feat: add SoundCloud client with client_id extraction

Made-with: Cursor
2026-03-12 01:36:03 -04:00
parent d2ab4e94c0
commit 49846f9d7e
2 changed files with 104 additions and 0 deletions
--- a/src/ntr_fetcher/soundcloud.py
+++ b/src/ntr_fetcher/soundcloud.py
@@ -0,0 +1,54 @@
+import json
+import logging
+import re
+from datetime import datetime
+from urllib.parse import parse_qs, urlparse
+
+import httpx
+
+from ntr_fetcher.models import Track
+
+logger = logging.getLogger(__name__)
+
+SOUNDCLOUD_BASE = "https://soundcloud.com"
+API_BASE = "https://api-v2.soundcloud.com"
+HYDRATION_PATTERN = re.compile(r"__sc_hydration\s*=\s*(\[.*?\])\s*;", re.DOTALL)
+
+
+def _build_cursor(until: datetime, user_id: int) -> str:
+    ts = until.strftime("%Y-%m-%dT%H:%M:%S.000Z")
+    padded_user = str(user_id).zfill(22)
+    return f"{ts},user-track-likes,000-{padded_user}-99999999999999999999"
+
+
+class SoundCloudClient:
+    def __init__(self, http_client: httpx.AsyncClient | None = None):
+        self._http = http_client or httpx.AsyncClient(timeout=15.0)
+        self._client_id: str | None = None
+
+    async def _extract_client_id(self) -> str:
+        if self._client_id is not None:
+            return self._client_id
+
+        resp = await self._http.get(SOUNDCLOUD_BASE)
+        resp.raise_for_status()
+        match = HYDRATION_PATTERN.search(resp.text)
+        if not match:
+            raise ValueError("Could not find __sc_hydration in SoundCloud HTML — cannot extract client_id")
+
+        hydration = json.loads(match.group(1))
+        for entry in hydration:
+            if entry.get("hydratable") == "apiClient":
+                self._client_id = entry["data"]["id"]
+                is_expiring = entry["data"].get("isExpiring", False)
+                if is_expiring:
+                    logger.warning("SoundCloud client_id is marked as expiring")
+                return self._client_id
+
+        raise ValueError("No apiClient entry in __sc_hydration — cannot extract client_id")
+
+    def invalidate_client_id(self) -> None:
+        self._client_id = None
+
+    async def close(self) -> None:
+        await self._http.aclose()
--- a/tests/test_soundcloud.py
+++ b/tests/test_soundcloud.py
@@ -0,0 +1,50 @@
+import re
+
+import pytest
+import httpx
+
+from ntr_fetcher.soundcloud import SoundCloudClient
+
+
+FAKE_HTML = """
+<html><head><script>
+window.__sc_hydration = [
+  {"hydratable": "user", "data": {}},
+  {"hydratable": "apiClient", "data": {"id": "test_client_id_abc123", "isExpiring": false}}
+];
+</script></head></html>
+"""
+
+FAKE_HTML_EXPIRING = """
+<html><head><script>
+window.__sc_hydration = [
+  {"hydratable": "apiClient", "data": {"id": "expiring_id_xyz", "isExpiring": true}}
+];
+</script></head></html>
+"""
+
+
+@pytest.mark.asyncio
+async def test_extract_client_id(httpx_mock):
+    httpx_mock.add_response(url="https://soundcloud.com", text=FAKE_HTML)
+    client = SoundCloudClient()
+    client_id = await client._extract_client_id()
+    assert client_id == "test_client_id_abc123"
+
+
+@pytest.mark.asyncio
+async def test_extract_client_id_caches(httpx_mock):
+    httpx_mock.add_response(url="https://soundcloud.com", text=FAKE_HTML)
+    client = SoundCloudClient()
+    id1 = await client._extract_client_id()
+    id2 = await client._extract_client_id()
+    assert id1 == id2
+    assert len(httpx_mock.get_requests()) == 1
+
+
+@pytest.mark.asyncio
+async def test_extract_client_id_bad_html(httpx_mock):
+    httpx_mock.add_response(url="https://soundcloud.com", text="<html>no hydration here</html>")
+    client = SoundCloudClient()
+    with pytest.raises(ValueError, match="client_id"):
+        await client._extract_client_id()