jebbs/data-extracter-extesion
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

code optimize

Browse Source
This commit is contained in:
jebbs
2018-09-27 11:57:21 +08:00
parent c3b765ca8e
commit fb2bb7b59e
2 changed files with 76 additions and 42 deletions
Download Patch File Download Diff File Expand all files Collapse all files

89
scripts/content.js
View File

@ -1,19 +1,22 @@
chrome.runtime.onMessage.addListener(
function (request, sender, sendResponse) {
if (!request.from) return;
let [ext, act] = request.from.split(":");
if (ext.toLowerCase() !== 'dataextracter') return;
// console.log(request);
switch (request.from) {
case "DataExtracter:Extract":
switch (act.toLowerCase()) {
case "extract":
let data = extractTabData(request.itemsSelector, request.fieldSelectors);
if (sendResponse) sendResponse(data);
break;
case "DataExtracter:GotoUrl":
case "gotourl":
window.location.replace(request.url);
if (sendResponse) sendResponse(request.url);
break;
case "DataExtracter:ReportIn":
case "reportin":
if (sendResponse) sendResponse(request.from);
break;
case "DataExtracter:QueryUrl":
case "queryurl":
if (sendResponse) sendResponse(window.location.href);
break;
default:
@ -22,19 +25,7 @@
}
);
function extractTabData(itemsSelector, fieldSelectors) {
return $(itemsSelector).toArray().map(
item => fieldSelectors.map(
selector => {
let [cls, attr] = selector.split('@').slice(0, 2);
return $(item).find(cls).toArray().map(find => attr ? find[attr] : find.textContent.trim()).join('\n')
}
)
);
}
function extract(...args) {
let sig = `
const sig = `
# DataExtracter Help
----------------------------
@ -66,7 +57,24 @@ extract(".list-item", ["a.title", "p.content"], "http://sample.com/?pn=\${page}"
### Extract link text and target (use 'selector@attribute')
extract('.list-item', ['a.title', 'a.title@href'])
### Collect links from page(s) & Extract data of each link (only available in console of extension background page)
extract('body',["a.title", "p.content"], await getData('.list-item', ['.item a@href'],["http://sample.com/abc"]))
`.trim();
function extractTabData(itemsSelector, fieldSelectors) {
return $(itemsSelector).toArray().map(
item => fieldSelectors.map(
selector => {
let [cls, attr] = selector.split('@').slice(0, 2);
return $(item).find(cls).toArray().map(find => attr ? find[attr] : find.textContent.trim()).join('\n')
}
)
);
}
function extract(...args) {
if (!testArgs(...args)) {
console.log(sig);
return;
@ -90,12 +98,41 @@ extract('.list-item', ['a.title', 'a.title@href'])
}
function testArgs(...args) {
if (args.length < 2) return false;
if (args.length == 2)
return (args[0] && args[1] && (typeof args[0] == "string") && (args[1] instanceof Array))
let urls = [];
if (args.length > 2) return (typeof args[2] == "string") && (
(args[3] instanceof Array) ||
(!isNaN(args[3]) && !isNaN(args[4]) && !isNaN(args[5]))
)
switch (args.length) {
case 0, 1:
return false;
case 2:
return args[0] && args[1] &&
(typeof args[0] == "string") &&
(args[1] instanceof Array) &&
testArrayVals(args[1], v => typeof v == "string");
case 3:
return args[0] && args[1] &&
typeof args[0] == "string" &&
args[1] instanceof Array &&
testArrayVals(args[1], v => typeof v == "string") &&
args[2] instanceof Array &&
testArrayVals(args[2], v => typeof v == "string");
case 4:
return args[0] && args[1] &&
typeof args[0] == "string" &&
args[1] instanceof Array &&
testArrayVals(args[1], v => typeof v == "string") &&
typeof args[2] == "string" &&
args[3] instanceof Array &&
testArrayVals(args[3], v => typeof v == "number");
case 5:
return args[0] && args[1] &&
typeof args[0] == "string" &&
args[1] instanceof Array &&
testArrayVals(args[1], v => typeof v == "string") &&
typeof args[2] == "string" &&
!isNaN(args[3]) && !isNaN(args[4]) && !isNaN(args[5]);
default:
return false;
}
function testArrayVals(arr, tester) {
return arr.reduce((p, c) => p && tester(c), true);
}
}

29
scripts/extract.js
View File

@ -75,10 +75,10 @@ function redirectTab(tab, url) {
if (url !== u) {
curUrl = u;
let req = {
from: "DataExtracter:GotoUrl",
from: "GotoUrl",
url: url
}
chrome.tabs.sendMessage(tab.id, req);
sendMessage(tab, req);
}
})
.then(() => queryUrl(tab, curUrl))
@ -94,13 +94,12 @@ function redirectTab(tab, url) {
*/
function extractTabData(tab, itemsSelector, fieldSelectors) {
let req = {
from: "DataExtracter:Extract",
from: "Extract",
itemsSelector: itemsSelector,
fieldSelectors: fieldSelectors
}
let failMsg = "extractTabData failed after 10 second.";
let cond = r => !!r;
return sendMessageAndDetect(tab, req, cond, failMsg);
return sendMessage(tab, req, cond);
}
/**
@ -110,11 +109,10 @@ function extractTabData(tab, itemsSelector, fieldSelectors) {
*/
function reportIn(tab) {
let req = {
from: "DataExtracter:ReportIn"
from: "ReportIn"
}
let failMsg = "reportIn failed after 10 second.";
let cond = r => r == req.from;
return sendMessageAndDetect(tab, req, cond, failMsg);
return sendMessage(tab, req, cond);
}
/**
@ -125,11 +123,10 @@ function reportIn(tab) {
*/
function queryUrl(tab, urlExcluded) {
let req = {
from: "DataExtracter:QueryUrl"
from: "QueryUrl"
}
let failMsg = "queryUrl failed after 10 second.";
let cond = url => url && (!urlExcluded || (urlExcluded && urlExcluded != url));
return sendMessageAndDetect(tab, req, cond, failMsg);
return sendMessage(tab, req, cond);
}
/**
@ -137,27 +134,27 @@ function queryUrl(tab, urlExcluded) {
* @param {object} tab the table where to send the message
* @param {object} req the request data.
* @param {function} cond success condition function, r:any=>boolean
* @param {string} failMsg message when failed after time out
* @param {number} failedTimeOut fail time out
* @param {number} detectInterval interval for detecting
* @return {Promise} a promise of the response.
*/
function sendMessageAndDetect(tab, req, cond, failMsg, failedTimeOut, detectInterval) {
function sendMessage(tab, req, cond, failedTimeOut, detectInterval) {
req.from = "DataExtracter:" + req.from;
failedTimeOut = failedTimeOut || 10000;
detectInterval = detectInterval || 500;
return new Promise((resolve, reject) => {
let timeOut;
let rejectTimeout = setTimeout(() => {
reject(failMsg);
reject(`${req.from} failed after ${failedTimeOut/1000} seconds.`);
clearTimeout(timeOut);
}, failedTimeOut);
loop();
function loop() {
chrome.tabs.sendMessage(tab.id, req, r => {
if (cond(r)) {
resolve(r);
if (!cond || cond(r)) {
clearTimeout(rejectTimeout);
resolve(r);
} else {
timeOut = setTimeout(() => {
loop();