xml2xml.js

#!/usr/bin/env node
// 
// example :
// find xml-orig/ -type f | ./xml2xml.js -c 10 --xslt mafeuille.xslt --dpath ./xml-dest --epath ./xml-error
//
// Requirements:
// npm install shelljs optimist lazy async
// 
// tentative non concluante :
// sudo apt-get install libxml2-dev
// sudo apt-get install libxslt-dev
// npm install node_xslt
// var xslt  = require('node_xslt')
// var stylesheet = xslt.readXsltFile('mafeuille.xsl');
// var document = '';
//   document = xslt.readXmlFile(task.orig);
//   var transformedString = xslt.transform(stylesheet, document);
//   fs.fileWriteSync(task.dest, transformedString);


"use strict";

var fs     = require('fs');
var shell  = require('shelljs');
var Lazy   = require('lazy');
var async  = require('async');

var start_time = new Date();

// get the command line argument 
var optimist = require('optimist')
  .usage('Usage: $0 -c [number] --xslt [filepath] --dpath [destinationpath] --epath [errorpath]')
  .demand('concurrency').alias('concurrency', 'c').default('concurrency', 4)
  .describe('concurrency', 'how many XML are converted in parallele')
  .demand('xslt').describe('xslt', 'the stylsheet tu use for the xml transformations')
  .demand('dpath').describe('dpath', 'where is stored transformed xml')
  .demand('epath').describe('epath', 'where is stored error xml')
;
var argv = optimist.argv;
if (argv.help) {
  optimist.showHelp();
  process.exit(0);
}

var xsltErrors = {
  0: 'No error (normal operation)',
  1: 'No argument',
  2: 'Too many parameters',
  3: 'Unknown option',
  4: 'Failed to parse the stylesheet',
  5: 'Error in the stylesheet',
  6: 'Error in one of the documents',
  7: 'Unsupported xsl:output method',
  8: 'String parameter contains both quote and double-quotes',
  9: 'Internal processing error',
  10: 'Processing was stopped by a terminating message',
  11: 'Could not write the result to the output file',
};

// count number of transformed xml files 
var nb = 0;

// create a queue to handle with concurrency several tasks
// one task is one xml transformation using xsltproc
var q = async.queue(function (task, callback) {
  
  shell.exec('xsltproc --nonet -o ' + task.dest + ' ' + argv.xslt + ' ' + task.orig, {silent: true, async: true}, function (code, output) {
    nb++;
    
    // if an error is returned by xsltproc, write it !
    if (code != 0) {
      fs.writeFile(task.error, output + '\n\n' + xsltErrors[code], callback);
      console.log(code);
    } else {
      // just a signal on the terminal to tell that code is running
      process.stdout.write('.');
      callback();
    }
  });    

}, argv.concurrency);

// called when queue is empty
q.drain = function() {
  var end_time = new Date();
  console.log('');
  console.log(nb + ' items have been processed');
  console.log('start time: ' + start_time);
  console.log('end time: ' + end_time);
  console.log('elapsed time: ' + (end_time - start_time)/1000 + ' seconds');
}

// split input stream line by line
// one line is one xml path
Lazy(process.stdin)
  .lines
  .map(String)
  .map(function (line) {
    q.push({
      orig:  line,
      dest:  line.replace(line.split('/')[0], argv.dpath),
      error: line.replace(line.split('/')[0], argv.epath)
    });
  });