Files
data-extracter-extesion/scripts/extract.js
2018-05-23 10:46:04 +08:00

155 lines
4.5 KiB
JavaScript

// function extract(itemsSelector, fieldSelectors, url, from, to, interval)
// function extract(itemsSelector, fieldSelectors, url, pages)
function extract(itemsSelector, fieldSelectors, url, ...args) {
let urls = [];
if (url) {
if (args[0] instanceof Array) {
urls = args[0].map(p => url.replace("${page}", p));
} else if (args.length >= 3) {
let from = args.shift();
let to = args.shift();
let interval = args.shift();
for (let i = from; i <= to; i += interval) {
urls.push(url.replace("${page}", i));
}
}
}
let data = [];
return new Promise((resolve, reject) => {
chrome.tabs.query({
active: true,
currentWindow: true
}, function (tabs) {
let pms;
let tab = tabs[0];
if (urls.length) {
pms = urls.reduce((p, url) => p.then(
results => {
data.push(...results);
return redirectTab(tab, url).then(
tab => extractTabData(tab, itemsSelector, fieldSelectors)
);
},
() => p
), Promise.resolve([]));
} else {
pms = extractTabData(tab, itemsSelector, fieldSelectors);
}
pms.then(
results => {
data.push(...results);
data.unshift(fieldSelectors);
saveFileAsk(data);
resolve("save done.")
},
err => reject(err)
);
});
});
}
function extractTabData(tab, itemsSelector, fieldSelectors) {
let done = false;
return new Promise((resolve, reject) => {
chrome.tabs.sendMessage(
tab.id, {
from: "doExtractRequest",
itemsSelector: itemsSelector,
fieldSelectors: fieldSelectors
},
response => {
done = true;
resolve(response);
}
);
setTimeout(() => {
if (!done) reject(`extractTabData failed after 2 second.`);
}, 2000);
});
}
function redirectTab(tab, url) {
return new Promise((resolve, reject) => {
let req = {
from: "doExtractGotoUrl",
url: url
}
let curUrl = "";
queryUrl(tab)
.then(u => {
if (url !== u) {
curUrl = u;
chrome.tabs.sendMessage(tab.id, req);
}
})
.then(() => queryUrl(tab, curUrl))
.then(() => reportIn(tab))
.then(tab => resolve(tab))
.catch(err => reject(err));
});
}
function reportIn(tab) {
let req = {
from: "doExtractReportIn"
}
return new Promise((resolve, reject) => {
let timeOut;
let rejectTimeout = setTimeout(() => {
reject(`reportIn failed after 10 second.`);
clearTimeout(timeOut);
}, 10000);
loop();
function loop() {
chrome.tabs.sendMessage(tab.id, req, r => {
if (r == req.from) {
resolve(tab);
clearTimeout(rejectTimeout);
} else {
timeOut = setTimeout(() => {
loop();
}, 500);
}
});
}
});
}
function queryUrl(tab, urlExcluded) {
let req = {
from: "doExtractQueryUrl"
}
return new Promise((resolve, reject) => {
let timeOut;
let rejectTimeout = setTimeout(() => {
reject(`queryUrl failed after 10 second.`);
clearTimeout(timeOut);
}, 10000);
loop();
function loop() {
chrome.tabs.sendMessage(tab.id, req, url => {
if (url && (!urlExcluded || (urlExcluded && urlExcluded != url))) {
resolve(url);
clearTimeout(rejectTimeout);
} else {
timeOut = setTimeout(() => {
loop();
}, 500);
}
});
}
});
}
function createTab(url) {
return new Promise((resolve, reject) => {
chrome.tabs.create({
active: true,
url: url
}, function (tab) {
resolve(tab);
});
});
}