From 8899fe0b9259748da2ef19d37028c317f39f37d3 Mon Sep 17 00:00:00 2001 From: renaisun <43090234+renaisun@users.noreply.github.com> Date: Thu, 9 Jun 2022 14:35:48 +0800 Subject: [PATCH] Add SINGLEFILE_ARGS to control single-file arguments --- archivebox/config.py | 2 ++ archivebox/extractors/singlefile.py | 2 ++ 2 files changed, 4 insertions(+) diff --git a/archivebox/config.py b/archivebox/config.py index cfe41b53..bd3a1688 100644 --- a/archivebox/config.py +++ b/archivebox/config.py @@ -177,6 +177,7 @@ CONFIG_SCHEMA: Dict[str, ConfigDefaultDict] = { '--compressed' ]}, 'GIT_ARGS': {'type': list, 'default': ['--recursive']}, + 'SINGLEFILE_ARGS': {'type': list, 'default' : None} }, 'SEARCH_BACKEND_CONFIG' : { @@ -391,6 +392,7 @@ DYNAMIC_CONFIG_SCHEMA: ConfigDefaultDict = { 'USE_SINGLEFILE': {'default': lambda c: c['USE_SINGLEFILE'] and c['SAVE_SINGLEFILE']}, 'SINGLEFILE_VERSION': {'default': lambda c: bin_version(c['SINGLEFILE_BINARY']) if c['USE_SINGLEFILE'] else None}, + 'SINGLEFILE_ARGS': {'default': lambda c: c['SINGLEFILE_ARGS'] or []}, 'USE_READABILITY': {'default': lambda c: c['USE_READABILITY'] and c['SAVE_READABILITY']}, 'READABILITY_VERSION': {'default': lambda c: bin_version(c['READABILITY_BINARY']) if c['USE_READABILITY'] else None}, diff --git a/archivebox/extractors/singlefile.py b/archivebox/extractors/singlefile.py index 3279960e..80ad90b1 100644 --- a/archivebox/extractors/singlefile.py +++ b/archivebox/extractors/singlefile.py @@ -17,6 +17,7 @@ from ..config import ( SAVE_SINGLEFILE, DEPENDENCIES, SINGLEFILE_VERSION, + SINGLEFILE_ARGS, CHROME_BINARY, ) from ..logging_util import TimedProgress @@ -47,6 +48,7 @@ def save_singlefile(link: Link, out_dir: Optional[Path]=None, timeout: int=TIMEO browser_args = '--browser-args={}'.format(json.dumps(browser_args[1:])) cmd = [ DEPENDENCIES['SINGLEFILE_BINARY']['path'], + *SINGLEFILE_ARGS, '--browser-executable-path={}'.format(CHROME_BINARY), browser_args, link.url,