genius.search_song returns 403 Forbidden (#288)

* genius.search_song returns 403 Forbidden Fixes #287 and Closes #289 by making User-Agent configurable and adding optional proxy support. * Add optional user_agent argument * UPDATE add[proxy], remove[headers] * Add proxy docstring --------- Co-authored-by: John William Ruth Miller <[email protected]> Co-authored-by: Vauth <[email protected]>
johnwmillr · Dec 22, 2024 · 7635cff · 7635cff
1 parent 795a81b
commit 7635cff
Show file tree

Hide file tree

Showing 4 changed files with 23 additions and 3 deletions.
diff --git a/lyricsgenius/__init__.py b/lyricsgenius/__init__.py
@@ -14,4 +14,4 @@
 __url__ = 'https://github.com/johnwmillr/LyricsGenius'
 __description__ = 'A Python wrapper around the Genius API'
 __license__ = 'MIT'
-__version__ = '3.1.2'
+__version__ = '3.2.0'
diff --git a/lyricsgenius/api/api.py b/lyricsgenius/api/api.py
@@ -36,6 +36,8 @@ class API(Sender):
         sleep_time (:obj:`str`, optional): time to wait between requests.
         retries (:obj:`int`, optional): Number of retries in case of timeouts and
             errors with a >= 500 response code. By default, requests are only made once.
+        user_agent (:obj:`str`, optional): User agent for the request header.
+        proxy (:obj:`dict[str, str]`, optional): Proxy settings.
 
     Attributes:
         response_format (:obj:`str`, optional): API response format (dom, plain, html).
@@ -55,13 +57,17 @@ def __init__(self,
                  timeout=5,
                  sleep_time=0.2,
                  retries=0,
+                 user_agent='',
+                 proxy=None,
                  ):
         super().__init__(
             access_token=access_token,
             response_format=response_format,
             timeout=timeout,
             sleep_time=sleep_time,
             retries=retries,
+            user_agent=user_agent,
+            proxy=proxy,
         )
 
     def account(self, text_format=None):
@@ -524,6 +530,7 @@ def __init__(
         timeout=5,
         sleep_time=0.2,
         retries=0,
+        user_agent='',
         **kwargs
     ):
 
@@ -538,5 +545,6 @@ def __init__(
             sleep_time=sleep_time,
             retries=retries,
             public_api_constructor=public_api_constructor,
+            user_agent=user_agent,
             **kwargs
         )
diff --git a/lyricsgenius/api/base.py b/lyricsgenius/api/base.py
@@ -1,5 +1,6 @@
 import time
 import os
+import platform
 from json.decoder import JSONDecodeError
 
 import requests
@@ -21,12 +22,17 @@ def __init__(
         sleep_time=0.2,
         retries=0,
         public_api_constructor=False,
+        user_agent='',
+        proxy=None,
     ):
         self._session = requests.Session()
+        user_agent_root = f'{platform.system()} {platform.release()}; Python {platform.python_version()}'
         self._session.headers = {
             'application': 'LyricsGenius',
-            'User-Agent': 'https://github.com/johnwmillr/LyricsGenius'
+            'User-Agent': f'({user_agent}) ({user_agent_root})' if user_agent else user_agent_root,
         }
+        if proxy:
+            self._session.proxies = proxy
         if access_token is None:
             access_token = os.environ.get('GENIUS_ACCESS_TOKEN')
 

diff --git a/lyricsgenius/genius.py b/lyricsgenius/genius.py
@@ -36,6 +36,8 @@ class Genius(API, PublicAPI):
             excluded terms with user's. Default excluded terms are listed below.
         retries (:obj:`int`, optional): Number of retries in case of timeouts and
             errors with a >= 500 response code. By default, requests are only made once.
+        user_agent (:obj:`str`, optional): User agent for the request header.
+        proxy (:obj:`dict[str, str]`, optional): Proxy settings.
 
     Attributes:
         verbose (:obj:`bool`, optional): Turn printed messages on or off.
@@ -71,14 +73,18 @@ def __init__(self, access_token=None,
                  skip_non_songs=True, excluded_terms=None,
                  replace_default_terms=False,
                  retries=0,
+                 user_agent='',
+                 proxy=None,
                  ):
         # Genius Client Constructor
         super().__init__(
             access_token=access_token,
             response_format=response_format,
             timeout=timeout,
             sleep_time=sleep_time,
-            retries=retries
+            retries=retries,
+            user_agent=user_agent,
+            proxy=proxy,
         )
 
         self.verbose = verbose