From 8c5555a64fab89997c59fc457e5c265b93b39a7f Mon Sep 17 00:00:00 2001
From: missinglink <insomnia@rcpt.at>
Date: Mon, 13 Jun 2022 12:43:18 +0200
Subject: [PATCH] feat(analysis): append ordinal suffix to numeric street names

---
 lib/analysis/Token.js    |  4 ++++
 lib/analysis/ordinals.js | 45 ++++++++++++++++++++++++++++++++++++++++
 lib/cleanup_v2.js        |  4 ++++
 test/cleanup_v2.js       | 25 ++++++++++++++++++++--
 4 files changed, 76 insertions(+), 2 deletions(-)
 create mode 100644 lib/analysis/ordinals.js

diff --git a/lib/analysis/Token.js b/lib/analysis/Token.js
index abdd9f49a..9969216c7 100644
--- a/lib/analysis/Token.js
+++ b/lib/analysis/Token.js
@@ -9,6 +9,10 @@ class Token {
     return _.isString(this.body) && !_.isEmpty(this.body);
   }
 
+  isNumeric() {
+    return /^\d+$/.test(this.body);
+  }
+
   findCase() {
     if (this.body === _.toLower(this.body)) { return Token.LOWERCASED; }
     if (this.body === _.toUpper(this.body)) { return Token.UPPERCASED; }
diff --git a/lib/analysis/ordinals.js b/lib/analysis/ordinals.js
new file mode 100644
index 000000000..870a3f39e
--- /dev/null
+++ b/lib/analysis/ordinals.js
@@ -0,0 +1,45 @@
+const _ = require('lodash');
+
+// The ordinal function replaces all numeric street names (ie. 30 street)
+// with a version including ordinals (ie. 30th street).
+// note: this is currently only configured for the English language
+
+function ordinals(opts) {
+  return (tokens) => {
+
+    // consider all but final token
+    for (var o = 0; o < tokens.length-1; o++) {
+
+      // token must be entirely numeric
+      if (!tokens[o].isNumeric()) { continue; }
+
+      // token must be followed by a street type token
+      if (!_.has(opts.dict.streetTypes, _.toLower(tokens[o+1].body))) { continue; }
+
+      // token must either be the leftmost token or be preceeded by a directional token
+      if(o !== 0) {
+        if (!_.has(opts.dict.directionalExpansions, _.toLower(tokens[o-1].body))) {
+          continue;
+        }
+      }
+
+      // append the english ordinal suffix
+      tokens[o].body += englishOrdinalSuffix(tokens[o].body);
+
+      // maximum of one replacement
+      break;
+    }
+
+    return tokens;
+  };
+}
+
+function englishOrdinalSuffix(i) {
+  const j = i % 10, k = i % 100;
+  if (j === 1 && k !== 11) { return 'st'; }
+  if (j === 2 && k !== 12) { return 'nd'; }
+  if (j === 3 && k !== 13) { return 'rd'; }
+  return 'th';
+}
+
+module.exports = ordinals;
diff --git a/lib/cleanup_v2.js b/lib/cleanup_v2.js
index 8a91f2633..45f612ab1 100644
--- a/lib/cleanup_v2.js
+++ b/lib/cleanup_v2.js
@@ -1,6 +1,7 @@
 const _ = require('lodash');
 const dictionary = require('./analysis/dictionary');
 const synonyms = require('./analysis/synonyms');
+const ordinals = require('./analysis/ordinals');
 const Token = require('./analysis/Token');
 
 /**
@@ -116,6 +117,9 @@ function cleanupStreetName(input) {
   // capitalize lowercased tokens (leaving mixed case tokens unchanged)
   tokens.forEach(token => token.selectivelyCapitalize());
 
+  // add ordinals to english numeric street names
+  tokens = ordinals({ dict })(tokens);
+
   // convert objects to strings and join by whitespace
   return tokens.map(token => token.body).join(' ');
 }
diff --git a/test/cleanup_v2.js b/test/cleanup_v2.js
index 2f4b48c65..27c6465a5 100644
--- a/test/cleanup_v2.js
+++ b/test/cleanup_v2.js
@@ -181,6 +181,27 @@ tape('contract english diagonals - last token position', (t) => {
   t.end();
 });
 
+// add missing English street name ordinals
+tape('add missing English street name ordinals', (t) => {
+  t.equal(analyzer('W 26 St'), 'West 26th Street');
+  t.equal(analyzer('W 26th St'), 'West 26th Street');
+  t.equal(analyzer('1 St'), '1st Street');
+  t.equal(analyzer('2 Rd'), '2nd Road');
+  t.equal(analyzer('3 Ave'), '3rd Avenue');
+  t.equal(analyzer('4 Ln'), '4th Lane');
+  t.equal(analyzer('11 St'), '11th Street');
+  t.equal(analyzer('12 Rd'), '12th Road');
+  t.equal(analyzer('13 Ave'), '13th Avenue');
+  t.equal(analyzer('14 Ln'), '14th Lane');
+  t.equal(analyzer('101 St'), '101st Street');
+  t.equal(analyzer('102 Rd'), '102nd Road');
+  t.equal(analyzer('103 Ave'), '103rd Avenue');
+  t.equal(analyzer('104 Ln'), '104th Lane');
+  t.equal(analyzer('no 1 st'), 'No 1 Street');
+  t.equal(analyzer('no #1 st'), 'No #1 Street');
+  t.end();
+});
+
 // --- NOOP inputs which should never change ---
 
 // no-ops, these inputs should not change regardless of the algorithm used
@@ -223,7 +244,7 @@ tape('misc', (t) => {
   t.equal(analyzer('YELLOWSTONE BLVD'), 'Yellowstone Boulevard');
   t.equal(analyzer('YESHIVA LN'), 'Yeshiva Lane');
   t.equal(analyzer('WYGANT PL'), 'Wygant Place');
-  t.equal(analyzer('W  262 ST'), 'West 262 Street');
+  t.equal(analyzer('W  262 ST'), 'West 262nd Street');
   t.equal(analyzer('W 26TH ST'), 'West 26th Street');
   t.equal(analyzer('WILLIE MC DONALD WAY'), 'Willie Mc Donald Way');
   t.equal(analyzer('West 93rd Street'), 'West 93rd Street');
@@ -232,7 +253,7 @@ tape('misc', (t) => {
   t.equal(analyzer('E  HAMPTON BLVD'), 'East Hampton Boulevard');
   t.equal(analyzer('MARATHON PKWY'), 'Marathon Parkway');
   t.equal(analyzer('ANDREWS AVE  S'), 'Andrews Avenue South');
-  t.equal(analyzer('W  13 ST'), 'West 13 Street');
+  t.equal(analyzer('W  13 ST'), 'West 13th Street');
   t.end();
 });