Browse Source

Merge pull request #457 from pelias/rename-address-prop

Rename address prop
pull/459/head
Julian Simioni 9 years ago
parent
commit
5e2e71fa15
  1. 25
      middleware/confidenceScore.js
  2. 10
      middleware/dedupe.js
  3. 10
      middleware/localNamingConventions.js
  4. 4
      middleware/renamePlacenames.js
  5. 2
      package.json
  6. 6
      query/autocomplete_defaults.js
  7. 6
      query/reverse_defaults.js
  8. 6
      query/search_defaults.js
  9. 2
      test/unit/fixture/dedupe_elasticsearch_nonascii_results.js
  10. 22
      test/unit/fixture/dedupe_elasticsearch_results.js
  11. 6
      test/unit/fixture/search_full_address.js
  12. 4
      test/unit/fixture/search_regions_address.js
  13. 4
      test/unit/middleware/localNamingConventions.js

25
middleware/confidenceScore.js

@ -100,9 +100,10 @@ function checkForDealBreakers(req, hit) {
return true; return true;
} }
if (check.assigned(req.clean.parsed_text.postalcode) && check.assigned(hit.address) && if (check.assigned(req.clean.parsed_text.postalcode) && check.assigned(hit.address_parts) &&
req.clean.parsed_text.postalcode !== hit.address.zip) { req.clean.parsed_text.postalcode !== hit.address_parts.zip) {
logger.debug('[confidence][deal-breaker]: postalcode !== zip (' + req.clean.parsed_text.postalcode + ' !== ' + hit.address.zip + ')'); logger.debug('[confidence][deal-breaker]: postalcode !== zip (' + req.clean.parsed_text.postalcode +
' !== ' + hit.address_parts.zip + ')');
return true; return true;
} }
} }
@ -155,8 +156,8 @@ function checkName(text, parsed_text, hit) {
*/ */
function checkQueryType(text, hit) { function checkQueryType(text, hit) {
if (check.assigned(text) && check.assigned(text.number) && if (check.assigned(text) && check.assigned(text.number) &&
(check.undefined(hit.address) || (check.undefined(hit.address_parts) ||
(check.assigned(hit.address) && check.undefined(hit.address.number)))) { (check.assigned(hit.address_parts) && check.undefined(hit.address_parts.number)))) {
return 0; return 0;
} }
return 1; return 1;
@ -206,10 +207,10 @@ function propMatch(textProp, hitProp, expectEnriched) {
* @param {string} [text.state] * @param {string} [text.state]
* @param {string} [text.country] * @param {string} [text.country]
* @param {object} hit * @param {object} hit
* @param {object} [hit.address] * @param {object} [hit.address_parts]
* @param {string|number} [hit.address.number] * @param {string|number} [hit.address_parts.number]
* @param {string} [hit.address.street] * @param {string} [hit.address_parts.street]
* @param {string|number} [hit.address.zip] * @param {string|number} [hit.address_parts.zip]
* @param {Array} [hit.parent.region_a] * @param {Array} [hit.parent.region_a]
* @param {Array} [hit.parent.country_a] * @param {Array} [hit.parent.country_a]
* @returns {number} * @returns {number}
@ -219,9 +220,9 @@ function checkAddress(text, hit) {
var res = 0; var res = 0;
if (check.assigned(text) && check.assigned(text.number) && check.assigned(text.street)) { if (check.assigned(text) && check.assigned(text.number) && check.assigned(text.street)) {
res += propMatch(text.number, (hit.address ? hit.address.number : null), false); res += propMatch(text.number, (hit.address_parts ? hit.address_parts.number : null), false);
res += propMatch(text.street, (hit.address ? hit.address.street : null), false); res += propMatch(text.street, (hit.address_parts ? hit.address_parts.street : null), false);
res += propMatch(text.postalcode, (hit.address ? hit.address.zip: null), true); res += propMatch(text.postalcode, (hit.address_parts ? hit.address_parts.zip: null), true);
res += propMatch(text.state, hit.parent.region_a[0], true); res += propMatch(text.state, hit.parent.region_a[0], true);
res += propMatch(text.country, hit.parent.country_a[0], true); res += propMatch(text.country, hit.parent.country_a[0], true);

10
middleware/dedupe.js

@ -54,12 +54,12 @@ function isDifferent(item1, item2) {
propMatch(item1, item2, 'name'); propMatch(item1, item2, 'name');
} }
if (item1.hasOwnProperty('address') && item2.hasOwnProperty('address')) { if (item1.hasOwnProperty('address_parts') && item2.hasOwnProperty('address_parts')) {
propMatch(item1.address, item2.address, 'number'); propMatch(item1.address_parts, item2.address_parts, 'number');
propMatch(item1.address, item2.address, 'street'); propMatch(item1.address_parts, item2.address_parts, 'street');
propMatch(item1.address, item2.address, 'zip'); propMatch(item1.address_parts, item2.address_parts, 'zip');
} }
else if (item1.address !== item2.address) { else if (item1.address_parts !== item2.address_parts) {
throw new Error('different'); throw new Error('different');
} }
} }

