feat: add RF4MAP and RF4 Posts research parsers

rf4_research/community_cli.py

@@ -6,7 +6,13 @@ import sys
 from dataclasses import asdict
 from urllib.request import Request, urlopen
 
-from .community_sources import parse_rf4db_catches, parse_rf4stat_fishing, parse_rf4stat_posts
+from .community_sources import (
+    parse_rf4db_catches,
+    parse_rf4map_point,
+    parse_rf4posts_spot,
+    parse_rf4stat_fishing,
+    parse_rf4stat_posts,
+)
 
 
 SOURCES = {
@@ -14,6 +20,10 @@ SOURCES = {
     "rf4stat-fishing": ("https://rf4-stat.ru/fishing/", parse_rf4stat_fishing),
     "rf4stat-posts": ("https://rf4-stat.ru/posts/", parse_rf4stat_posts),
 }
+DETAIL_SOURCES = {
+    "rf4map-point": parse_rf4map_point,
+    "rf4posts-spot": parse_rf4posts_spot,
+}
 USER_AGENT = "RF4-Spotter/0.1 (authorized data integration)"
 
 
@@ -27,14 +37,17 @@ def fetch_html(url: str, *, timeout: float = 30) -> str:
 
 def main(argv: list[str] | None = None) -> int:
     parser = argparse.ArgumentParser(description="Fetch one authorized RF4 community source page")
-    parser.add_argument("source", choices=SOURCES)
+    parser.add_argument("source", choices=(*SOURCES, *DETAIL_SOURCES))
     parser.add_argument("--url", help="Override the configured public page URL")
     parser.add_argument("--limit", type=int, default=100, choices=range(1, 501), metavar="1..500")
     args = parser.parse_args(argv)
-    default_url, parse = SOURCES[args.source]
+    if args.source in DETAIL_SOURCES and not args.url:
+        parser.error(f"--url is required for {args.source}")
+    default_url, parse = SOURCES.get(args.source, (None, DETAIL_SOURCES.get(args.source)))
     url = args.url or default_url
     try:
-        records = parse(fetch_html(url))[:args.limit]
+        html = fetch_html(url)
+        records = (parse(html, source_url=url) if args.source in DETAIL_SOURCES else parse(html))[:args.limit]
     except Exception as exc:
         print(f"community source failed: {exc}", file=sys.stderr)
         return 1