Brotli: Don't accept brotli if library can't be loaded. (#444)

* brotli: if the brotli module can not be loaded, print warning
and also remove `br` from any Accept-Encoding header to avoid recording with brotli, addresses #434

pywb/rewrite/rewriteinputreq.py

@@ -6,6 +6,14 @@
 import re
 
 
+try: # pragma: no cover
+    import brotli
+    has_brotli = True
+except Exception:  # pragma: no cover
+    has_brotli = False
+    print('Warning: brotli module could not be loaded, will not be able to replay brotli-encoded content')
+
+
 #=============================================================================
 class RewriteInputRequest(DirectWSGIInputRequest):
     RANGE_ARG_RX = re.compile('.*.googlevideo.com/videoplayback.*([&?]range=(\d+)-(\d+))')
@@ -79,6 +87,12 @@ def get_req_headers(self):
                 if self.splits:
                     value = self.splits.scheme
 
+            elif not has_brotli and name == 'HTTP_ACCEPT_ENCODING' and 'br' in value:
+                # if brotli not available, remove 'br' from accept-encoding to avoid
+                # capture brotli encoded content
+                name = 'Accept-Encoding'
+                value = ','.join([enc for enc in value.split(',') if enc.strip() != 'br'])
+
             elif name.startswith('HTTP_'):
                 name = name[5:].title().replace('_', '-')
 

tests/test_record_replay.py

@@ -3,10 +3,15 @@
 from pywb.manager.autoindex import AutoIndexer
 from pywb.warcserver.test.testutils import to_path, HttpBinLiveTests, TEST_WARC_PATH, TEST_CDX_PATH
 
+from warcio import ArchiveIterator
+
 import os
 import time
 import json
 
+from mock import patch
+import pytest
+
 
 # ============================================================================
 class TestRecordReplay(HttpBinLiveTests, CollsDirMixin, BaseConfigTest):
@@ -153,6 +158,20 @@ def test_init_and_rec(self):
         assert names[0].startswith('pywb-rec-test-')
         assert names[0].endswith('.warcgz')
 
+        TestRecordCustomConfig.warc_name = os.path.join(dir_name, names[0])
+
+    @patch('pywb.rewrite.rewriteinputreq.has_brotli', False)
+    def test_no_brotli(self):
+        res = self.testapp.get('/test-new/record/mp_/http://httpbin.org/get?C=D',
+                               headers={'Accept-Encoding': 'gzip, deflate, br'})
+        assert '"C": "D"' in res.text
+
+        with open(self.warc_name, 'rb') as fh:
+            for record in ArchiveIterator(fh):
+                last_record = record
+
+        assert record.http_headers['Accept-Encoding'] == 'gzip, deflate'
+
 
 # ============================================================================
 class TestRecordFilter(HttpBinLiveTests, CollsDirMixin, BaseConfigTest):
@@ -174,17 +193,17 @@ def setup_class(cls):
                      }
         super(TestRecordFilter, cls).setup_class('config_test_record.yaml', custom_config=rec_custom)
         manager(['init', 'test-new'])
-        
+
     def test_skip_existing(self):
         dir_name = os.path.join(self.root_dir, '_test_colls', 'test-new', 'archive')
         assert os.path.isdir(dir_name)
         res = self.testapp.get('/fallback/cdx?url=http://example.com/?example=1')
         assert res.text != ''
-        
+
         res = self.testapp.get('/test-new/record/mp_/http://example.com/?example=1')
         assert 'Example Domain' in res.text
         assert os.listdir(dir_name) == []
-    
+
     def test_record_new(self):
         dir_name = os.path.join(self.root_dir, '_test_colls', 'test-new', 'archive')
         assert os.path.isdir(dir_name)