[refactor] soundcloud engine

Closes: https://github.com/searxng/searxng/issues/4226
Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>

1 files changed, 112 insertions, 68 deletions

diff --git a/searx/engines/soundcloud.py b/searx/engines/soundcloud.py
index 3281ea398..3fee87b27 100644
--- a/searx/engines/soundcloud.py
+++ b/searx/engines/soundcloud.py
@@ -1,102 +1,146 @@
 # SPDX-License-Identifier: AGPL-3.0-or-later
-"""
- Soundcloud (Music)
-"""
+"""SoundCloud is a German audio streaming service."""
 
 import re
-from json import loads
 from urllib.parse import quote_plus, urlencode
-from lxml import html
+import datetime
+
 from dateutil import parser
+from lxml import html
+
 from searx.network import get as http_get
 
-# about
 about = {
-    "website": 'https://soundcloud.com',
-    "wikidata_id": 'Q568769',
-    "official_api_documentation": 'https://developers.soundcloud.com/',
-    "use_official_api": True,
+    "website": "ttps://soundcloud.com",
+    "wikidata_id": "Q568769",
+    "official_api_documentation": "https://developers.soundcloud.com/docs/api/guide",
+    "use_official_api": False,
     "require_api_key": False,
     "results": 'JSON',
 }
 
-# engine dependent config
-categories = ['music']
+categories = ["music"]
 paging = True
 
-# search-url
-# missing attribute: user_id, app_version, app_locale
-url = 'https://api-v2.soundcloud.com/'
-search_url = (
-    url + 'search?{query}'
-    '&variant_ids='
-    '&facet=model'
-    '&limit=20'
-    '&offset={offset}'
-    '&linked_partitioning=1'
-    '&client_id={client_id}'
-)  # noqa
+search_url = "https://api-v2.soundcloud.com/search"
+"""This is not the offical (developer) url, it is the API which is used by the
+HTML frontend of the common WEB site.
+"""
 
 cid_re = re.compile(r'client_id:"([^"]*)"', re.I | re.U)
-guest_client_id = ''
+guest_client_id = ""
+results_per_page = 10
+
+soundcloud_facet = "model"
+
+app_locale_map = {
+    "de": "de",
+    "en": "en",
+    "es": "es",
+    "fr": "fr",
+    "oc": "fr",
+    "it": "it",
+    "nl": "nl",
+    "pl": "pl",
+    "szl": "pl",
+    "pt": "pt_BR",
+    "pap": "pt_BR",
+    "sv": "sv",
+}
 
 
-def get_client_id():
-    resp = http_get("https://soundcloud.com")
+def request(query, params):
+
+    # missing attributes: user_id, app_version
+    # - user_id=451561-497874-703312-310156
+    # - app_version=1740727428
 
-    if resp.ok:
-        tree = html.fromstring(resp.content)
-        # script_tags has been moved from /assets/app/ to /assets/ path.  I
-        # found client_id in https://a-v2.sndcdn.com/assets/49-a0c01933-3.js
-        script_tags = tree.xpath("//script[contains(@src, '/assets/')]")
-        app_js_urls = [script_tag.get('src') for script_tag in script_tags if script_tag is not None]
+    args = {
+        "q": query,
+        "offset": (params['pageno'] - 1) * results_per_page,
+        "limit": results_per_page,
+        "facet": soundcloud_facet,
+        "client_id": guest_client_id,
+        "app_locale": app_locale_map.get(params["language"].split("-")[0], "en"),
+    }
+
+    params['url'] = f"{search_url}?{urlencode(args)}"
+    return params
+
+
+def response(resp):
+    results = []
+    data = resp.json()
+
+    for result in data.get("collection", []):
+
+        if result["kind"] in ("track", "playlist"):
+            url = result.get("permalink_url")
+            if not url:
+                continue
+            uri = quote_plus(result.get("uri"))
+            content = [
+                result.get("description"),
+                result.get("label_name"),
+            ]
+            res = {
+                "url": url,
+                "title": result["title"],
+                "content": " / ".join([c for c in content if c]),
+                "publishedDate": parser.parse(result["last_modified"]),
+                "iframe_src": "https://w.soundcloud.com/player/?url=" + uri,
+                "views": result.get("likes_count"),
+            }
+            thumbnail = result["artwork_url"] or result["user"]["avatar_url"]
+            res["thumbnail"] = thumbnail or None
+            length = int(result.get("duration", 0) / 1000)
+            if length:
+                length = datetime.timedelta(seconds=length)
+                res["length"] = length
+            res["views"] = result.get("playback_count", 0) or None
+            res["author"] = result.get("user", {}).get("full_name") or None
+            results.append(res)
 
-        # extracts valid app_js urls from soundcloud.com content
-        for app_js_url in app_js_urls[::-1]:
-            # gets app_js and searches for the clientid
-            resp = http_get(app_js_url)
-            if resp.ok:
-                cids = cid_re.search(resp.content.decode())
-                if cids is not None and len(cids.groups()):
-                    return cids.groups()[0]
-    logger.warning("Unable to fetch guest client_id from SoundCloud, check parser!")
-    return ""
+    return results
 
 
 def init(engine_settings=None):  # pylint: disable=unused-argument
     global guest_client_id  # pylint: disable=global-statement
-    # api-key
     guest_client_id = get_client_id()
 
 
-# do search-request
-def request(query, params):
-    offset = (params['pageno'] - 1) * 20
+def get_client_id() -> str:
 
-    params['url'] = search_url.format(query=urlencode({'q': query}), offset=offset, client_id=guest_client_id)
+    client_id = ""
+    url = "https://soundcloud.com"
+    resp = http_get(url, timeout=10)
 
-    return params
+    if not resp.ok:
+        logger.error("init: GET %s failed", url)
+        return client_id
 
+    tree = html.fromstring(resp.content)
+    script_tags = tree.xpath("//script[contains(@src, '/assets/')]")
+    app_js_urls = [tag.get("src") for tag in script_tags if tag is not None]
 
-def response(resp):
-    results = []
-    search_res = loads(resp.text)
+    # extracts valid app_js urls from soundcloud.com content
 
-    # parse results
-    for result in search_res.get('collection', []):
+    for url in app_js_urls[::-1]:
 
-        if result['kind'] in ('track', 'playlist'):
-            uri = quote_plus(result['uri'])
-            res = {
-                'url': result['permalink_url'],
-                'title': result['title'],
-                'content': result['description'] or '',
-                'publishedDate': parser.parse(result['last_modified']),
-                'iframe_src': "https://w.soundcloud.com/player/?url=" + uri,
-            }
-            thumbnail = result['artwork_url'] or result['user']['avatar_url']
-            if thumbnail:
-                res['thumbnail'] = thumbnail
-            results.append(res)
+        # gets app_js and search for the client_id
+        resp = http_get(url)
 
-    return results
+        if not resp.ok:
+            logger.error("init: app_js GET %s failed", url)
+            continue
+
+        cids = cid_re.search(resp.content.decode())
+        if cids and len(cids.groups()):
+            client_id = cids.groups()[0]
+            break
+
+    if client_id:
+        logger.info("using client_id '%s' for soundclud queries", client_id)
+    else:
+        logger.warning("missing valid client_id for soundclud queries")
+    return client_id