Adds a User-agent header to curl module requests

peru/resources/plugins/curl/curl_plugin.py

@@ -9,10 +9,26 @@
 import tarfile
 from urllib.error import HTTPError, URLError
 from urllib.parse import urlsplit
+from urllib.request import Request
+import peru.main
 import urllib.request
 import zipfile
 
 
+def add_user_agent_to_request(request):
+    components = [
+        "peru/%s" % peru.main.get_version(),
+        urllib.request.URLopener.version
+    ]
+    request.add_header("User-agent", " ".join(components))
+    return request
+
+
+def build_request(url):
+    request = Request(url)
+    return add_user_agent_to_request(request)
+
+
 def get_request_filename(request):
     '''Figure out the filename for an HTTP download.'''
     # Check to see if a filename is specified in the HTTP headers.
@@ -80,7 +96,7 @@ def plugin_sync(url, sha1):
         # Download directly to the destination dir.
         download_dir = dest
 
-    with urllib.request.urlopen(url) as request:
+    with urllib.request.urlopen(build_request(url)) as request:
         filename = os.environ['PERU_MODULE_FILENAME']
         if not filename:
             filename = get_request_filename(request)
@@ -151,7 +167,7 @@ def __init__(self, message):
 
 def plugin_reup(url, sha1):
     reup_output = os.environ['PERU_REUP_OUTPUT']
-    with urllib.request.urlopen(url) as request:
+    with urllib.request.urlopen(build_request(url)) as request:
         digest = download_file(request, None)
     with open(reup_output, 'w') as output_file:
         print('sha1:', digest, file=output_file)

tests/test_curl_plugin.py

@@ -2,6 +2,7 @@
 import importlib.machinery
 import io
 from os.path import abspath, join, dirname
+import urllib
 
 import peru
 import shared
@@ -100,3 +101,14 @@ def test_evil_archives(self):
             tar_archive = shared.test_resources / (case + '.tar')
             with self.assertRaises(curl_plugin.EvilArchiveError):
                 curl_plugin.extract_tar(str(tar_archive), dest)
+
+    def test_request_has_user_agent_header(self):
+        actual = curl_plugin.build_request("http://example.test")
+        print(actual.header_items())
+        self.assertTrue(actual.has_header("User-agent"))
+        ua_header = actual.get_header("User-agent")
+        peru_component, urllib_component = ua_header.split(' ')
+        _, peru_version = peru_component.split('/')
+        _, urllib_version = urllib_component.split('/')
+        self.assertEqual(peru.main.get_version(), peru_version)
+        self.assertEqual(urllib.request.__version__, urllib_version)