fix config file atomic writing bugs

archivebox/config/__init__.py

@@ -324,6 +324,8 @@ def load_config_file(out_dir: str=None) -> Optional[Dict[str, str]]:
 def write_config_file(config: Dict[str, str], out_dir: str=None) -> ConfigDict:
     """load the ini-formatted config file from OUTPUT_DIR/Archivebox.conf"""
 
+    from ..system import atomic_write
+
     out_dir = out_dir or os.path.abspath(os.getenv('OUTPUT_DIR', '.'))
     config_path = os.path.join(out_dir, CONFIG_FILENAME)
     
@@ -362,8 +364,9 @@ def write_config_file(config: Dict[str, str], out_dir: str=None) -> ConfigDict:
         else:
             config_file['SERVER_CONFIG'] = {'SECRET_KEY': random_secret_key}
 
-
-    atomic_write(config_path, '\n'.join((CONFIG_HEADER, config_file)))
+    with open(config_path, 'w+') as new:
+        config_file.write(new)
+    
     try:
         # validate the config by attempting to re-parse it
         CONFIG = load_all_config()

archivebox/extractors/__init__.py

@@ -45,7 +45,7 @@ def archive_link(link: Link, overwrite: bool=False, out_dir: Optional[str]=None)
         ('media', should_save_media, save_media),
         ('archive_org', should_save_archive_dot_org, save_archive_dot_org),
     )
-    
+
     out_dir = out_dir or link.link_dir
     try:
         is_new = not os.path.exists(out_dir)
@@ -61,7 +61,7 @@ def archive_link(link: Link, overwrite: bool=False, out_dir: Optional[str]=None)
             try:
                 if method_name not in link.history:
                     link.history[method_name] = []
-                
+
                 if should_run(link, out_dir) or overwrite:
                     log_archive_method_started(method_name)
 
@@ -83,7 +83,7 @@ def archive_link(link: Link, overwrite: bool=False, out_dir: Optional[str]=None)
 
         write_link_details(link, out_dir=link.link_dir)
         patch_main_index(link)
-        
+
         # # If any changes were made, update the main links index json and html
         # was_changed = stats['succeeded'] or stats['failed']
         # if was_changed:

archivebox/extractors/wget.py

@@ -78,6 +78,7 @@ def save_wget(link: Link, out_dir: Optional[str]=None, timeout: int=TIMEOUT) ->
         *([] if CHECK_SSL_VALIDITY else ['--no-check-certificate', '--no-hsts']),
         link.url,
     ]
+
     status = 'succeeded'
     timer = TimedProgress(timeout, prefix='      ')
     try:
@@ -111,7 +112,7 @@ def save_wget(link: Link, out_dir: Optional[str]=None, timeout: int=TIMEOUT) ->
                 raise ArchiveError('500 Internal Server Error', hints)
             raise ArchiveError('Got an error from the server', hints)
 
-        # chmod_file(output, cwd=out_dir)
+        chmod_file(output, cwd=out_dir)
     except Exception as err:
         status = 'failed'
         output = err

archivebox/main.py

@@ -882,7 +882,7 @@ def config(config_options_str: Optional[str]=None,
                 print('    {}'.format(printable_config(side_effect_changes, prefix='    ')))
         if failed_options:
             stderr()
-            stderr('[X] These options failed to set:', color='red')
+            stderr('[X] These options failed to set (check for typos):', color='red')
             stderr('    {}'.format('\n    '.join(failed_options)))
         raise SystemExit(bool(failed_options))
     elif reset:

archivebox/system.py

@@ -16,8 +16,7 @@ from .util import enforce_types, ExtendedEncoder
 from .config import OUTPUT_PERMISSIONS
 
 
-
-def run(*args, input=None, capture_output=True, text=False, timeout=None, check=False, **kwargs):
+def run(*args, input=None, capture_output=True, text=False, **kwargs):
     """Patched of subprocess.run to fix blocking io making timeout=innefective"""
 
     if input is not None:
@@ -29,12 +28,13 @@ def run(*args, input=None, capture_output=True, text=False, timeout=None, check=
             raise ValueError('stdout and stderr arguments may not be used '
                              'with capture_output.')
 
-    return subprocess_run(*args, input=input, capture_output=capture_output, text=text, timeout=timeout, check=check, **kwargs)
+    return subprocess_run(*args, input=input, capture_output=capture_output, text=text, **kwargs)
+
 
 @enforce_types
 def atomic_write(path: Union[Path, str], contents: Union[dict, str, bytes], overwrite: bool=True) -> None:
     """Safe atomic write to filesystem by writing to temp file + atomic rename"""
-    
+
     mode = 'wb+' if isinstance(contents, bytes) else 'w'
 
     # print('\n> Atomic Write:', mode, path, len(contents), f'overwrite={overwrite}')
@@ -44,8 +44,9 @@ def atomic_write(path: Union[Path, str], contents: Union[dict, str, bytes], over
         elif isinstance(contents, (bytes, str)):
             f.write(contents)
 
+
 @enforce_types
-def chmod_file(path: str, cwd: str='.', permissions: str=OUTPUT_PERMISSIONS, timeout: int=30) -> None:
+def chmod_file(path: str, cwd: str='.', permissions: str=OUTPUT_PERMISSIONS) -> None:
     """chmod -R <permissions> <cwd>/<path>"""
 
     root = Path(cwd) / path
@@ -93,6 +94,7 @@ def get_dir_size(path: str, recursive: bool=True, pattern: Optional[str]=None) -
 
 CRON_COMMENT = 'archivebox_schedule'
 
+
 @enforce_types
 def dedupe_cron_jobs(cron: CronTab) -> CronTab:
     deduped: Set[Tuple[str, str]] = set()