Grab categories from Amazon aStores
This task use javascript sandbox with jsoup support to grab categories from Amazon aStores.
Grab categories from Amazon aStores
- Create javascript sandbox with jsoup support
- Create javascript as following
javascript
| 1 | function main(env, args) { |
| 2 | var astore = 'paesia'; |
| 3 | try { |
| 4 | var categories = grabCategory(astore, env); |
| 5 | var map = env.newHashMap(); |
| 6 | for (var i = 0; i < categories.size(); i++) { |
| 7 | var cat = categories.get(i); |
| 8 | map.put(cat.get('node'), cat); |
| 9 | } |
| 10 | for (var i = 0; i < categories.size(); i++) { |
| 11 | var cat = categories.get(i); |
| 12 | var node = cat.get('node'); |
| 13 | var title = cat.get('title'); |
| 14 | var parentNode = cat.get('parent'); |
| 15 | var parentCat = map.get(parentNode); |
| 16 | var parentTitle = ''; |
| 17 | if (parentCat != null) { |
| 18 | parentTitle = parentCat.get('title'); |
| 19 | } |
| 20 | var line = ''; |
| 21 | line += '\r\nTitle: ' + title; |
| 22 | line += '\r\nNode: ' + node; |
| 23 | line += '\r\nParent: ' + parentNode; |
| 24 | line += '\r\nParent Title: ' + parentTitle; |
| 25 | env.info(line); |
| 26 | } |
| 27 | } catch (e) { |
| 28 | env.error(e); |
| 29 | } |
| 30 | } |
| 31 | |
| 32 | function grabCategory(astore, env) { |
| 33 | var tag = env.newArrayList(); |
| 34 | try { |
| 35 | var nodelist = env.newArrayList(); |
| 36 | var alink = env.newURL('http://astore.amazon.com/' + astore + '-20'); |
| 37 | var doc = env.newJsoup().parse(alink, 60000); |
| 38 | var elements = doc.select('#searchbrowse a'); |
| 39 | for (var i = 0; i < elements.size(); i++) { |
| 40 | var element = elements.get(i); |
| 41 | var title = element.text(); |
| 42 | var url = element.attr('href'); |
| 43 | var pos = url.lastIndexOf('node='); |
| 44 | if (pos < 0) continue; |
| 45 | var node = url.substring(pos + 5); |
| 46 | pos = node.indexOf('&'); |
| 47 | if (pos >= 0) { |
| 48 | node = node.substring(0, pos); |
| 49 | } |
| 50 | var item = env.newHashMap(); |
| 51 | item.put('title', title); |
| 52 | item.put('node', node); |
| 53 | item.put('parent', ''); |
| 54 | tag.add(item); |
| 55 | nodelist.add(node); |
| 56 | } |
| 57 | var no = 0; |
| 58 | while (no < nodelist.size()) { |
| 59 | alink = env.newURL('http://astore.amazon.com/' + astore + '-20?node=' + nodelist.get(no)); |
| 60 | doc = env.newJsoup().parse(alink, 60000); |
| 61 | elements = doc.select('#searchbrowse .indent a'); |
| 62 | for (var i = 0; i < elements.size(); i++) { |
| 63 | var element = elements.get(i); |
| 64 | var title = element.text(); |
| 65 | var url = element.attr('href'); |
| 66 | var pos = url.lastIndexOf('node='); |
| 67 | if (pos < 0) continue; |
| 68 | var node = url.substring(pos + 5); |
| 69 | pos = node.indexOf('&'); |
| 70 | if (pos >= 0) { |
| 71 | node = node.substring(0, pos); |
| 72 | } |
| 73 | if (nodelist.indexOf(node) >= 0) continue; |
| 74 | var item = env.newHashMap(); |
| 75 | item.put('title', title); |
| 76 | item.put('node', node); |
| 77 | item.put('parent', nodelist.get(no)); |
| 78 | tag.add(item); |
| 79 | nodelist.add(node); |
| 80 | } |
| 81 | no++; |
| 82 | } |
| 83 | } catch (e) { |
| 84 | env.error(e); |
| 85 | } |
| 86 | return tag; |
| 87 | } |
function main(env, args) {
var astore = 'paesia';
try {
var categories = grabCategory(astore, env);
var map = env.newHashMap();
for (var i = 0; i < categories.size(); i++) {
var cat = categories.get(i);
map.put(cat.get('node'), cat);
}
for (var i = 0; i < categories.size(); i++) {
var cat = categories.get(i);
var node = cat.get('node');
var title = cat.get('title');
var parentNode = cat.get('parent');
var parentCat = map.get(parentNode);
var parentTitle = '';
if (parentCat != null) {
parentTitle = parentCat.get('title');
}
var line = '';
line += '\r\nTitle: ' + title;
line += '\r\nNode: ' + node;
line += '\r\nParent: ' + parentNode;
line += '\r\nParent Title: ' + parentTitle;
env.info(line);
}
} catch (e) {
env.error(e);
}
}
function grabCategory(astore, env) {
var tag = env.newArrayList();
try {
var nodelist = env.newArrayList();
var alink = env.newURL('http://astore.amazon.com/' + astore + '-20');
var doc = env.newJsoup().parse(alink, 60000);
var elements = doc.select('#searchbrowse a');
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('node=');
if (pos < 0) continue;
var node = url.substring(pos + 5);
pos = node.indexOf('&');
if (pos >= 0) {
node = node.substring(0, pos);
}
var item = env.newHashMap();
item.put('title', title);
item.put('node', node);
item.put('parent', '');
tag.add(item);
nodelist.add(node);
}
var no = 0;
while (no < nodelist.size()) {
alink = env.newURL('http://astore.amazon.com/' + astore + '-20?node=' + nodelist.get(no));
doc = env.newJsoup().parse(alink, 60000);
elements = doc.select('#searchbrowse .indent a');
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('node=');
if (pos < 0) continue;
var node = url.substring(pos + 5);
pos = node.indexOf('&');
if (pos >= 0) {
node = node.substring(0, pos);
}
if (nodelist.indexOf(node) >= 0) continue;
var item = env.newHashMap();
item.put('title', title);
item.put('node', node);
item.put('parent', nodelist.get(no));
tag.add(item);
nodelist.add(node);
}
no++;
}
} catch (e) {
env.error(e);
}
return tag;
}
No comments:
Post a Comment