def test_hackernews(self):
from sasoup.baserules import xpath, xpaths, xpathz, search, dpath, base, addon, fields, which
hacker_news = {
'fields_rules': {
'news': xpaths("//body/center/table/tr", evalx="result"),
},
'result_rules': {
'title': xpath("//span[@class='pagetop']/b/a"),
'google': search(r'\>(Google.+?)\<'),
'html5': search(r'\>(HTML5.+?)\<'),
'facebook': xpath("//a[contains(text(),'Facebook')]"),
'titles': xpathz(
'news',
xpaths(".//td[@class='title']/a")),
'tsmps': xpathz(
'news',
xpaths(".//td[@class='subtext']/node()[4]")),
'points': xpathz(
'news',
xpaths(".//td[@class='subtext']/span")),
},
'result_filters': {
'titles': (None, "result[2]"),
'tsmps': (None, "result[2]"),
'points': (None, "result[2]"),
}
}
url = 'https://news.ycombinator.com/'
html = url_get(url)
results = dict(Parser(html, hacker_news, 'utf-8').parse())
for key, result in results.items():
print '{} : {}'.format(key, result)
"shopUrl": which(
xpath("//a[contains(@class, 'seller-name')]/@href"), xpath("//a[contains(@class,'enter-shop')]/@href")
),
"itemImg": xpath("//img[@id='J_ImgBooth']/@data-src"),
"itemTitle": fields("itemViewed", dpath("['title']")),
"initPrice": fields("itemViewed", dpath("['price']")),
"promoInfo": None,
"postageInfo": None,
"monthlyTrade": ajax("apiItemInfo", dpath("['quantity']['quanity']")),
"itemRate": None,
"bonus": None,
"favNum": ajax("saveCounts", dpath("['{apiItemCollectsKey}']")),
"totalSoldOut": None,
"attrList": xpaths(
"//ul[@class='attributes-list']/li",
# evalx="re.split(u'[:\uff1a]', _strip(result.text))",
evalx="result.text.partition(':' if result.text.partition(':')[0] < result.text.partition(u'\uff1a')[0] else u'\uff1a')[::2]",
),
"reviewCount": ajax("saveCounts", dpath("['{apiItemViewsKey}']")),
"starts": addon("starts"),
"ends": addon("ends"),
"userTag": None,
"cid": None,
"location": ajax("wholeSibUrl", dpath("['location']")),
"brand": None,
"gradeAvg": None,
"peopleNum": None,
"periodSoldQuantity": None,
"rateTotal": None,
"spuId": None,
"totalSoldQuantity": None,
# encoding: utf8
from sasoup.baserules import xpath, xpaths, xpathz, search, dpath, base, addon, fields, which, ajaxurl, ajax, RespType
appsorules = {
'fields_rules': {
'items': xpaths("//div[contains(@id, 'liveblog-entry-')]/div", evalx="result"),
},
'result_rules': {
'tsmp': xpathz(
'items',
xpaths(".//p[1]/strong", evalx="result.text")),
'title': xpathz(
'items',
xpaths(".//p[2]/a/strong", evalx="result.text")),
'link': xpathz(
'items',
xpaths(".//p[2]/a/@href")),
'desc': xpathz(
'items',
xpaths(".//p[3]", evalx="result.text")),
'intro': xpathz(
'items',
xpaths(".//p[4]", evalx="result.text")),
},
}
# encoding: utf8
from sasoup.baserules import xpath, xpaths, xpathz, search, dpath, base, addon, fields, which, next, ajaxurl, ajax, RespType
from sasoup.baserules import init_rules
rules = {
'url': 'http://www.etao.com',
'fields_rules': {
'feedList': xpaths("//div[@id='J_FeedList']//div[@id]", evalx="result"),
},
'result_rules': {
'feed': (
xpathz('feedList', xpath(".//h3[@class='feed-title']/a/@title", evalx="_strip(result)")),
xpathz('feedList', xpath(".//h3[@class='feed-title']/a/strong/text()", evalx="_strip(result)")),
xpathz('feedList', xpath(".//div[@class='feed-desc']/p/text()", evalx="_strip(result)")),
),
'cats': next(
xpaths("//div[contains(@class,'J_PCMain')/li]", evalx="result"),
xpath(".//h3/a/text()", evalx="result"),
),
},
}
rules = init_rules(rules)
# encoding: utf8
from sasoup.baserules import xpath, xpaths, xpathz, search, dpath, base, addon, fields, which, ajaxurl, ajax, RespType
"""
jdrules:
cat: 顶级类目
electrontics_cat: 家电通讯类目列表
electrontics_items: 家电通讯类目特价商品
digitals_cat: 电脑数码类目列表
digitals_items: 电脑数码类目特价商品
"""
jdrules = {
'fields_rules': {
'catItems': xpaths("//div[@id='_JD_ALLSORT']/div", evalx="result"),
'cat1Cats': xpaths("//div[@id='electronics']/div[contains(@class,'catalogue')]//ul/li", evalx="result"),
'cat1Items': xpaths("//div[@id='electronics']/div[contains(@class,'plist')]/div[2]//li", evalx="result"),
'cat2Cats': xpaths("//div[@id='digitals']/div[contains(@class,'catalogue')]//ul/li", evalx="result"),
'cat2Items': xpaths("//div[@id='digitals']/div[contains(@class,'plist')]/div[2]//li", evalx="result"),
},
'result_rules': {
'cat': xpathz(
'catItems',
xpaths(".//h3/a", evalx="result.text")),
'electrontics_cat': xpathz(
'cat1Cats',
xpath(".//a", evalx="result.text")),
'electrontics_items': xpathz(
'cat1Items',
xpaths("./div[@class='p-name']/a|./div[@class='p-price']/span", evalx="result.text")),
'digitals_cat': xpathz(
'cat2Cats',
