Init repo

Author: Julien Balian

Diff for: .gitignore

@@ -0,0 +1,24 @@
+# Projects files
+keys.sh
+
+# OS Artifacts
+*.DS_Store
+
+# Vagrant Artifacts
+.vagrant/*
+
+# NPM Artifacts
+node_modules/*
+npm-debug.log
+libpeerconnection.log
+
+# Docco artifacts
+docs/*
+
+# IDE Artifacts
+.idea/*
+*.iml
+*.sublime-*
+
+# Jekyll artifacts
+_site/*

Diff for: .jshintrc

@@ -0,0 +1,75 @@
+
+{
+  "bitwise"       : true,
+  "curly"         : true,
+  "eqeqeq"        : true,
+  "immed"         : true,
+  "latedef"       : true,
+  "newcap"        : true,
+  "noarg"         : true,
+  "noempty"       : true,
+  "nonew"         : true,
+  "plusplus"      : true,
+  "regexp"        : true,
+  "undef"         : true,
+  "strict"        : true,
+  "trailing"      : true,
+  "unused"        : true,
+  "loopfunc"      : true,
+  
+  "asi"           : false,
+  "boss"          : false,
+  "debug"         : false,
+  "eqnull"        : false,
+  "es5"           : false,
+  "esnext"        : false,
+  "evil"          : false,
+  "expr"          : false,
+  "forin"         : false,
+  "funcscope"     : false,
+  "globalstrict"  : false,
+  "quotmark"      : false,
+  "iterator"      : false,
+  "lastsemic"     : false,
+  "laxbreak"      : false,
+  "laxcomma"      : false,
+  "multistr"      : false,
+  "onecase"       : false,
+  "proto"         : false,
+  "regexdash"     : false,
+  "scripturl"     : false,
+  "smarttabs"     : false,
+  "shadow"        : false,
+  "sub"           : false,
+  "supernew"      : false,
+  "validthis"     : false,
+  
+  "browser"       : false,
+  "couch"         : false,
+  "devel"         : false,
+  "dojo"          : false,
+  "jquery"        : false,
+  "mootools"      : false,
+  "node"          : true,
+  "nonstandard"   : false,
+  "prototypejs"   : false,
+  "rhino"         : false,
+  "wsh"           : false,
+  
+  "nomen"         : false,
+  "onevar"        : false,
+  "passfail"      : false,
+  "white"         : false,
+  
+  "maxerr"        : 100,
+  "predef"        : [
+    "describe",
+    "it",
+    "after",
+    "before",
+    "beforeEach"
+  ],
+  "indent"        : 2,
+  "maxstatements" : 20,
+  "maxcomplexity" : 10
+}

Diff for: Berksfile

@@ -0,0 +1,5 @@
+site :opscode
+
+cookbook 'apt', '~> 2.1.0'
+cookbook 'java', '~> 1.12.0'
+cookbook 'nodejs', '~> 1.3.0'

Diff for: Berksfile.lock

@@ -0,0 +1,25 @@
+{
+  "sources": {
+    "apt": {
+      "locked_version": "2.1.0"
+    },
+    "java": {
+      "locked_version": "1.12.0"
+    },
+    "nodejs": {
+      "locked_version": "1.3.0"
+    },
+    "windows": {
+      "locked_version": "1.10.0"
+    },
+    "chef_handler": {
+      "locked_version": "1.1.4"
+    },
+    "yum": {
+      "locked_version": "2.3.4"
+    },
+    "build-essential": {
+      "locked_version": "1.4.2"
+    }
+  }
+}

Diff for: README.md

@@ -0,0 +1,10 @@
+# pdf2htmlEX 0.10 Cluestr Hydrater
+> Visit http://cluestr.com for details about Cluestr.
+
+Cluestr Hydrater for pdf files
+
+# How to install?
+Vagrant up everything (`vagrant up`, `vagrant ssh`).
+
+
+Support: `[email protected]`.

