beatport: estrazione __NEXT_DATA__ + eccezioni tipizzate

core/beatport.py

@@ -7,6 +7,8 @@ Vedi docs/superpowers/specs/2026-07-14-beatport-charts-design.md.
 
 from __future__ import annotations
 
+import json
+import re
 from dataclasses import dataclass
 
 
@@ -33,6 +35,35 @@ class BeatportTrack:
         return f"{self.artists} {self.title}"
 
 
+class BeatportError(Exception):
+    """Base per errori Beatport."""
+
+
+class BeatportUnreachableError(BeatportError):
+    """Rete o server Beatport non raggiungibile / 5xx."""
+
+
+class BeatportParseError(BeatportError):
+    """HTML/JSON ricevuto ma non conforme allo schema atteso."""
+
+
+_NEXT_DATA_RE = re.compile(
+    r'<script id="__NEXT_DATA__"[^>]*>(.+?)</script>',
+    re.DOTALL,
+)
+
+
+def _extract_next_data(html: str) -> dict:
+    """Estrae il payload JSON dallo script <__NEXT_DATA__> di Next.js."""
+    m = _NEXT_DATA_RE.search(html)
+    if not m:
+        raise BeatportParseError("__NEXT_DATA__ non trovato nella pagina Beatport")
+    try:
+        return json.loads(m.group(1))
+    except json.JSONDecodeError as e:
+        raise BeatportParseError(f"__NEXT_DATA__ JSON malformato: {e}") from e
+
+
 # Mappa slug URL Beatport -> (numeric_id, display_name)
 # Enumerata via scripts/refresh_beatport_genres.py (Task 1).
 # Copia dell'output di /tmp/genres_output.txt (31 generi).

tests/test_beatport.py

@@ -66,3 +66,20 @@ class TestBeatportTrack:
         )
         with pytest.raises(Exception):
             t.title = "Z"  # frozen=True impedisce mutazione
+
+
+class TestExtractNextData:
+    def test_extracts_json_from_valid_html(self, fixtures_dir):
+        html = (fixtures_dir / "beatport_melodic_top100.html").read_text()
+        data = beatport._extract_next_data(html)
+        assert isinstance(data, dict)
+        assert "props" in data
+
+    def test_missing_script_raises(self):
+        with pytest.raises(beatport.BeatportParseError, match="__NEXT_DATA__ non trovato"):
+            beatport._extract_next_data("<html><body>nulla</body></html>")
+
+    def test_malformed_json_raises(self):
+        broken = '<script id="__NEXT_DATA__" type="application/json">{not: valid}</script>'
+        with pytest.raises(beatport.BeatportParseError, match="JSON malformato"):
+            beatport._extract_next_data(broken)