10
middleware/localNamingConventions.js

@ -14,9 +14,9 @@ function applyLocalNamingConventions(req, res, next) {
res.data.filter(function(place){ res.data.filter(function(place){
// only relevant for German addresses // only relevant for German addresses
if( place.parent.country_a.indexOf('DEU') === -1 ){ return false; } if( place.parent.country_a.indexOf('DEU') === -1 ){ return false; }
if( !place.hasOwnProperty('address') ){ return false; } if( !place.hasOwnProperty('address_parts') ){ return false; }
if( !place.address.hasOwnProperty('number') ){ return false; } if( !place.address_parts.hasOwnProperty('number') ){ return false; }
if( !place.address.hasOwnProperty('street') ){ return false; } if( !place.address_parts.hasOwnProperty('street') ){ return false; }
return true; return true;
}) })
.forEach( flipNumberAndStreet ); .forEach( flipNumberAndStreet );
@ -27,8 +27,8 @@ function applyLocalNamingConventions(req, res, next) {
// DE address should have the housenumber and street name flipped // DE address should have the housenumber and street name flipped
// eg. '101 Grolmanstraße' -> 'Grolmanstraße 101' // eg. '101 Grolmanstraße' -> 'Grolmanstraße 101'
function flipNumberAndStreet(place) { function flipNumberAndStreet(place) {
var standard = ( place.address.number + ' ' + place.address.street ), var standard = ( place.address_parts.number + ' ' + place.address_parts.street ),
flipped = ( place.address.street + ' ' + place.address.number ); flipped = ( place.address_parts.street + ' ' + place.address_parts.number );
// flip street name and housenumber // flip street name and housenumber
if( place.name.default === standard ){ if( place.name.default === standard ){

4
middleware/renamePlacenames.js

@ -58,9 +58,9 @@ function renamePlacenames(req, res, next) {
* Rename the fields in one record * Rename the fields in one record
*/ */
function renameOneRecord(place) { function renameOneRecord(place) {
if (place.address) { if (place.address_parts) {
Object.keys(ADDRESS_PROPS).forEach(function (prop) { Object.keys(ADDRESS_PROPS).forEach(function (prop) {
place[ADDRESS_PROPS[prop]] = place.address[prop]; place[ADDRESS_PROPS[prop]] = place.address_parts[prop];
}); });
} }

2
package.json

@ -53,7 +53,7 @@
"morgan": "1.7.0", "morgan": "1.7.0",
"pelias-config": "^1.0.1", "pelias-config": "^1.0.1",
"pelias-logger": "^0.0.8", "pelias-logger": "^0.0.8",
"pelias-query": "^6.1.0", "pelias-query": "6.2.0",
"pelias-suggester-pipeline": "2.0.4", "pelias-suggester-pipeline": "2.0.4",
"stats-lite": "1.0.3", "stats-lite": "1.0.3",
"through2": "2.0.1" "through2": "2.0.1"

6
query/autocomplete_defaults.js

@ -39,15 +39,15 @@ module.exports = _.merge({}, peliasQuery.defaults, {
'function_score:boost_mode': 'multiply', 'function_score:boost_mode': 'multiply',
'address:housenumber:analyzer': 'peliasHousenumber', 'address:housenumber:analyzer': 'peliasHousenumber',
'address:housenumber:field': 'address.number', 'address:housenumber:field': 'address_parts.number',
'address:housenumber:boost': 2, 'address:housenumber:boost': 2,
'address:street:analyzer': 'peliasStreet', 'address:street:analyzer': 'peliasStreet',
'address:street:field': 'address.street', 'address:street:field': 'address_parts.street',
'address:street:boost': 5, 'address:street:boost': 5,
'address:postcode:analyzer': 'peliasZip', 'address:postcode:analyzer': 'peliasZip',
'address:postcode:field': 'address.zip', 'address:postcode:field': 'address_parts.zip',
'address:postcode:boost': 2000, 'address:postcode:boost': 2000,
'admin:country_a:analyzer': 'standard', 'admin:country_a:analyzer': 'standard',

