searchenginewatch.com.script

垂直搜索的网络爬虫

FIND_NODE	... #comment $value='-Article Body-'	[next][END_OF_DOCUMENT]
SAVE_POS	$content_start	[next][END_OF_DOCUMENT]
FIND_NODE	h1 ...	[next][END_OF_DOCUMENT]
STORE_TEXT	title	-1	-1	[next][END_OF_DOCUMENT]
FIND_NODE	... font	[next][END_OF_DOCUMENT]
SAVE_TEXT	$meta	-1	-1	[next][END_OF_DOCUMENT]
STORE_NODE	meta	$meta	[next][END_OF_DOCUMENT]
REGEXP	$meta	/By\s+(.+?),?\s+(\S+ \d+,\s*\d+)/	[next][GOTO_TASK	11]
STORE_NODE	author	$1	[next][END_OF_DOCUMENT]
STORE_ISODATE	iso-date	MMM dd',' yyyy	$2	[next][next]
FIND_NODE	... h2 $firstChildNode='#text' $firstChildNodeValue='Search Headlines'	[next][GOTO_TASK	13]
SAVE_POS	$content_end	[GOTO_TASK	15][next]
FIND_NODE	... #comment $value='content_stop'	[next][END_OF_DOCUMENT]
SAVE_POS	$content_end	[next][END_OF_DOCUMENT]
STORE_TEXT	content	$content_start	$content_end	[next][END_OF_DOCUMENT]
STORE_LINKS	$content_start	$content_end	[next][END_OF_DOCUMENT]
END_OF_ARTICLE		[next][END_OF_DOCUMENT]
GOTO_TASK	1	[][]