Merge pull request #13 from owogawc/california

Add parsing for California

lottery_data_scraper/california.py

@@ -0,0 +1,63 @@
+import locale
+import logging
+import json
+import operator
+import html2text
+
+from lottery_data_scraper.schemas import GameSchema
+from lottery_data_scraper.util import fetch_html
+
+# Set local for currency conversion and formatting
+# because California only gives prize values and our schema
+# expects a string representation of the prize.
+# https://docs.python.org/3/library/locale.html
+locale.setlocale(locale.LC_ALL, "en_US.utf8")
+
+logger = logging.getLogger(__name__)
+h = html2text.HTML2Text()
+
+BASE_URL = "https://www.calottery.com"
+SCRATCHER_URL = "https://www.calottery.com/api/games/scratchers"
+
+
+def num_tx_initial(game):
+    grand_prize = game["topPrizeTier"]
+    return grand_prize["odds"] * grand_prize["totalNumberOfPrizes"]
+
+
+def fetch_games():
+    response = json.loads(fetch_html(SCRATCHER_URL))
+    games = []
+    for game_ in response["games"]:
+        prizes = []
+        for prize_ in game_["prizeTiers"]:
+            prize = {
+                "available": prize_["numberOfPrizesPending"],
+                "claimed": prize_["numberOfPrizesCashed"],
+                "value": prize_["value"],
+                "prize": locale.currency(prize_["value"], grouping=True)[
+                    :-3
+                ],  # -3 to drop the cents
+            }
+            prizes.append(prize)
+        grand_prize = sorted(game_["prizeTiers"], key=operator.itemgetter("value"))[-1]
+        game = {
+            "game_id": game_["gameNumber"],
+            "name": game_["name"],
+            "desription": h.handle(game_["description"]),
+            "image_urls": [game_["unScratchedImage"], game_["scratchedImage"]],
+            "how_to_play": h.handle(game_["howToPlay"]),
+            "num_tx_initial": num_tx_initial(game_),
+            "price": game_["price"],
+            "prizes": prizes,
+            "state": "tx",
+            "url": BASE_URL + game_["productPage"],
+        }
+        games.append(game)
+    return games
+
+
+if __name__ == "__main__":
+    games = fetch_games()
+    schema = GameSchema(many=True)
+    print(schema.dumps(games))

tests/test_california.py

@@ -0,0 +1,19 @@
+import json
+import subprocess
+import unittest
+
+
+class TestCalifornia(unittest.TestCase):
+    def test_all(self):
+        result = subprocess.run(
+            ["python3", "-m", "lottery_data_scraper.california"], capture_output=True
+        )
+        data = json.loads(result.stdout)
+        self.assertEqual(
+            data[0]["game_id"], "1405", "Expected the first game to be PAC-MAN, #1405."
+        )
+        self.assertEqual(
+            data[0]["num_tx_initial"],
+            37080000,
+            "Expected 37,080,000 tickets for PAC-MAN #1405.",
+        )