fix: fixed age and birthday support for all locales (#111)

Makefile

@@ -74,9 +74,11 @@ else
 endif
 
 # Run OverFastAPI application (production mode)
-up: down
+up:
 	@echo "Building OverFastAPI (production mode)..."
 	docker compose build
+	@echo "Stopping OverFastAPI and cleaning containers..."
+	docker compose down -v --remove-orphans
 	@echo "Launching OverFastAPI (production mode)..."
 	docker compose up -d
 

app/parsers/helpers.py

@@ -4,7 +4,7 @@
 import unicodedata
 from functools import cache
 
-from app.common.enums import CompetitiveDivision, CompetitiveRole, HeroKey, Role
+from app.common.enums import CompetitiveDivision, CompetitiveRole, HeroKey, Locale, Role
 from app.common.helpers import read_csv_data_file
 from app.config import settings
 
@@ -308,13 +308,35 @@ def get_plural_stat_key(stat_key: str) -> str:
     return stat_keys_mapping.get(stat_key, stat_key)
 
 
-def get_birthday_and_age(birthday_age_text: str) -> tuple[str | None, int | None]:
+def get_birthday_and_age(text: str, locale: Locale) -> tuple[str | None, int | None]:
     """Get birthday and age from text for a given hero"""
-    result = re.match(r"^(.*) \(Age: (\d+)\)$", birthday_age_text)
+
+    # Regex matching the birthday for every known locale
+    birthday_regex = r"^(.*) [\(（].*[:：] ?(\d+).*[\)）]$"
+
+    result = re.match(birthday_regex, text)
     if not result:
         return None, None
 
-    birthday = result[1] if result[1] != "Unknown" else None
+    # Text corresponding to "Unknown" in the locale of the page
+    unknown_texts = {
+        Locale.GERMAN: "Unbekannt",
+        Locale.ENGLISH_EU: "Unknown",
+        Locale.ENGLISH_US: "Unknown",
+        Locale.SPANISH_EU: "Desconocido",
+        Locale.SPANISH_LATIN: "Desconocido",
+        Locale.FRENCH: "Inconnu",
+        Locale.ITALIANO: "Sconosciuto",
+        Locale.JAPANESE: "不明",
+        Locale.KOREAN: "알 수 없음",
+        Locale.POLISH: "Nieznane",
+        Locale.PORTUGUESE_BRAZIL: "Desconhecido",
+        Locale.RUSSIAN: "Неизвестно",
+        Locale.CHINESE_TAIWAN: "未知",
+    }
+    unknown_text = unknown_texts.get(locale, "Unknown")
+
+    birthday = result[1] if result[1] != unknown_text else None
     age = int(result[2]) if result[2] else None
 
     return birthday, age

app/parsers/hero_parser.py

@@ -5,7 +5,7 @@
 from bs4 import Tag
 from fastapi import status
 
-from app.common.enums import MediaType
+from app.common.enums import Locale, MediaType
 from app.common.exceptions import ParserBlizzardError
 from app.config import settings
 
@@ -23,6 +23,10 @@ class HeroParser(APIParser):
         404,  # Hero Not Found response, we want to handle it here
     ]
 
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        self.locale = kwargs.get("locale") or Locale.ENGLISH_US
+
     def get_blizzard_url(self, **kwargs) -> str:
         return f"{super().get_blizzard_url(**kwargs)}/{kwargs.get('hero_key')}"
 
@@ -49,11 +53,12 @@ def parse_data(self) -> dict:
             "story": self.__get_story(lore_section),
         }
 
