add zero_prefixed_house_numbers script

Document.js

@@ -42,6 +42,7 @@ function Document( source, layer, source_id ){
   this.addPostProcessingScript( require('./post/intersections') );
   this.addPostProcessingScript( require('./post/seperable_street_names').post );
   this.addPostProcessingScript( require('./post/alphanumeric_postcodes') );
+  this.addPostProcessingScript( require('./post/zero_prefixed_house_numbers') );
   this.addPostProcessingScript( require('./post/deduplication') );
   this.addPostProcessingScript( require('./post/language_field_trimming') );
   this.addPostProcessingScript( require('./post/popularity') );

post/zero_prefixed_house_numbers.js

@@ -0,0 +1,28 @@
+/**
+ * Zero prefixed house number post-processing script strips leading zeros from
+ * house numbers. eg. house number `001` -> `1`.
+ *
+ * This functionality was previously handled in elasticsearch
+ * using a `removeAllZeroNumericPrefix` filter, see:
+ * - https://github.com/pelias/schema/issues/503
+ */
+
+const _ = require('lodash');
+
+function housenumber(doc) {
+
+  // ensure housenumber is set
+  let houseno = doc.getAddress('number');
+  if( !_.isString(houseno) || _.isEmpty(houseno) ){ return; }
+
+  // trim leading/trailing whitespace
+  houseno = houseno.trim();
+
+  // only applies to housenumbers prefixed with a zero
+  if ( houseno.length < 2 || !houseno.startsWith('0') ) { return; }
+
+  // trim leading zeros
+  doc.setAddress('number', houseno.replace(/^0+/, ''));
+}
+
+module.exports = housenumber;

test/document/post.js

@@ -3,12 +3,13 @@ const Document = require('../../Document');
 const intersections = require('../../post/intersections');
 const seperable_street_names = require('../../post/seperable_street_names').post;
 const alphanumeric_postcodes = require('../../post/alphanumeric_postcodes');
+const zero_prefixed_house_numbers = require('../../post/zero_prefixed_house_numbers');
 const deduplication = require('../../post/deduplication');
 const language_field_trimming = require('../../post/language_field_trimming');
 const popularity = require('../../post/popularity');
 const DEFAULT_SCRIPTS = [
   intersections, seperable_street_names, alphanumeric_postcodes, 
-  deduplication, language_field_trimming, popularity
+  zero_prefixed_house_numbers, deduplication, language_field_trimming, popularity
 ];
 
 module.exports.tests = {};

test/post/zero_prefixed_house_numbers.js

@@ -0,0 +1,90 @@
+const Document = require('../../Document');
+const housenumber = require('../../post/zero_prefixed_house_numbers');
+
+module.exports.tests = {};
+
+module.exports.tests.noop = function(test) {
+  test('noop: house number not set', function(t) {
+    const doc = new Document('mysource', 'mylayer', 'myid');
+
+    housenumber(doc);
+
+    // no action taken
+    t.equal(doc.getAddress('number'), undefined, 'not set');
+
+    t.end();
+  });
+
+  test('noop: house no leading zero', function(t) {
+    const doc = new Document('mysource', 'mylayer', 'myid');
+    doc.setAddress('number', '10');
+
+    housenumber(doc);
+
+    // no action taken
+    t.equal(doc.getAddress('number'), '10', 'no change');
+
+    t.end();
+  });
+
+  test('noop: house no is literally zero', function(t) {
+    const doc = new Document('mysource', 'mylayer', 'myid');
+    doc.setAddress('number', '0');
+
+    housenumber(doc);
+
+    // no action taken
+    t.equal(doc.getAddress('number'), '0', 'no change');
+
+    t.end();
+  });
+};
+
+module.exports.tests.strip_prefix = function(test) {
+  test('strip: house number with zero prefix', function(t) {
+    const doc = new Document('mysource', 'mylayer', 'myid');
+    doc.setAddress('number', '010');
+
+    housenumber(doc);
+
+    // prefix removed
+    t.equal(doc.getAddress('number'), '10', 'prefix removed');
+
+    t.end();
+  });
+
+  test('strip: house number with multiple zero prefix', function(t) {
+    const doc = new Document('mysource', 'mylayer', 'myid');
+    doc.setAddress('number', '000000001000');
+
+    housenumber(doc);
+
+    // prefix removed
+    t.equal(doc.getAddress('number'), '1000', 'prefix removed');
+
+    t.end();
+  });
+
+  test('strip: house number with multiple zero prefix plus multiple whitespace', function(t) {
+    const doc = new Document('mysource', 'mylayer', 'myid');
+    doc.setAddress('number', '       000000009990');
+
+    housenumber(doc);
+
+    // prefix removed
+    t.equal(doc.getAddress('number'), '9990', 'prefix removed');
+
+    t.end();
+  });
+};
+
+module.exports.all = function (tape, common) {
+
+  function test(name, testFunction) {
+    return tape('post/zero_prefixed_house_numbers: ' + name, testFunction);
+  }
+
+  for( var testCase in module.exports.tests ){
+    module.exports.tests[testCase](test, common);
+  }
+};

test/run.js

@@ -24,6 +24,7 @@ const tests = [
   require('./document/post.js'),
   require('./post/intersections.js'),
   require('./post/alphanumeric_postcodes.js'),
+  require('./post/zero_prefixed_house_numbers.js'),
   require('./post/deduplication.js'),
   require('./post/seperable_street_names.js'),
   require('./post/language_field_trimming.js'),