瀏覽代碼

Merge pull request #193 from sbrl/patch-1

wget: Remove unsupported compression option
Nick Sweeting 6 年之前
父節點
當前提交
e797a7a5f3
共有 2 個文件被更改,包括 3 次插入1 次删除
  1. 2 1
      archivebox/archive_methods.py
  2. 1 0
      archivebox/config.py

+ 2 - 1
archivebox/archive_methods.py

@@ -33,6 +33,7 @@ from config import (
     WGET_USER_AGENT,
     CHECK_SSL_VALIDITY,
     COOKIES_FILE,
+    WGET_AUTO_COMPRESSION
 )
 from util import (
     domain,
@@ -224,10 +225,10 @@ def fetch_wget(link_dir, link, timeout=TIMEOUT):
         '--backup-converted',
         '--span-hosts',
         '--no-parent',
-        '--compression=auto',
         '-e', 'robots=off',
         '--restrict-file-names=unix',
         '--timeout={}'.format(timeout),
+        *(('--compression=auto',) if WGET_AUTO_COMPRESSION else ()),
         *(() if FETCH_WARC else ('--timestamping',)),
         *(('--warc-file={}'.format(warc_path),) if FETCH_WARC else ()),
         *(('--page-requisites',) if FETCH_WGET_REQUISITES else ()),

+ 1 - 0
archivebox/config.py

@@ -72,6 +72,7 @@ TEMPLATES_DIR = os.path.join(PYTHON_PATH, 'templates')
 CHROME_SANDBOX = os.getenv('CHROME_SANDBOX', 'True').lower() == 'true'
 USE_CHROME = FETCH_PDF or FETCH_SCREENSHOT or FETCH_DOM
 USE_WGET = FETCH_WGET or FETCH_WGET_REQUISITES or FETCH_WARC
+WGET_AUTO_COMPRESSION = USE_WGET and WGET_BINARY and (not run([WGET_BINARY, "--compression=auto", "--help"], stdout=DEVNULL).returncode)
 
 ########################### Environment & Dependencies #########################