eCFR test harness for citation

test_harness.js

#!/usr/bin/env node

var Citation = require('../citation');
var fs = require('fs');
var sax = require('sax');
var unzip = require('unzip2');

function parserFactory() {
  var stream = sax.createStream(true, {trim: true});
  var in_contents = 0;
  var title;
  stream.on('opentag', function(node) {
    if (node.name == 'CONTENTS') {
      in_contents++;
    }
  });
  stream.on('closetag', function(name) {
    if (name == 'CONTENTS') {
      in_contents--;
    }
  });
  stream.on('text', function(text) {
    if (in_contents != 0) {
      return;
    }
    var tag_name = stream._parser.tag.name;
    if (tag_name == 'SECTNO') {
      var input = title + ' CFR ' + text;
      var output = Citation.find(input).citations;
      if (output.length == 0) {
        console.log("Did not parse " + input);
      } else if (output.length == 1) {
        if (input == output[0].match) {
        } else {
          console.log("Incorrect match found for " + input);
        }
      } else {
        console.log("More than one match found for " + input);
        console.log(output);
      }
    } else if (tag_name == 'TITLENUM') {
      title = text.replace('Title ', '');
    }
  });
  return stream
}

if (process.argv.length <= 2) {
  console.error("Usage: Download a bulk data zip file from http://www.gpo.gov/fdsys/bulkdata/CFR, then pass its name as a command line argument");
}

process.argv.slice(2).forEach(function(path) {
  fs.createReadStream(path)
    .pipe(unzip.Parse())
    .on('entry', function(entry) {
      var name = entry.path;
      var type = entry.type;
      if (type == 'File') {
        console.log(name);
        var parser = parserFactory();
        entry.pipe(parser);
      } else {
        entry.autodrain();
      }
    });
});