enforce utf8 on more file reads

archivebox/config.py

@@ -298,7 +298,7 @@ DYNAMIC_CONFIG_SCHEMA: ConfigDefaultDict = {
     'URL_BLACKLIST_PTN':        {'default': lambda c: c['URL_BLACKLIST'] and re.compile(c['URL_BLACKLIST'] or '', re.IGNORECASE | re.UNICODE | re.MULTILINE)},
 
     'ARCHIVEBOX_BINARY':        {'default': lambda c: sys.argv[0] or bin_path('archivebox')},
-    'VERSION':                  {'default': lambda c: json.loads((Path(c['PACKAGE_DIR']) / 'package.json').read_text().strip())['version']},
+    'VERSION':                  {'default': lambda c: json.loads((Path(c['PACKAGE_DIR']) / 'package.json').read_text(encoding='utf-8').strip())['version']},
 
     'PYTHON_BINARY':            {'default': lambda c: sys.executable},
     'PYTHON_ENCODING':          {'default': lambda c: sys.stdout.encoding.upper()},

archivebox/core/models.py

@@ -201,7 +201,7 @@ class Snapshot(models.Model):
     @cached_property
     def headers(self) -> Optional[dict]:
         try:
-            return json.loads((Path(self.link_dir) / 'headers.json').read_text().strip())
+            return json.loads((Path(self.link_dir) / 'headers.json').read_text(encoding='utf-8').strip())
         except Exception:
             pass
         return None