From 7d28b15f3cebee2d5739266f82cb3f0daa0f17f4 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20Fio=C5=82ka?=
 <35661939+JanF01@users.noreply.github.com>
Date: Tue, 15 Aug 2023 10:44:01 +0200
Subject: [PATCH] feat(poland): classifying more addresses for poland (#174)

---
 classifier/PlaceClassifier.js                 |  2 +-
 classifier/PostcodeClassifier.js              |  2 +-
 classifier/StreetPrefixClassifier.js          |  2 +-
 classifier/StreetSuffixClassifier.js          |  2 +-
 classifier/scheme/street.js                   | 38 +++++++++++++++++++
 .../dictionaries/libpostal/pl/place_names.txt |  1 +
 .../dictionaries/libpostal/pl/synonyms.txt    |  1 +
 test/address.pol.test.js                      | 22 +++++++++++
 8 files changed, 66 insertions(+), 4 deletions(-)
 create mode 100644 resources/pelias/dictionaries/libpostal/pl/place_names.txt
 create mode 100644 resources/pelias/dictionaries/libpostal/pl/synonyms.txt

diff --git a/classifier/PlaceClassifier.js b/classifier/PlaceClassifier.js
index 67967896..9820b6bc 100644
--- a/classifier/PlaceClassifier.js
+++ b/classifier/PlaceClassifier.js
@@ -9,7 +9,7 @@ class PlaceClassifier extends WordClassifier {
   setup () {
     // load index tokens
     this.index = {}
-    libpostal.load(this.index, ['fr', 'de', 'en'], 'place_names.txt')
+    libpostal.load(this.index, ['fr', 'de', 'en', 'pl'], 'place_names.txt')
     libpostal.generatePlurals(this.index)
   }
 
diff --git a/classifier/PostcodeClassifier.js b/classifier/PostcodeClassifier.js
index c263c30e..5352c2d4 100644
--- a/classifier/PostcodeClassifier.js
+++ b/classifier/PostcodeClassifier.js
@@ -10,7 +10,7 @@ const dictPath = path.join(__dirname, `../resources/chromium-i18n/ssl-address`)
 // const countryCodes = fs.readdirSync(dictPath)
 //   .filter(p => p.endsWith('.json'))
 //   .map(p => p.split('.')[0])
-const countryCodes = ['us', 'gb', 'fr', 'de', 'es', 'pt', 'au', 'nz', 'kr', 'jp', 'in', 'ru', 'br', 'nl']
+const countryCodes = ['us', 'gb', 'fr', 'de', 'es', 'pt', 'au', 'nz', 'kr', 'jp', 'in', 'ru', 'br', 'nl', 'pl']
 
 class PostcodeClassifier extends WordClassifier {
   setup () {
diff --git a/classifier/StreetPrefixClassifier.js b/classifier/StreetPrefixClassifier.js
index a0f8bde4..fd6a8fbd 100644
--- a/classifier/StreetPrefixClassifier.js
+++ b/classifier/StreetPrefixClassifier.js
@@ -7,7 +7,7 @@ const libpostal = require('../resources/libpostal/libpostal')
 
 // prefix languages
 // languages which use a street prefix instead of a suffix
-const prefix = ['fr', 'ca', 'es', 'pt', 'ro']
+const prefix = ['fr', 'ca', 'es', 'pt', 'ro', 'pl']
 
 class StreetPrefixClassifier extends WordClassifier {
   setup () {
diff --git a/classifier/StreetSuffixClassifier.js b/classifier/StreetSuffixClassifier.js
index 6f34bc55..0f3fb33b 100644
--- a/classifier/StreetSuffixClassifier.js
+++ b/classifier/StreetSuffixClassifier.js
@@ -7,7 +7,7 @@ const libpostal = require('../resources/libpostal/libpostal')
 
 // prefix languages
 // languages which use a street prefix instead of a suffix
-const prefix = ['fr', 'ca', 'es', 'pt', 'ro']
+const prefix = ['fr', 'ca', 'es', 'pt', 'ro', 'pl']
 
 class StreetSuffixClassifier extends WordClassifier {
   setup () {
diff --git a/classifier/scheme/street.js b/classifier/scheme/street.js
index 9df7ae43..d06f1a89 100644
--- a/classifier/scheme/street.js
+++ b/classifier/scheme/street.js
@@ -216,6 +216,44 @@ module.exports = [
       }
     ]
   },
+  {
+    // Aleja Wojska Polskiego
+    confidence: 0.91,
+    Class: StreetClassification,
+    scheme: [
+      {
+        is: ['StreetPrefixClassification'],
+        not: ['StreetClassification', 'IntersectionClassification']
+      },
+      {
+        is: ['PlaceClassification'],
+        not: ['StreetClassification', 'IntersectionClassification']
+      },
+      {
+        is: ['AlphaClassification', 'GivenNameClassification', 'PersonClassification'],
+        not: ['StreetClassification', 'StreetPrefixClassification']
+      }
+    ]
+  },
+  {
+    // Aleja 11 Listopada
+    confidence: 0.84,
+    Class: StreetClassification,
+    scheme: [
+      {
+        is: ['StreetPrefixClassification'],
+        not: ['StreetClassification', 'IntersectionClassification']
+      },
+      {
+        is: ['NumericClassification'],
+        not: ['StreetClassification', 'IntersectionClassification']
+      },
+      {
+        is: ['AlphaClassification', 'GivenNameClassification', 'PersonClassification'],
+        not: ['StreetClassification', 'StreetPrefixClassification']
+      }
+    ]
+  },
   {
     // Boulevard du Général Charles De Gaulle
     confidence: 0.81,
diff --git a/resources/pelias/dictionaries/libpostal/pl/place_names.txt b/resources/pelias/dictionaries/libpostal/pl/place_names.txt
new file mode 100644
index 00000000..528e8d14
--- /dev/null
+++ b/resources/pelias/dictionaries/libpostal/pl/place_names.txt
@@ -0,0 +1 @@
+wojsko|wojska|wojsk
diff --git a/resources/pelias/dictionaries/libpostal/pl/synonyms.txt b/resources/pelias/dictionaries/libpostal/pl/synonyms.txt
new file mode 100644
index 00000000..12a0f279
--- /dev/null
+++ b/resources/pelias/dictionaries/libpostal/pl/synonyms.txt
@@ -0,0 +1 @@
+polskiego|polski|pol
diff --git a/test/address.pol.test.js b/test/address.pol.test.js
index 981b3f48..19c52b48 100644
--- a/test/address.pol.test.js
+++ b/test/address.pol.test.js
@@ -5,6 +5,28 @@ const testcase = (test, common) => {
     { street: 'Szewska' }, { housenumber: '6' },
     { locality: 'Kraków' }
   ])
+
+  assert('aleja Wojska Polskiego 178', [
+    { street: 'aleja Wojska Polskiego' }, { housenumber: '178' }
+  ])
+
+  assert('aleja 29 listopada 11', [
+    { street: 'aleja 29 listopada' }, { housenumber: '11' }
+  ])
+
+  assert('aleja Wojska 178', [
+    { street: 'aleja Wojska' }, { housenumber: '178' }
+  ])
+
+  assert('Ulica Strzelecka 12, Nowy Sącz', [
+    { street: 'Ulica Strzelecka' }, { housenumber: '12' },
+    { locality: 'Nowy Sącz' }
+  ])
+
+  assert('Żorska 11, 47-400', [
+    { street: 'Żorska' }, { housenumber: '11' },
+    { postcode: '47-400' }
+  ])
 }
 
 module.exports.all = (tape, common) => {