feat(Document): strip URLs from name/address

pelias · Dec 13, 2024 · 1cfe496 · 1cfe496
1 parent bdbcac5
commit 1cfe496
Show file tree

Hide file tree

Showing 4 changed files with 26 additions and 12 deletions.
diff --git a/Document.js b/Document.js
@@ -242,7 +242,8 @@ Document.prototype.setName = function( prop, value ){
 
   validate.type('string', value);
   validate.truthy(value);
-  validate.regex.nomatch(value, /https?:\/\//);
+  value = transform.stripURLs(value);
+  if (!value.length) { return; }
 
   if( Array.isArray( this.name[ prop ] ) ){
     this.name[ prop ][ 0 ] = value;
@@ -257,7 +258,8 @@ Document.prototype.setNameAlias = function( prop, value ){
 
   validate.type('string', value);
   validate.truthy(value);
-  validate.regex.nomatch(value, /https?:\/\//);
+  value = transform.stripURLs(value);
+  if (!value.length) { return; }
 
   // is this the first time setting this prop? ensure it's an array
   if( !this.hasName( prop ) ){
@@ -405,7 +407,8 @@ Document.prototype.setAddress = function( prop, value ){
   validate.type('string', value);
   validate.truthy(value);
   validate.property(addressFields, prop);
-  validate.regex.nomatch(value, /https?:\/\//);
+  value = transform.stripURLs(value);
+  if (!value.length) { return; }
 
   if( Array.isArray( this.address_parts[ prop ] ) ){
     this.address_parts[ prop ][ 0 ] = value;
@@ -421,7 +424,8 @@ Document.prototype.setAddressAlias = function( prop, value ){
   validate.type('string', value);
   validate.truthy(value);
   validate.property(addressFields, prop);
-  validate.regex.nomatch(value, /https?:\/\//);
+  value = transform.stripURLs(value);
+  if (!value.length) { return; }
 
   // is this the first time setting this prop? ensure it's an array
   if( !this.hasAddress( prop ) ){

diff --git a/test/document/address.js b/test/document/address.js
@@ -51,8 +51,10 @@ module.exports.tests.setAddress = function(test) {
   });
   test('setAddress - http regex', function (t) {
     var doc = new Document('mysource', 'mylayer', 'myid');
-    t.throws(doc.setAddress.bind(doc, 'number', 'http://www.pelias.io'), /invalid regex/, 'regex failure');
-    t.throws(doc.setAddress.bind(doc, 'number', 'AAhttp://www.pelias.ioBB'), /invalid regex/, 'regex failure');
+    t.doesNotThrow(doc.setAddress.bind(doc, 'number', 'http://www.pelias.io'), /invalid regex/);
+    t.doesNotThrow(doc.setAddress.bind(doc, 'number', 'AAhttp://www.pelias.ioBB'), /invalid regex/);
+    t.doesNotThrow(doc.setAddress.bind(doc, 'number', 'AA http://www.pelias.io BB'), /invalid regex/);
+    t.deepEqual(doc.address_parts.number, 'AA  BB', 'URLs removed');
     t.end();
   });
 };
@@ -116,8 +118,10 @@ module.exports.tests.setAddressAlias = function(test) {
   });
   test('setAddressAlias - http regex', function (t) {
     var doc = new Document('mysource', 'mylayer', 'myid');
-    t.throws(doc.setAddressAlias.bind(doc, 'number', 'http://www.pelias.io'), /invalid regex/, 'regex failure');
-    t.throws(doc.setAddressAlias.bind(doc, 'number', 'AAhttp://www.pelias.ioBB'), /invalid regex/, 'regex failure');
+    t.doesNotThrow(doc.setAddressAlias.bind(doc, 'number', 'http://www.pelias.io'), /invalid regex/);
+    t.doesNotThrow(doc.setAddressAlias.bind(doc, 'number', 'AAhttp://www.pelias.ioBB'), /invalid regex/);
+    t.doesNotThrow(doc.setAddressAlias.bind(doc, 'number', 'AA http://www.pelias.io BB'), /invalid regex/);
+    t.deepEqual(doc.address_parts.number, ['AA', 'AA', 'AA  BB'], 'URLs removed');
     t.end();
   });
 };

diff --git a/test/document/name.js b/test/document/name.js
@@ -41,8 +41,10 @@ module.exports.tests.setName = function(test) {
   });
   test('setName - http regex', function (t) {
     var doc = new Document('mysource', 'mylayer', 'myid');
-    t.throws(doc.setName.bind(doc, 'default', 'http://www.pelias.io'), /invalid regex/, 'regex failure');
-    t.throws(doc.setName.bind(doc, 'default', 'AAhttp://www.pelias.ioBB'), /invalid regex/, 'regex failure');
+    t.doesNotThrow(doc.setName.bind(doc, 'default', 'http://www.pelias.io'), /invalid regex/);
+    t.doesNotThrow(doc.setName.bind(doc, 'default', 'AAhttp://www.pelias.ioBB'), /invalid regex/);
+    t.doesNotThrow(doc.setName.bind(doc, 'default', 'AA http://www.pelias.io BB'), /invalid regex/);
+    t.deepEqual(doc.name.default, 'AA  BB', 'URLs removed');
     t.end();
   });
 };
@@ -106,8 +108,10 @@ module.exports.tests.setNameAlias = function(test) {
   });
   test('setNameAlias - http regex', function (t) {
     var doc = new Document('mysource', 'mylayer', 'myid');
-    t.throws(doc.setNameAlias.bind(doc, 'default', 'http://www.pelias.io'), /invalid regex/, 'regex failure');
-    t.throws(doc.setNameAlias.bind(doc, 'default', 'AAhttp://www.pelias.ioBB'), /invalid regex/, 'regex failure');
+    t.doesNotThrow(doc.setNameAlias.bind(doc, 'default', 'http://www.pelias.io'), /invalid regex/);
+    t.doesNotThrow(doc.setNameAlias.bind(doc, 'default', 'AAhttp://www.pelias.ioBB'), /invalid regex/);
+    t.doesNotThrow(doc.setNameAlias.bind(doc, 'default', 'AA http://www.pelias.io BB'), /invalid regex/);
+    t.deepEqual(doc.name.default, ['AA', 'AA', 'AA  BB'], 'URLs removed');
     t.end();
   });
 };

diff --git a/util/transform.js b/util/transform.js
@@ -31,3 +31,5 @@ module.exports.toULLR = function( val ) {
     max_lon: val.lowerRight.lon
   });
 };
+
+module.exports.stripURLs = (val) => val.replace(/(?:https?|ftp):\/\/\S*/g, '').trim();
-Original file line number
+Diff line change
@@ Expand Up / @@ -31,3 +31,5 @@ module.exports.toULLR = function( val ) { @@
         max_lon: val.lowerRight.lon
       });
     };
+    module.exports.stripURLs = (val) => val.replace(/(?:https?|ftp):\/\/\S*/g, '').trim();