modular parser framework

openmensa_parsers/parsers/__init__.py

@@ -0,0 +1,5 @@
+"""Parser/provider implementations for OpenMensa feed sources."""
+
+from openmensa_parsers.parsers.registry import create_parser, get_parser_class
+
+__all__ = ["create_parser", "get_parser_class"]

openmensa_parsers/parsers/base.py

@@ -0,0 +1,61 @@
+"""Shared parser contract for city-specific OpenMensa parsers."""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+from typing import Any, Protocol
+
+from openmensa_parsers.config import Canteen
+from openmensa_parsers.xml_types.canteen_xml import CanteenXML
+from openmensa_parsers.xml_types.feed_xml import FeedXML, ScheduleXML
+
+
+@dataclass(frozen=True)
+class FeedDefinition:
+    """Default feed metadata used when publishing a parser result."""
+
+    source: str
+    name: str = "full"
+    priority: int = 0
+    schedule: dict[str, Any] = field(
+        default_factory=lambda: {"hour": "8-14", "retry": "30 1"}
+    )
+
+
+class OpenMensaParser(Protocol):
+    """Contract implemented by each city/source parser."""
+
+    id: str
+    feed: FeedDefinition
+
+    def fetch(self) -> Any:
+        """Download or load source-specific raw data."""
+
+    def parse(
+        self,
+        config: dict[str, Canteen],
+        raw_data: Any,
+    ) -> dict[str, CanteenXML]:
+        """Convert raw source data into OpenMensa canteen structures."""
+
+    def create_feed(self, canteen: Canteen, url: str) -> FeedXML:
+        """Build the OpenMensa feed metadata for one configured canteen."""
+
+
+class BaseOpenMensaParser:  # pylint: disable=too-few-public-methods
+    """Base helper for parsers that use the standard OpenMensa feed block."""
+
+    id = "base"
+    feed: FeedDefinition
+
+    def create_feed(self, _canteen: Canteen, url: str) -> FeedXML:
+        """Create a standard feed tag for a configured canteen."""
+        schedule_data = dict(self.feed.schedule)
+        schedule = ScheduleXML(**schedule_data)
+        return FeedXML(
+            name=self.feed.name,
+            priority=self.feed.priority,
+            source=self.feed.source,
+            url=url,
+            schedule=schedule,
+        )

openmensa_parsers/parsers/potsdam.py

@@ -0,0 +1,61 @@
+"""Potsdam parser/provider implementation."""
+
+from __future__ import annotations
+
+import logging
+
+from openmensa_parsers.config import Canteen
+from openmensa_parsers.parsers.base import BaseOpenMensaParser, FeedDefinition
+from openmensa_parsers.webspeiseplan_api import (
+    WebspeiseplanAPI,
+    WebspeiseplanData,
+)
+from openmensa_parsers.webspeiseplan_parser import WebspeiseplanParser
+from openmensa_parsers.xml_types.canteen_xml import CanteenXML
+
+
+class PotsdamParser(BaseOpenMensaParser):
+    """Parser for Studentenwerk Potsdam's Webspeiseplan source."""
+
+    id = "potsdam"
+    BASE_URL = "https://swp.webspeiseplan.de"
+    feed = FeedDefinition(source=BASE_URL)
+
+    def __init__(
+        self,
+        api: WebspeiseplanAPI | None = None,
+        parser: WebspeiseplanParser | None = None,
+    ) -> None:
+        """Initialize the Potsdam parser with fetch and parse helpers."""
+        self.api = WebspeiseplanAPI(self.BASE_URL) if api is None else api
+        self.parser = WebspeiseplanParser() if parser is None else parser
+        self.logger = logging.getLogger(__name__)
+
+    def fetch(self) -> WebspeiseplanData:
+        """Download all data required by the Potsdam parser."""
+        return self.api.fetch_all()
+
+    def parse(
+        self,
+        config: dict[str, Canteen],
+        raw_data: WebspeiseplanData,
+    ) -> dict[str, CanteenXML]:
+        """Convert Potsdam Webspeiseplan data into canteen structures."""
+        parsed: dict[str, CanteenXML] = {}
+        for canteen_key, configured_canteen in config.items():
+            source_name = configured_canteen.name
+            if source_name not in raw_data.outlets:
+                self.logger.warning("%s not found in keys", source_name)
+                continue
+
+            outlet = dict(raw_data.outlets[source_name])
+            menus = raw_data.menus[source_name]
+            categories = raw_data.meal_categories[source_name]
+            locations = raw_data.locations[source_name]
+            outlet["isPublic"] = locations["isPublic"]
+
+            canteen = self.parser.parse_canteen_meta_times(outlet)
+            for meal_data in self.parser.parse_meals(menus, categories):
+                canteen.add_meal(**meal_data)
+            parsed[canteen_key] = canteen
+        return parsed

openmensa_parsers/parsers/registry.py

@@ -0,0 +1,25 @@
+"""Registry for city/source parser implementations."""
+
+from __future__ import annotations
+
+from openmensa_parsers.parsers.base import OpenMensaParser
+from openmensa_parsers.parsers.potsdam import PotsdamParser
+
+
+PARSER_CLASSES: dict[str, type[OpenMensaParser]] = {
+    PotsdamParser.id: PotsdamParser,
+}
+
+
+def get_parser_class(parser_id: str) -> type[OpenMensaParser]:
+    try:
+        return PARSER_CLASSES[parser_id]
+    except KeyError as exc:
+        configured = ", ".join(sorted(PARSER_CLASSES))
+        raise KeyError(
+            f"Unknown parser {parser_id!r}; configured parsers: {configured}"
+        ) from exc
+
+
+def create_parser(parser_id: str) -> OpenMensaParser:
+    return get_parser_class(parser_id)()