old_archive.path.parent / f"{old_archive.path.stem}.cbz"

perdoo/services/marvel.py

@@ -25,7 +25,7 @@
 
 class Marvel(BaseService[Series, Comic]):
     def __init__(self: Marvel, settings: MarvelSettings):
-        cache = SqliteCache(db_name=str(get_cache_dir() / "mokkari.sqlite"), expire=14)
+        cache = SqliteCache(db_name=str(get_cache_dir() / "esak.sqlite"), expire=14)
         self.session = Esak(
             public_key=settings.public_key, private_key=settings.private_key, cache=cache
         )
@@ -40,7 +40,7 @@ def _get_series_id(self: Marvel, title: str | None) -> int | None:
             if not options:
                 LOGGER.warning("Unable to find any Series with the title: '%s'", title)
             index = create_menu(
-                options=[f"{x.id} | {x.title} ({x.start_year})" for x in options],
+                options=[f"{x.id} | {x.title}" for x in options],
                 title="Marvel Series",
                 default="None of the Above",
             )

perdoo/settings.py

@@ -77,6 +77,28 @@ def __str__(self: OutputFormat) -> str:
         return self.value
 
 
+class Service(Enum):
+    COMICVINE = "Comicvine"
+    LEAGUE_OF_COMIC_GEEKS = "League of Comic Geeks"
+    MARVEL = "Marvel"
+    METRON = "Metron"
+
+    @staticmethod
+    def load(value: str) -> Service:
+        for entry in Service:
+            if entry.value.casefold() == value.casefold():
+                return entry
+        raise ValueError(f"`{value}` isn't a valid Service")
+
+    def __lt__(self: Service, other) -> int:  # noqa: ANN001
+        if not isinstance(other, type(self)):
+            raise NotImplementedError
+        return self.value < other.value
+
+    def __str__(self: Service) -> str:
+        return self.value
+
+
 class Output(SettingsModel):
     create_comic_info: bool = True
     create_metron_info: bool = True
@@ -99,6 +121,12 @@ class Settings(SettingsModel):
     league_of_comic_geeks: LeagueofComicGeeks = LeagueofComicGeeks()
     marvel: Marvel = Marvel()
     metron: Metron = Metron()
+    service_order: list[Service] = [
+        Service.METRON,
+        Service.MARVEL,
+        Service.COMICVINE,
+        Service.LEAGUE_OF_COMIC_GEEKS,
+    ]
     output: Output = Output()
 
     @classmethod
@@ -113,6 +141,7 @@ def save(self: Settings) -> Settings:
         with self._filename.open("wb") as stream:
             content = self.dict(by_alias=False)
             content["collection_folder"] = str(content["collection_folder"])
+            content["service_order"] = [str(x) for x in content["service_order"]]
             content["output"]["format"] = str(content["output"]["format"])
             tomlwriter.dump(content, stream)
         return self