-    @staticmethod
-    def __get_summary(overview_section: Tag) -> dict:
+    def __get_summary(self, overview_section: Tag) -> dict:
         header_section = overview_section.find("blz-header")
         extra_list_items = overview_section.find("blz-list").find_all("blz-list-item")
-        birthday, age = get_birthday_and_age(extra_list_items[2].get_text())
+        birthday, age = get_birthday_and_age(
+            text=extra_list_items[2].get_text(), locale=self.locale
+        )
 
         return {
             "name": header_section.find("h2").get_text(),

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "overfast-api"
-version = "2.28.0"
+version = "2.28.1"
 description = "Overwatch API giving data about heroes, maps, and players statistics."
 license = "MIT"
 authors = ["Valentin PORCHET <[email protected]>"]
@@ -116,6 +116,8 @@ ignore = [
     "Q003",    # avoidable escape quote
     "W191"     # tab indentation detected instead of spaces
 ]
+# Allow some confusable UTF8 chars (used in regexp)
+allowed-confusables = ["（", "）", "："]
 
 [tool.ruff.lint.isort]
 # Consider app as first-party for imports in tests

tests/parsers/test_helpers.py

@@ -1,6 +1,6 @@
 import pytest
 
-from app.common.enums import CompetitiveDivision, CompetitiveRole, HeroKey
+from app.common.enums import CompetitiveDivision, CompetitiveRole, HeroKey, Locale
 from app.config import settings
 from app.parsers import helpers
 
@@ -254,17 +254,34 @@ def test_get_hero_role(hero_key: HeroKey):
 
 
 @pytest.mark.parametrize(
-    ("input_str", "result"),
+    ("input_str", "locale", "result"),
     [
         # Classic cases
-        ("Aug 19 (Age: 37)", ("Aug 19", 37)),
-        ("May 9 (Age: 1)", ("May 9", 1)),
+        ("Aug 19 (Age: 37)", Locale.ENGLISH_US, ("Aug 19", 37)),
+        ("May 9 (Age: 1)", Locale.ENGLISH_US, ("May 9", 1)),
         # Specific unknown case (bastion)
-        ("Unknown (Age: 32)", (None, 32)),
+        ("Unknown (Age: 32)", Locale.ENGLISH_US, (None, 32)),
+        # Specific venture case (not the same spacing)
+        ("Aug 6 (Age : 26)", Locale.ENGLISH_US, ("Aug 6", 26)),
+        ("Aug 6 (Age : 26)", Locale.ENGLISH_EU, ("Aug 6", 26)),
+        # Other languages than english
+        ("6. Aug. (Alter: 26)", Locale.GERMAN, ("6. Aug.", 26)),
+        ("6 ago (Edad: 26)", Locale.SPANISH_EU, ("6 ago", 26)),
+        ("6 ago (Edad: 26)", Locale.SPANISH_LATIN, ("6 ago", 26)),
+        ("6 août (Âge : 26 ans)", Locale.FRENCH, ("6 août", 26)),
+        ("6 ago (Età: 26)", Locale.ITALIANO, ("6 ago", 26)),
+        ("8月6日 （年齢: 26）", Locale.JAPANESE, ("8月6日", 26)),
+        ("8월 6일 (나이: 26세)", Locale.KOREAN, ("8월 6일", 26)),
+        ("6 sie (Wiek: 26 lat)", Locale.POLISH, ("6 sie", 26)),
+        ("6 de ago. (Idade: 26)", Locale.PORTUGUESE_BRAZIL, ("6 de ago.", 26)),
+        ("6 авг. (Возраст: 26)", Locale.RUSSIAN, ("6 авг.", 26)),
+        ("8月6日 （年齡：26）", Locale.CHINESE_TAIWAN, ("8月6日", 26)),
         # Invalid case
-        ("Unknown", (None, None)),
+        ("Unknown", Locale.ENGLISH_US, (None, None)),
     ],
 )
-def test_get_birthday_and_age(input_str: str, result: tuple[str | None, int | None]):
+def test_get_birthday_and_age(
+    input_str: str, locale: Locale, result: tuple[str | None, int | None]
+):
     """Get birthday and age from text for a given hero"""
-    assert helpers.get_birthday_and_age(input_str) == result
+    assert helpers.get_birthday_and_age(input_str, locale) == result