Grab products from Amazon aStores
This task use javascript sandbox with jsoup support to grab products from Amazon aStores.
Grab products from Amazon aStores
- Create javascript sandbox with jsoup support
- Create javascript as following
javascript
function main(env, args) {
var astore = 'paesia';
var node = '100';
var maxpage = 2;
try {
var products = grabProduct(astore, node, maxpage, env);
var links = args.get('links');
for (var i = 0; i < products.size(); i++) {
links.add(products.get(i));
}
} catch (e) {
env.error(e);
}
}
function grabProduct(astore, node, maxpage, env) {
var tag = env.newArrayList();
for (var no = 1; no <= maxpage; no++) {
try {
var alink = env.newURL('http://astore.amazon.com/' + astore + '-20?node=' + node + '&page=' + no);
var doc = env.newJsoup().parse(alink, 60000);
var elements = doc.select('#featuredProducts .textrow a');
var map = env.newHashMap();
for (var i = 0; i < elements.size(); i++) {
var element = elements.get(i);
var title = element.text();
var url = element.attr('href');
var pos = url.lastIndexOf('/detail/');
if (pos < 0) continue;
var code = url.substring(pos + 8);
var url = env.newURL(alink, url) + '';
var item = env.newHashMap();
item.put('code', code);
item.put('title', title);
item.put('url', url);
map.put(code, item);
}
elements = doc.select('#featuredProducts .imagerow a');
for (var i = 0; i < elements.size(); i++) {
var element = elements.get(i);
var url = element.attr('href');
var pos = url.lastIndexOf('/detail/');
if (pos < 0) continue;
var code = url.substring(pos + 8);
var item = map.get(code);
if (item == null) continue;
var child = element.select('img').first();
if (child == null) continue;
var title = child.attr('alt');
var smimg = child.attr('src');
if (title.length() > 0) {
item.put('title', title);
}
item.put('small-image', smimg);
}
var keys = env.getKeys(map);
for (var i = 0; i < keys.size(); i++) {
try {
var item = map.get(keys.get(i));
alink = env.newURL(item.get('url'));
doc = env.newJsoup().parse(alink, 60000);
var element = doc.select('#detailImage img').first();
if (element != null) {
item.put('large-image', element.attr('src'));
}
element = doc.select('#productDescription').first();
if (element != null) {
var desc = element.html();
var pattern = '<h2>Product Description</h2>';
var pos = desc.indexOf(pattern);
if (pos >= 0) {
desc = desc.substring(pos + pattern.length);
}
var bdoc = env.newJsoup().parse(desc, item.get('url'));
buildURL(bdoc, item.get('url'), env);
desc = bdoc.select('body').first().html();
if (desc.indexOf('<html') < 0) {
item.put('description', desc);
}
}
element = doc.select('#productDetails').first();
if (element != null) {
var desc = element.html();
var pattern = '<h2>Product Details</h2>';
var pos = desc.indexOf(pattern);
if (pos >= 0) {
desc = desc.substring(pos + pattern.length);
}
var bdoc = env.newJsoup().parse(desc, item.get('url'));
buildURL(bdoc, item.get('url'), env);
desc = bdoc.select('body').first().html();
if (desc.indexOf('<html') < 0) {
item.put('details', desc);
}
}
element = doc.select('#editorialReviews').first();
if (element != null) {
var desc = element.html();
var bdoc = env.newJsoup().parse(desc, item.get('url') + '');
buildURL(bdoc, item.get('url'), env);
desc = bdoc.select('body').first().html();
if (desc.indexOf('<html') < 0) {
item.put('editorial-reviews', desc);
}
}
element = doc.select('#detailListPrice').first();
if (element != null) {
item.put('list-price', element.text());
}
element = doc.select('#detailOfferPrice').first();
if (element != null) {
item.put('offer-price', element.text());
}
element = doc.select('#addToCartForm a').first();
if (element != null) {
item.put('buy-url', element.attr('href'));
}
} catch (e) {
env.error(e);
}
}
for (var i = 0; i < keys.size(); i++) {
tag.add(map.get(keys.get(i)));
}
} catch (e) {
env.error(e);
}
}
return tag;
}
function buildURL(doc, baseUrl, env) {
baseUrl = env.newURL(baseUrl);
var elements = doc.select('a');
for (var i = 0; i < elements.size(); i++) {
var element = elements.get(i);
var url = env.newURL(baseUrl, element.attr('href'));
element.attr('href', url + '');
}
elements = doc.select('img');
for (var i = 0; i < elements.size(); i++) {
var element = elements.get(i);
var url = env.newURL(baseUrl, element.attr('src'));
element.attr('src', url + '');
}
}
No comments:
Post a Comment