From 1cfe49610692b2a4d488465872e622a438aafd93 Mon Sep 17 00:00:00 2001
From: Peter Johnson <insomnia@rcpt.at>
Date: Wed, 11 Dec 2024 12:47:02 +0100
Subject: [PATCH] feat(Document): strip URLs from name/address

---
 Document.js              | 12 ++++++++----
 test/document/address.js | 12 ++++++++----
 test/document/name.js    | 12 ++++++++----
 util/transform.js        |  2 ++
 4 files changed, 26 insertions(+), 12 deletions(-)

diff --git a/Document.js b/Document.js
index 2ae9702..8a35348 100644
--- 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
index 2f10d91..4327643 100644
--- 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
index ddddfc3..9e0ee70 100644
--- 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
index 22223bd..7ca143b 100644
--- 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();
\ No newline at end of file