docparse-upload-process/lib/fetch.js

process upload api request for the docparse server

var querystring = require('querystring');
var request = require('request');
var inspect = require('eyespect').inspector();
var mongoose = require('mongoose');
var Upload = require('docparse-upload');
var User = require('docparse-user');
var async = require('async');
var userInUpload = require('./userInUpload');
var rk = require('required-keys');

function keysExist(data, cb) {
  var keys = [
    'userID',
    'hash',
    'config'
  ]
  rk.nonNull(data, keys, function (err, reply) {
    if (err) { return cb(err); }
    cb();
  });
}

function findUpload(data, cb) {
  Upload.findOne({hash: data.hash}, function (err, reply) {
    if (err) { return cb(err); }
    if (!reply) {
      return cb('upload not found with hash: '+ data.hash);
    }
    data.upload = reply;
    cb();
  });
}
function findUser(data, cb) {
  User.findById(data.userID, function (err, reply) {
    if (err || !reply) {
      return cb('user not found with id: '+ data.userID);
    }
    data.user = reply;
    cb();
  });
}
function fetchUpload(data, cb) {
  var upload = data.upload;
  if (upload.ocr_status.complete) {
    inspect('upload ocr status complete');
    return cb();
  }
  // start updating the upload in the background
  updateUpload(data, cb);
}

/**
 * Issue a fetch request the remote pdfer service for the latest information about this upload
 */
function updateUpload(data, cb) {
  var config = data.config
  var upload = data.upload;
  var username = config.get('pdfer:username');
  var password = config.get('pdfer:password');
  var host = config.get('pdfer:host');
  var port = config.get('pdfer:port');

  var url = 'http://'+username+':'+password+'@'+host + ':' + port
    +'/api/fetch/'+data.hash;
  var options = {
    url: url,
    method: 'get',
    json: true
  }
  request(options, function (err, response, body) {
    if (err) { return cb(err); }
    var statusCode = response.statusCode;
    if (statusCode !== 200) {
      var message = 'error updating upload status in fetch request, bad status code: ' + statusCode;
      inspect(body, message);
      return cb(message);
    }
    upload.status = body.status;

    if (body.parsed) {
      upload.ocr_status.complete = true;
      upload.ocr_status.running = true;
      upload.text_pages = body.text_pages;
      upload.num_page = body.text_pages.length;
      upload.status = 'ocr complete, waiting to parse'
    }
    upload.save(cb);
  });
}


module.exports = function(data, callback) {
  var upload;
  var output;
  async.series([
    function (cb) {
      keysExist(data, cb)
    },
    function (cb) {
      findUpload(data, cb)
    },
    function (cb) {
      findUser(data, cb)
    },
    function(cb) {
      userInUpload(data, function (err, reply) {
        if (err) { return cb(err); }
        if (!reply) {
          return cb('you are not permitted to fetch this upload');
        }
        cb();
      });
    },
    function(cb) {
      fetchUpload(data, cb)
    }
  ], function (err) {
    if (err) { return callback(err); }
    return callback(null, data.upload);
  });
}