Diff for: Vagrantfile

@@ -0,0 +1,71 @@
+# -*- mode: ruby -*-
+# vi: set ft=ruby :
+
+# TODO: use our recipe for tika
+$script = <<SCRIPT
+//>script inline
+sudo apt-get update;
+sudo apt-get install -y autoconf libtool \
+ libpng12-dev zlibc zlib1g-dev libtiff-dev libungif4-dev libjpeg-dev \
+ libxml2-dev libuninameslist-dev xorg-dev subversion cvs gettext git \
+ libpango1.0-dev libcairo2-dev python-dev uuid-dev libreadline-dev \
+ cmake util-linux pkg-config libpng-dev \
+ poppler-data git libfontconfig-dev gettext libcairo2-dev \
+ libtool;
+
+mkdir deps;cd deps;
+wget http://poppler.freedesktop.org/poppler-0.24.2.tar.xz;
+tar -xvJf poppler-0.24.2.tar.xz;
+cd poppler-0.24.2;
+./configure --enable-xpdf-headers;
+make;
+sudo make install;
+cd ..;
+
+# may need a FreeType Install
+# and libSpiro
+
+git clone https://github.com/coolwanglu/fontforge.git;
+cd fontforge;
+./autogen.sh;
+./configure;
+make;
+sudo make install;
+
+cd ..;
+git clone git://github.com/coolwanglu/pdf2htmlEX.git;
+cd pdf2htmlEX;
+cmake . && make && sudo make install;
+echo "\nexport LD_LIBRARY_PATH=/usr/local/lib:$LD_LIBRARY_PATH;" > /home/vagrant/.bashrc
+SCRIPT
+
+Vagrant.configure("2") do |config|
+  config.vm.hostname = "cluestrhydraterpdf"
+
+  config.vm.box = "precise64"
+  config.vm.box_url = "http://files.vagrantup.com/precise64.box"
+
+  config.vm.network :forwarded_port, host: 8000, guest: 8000
+
+  config.berkshelf.berksfile_path = "./Berksfile"
+  config.berkshelf.enabled = true
+  config.omnibus.chef_version = '11.6.0'
+
+  config.vm.provision :chef_solo do |chef|
+    chef.run_list = [
+      "recipe[apt]",
+      "recipe[java]",
+      "recipe[nodejs]",
+    ]
+
+    chef.json = {
+      :java => {
+        :install_flavor => "openjdk",
+        :jdk_version => "7"
+      }
+    }
+  end
+
+  config.vm.provision :shell,
+      :inline => $script
+end

Diff for: app.js

@@ -0,0 +1,25 @@
+'use strict';
+
+// Load configuration and initialize server
+var restify = require('restify');
+var async = require('async');
+
+var configuration = require('./config/configuration.js');
+var lib = require('./lib/hydrater-pdf');
+
+var handlers = lib.handlers;
+var server = restify.createServer();
+
+
+// Middleware Goes Here
+server.use(restify.acceptParser(server.acceptable));
+server.use(restify.queryParser());
+server.use(restify.bodyParser());
+
+server.queue = async.queue(lib.helpers.hydrate, configuration.concurrency);
+
+// Load routes
+require('./config/routes.js')(server, handlers);
+
+// Expose the server
+module.exports = server;

Diff for: bin/server

@@ -0,0 +1,11 @@
+#!/bin/env node
+// # bin/server
+// Launch a node app
+
+// Load configuration
+var config = require("../config/configuration.js");
+var server = require('../app.js');
+
+// Start the server
+var spawner = require('sspawn')(server, {port: config.port, workers: config.workers});
+spawner.start();

Diff for: config/configuration.js

@@ -0,0 +1,28 @@
+/**
+ * @file Defines the hydrater settings.
+ */
+
+// node_env can either be "development" or "production"
+var node_env = process.env.NODE_ENV || "development";
+var default_port = 8000;
+
+var default_pdf_version = "1.4";
+var default_pdf_path = "/etc/pdf-" + default_pdf_version + "/pdf-app-" + default_pdf_version + ".jar";
+
+// Number of pdf instance to run simultaneously per cluster
+var default_concurrency = 1;
+
+if(node_env === "production") {
+  default_port = 80;
+}
+
+// Exports configuration
+module.exports = {
+  env: node_env,
+  port: process.env.PORT || default_port,
+  workers: process.env.WORKERS || 2,
+
+  pdf_version: process.env.pdf_VERSION || default_pdf_version,
+  pdf_path: process.env.pdf_PATH || default_pdf_path,
+  concurrency: process.env.pdf_CONCURRENCY || default_concurrency
+};

Diff for: config/routes.js

@@ -0,0 +1,11 @@
+'use strict';
+/**
+ * @file Defines the routes available on the server
+
+ * Will define all availables exposed HTTP paths, and their methods (GET / POST / ...).
+ */
+
+// Routes client requests to handlers
+module.exports = function router(server, handlers) {
+  server.post('/hydrate', handlers.hydrater);
+};

Diff for: lib/hydrater-pdf/handlers/hydrater.js

@@ -0,0 +1,38 @@
+'use strict';
+
+/**
+ * @file Define the pdf hydrater endpoint
+ *
+ * Extract generic information from files
+ *
+ */
+
+var restify = require('restify');
+
+
+/**
+ * This handler receives a document on a POST request and process the document
+ * to find the content, the language and the associated meta-datas
+ *
+ * @param {Object} req Request object from the client
+ * @param {Object} res Response we want to return
+ * @param {Function} next Callback to call once res has been populated.
+ */
+module.exports = function(req, res, next) {
+  if(!req.params.file_path) {
+    return next(new restify.BadMethodError('No file to process'));
+  } else if(!req.params.callback) {
+    return next(new restify.BadMethodError('No specified callback'));
+  }
+
+  res.send(204);
+  next();
+
+  // Push to queue
+  require('../../../app.js').queue.push({
+    file_path: req.params.file_path,
+    callback: req.params.callback
+  });
+
+  console.log("Queuing: ", req.params.file_path);
+};