Merge branch 'main' into upgrade-dependencies

README.rst

@@ -60,19 +60,15 @@ Installation for Deployment
 
 To install pywb for usage, you can use:
 
-```shell
-pip install pywb
-```
+``pip install pywb``
 
 Note: depending on your Python installation, you may have to use `pip3` instead of `pip`.
 
 
 Installation from local copy
 ----------------------------
 
-```shell
-git clone https://github.com/webrecorder/pywb
-```
+``git clone https://github.com/webrecorder/pywb``
 
 To install from a locally cloned copy, install with ``pip install -e .`` or ``python setup.py install``.
 

pywb/apps/frontendapp.py

@@ -667,10 +667,12 @@ def handle_request(self, environ, start_response):
             # store original script_name (original prefix) before modifications are made
             environ['ORIG_SCRIPT_NAME'] = environ.get('SCRIPT_NAME')
 
-            lang = args.pop('lang', self.default_locale)
+            lang = args.pop('lang', '')
             if lang:
                 shift_path_info(environ)
                 environ['pywb_lang'] = lang
+            elif self.default_locale:
+                environ['pywb_lang'] = self.default_locale
 
             response = endpoint(environ, **args)
 

pywb/manager/manager.py

@@ -12,7 +12,7 @@
 from pkg_resources import resource_string, get_distribution
 
 from argparse import ArgumentParser, RawTextHelpFormatter
-from tempfile import mkdtemp
+from tempfile import mkdtemp, TemporaryDirectory
 from zipfile import ZipFile
 
 from pywb.utils.loaders import load_yaml_config
@@ -213,35 +213,35 @@ def _add_wacz_uncompressed(self, wacz):
         # delete temporary files
         shutil.rmtree(temp_dir)
 
-    @staticmethod
-    def _add_wacz_index(collection_index_path, wacz_index_path, filename_mapping):
+    def _add_wacz_index(self, collection_index_path, wacz_index_path, filename_mapping):
         from pywb.warcserver.index.cdxobject import CDXObject
 
-        # copy collection index to temporary directory
-        tempdir = mkdtemp()
-        collection_index_name = os.path.basename(collection_index_path)
-        collection_index_temp_path = os.path.join(tempdir, collection_index_name)
-
-        if os.path.exists(collection_index_path):
-            shutil.copy2(collection_index_path, collection_index_temp_path)
+        # rewrite wacz index to temporary index file
+        tempdir = TemporaryDirectory()
+        wacz_index_name = os.path.basename(wacz_index_path)
+        rewritten_index_path = os.path.join(tempdir.name, wacz_index_name)
 
-        with open(collection_index_temp_path, 'a') as collection_index_temp_file:
+        with open(rewritten_index_path, 'w') as rewritten_index:
             if wacz_index_path.endswith('.gz'):
-                wacz_index_file = gzip.open(wacz_index_path, 'rb')
+                wacz_index = gzip.open(wacz_index_path, 'rb')
             else:
-                wacz_index_file = open(wacz_index_path, 'rb')
-            collection_index_temp_file.write('\n')
-            for line in wacz_index_file.readlines():
+                wacz_index = open(wacz_index_path, 'rb')
+
+            for line in wacz_index:
                 cdx_object = CDXObject(cdxline=line)
                 if cdx_object['filename'] in filename_mapping:
                     cdx_object['filename'] = filename_mapping[cdx_object['filename']]
-                collection_index_temp_file.write(cdx_object.to_cdxj())
+                rewritten_index.write(cdx_object.to_cdxj())
+
+        if not os.path.isfile(collection_index_path):
+            shutil.move(rewritten_index_path, collection_index_path)
+            return
 
-            wacz_index_file.close()
+        temp_coll_index_path = collection_index_path + '.tmp.' + timestamp20_now()
+        self._merge_indices(collection_index_path, rewritten_index_path, temp_coll_index_path)
+        shutil.move(temp_coll_index_path, collection_index_path)
 
