From 6310e2bb06bad388f67527592c73da1272cc93a8 Mon Sep 17 00:00:00 2001 From: Gabriel Dunne Date: Fri, 22 Mar 2013 02:35:16 -0700 Subject: [PATCH] writing to file --- phantom-scrape-thumbnail.js | 11 +++++++++++ scrape-mult.js => phantom-scrape.js | 6 ------ scrape-from-json.js | 28 ++++++++++++++++++++++++++++ scrape-id.js | 16 ++++++++++++++++ scrape.js | 15 --------------- 5 files changed, 55 insertions(+), 21 deletions(-) create mode 100644 phantom-scrape-thumbnail.js rename scrape-mult.js => phantom-scrape.js (81%) create mode 100644 scrape-from-json.js create mode 100644 scrape-id.js delete mode 100644 scrape.js diff --git a/phantom-scrape-thumbnail.js b/phantom-scrape-thumbnail.js new file mode 100644 index 0000000..9724f6a --- /dev/null +++ b/phantom-scrape-thumbnail.js @@ -0,0 +1,11 @@ +var system = require('system'), + page = require('webpage').create(); + +page.open(system.args[1], function () { + var th = page.evaluate(function () { + return document.getElementById('thumbnail').src; + }); + console.log(th); + //console.log(JSON.stringify({ thumbnail : th })); + phantom.exit(); +}); diff --git a/scrape-mult.js b/phantom-scrape.js similarity index 81% rename from scrape-mult.js rename to phantom-scrape.js index 55c3afe..596c703 100644 --- a/scrape-mult.js +++ b/phantom-scrape.js @@ -1,17 +1,11 @@ var system = require('system'); var page = require('webpage').create(); - -var prelinger_ids = require('./prelinger_identifiers.json'); -var ids = prelinger_ids.identifiers; var url_details = "http://archive.org/details/"; if (system.args.length === 1) { console.log('Identifier Required'); } else { var id = system.args[1]; - - - page.open(url_details + id, function () { var th = page.evaluate(function () { return document.getElementById('thumbnail').src; diff --git a/scrape-from-json.js b/scrape-from-json.js new file mode 100644 index 0000000..ef857f4 --- /dev/null +++ b/scrape-from-json.js @@ -0,0 +1,28 @@ +var exec = require('child_process').exec; +var fs = require("fs"); + +// fs.readFileSync('./input.txt').toString().split('\n').forEach(function (line) { +// console.log(line); +// fs.appendFileSync("./output.txt", line.toString() + "\n"; +// }); + +var prelinger = require('./prelinger_identifiers.json'); + +var output = { "clips" : [] }; +for (var i = 0; i < prelinger.identifiers.length; i++) { + var id = prelinger.identifiers[i]; + output.clips[i] = { "id" : id }; +} + +fs.open("./output.txt", 'w', 0666, function(err, fd) { + fs.writeSync(fd, JSON.stringify(output) + "\n", null, undefined, function(err, written) { + }) +}); + +// "http://archive.org/details/" + + +// var id = '0924_Greatest_Name_in_Wine_The_11_00_13_15'; +// cmd = spawn('phantomjs' ['phantom-scrape-id.js']); + + + diff --git a/scrape-id.js b/scrape-id.js new file mode 100644 index 0000000..596c703 --- /dev/null +++ b/scrape-id.js @@ -0,0 +1,16 @@ +var system = require('system'); +var page = require('webpage').create(); +var url_details = "http://archive.org/details/"; + +if (system.args.length === 1) { + console.log('Identifier Required'); +} else { + var id = system.args[1]; + page.open(url_details + id, function () { + var th = page.evaluate(function () { + return document.getElementById('thumbnail').src; + }); + console.log(th); + phantom.exit(); + }); +} diff --git a/scrape.js b/scrape.js deleted file mode 100644 index edb3e47..0000000 --- a/scrape.js +++ /dev/null @@ -1,15 +0,0 @@ - -var identifier = "animals_in_the_service_of_man_2"; -var url_details = "http://archive.org/details/"; - -var page = require('webpage').create(); -page.open(url_details + identifier, function () { - // page.render('page.png'); - - var ua = page.evaluate(function () { - return document.getElementById('thumbnail').src; - }); - - console.log(ua); - phantom.exit(); -}); -- 2.34.1