If the ComicCacher is to be a generic cache for talkers it must assume
very little. Current assumptions:
- There are issues that can be queried individually by an "Issue ID" and they have a relation to a single series
- There are series that can be queried individually by an "Series ID" and they have a relation to zero or more issues
- There are Searches that can be queried by the search term and they have a relation to zero or more series
Each series and issue have a boolean `complete` attribute which is up to the talker to decide what it means.
Data is returned as a tuple ([series, complete] or [issue, complete]) or a list of tuples
An issue consists of an ID, an series ID and a binary data attribute which is up to the talker to determine what it means.
An series consists of in ID and a binary data attribute which is up to the talker to determine what it means.
The data attribute is binary to allow for compression and efficient storage of binary data (e.g. pickle) it is suggested to store it as json or similar text format encoded with utf-8. If the talker is using a website API it is suggested to store the raw response from the server.
All caches automatically expire 7 days after insertion.
I could not find a good reason for ComicIssue to exist other than that
it had more attributes than GenericMetadata, so it has been replaced.
New attributes for GenericMetadata:
series_id: a string uniquely identifying the series to tag_origin
series_aliases: alternate series names that are not the canonical name
title_aliases: alternate issue titles that are not the canonical name
alternate_images: a list of urls to alternate cover images
Updated attributes for GenericMetadata:
genre -> genres: str -> list[str]
comments -> description: str -> str
story_arc -> story_arcs: str -> list[str]
series_group -> series_groups: str -> list[str]
character -> characters: str -> list[str]
team -> teams: str -> list[str]
location -> locations: str -> list[str]
tag_origin -> tag_origin: str -> TagOrigin (tuple[str, str])
ComicSeries has been relocated to the ComicAPI package, currently has no
usage within ComicAPI.
CreditMetadata has been renamed to Credit and has replaced Credit from
ComicTalker.
fetch_series has been added to ComicTalker, this is currently only used
in the GUI when a series is selected and does not already contain the
needed fields, this function should always be cached.
A new split function has been added to ComicAPI, all uses of split on
single characters have been updated to use this
cleanup_html and the corresponding setting are now only used in
ComicTagger proper, for display we want any html directly from the
upstream. When applying the metadata we then strip the description of
any html.
A new conversion has been added to the MetadataFormatter:
j: joins any lists into a string with ', '. Note this is a valid
operation on strings as well, it will add ', ' in between every
character.
parse_settings now assigns the given ComicTaggerPaths object to the
result ensuring that the correct path is always used.
IssueString.as_string always returns a string this is a problem for
GenericMetadata. When the overlay function is used it checks
specifically for the value None this allows the -m option to unset
attributes however the issue attribute would get set to the empty
string when loading ComicRack tags regardless of if there was a value
stored in the file. Fixes #465 and #480
