[archive] new site support

src/you_get/common.py

@@ -961,6 +961,7 @@ def url_to_module(url):
     from .extractors import (
         acfun,
         alive,
+        archive,
         baidu,
         baomihua,
         bilibili,
@@ -1040,6 +1041,7 @@ def url_to_module(url):
         '163': netease,
         '56': w56,
         'acfun': acfun,
+        'archive': archive,
         'baidu': baidu,
         'baomihua': baomihua,
         'bilibili': bilibili,

src/you_get/extractors/__init__.py

@@ -2,6 +2,7 @@
 
 from .acfun import *
 from .alive import *
+from .archive import *
 from .baidu import *
 from .bilibili import *
 from .blip import *

src/you_get/extractors/archive.py

@@ -0,0 +1,19 @@
+#!/usr/bin/env python
+
+__all__ = ['archive_download']
+
+from ..common import *
+
+def archive_download(url, output_dir='.', merge=True, info_only=False):
+    html = get_html(url)
+    title = r1(r'<meta property="og:title" content="([^"]*)"', html)
+    source = r1(r'<meta property="og:video" content="([^"]*)"', html)
+    mime, ext, size = url_info(source)
+
+    print_info(site_info, title, mime, size)
+    if not info_only:
+        download_urls([source], title, ext, size, output_dir, merge=merge)
+
+site_info = "Archive.org"
+download = archive_download
+download_playlist = playlist_not_supported('archive')