backend/plugins: Add scraper for info.yaml from schale.network

1 files changed, 82 insertions, 0 deletions

diff --git a/src/hircine/plugins/scrapers/schale_network.py b/src/hircine/plugins/scrapers/schale_network.py
new file mode 100644
index 0000000..e38cfe8
--- /dev/null
+++ b/src/hircine/plugins/scrapers/schale_network.py
@@ -0,0 +1,82 @@
+import re
+
+import yaml
+
+import hircine.enums as enums
+from hircine.scraper import Scraper
+from hircine.scraper.types import (
+    Artist,
+    Censorship,
+    Circle,
+    Direction,
+    Language,
+    Tag,
+    Title,
+)
+from hircine.scraper.utils import open_archive_file, parse_dict
+
+SOURCE_REGEX = re.compile(r"^SchaleNetwork:")
+
+
+class SchaleNetworkScraper(Scraper):
+    """
+    A scraper for ``info.yaml`` files found in archives downloaded from
+    *schale.network*.
+
+    .. list-table::
+       :align: left
+
+       * - **Requires**
+         - ``info.yaml`` in the archive or as a sidecar.
+       * - **Source**
+         - ``schale.network``
+    """
+
+    name = "schale.network info.yaml"
+    source = "schale.network"
+
+    def __init__(self, comic):
+        super().__init__(comic)
+
+        self.data = self.load()
+        source = self.data.get("source")
+
+        if source and re.match(SOURCE_REGEX, source):
+            self.is_available = True
+
+    def load(self):
+        try:
+            with open_archive_file(self.comic.archive, "info.yaml") as yif:
+                return yaml.safe_load(yif)
+        except Exception:
+            return {}
+
+    def scrape(self):
+        parsers = {
+            "title": Title,
+            "artist": Artist,
+            "circle": Circle,
+            "general": Tag.from_string,
+            "male": lambda s: Tag(namespace="male", tag=s),
+            "female": lambda s: Tag(namespace="female", tag=s),
+            "mixed": lambda s: Tag(namespace="mixed", tag=s),
+            "language": self.parse_language,
+            "other": self.parse_other,
+        }
+
+        yield from parse_dict(parsers, self.data)
+
+        yield Direction(enums.Direction.RIGHT_TO_LEFT)
+
+    def parse_language(self, input):
+        if not input or input in ["translated"]:
+            return
+
+        return Language.from_name(input)
+
+    def parse_other(self, input):
+        match input:
+            case "uncensored":
+                return Censorship(value=enums.Censorship.NONE)
+            case _:
+                return Tag.from_string(input)