dyn collection and all coll improvements: (#69)

support dynamic collections, all collection with remote archives (eg. s3:// paths)
- warcserver: allow custom dynamic collections index and archive path templates via 'dyn_index_path' and 'dyn_archive_path'
- pathresolver: allow resolving wildcard path prefixes with collection, to support remote paths and avoid globbing
- warcserver: don't add fixed collections dir to source to support resolving wildcard
- pathresolver: add wildcard resolving s3 path test
- referrer unrewrite: ensure referrer not empty

pywb/apps/rewriterapp.py

@@ -556,8 +556,9 @@ def unrewrite_referrer(self, environ, full_prefix):
 
         if referrer.startswith(full_prefix):
             referrer = referrer[len(full_prefix):]
-            environ['HTTP_REFERER'] = WbUrl(referrer).url
-            return True
+            if referrer:
+                environ['HTTP_REFERER'] = WbUrl(referrer).url
+                return True
 
         return False
 

pywb/warcserver/resource/pathresolvers.py

@@ -33,13 +33,18 @@ def __init__(self, template):
 
     def __call__(self, filename, cdx):
         full_path = self.template
+
         if hasattr(cdx, '_formatter') and cdx._formatter:
             full_path = cdx._formatter.format(full_path)
 
         path = full_path + filename
         if '*' not in path:
             return path
 
+        res_path = self.resolve_coll(path, cdx.get('source'))
+        if res_path:
+            return res_path
+
         if '://' in path:
             return path
 
@@ -49,6 +54,13 @@ def __call__(self, filename, cdx):
         else:
             return path
 
+    def resolve_coll(self, path, source):
+        if not source:
+            return
+
+        coll = source.split('/', 1)[0]
+        return path.replace('*', coll)
+
     def __repr__(self):
         return "PrefixResolver('{0}')".format(self.template)
 

pywb/warcserver/resource/test/test_pathresolvers.py

@@ -30,6 +30,15 @@ def test_resolver_dir_wildcard(self):
         assert len(res) == 1
         assert res[0] == os.path.join(get_test_dir(), 'warcs', 'example.warc.gz')
 
+    def test_resolver_dir_wildcard_with_coll(self):
+        resolver = DefaultResolverMixin.make_best_resolver('s3://bucket/colls/*/archives/')
+
+        cdx = CDXObject()
+        cdx['source'] = 'my-coll/indexes/index.cdxj'
+
+        res = resolver('example.warc.gz', cdx)
+        assert res == 's3://bucket/colls/my-coll/archives/example.warc.gz'
+
     def test_resolver_dir_wildcard_as_file_url(self):
         url = to_file_url(get_test_dir()) +  '/*/'
         resolver = DefaultResolverMixin.make_best_resolver(url)

pywb/warcserver/warcserver.py

@@ -84,14 +84,14 @@ def load_auto_colls(self):
             print('No Root Dir, Skip Auto Colls!')
             return
 
-        #indexes_templ = os.path.join('{coll}', 'indexes') + os.path.sep
-        self.indexes_templ = self.AUTO_DIR_INDEX_PATH.replace('/', os.path.sep)
+        self.indexes_templ = self.config.get('dyn_index_path', self.AUTO_DIR_INDEX_PATH).replace('/', os.path.sep)
+
         dir_source = CacheDirectoryIndexSource(base_prefix=self.root_dir,
-                                               base_dir=self.indexes_templ,
-                                               name=self.root_dir)
+                                               base_dir=self.indexes_templ)
 
-        self.archive_templ = self.AUTO_DIR_ARCHIVE_PATH.replace('/', os.path.sep)
-        self.archive_templ = os.path.join(self.root_dir, self.archive_templ)
+        self.archive_templ = self.config.get('dyn_archive_path', self.AUTO_DIR_ARCHIVE_PATH).replace('/', os.path.sep)
+        if '://' not in self.archive_templ:
+            self.archive_templ = os.path.join(self.root_dir, self.archive_templ)
 
         handler = DefaultResourceHandler(dir_source, self.archive_templ)
 

tests/test_record_replay.py

@@ -93,9 +93,9 @@ def test_cdx_all_coll(self):
         assert cdxj_lines[1]['url'] == 'http://httpbin.org/get?C=D'
         assert cdxj_lines[2]['url'] == 'http://httpbin.org/get?C=D'
 
-        assert cdxj_lines[0]['source'] == to_path('_test_colls:test/indexes/autoindex.cdxj')
-        assert cdxj_lines[1]['source'] == to_path('_test_colls:test2/indexes/autoindex.cdxj')
-        assert cdxj_lines[2]['source'] == to_path('_test_colls:test/indexes/autoindex.cdxj')
+        assert cdxj_lines[0]['source'] == to_path('test/indexes/autoindex.cdxj')
+        assert cdxj_lines[1]['source'] == to_path('test2/indexes/autoindex.cdxj')
+        assert cdxj_lines[2]['source'] == to_path('test/indexes/autoindex.cdxj')
 
         assert cdxj_lines[0]['filename'] == cdxj_lines[2]['filename']
 
@@ -104,7 +104,7 @@ def test_timemap_all_coll(self):
         link_lines = res.text.rstrip().split('\n')
         assert len(link_lines) == 5
 
-        assert to_path('_test_colls:test2/indexes/autoindex.cdxj') in link_lines[3]
-        assert to_path('_test_colls:test/indexes/autoindex.cdxj') in link_lines[4]
+        assert to_path('test2/indexes/autoindex.cdxj') in link_lines[3]
+        assert to_path('test/indexes/autoindex.cdxj') in link_lines[4]