2020-06-17 17:37:01 -04:00
|
|
|
LET baseURL = 'https://www.amazon.com/'
|
|
|
|
LET amazon = DOCUMENT(baseURL, { driver: "cdp" })
|
2018-11-12 19:58:12 -05:00
|
|
|
|
|
|
|
INPUT(amazon, '#twotabsearchtextbox', @criteria)
|
|
|
|
CLICK(amazon, '.nav-search-submit input[type="submit"]')
|
|
|
|
WAIT_NAVIGATION(amazon)
|
|
|
|
|
|
|
|
LET resultListSelector = '#s-results-list-atf'
|
2020-06-17 17:37:01 -04:00
|
|
|
LET resultItemSelector = '[data-component-type="s-search-result"]'
|
|
|
|
LET nextBtnSelector = 'ul.a-pagination .a-last a'
|
|
|
|
LET priceWholeSelector = '.a-price-whole'
|
|
|
|
LET priceFracSelector = '.a-price-fraction'
|
2018-11-12 19:58:12 -05:00
|
|
|
|
|
|
|
LET result = (
|
|
|
|
FOR pageNum IN PAGINATION(amazon, nextBtnSelector)
|
2020-06-17 17:37:01 -04:00
|
|
|
LIMIT @pages
|
2018-11-12 19:58:12 -05:00
|
|
|
|
2020-06-17 17:37:01 -04:00
|
|
|
LET wait = pageNum > 0 ? WAIT_NAVIGATION(amazon, 20000) : false
|
2018-11-12 19:58:12 -05:00
|
|
|
LET waitSelector = wait ? WAIT_ELEMENT(amazon, resultListSelector) : false
|
|
|
|
|
|
|
|
LET items = (
|
|
|
|
FOR el IN ELEMENTS(amazon, resultItemSelector)
|
2020-06-17 17:37:01 -04:00
|
|
|
LET hasPrice = ELEMENT_EXISTS(el, priceWholeSelector)
|
|
|
|
LET priceWholeTxt = hasPrice ? FIRST(REGEX_MATCH(INNER_TEXT(el, priceWholeSelector), "[0-9]+")) : "0"
|
|
|
|
LET priceFracTxt = hasPrice ? FIRST(REGEX_MATCH(INNER_TEXT(el, priceFracSelector), "[0-9]+")) : "00"
|
|
|
|
LET price = TO_FLOAT(priceWholeTxt + "." + priceFracTxt)
|
|
|
|
LET anchor = ELEMENT(el, "a")
|
|
|
|
|
|
|
|
PRINT(priceWholeTxt, priceFracTxt)
|
2018-11-12 19:58:12 -05:00
|
|
|
|
|
|
|
RETURN {
|
2020-06-17 17:37:01 -04:00
|
|
|
url: baseURL + anchor.attributes.href,
|
2018-11-12 19:58:12 -05:00
|
|
|
title: INNER_TEXT(el, 'h2'),
|
2020-06-17 17:37:01 -04:00
|
|
|
price
|
2018-11-12 19:58:12 -05:00
|
|
|
}
|
|
|
|
)
|
|
|
|
|
|
|
|
RETURN items
|
|
|
|
)
|
|
|
|
|
|
|
|
RETURN FLATTEN(result)
|