6
query/reverse_defaults.js

@ -39,15 +39,15 @@ module.exports = _.merge({}, peliasQuery.defaults, {
'function_score:boost_mode': 'replace', 'function_score:boost_mode': 'replace',
'address:housenumber:analyzer': 'peliasHousenumber', 'address:housenumber:analyzer': 'peliasHousenumber',
'address:housenumber:field': 'address.number', 'address:housenumber:field': 'address_parts.number',
'address:housenumber:boost': 2, 'address:housenumber:boost': 2,
'address:street:analyzer': 'peliasStreet', 'address:street:analyzer': 'peliasStreet',
'address:street:field': 'address.street', 'address:street:field': 'address_parts.street',
'address:street:boost': 5, 'address:street:boost': 5,
'address:postcode:analyzer': 'peliasZip', 'address:postcode:analyzer': 'peliasZip',
'address:postcode:field': 'address.zip', 'address:postcode:field': 'address_parts.zip',
'address:postcode:boost': 3, 'address:postcode:boost': 3,
'admin:country_a:analyzer': 'standard', 'admin:country_a:analyzer': 'standard',

6
query/search_defaults.js

@ -39,15 +39,15 @@ module.exports = _.merge({}, peliasQuery.defaults, {
'function_score:boost_mode': 'replace', 'function_score:boost_mode': 'replace',
'address:housenumber:analyzer': 'peliasHousenumber', 'address:housenumber:analyzer': 'peliasHousenumber',
'address:housenumber:field': 'address.number', 'address:housenumber:field': 'address_parts.number',
'address:housenumber:boost': 2, 'address:housenumber:boost': 2,
'address:street:analyzer': 'peliasStreet', 'address:street:analyzer': 'peliasStreet',
'address:street:field': 'address.street', 'address:street:field': 'address_parts.street',
'address:street:boost': 5, 'address:street:boost': 5,
'address:postcode:analyzer': 'peliasZip', 'address:postcode:analyzer': 'peliasZip',
'address:postcode:field': 'address.zip', 'address:postcode:field': 'address_parts.zip',
'address:postcode:boost': 20, 'address:postcode:boost': 20,
'admin:country_a:analyzer': 'standard', 'admin:country_a:analyzer': 'standard',

2
test/unit/fixture/dedupe_elasticsearch_nonascii_results.js

@ -55,7 +55,7 @@ module.exports = [
'name': { 'name': {
'default': '万里加油站' 'default': '万里加油站'
}, },
'address': { 'address_parts': {
'street': 'S308', 'street': 'S308',
'postalcode': '312044' 'postalcode': '312044'
}, },

22
test/unit/fixture/dedupe_elasticsearch_results.js

@ -4,7 +4,7 @@ module.exports = [
'lon': -76.207456, 'lon': -76.207456,
'lat': 40.039265 'lat': 40.039265
}, },
'address': {}, 'address_parts': {},
'parent': { 'parent': {
'localadmin': ['East Lampeter'], 'localadmin': ['East Lampeter'],
'region_a': ['PA'], 'region_a': ['PA'],
@ -31,7 +31,7 @@ module.exports = [
'lon': -76.207456, 'lon': -76.207456,
'lat': 40.039265 'lat': 40.039265
}, },
'address': {}, 'address_parts': {},
'parent': { 'parent': {
'localadmin': ['East Lampeter'], 'localadmin': ['East Lampeter'],
'region_a': ['PA'], 'region_a': ['PA'],
@ -58,7 +58,7 @@ module.exports = [
'lon': -76.23246, 'lon': -76.23246,
'lat': 39.99288 'lat': 39.99288
}, },
'address': {}, 'address_parts': {},
'parent': { 'parent': {
'localadmin': ['West Lampeter'], 'localadmin': ['West Lampeter'],
'region_a': ['PA'], 'region_a': ['PA'],
@ -85,7 +85,7 @@ module.exports = [
'lon': -76.20746, 'lon': -76.20746,
'lat': 40.03927 'lat': 40.03927
}, },
'address': {}, 'address_parts': {},
'parent': { 'parent': {
'localadmin': ['East Lampeter'], 'localadmin': ['East Lampeter'],
'region_a': ['PA'], 'region_a': ['PA'],
@ -112,7 +112,7 @@ module.exports = [
'lon': -76.232457, 'lon': -76.232457,
'lat': 39.992877 'lat': 39.992877
}, },
'address': {}, 'address_parts': {},
'parent': { 'parent': {
'region': ['Pennsylvania'], 'region': ['Pennsylvania'],
'locality': ['Lampeter'], 'locality': ['Lampeter'],
@ -139,7 +139,7 @@ module.exports = [
'lon': -76.207456, 'lon': -76.207456,
'lat': 40.038987 'lat': 40.038987
}, },
'address': {}, 'address_parts': {},
'parent': { 'parent': {
'region': ['Pennsylvania'], 'region': ['Pennsylvania'],
'locality': ['Smoketown'], 'locality': ['Smoketown'],
@ -166,7 +166,7 @@ module.exports = [
'lon': -76.20746, 'lon': -76.20746,
'lat': 40.03899 'lat': 40.03899
}, },
'address': {}, 'address_parts': {},
'parent': { 'parent': {
'region': ['Pennsylvania'], 'region': ['Pennsylvania'],
'locality': ['Smoketown'], 'locality': ['Smoketown'],
@ -193,7 +193,7 @@ module.exports = [
'lon': -94.167445, 'lon': -94.167445,
'lat': 38.762788 'lat': 38.762788
}, },
'address': {}, 'address_parts': {},
'parent': { 'parent': {
'region': ['Missouri'], 'region': ['Missouri'],
'locality': ['Strasburg'], 'locality': ['Strasburg'],
@ -219,7 +219,7 @@ module.exports = [
'lon': -78.36317, 'lon': -78.36317,
'lat': 38.98445 'lat': 38.98445
}, },
'address': {}, 'address_parts': {},
'name': { 'name': {
'default': 'Strasburg High School' 'default': 'Strasburg High School'
}, },
@ -245,7 +245,7 @@ module.exports = [
'lon': -100.16516, 'lon': -100.16516,
'lat': 46.13427 'lat': 46.13427
}, },
'address': {}, 'address_parts': {},
'name': { 'name': {
'default': 'Strasburg High School' 'default': 'Strasburg High School'
}, },
@ -271,7 +271,7 @@ module.exports = [
'lon': -81.532392, 'lon': -81.532392,
'lat': 40.597578 'lat': 40.597578
}, },
'address': {}, 'address_parts': {},
'name': { 'name': {
'default': 'Strasburg High School' 'default': 'Strasburg High School'
}, },