-        # copy temporary index back to original location and delete temporary directory
-        shutil.move(collection_index_temp_path, collection_index_path)
-        shutil.rmtree(tempdir)
+        tempdir.cleanup()
 
     def reindex(self):
         cdx_file = os.path.join(self.indexes_dir, self.DEF_INDEX_FILE)
@@ -294,20 +294,24 @@ def _index_merge_warcs(self, new_warcs, index_file, rel_root=None):
 
         merged_file = temp_file + '.merged'
 
-        last_line = None
-
-        with open(cdx_file, 'rb') as orig_index:
-            with open(temp_file, 'rb') as new_index:
-                with open(merged_file, 'w+b') as merged:
-                    for line in heapq.merge(orig_index, new_index):
-                        if last_line != line:
-                            merged.write(line)
-                            last_line = line
+        self._merge_indices(cdx_file, temp_file, merged_file)
 
         shutil.move(merged_file, cdx_file)
         #os.rename(merged_file, cdx_file)
         os.remove(temp_file)
 
+    @staticmethod
+    def _merge_indices(index1, index2, dest):
+        last_line = None
+
+        with open(index1, 'rb') as index1_f:
+            with open(index2, 'rb') as index2_f:
+                with open(dest, 'wb') as dest_f:
+                    for line in heapq.merge(index1_f, index2_f):
+                        if last_line != line:
+                            dest_f.write(line)
+                            last_line = line
+
     def set_metadata(self, namevalue_pairs):
         metadata_yaml = os.path.join(self.curr_coll_dir, 'metadata.yaml')
         metadata = None

pywb/rewrite/templateview.py

@@ -178,7 +178,7 @@ def switch_locale(context, locale):
 
             request_uri = environ.get('REQUEST_URI', environ.get('PATH_INFO'))
 
-            if curr_loc:
+            if curr_loc and request_uri.startswith('/' + curr_loc + '/'):
                 return request_uri.replace(curr_loc, locale, 1)
 
             app_prefix = environ.get('pywb.app_prefix', '')
@@ -196,11 +196,11 @@ def get_locale_prefixes(context):
             orig_prefix = environ.get('pywb.app_prefix', '')
             coll = environ.get('SCRIPT_NAME', '')
 
-            if orig_prefix:
+            if orig_prefix and coll.startswith(orig_prefix):
                 coll = coll[len(orig_prefix):]
 
             curr_loc = environ.get('pywb_lang', '')
-            if curr_loc:
+            if curr_loc and coll.startswith('/' + curr_loc):
                 coll = coll[len(curr_loc) + 1:]
 
             for locale in loc_map.keys():

pywb/templates/error.html

@@ -3,7 +3,7 @@
 {% block body %}
 <div class="container text-danger error">
     <div class="row justify-content-center">
-        <h2 class="display-2">Pywb Error</h2>
+        <h2 class="display-2">{{ _('Pywb Error') }}</h2>
     </div>
     <div class="row">
         <div class="col-12 text-center">

tests/test_manager.py

@@ -75,10 +75,15 @@ def test_merge_wacz_index(self, tmp_path):
                                 {'example.warc.gz': 'rewritten.warc.gz'})
         with open(os.path.join(manager.indexes_dir, manager.DEF_INDEX_FILE), 'r') as f:
             index_content = f.read()
+            index_content = index_content.strip()
 
         assert 'example.warc.gz' not in index_content
         assert 'rewritten.warc.gz' in index_content
 
+        # check that collection index is sorted
+        index_lines = index_content.split('\n')
+        assert sorted(index_lines) == index_lines
+
     def test_merge_wacz_index_gzip(self, tmp_path):
         manager = self.get_test_collections_manager(tmp_path)
         manager._add_wacz_index(os.path.join(manager.indexes_dir, manager.DEF_INDEX_FILE),