From 8f94f6ee77ae34ca9f3cc9f8cd9ae10fbbb032d6 Mon Sep 17 00:00:00 2001
From: Valentyn Bondarenko <murica.it@gmail.com>
Date: Mon, 4 Dec 2023 13:39:17 +0100
Subject: [PATCH] Task 1 - fix nonascii chars parsing

---
 otodom/task_1/Valentyn/settings.json    |  9 +++++++++
 otodom/task_1/Valentyn/src/HouseItem.py | 26 ++++++++++++++++++-------
 2 files changed, 28 insertions(+), 7 deletions(-)
 create mode 100644 otodom/task_1/Valentyn/settings.json

diff --git a/otodom/task_1/Valentyn/settings.json b/otodom/task_1/Valentyn/settings.json
new file mode 100644
index 0000000..c631d14
--- /dev/null
+++ b/otodom/task_1/Valentyn/settings.json
@@ -0,0 +1,9 @@
+{
+	"base_url": "str",
+	"price_min": "str",
+	"price_max": "str",
+	"city": "str",
+	"property_type": "str",
+	"only_for_sale": "bool",
+	"only_for_rent": "bool"
+}
\ No newline at end of file
diff --git a/otodom/task_1/Valentyn/src/HouseItem.py b/otodom/task_1/Valentyn/src/HouseItem.py
index 017a416..8f3b515 100644
--- a/otodom/task_1/Valentyn/src/HouseItem.py
+++ b/otodom/task_1/Valentyn/src/HouseItem.py
@@ -1,3 +1,6 @@
+import unicodedata
+
+
 class HouseItem:
     base_url = "https://www.otodom.pl"
 
@@ -19,6 +22,11 @@ def __init__(self, url: str):
             },
         }
 
+    def convert_to_ascii(self, text):
+        normalized = unicodedata.normalize("NFKD", text.replace('ł', 'l'))
+        ascii_text = normalized.encode("ascii", "ignore").decode("ascii")
+        return ascii_text
+
     def setPrice(self, priceStr: str):
         clean_price = priceStr.strip().replace(" zł", "").replace(" ", "")
         try:
@@ -30,11 +38,11 @@ def setPrice(self, priceStr: str):
             return self
 
     def setTitle(self, title: str):
-        self.dictionary["title"] = title
+        self.dictionary["title"] = self.convert_to_ascii(title)
         return self
 
     def setArea(self, area: str):
-        cleanNumber = area.strip().split(" ")[0].replace(",", ".")
+        cleanNumber = area.strip().split(" ")[0].replace(",", ".").replace("\xa0", "")
         self.dictionary["area"] = int(float(cleanNumber)) if cleanNumber != "" else None
         return self
 
@@ -45,17 +53,21 @@ def setRooms(self, rooms: str):
     def setLocalization(self, address: str):
         addressList = address.split(", ")
         if len(addressList) >= 5:
-            self.dictionary["localization"]["street"] = addressList[-5]
+            street = self.convert_to_ascii(addressList[-5])
+            self.dictionary["localization"]["street"] = street
         if len(addressList) >= 4:
-            self.dictionary["localization"]["district"] = addressList[-3]
+            district = self.convert_to_ascii(addressList[-3])
+            self.dictionary["localization"]["district"] = district
         if len(addressList) >= 3:
-            self.dictionary["localization"]["city"] = addressList[-2]
+            city = self.convert_to_ascii(addressList[-2])
+            self.dictionary["localization"]["city"] = city
         if len(addressList) >= 1:
-            self.dictionary["localization"]["province"] = addressList[-1]
+            province = self.convert_to_ascii(addressList[-1])
+            self.dictionary["localization"]["province"] = province
         return self
 
     def setEstateAgency(self, agency: str):
-        self.dictionary["estate_agency"] = agency
+        self.dictionary["estate_agency"] = self.convert_to_ascii(agency)
         return self
 
     def toDictionary(self):