6
test/unit/fixture/search_full_address.js

@ -77,7 +77,7 @@ module.exports = {
} }
},{ },{
'match': { 'match': {
'address.number': { 'address_parts.number': {
'query': '123', 'query': '123',
'boost': vs['address:housenumber:boost'], 'boost': vs['address:housenumber:boost'],
'analyzer': vs['address:housenumber:analyzer'] 'analyzer': vs['address:housenumber:analyzer']
@ -85,7 +85,7 @@ module.exports = {
} }
}, { }, {
'match': { 'match': {
'address.street': { 'address_parts.street': {
'query': 'main st', 'query': 'main st',
'boost': vs['address:street:boost'], 'boost': vs['address:street:boost'],
'analyzer': vs['address:street:analyzer'] 'analyzer': vs['address:street:analyzer']
@ -93,7 +93,7 @@ module.exports = {
} }
}, { }, {
'match': { 'match': {
'address.zip': { 'address_parts.zip': {
'query': '10010', 'query': '10010',
'boost': vs['address:postcode:boost'], 'boost': vs['address:postcode:boost'],
'analyzer': vs['address:postcode:analyzer'] 'analyzer': vs['address:postcode:analyzer']

4
test/unit/fixture/search_regions_address.js

@ -77,7 +77,7 @@ module.exports = {
} }
},{ },{
'match': { 'match': {
'address.number': { 'address_parts.number': {
'query': '1', 'query': '1',
'boost': vs['address:housenumber:boost'], 'boost': vs['address:housenumber:boost'],
'analyzer': vs['address:housenumber:analyzer'] 'analyzer': vs['address:housenumber:analyzer']
@ -85,7 +85,7 @@ module.exports = {
} }
}, { }, {
'match': { 'match': {
'address.street': { 'address_parts.street': {
'query': 'water st', 'query': 'water st',
'boost': vs['address:street:boost'], 'boost': vs['address:street:boost'],
'analyzer': vs['address:street:analyzer'] 'analyzer': vs['address:street:analyzer']

4
test/unit/middleware/localNamingConventions.js

@ -10,7 +10,7 @@ module.exports.tests.flipNumberAndStreet = function(test, common) {
'_type': 'test', '_type': 'test',
'name': { 'default': '1 Main St' }, 'name': { 'default': '1 Main St' },
'center_point': { 'lon': -7.131521, 'lat': 54.428866 }, 'center_point': { 'lon': -7.131521, 'lat': 54.428866 },
'address': { 'address_parts': {
'zip': 'BT77 0BG', 'zip': 'BT77 0BG',
'number': '1', 'number': '1',
'street': 'Main St' 'street': 'Main St'
@ -27,7 +27,7 @@ module.exports.tests.flipNumberAndStreet = function(test, common) {
'_type': 'test', '_type': 'test',
'name': { 'default': '23 Grolmanstraße' }, 'name': { 'default': '23 Grolmanstraße' },
'center_point': { 'lon': 13.321487, 'lat': 52.506781 }, 'center_point': { 'lon': 13.321487, 'lat': 52.506781 },
'address': { 'address_parts': {
'zip': '10623', 'zip': '10623',
'number': '23', 'number': '23',
'street': 'Grolmanstraße' 'street': 'Grolmanstraße'

Loading…
Cancel
Save