1
0
mirror of https://github.com/MontFerret/ferret.git synced 2025-02-01 13:07:49 +02:00
ferret/examples/pagination_uncontrolled.fql

42 lines
1.5 KiB
Plaintext
Raw Normal View History

LET baseURL = 'https://www.amazon.com/'
LET amazon = DOCUMENT(baseURL, { driver: "cdp" })
INPUT(amazon, '#twotabsearchtextbox', @criteria)
CLICK(amazon, '#nav-search-submit-button')
WAITFOR EVENT "navigation" IN amazon
FILTER current.url =~ "www\.amazon\.com\/s\?k="
TIMEOUT 50000
LET resultListSelector = '[data-component-type="s-search-results"]'
LET resultItemSelector = '[data-component-type="s-search-result"]'
LET nextBtnSelector = '.s-pagination-next:not(.s-pagination-disabled)'
LET priceWholeSelector = '.a-price-whole'
LET priceFracSelector = '.a-price-fraction'
LET result = (
FOR pageNum IN PAGINATION(amazon, nextBtnSelector)
LIMIT @pages
LET waitSelector = pageNum > 0 ? WAIT_ELEMENT(amazon, resultListSelector) : false
LET items = (
FOR el IN ELEMENTS(amazon, resultItemSelector)
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")
RETURN {
page: pageNum,
url: baseURL + anchor.attributes.href,
title: INNER_TEXT(el, 'h2'),
price
}
)
RETURN items
)
RETURN FLATTEN(result)