|
| 1 | +import time |
| 2 | +import json |
| 3 | + |
| 4 | +import dagster as dg |
| 5 | +import polars as pl |
| 6 | + |
| 7 | +from pydantic import BaseModel, HttpUrl, ValidationError |
| 8 | +from typing import Optional |
| 9 | +from termcolor import colored |
| 10 | + |
| 11 | +import requests |
| 12 | + |
| 13 | +class Series(BaseModel): |
| 14 | + id: str |
| 15 | + name: str |
| 16 | + logo: Optional[HttpUrl] = None |
| 17 | + |
| 18 | + |
| 19 | +class Set(BaseModel): |
| 20 | + series_id: str |
| 21 | + set_id: str |
| 22 | + set_name: str |
| 23 | + official_card_count: int | None |
| 24 | + total_card_count: int | None |
| 25 | + logo: Optional[str] = None |
| 26 | + symbol: Optional[str] = None |
| 27 | + |
| 28 | + |
| 29 | +@dg.asset(kinds={"API", "Polars", "Pydantic"}) |
| 30 | +def extract_series_data() -> pl.DataFrame: |
| 31 | + url: str = "https://api.tcgdex.net/v2/en/series" |
| 32 | + data = requests.get(url).json() |
| 33 | + |
| 34 | + # Pydantic validation |
| 35 | + try: |
| 36 | + validated: list[Series] = [Series(**item) for item in data] |
| 37 | + print( |
| 38 | + colored(" ✓", "green"), "Pydantic validation passed for all series entries." |
| 39 | + ) |
| 40 | + except ValidationError as e: |
| 41 | + print(colored(" ✖", "red"), "Pydantic validation failed.") |
| 42 | + print(e) |
| 43 | + raise |
| 44 | + |
| 45 | + filtered = [s.model_dump(mode="json") for s in validated if s.id in ["swsh", "sv"]] |
| 46 | + return pl.DataFrame(filtered) |
| 47 | + |
| 48 | + |
| 49 | +@dg.asset(kinds={"API", "Polars", "Pydantic"}) |
| 50 | +def extract_set_data() -> pl.DataFrame: |
| 51 | + url_list = [ |
| 52 | + "https://api.tcgdex.net/v2/en/series/swsh", |
| 53 | + "https://api.tcgdex.net/v2/en/series/sv" |
| 54 | + ] |
| 55 | + |
| 56 | + flat: list[dict] = [] |
| 57 | + |
| 58 | + for url in url_list: |
| 59 | + data = requests.get(url).json() |
| 60 | + series_id = data.get("id") |
| 61 | + |
| 62 | + for s in data.get("sets", []): |
| 63 | + entry = { |
| 64 | + "series_id": series_id, |
| 65 | + "set_id": s.get("id"), |
| 66 | + "set_name": s.get("name"), |
| 67 | + "official_card_count": s.get("cardCount", {}).get("official"), |
| 68 | + "total_card_count": s.get("cardCount", {}).get("total"), |
| 69 | + "logo": s.get("logo"), |
| 70 | + "symbol": s.get("symbol") |
| 71 | + } |
| 72 | + flat.append(entry) |
| 73 | + |
| 74 | + # Pydantic validation |
| 75 | + try: |
| 76 | + validated: list[Set] = [Set(**item) for item in flat] |
| 77 | + print( |
| 78 | + colored(" ✓", "green"), |
| 79 | + "Pydantic validation passed for all set entries." |
| 80 | + ) |
| 81 | + except ValidationError as e: |
| 82 | + print(colored(" ✖", "red"), "Pydantic validation failed.") |
| 83 | + print(e) |
| 84 | + raise |
| 85 | + |
| 86 | + return pl.DataFrame([s.model_dump(mode="json") for s in validated]) |
| 87 | + |
| 88 | + |
| 89 | +@dg.asset(kinds={"API"}) |
| 90 | +def extract_card_url_from_set() -> list: |
| 91 | + urls = [ |
| 92 | + "https://api.tcgdex.net/v2/en/sets/sv01", |
| 93 | + "https://api.tcgdex.net/v2/en/sets/sv02", |
| 94 | + ] |
| 95 | + |
| 96 | + all_card_urls = [] # Initialize empty list to collect all URLs |
| 97 | + |
| 98 | + for url in urls: |
| 99 | + try: |
| 100 | + r = requests.get(url) |
| 101 | + r.raise_for_status() |
| 102 | + |
| 103 | + data = r.json()["cards"] |
| 104 | + |
| 105 | + set_card_urls = [f"https://api.tcgdex.net/v2/en/cards/{card['id']}" for card in data] |
| 106 | + all_card_urls.extend(set_card_urls) # Add all URLs from this set |
| 107 | + |
| 108 | + time.sleep(0.1) |
| 109 | + |
| 110 | + except requests.RequestException as e: |
| 111 | + print(f"Failed to fetch set {url}: {e}") |
| 112 | + |
| 113 | + return all_card_urls |
| 114 | + |
| 115 | + |
| 116 | +@dg.asset(deps=[extract_card_url_from_set], kinds={"API"}) |
| 117 | +def extract_card_info() -> list: |
| 118 | + card_url_list = extract_card_url_from_set() |
| 119 | + cards_list = [] |
| 120 | + |
| 121 | + for url in card_url_list: |
| 122 | + try: |
| 123 | + r = requests.get(url) |
| 124 | + r.raise_for_status() |
| 125 | + data = r.json() |
| 126 | + cards_list.append(data) |
| 127 | + time.sleep(0.1) |
| 128 | + except requests.RequestException as e: |
| 129 | + print(f"Failed to fetch {url}: {e}") |
| 130 | + |
| 131 | + return cards_list |
| 132 | + |
| 133 | + |
| 134 | +@dg.asset(deps=[extract_card_info], kinds={"Polars"}) |
| 135 | +def create_card_dataframe() -> pl.DataFrame: |
| 136 | + cards_list = extract_card_info() |
| 137 | + |
| 138 | + all_flat_cards = [] |
| 139 | + |
| 140 | + for card in cards_list: |
| 141 | + flat = {} |
| 142 | + |
| 143 | + # Copy top-level scalar values |
| 144 | + scalar_keys = ['category', 'hp', 'id', 'illustrator', 'image', 'localId', |
| 145 | + 'name', 'rarity', 'regulationMark', 'retreat', 'stage'] |
| 146 | + for key in scalar_keys: |
| 147 | + flat[key] = card.get(key) |
| 148 | + |
| 149 | + # Flatten nested dicts with prefixes |
| 150 | + for key, value in card.get("legal", {}).items(): |
| 151 | + flat[f"legal_{key}"] = value |
| 152 | + |
| 153 | + for key, value in card.get("set", {}).items(): |
| 154 | + if isinstance(value, dict): |
| 155 | + for sub_key, sub_val in value.items(): |
| 156 | + flat[f"set_{key}_{sub_key}"] = sub_val |
| 157 | + else: |
| 158 | + flat[f"set_{key}"] = value |
| 159 | + |
| 160 | + # Flatten types (list of strings) |
| 161 | + flat["types"] = ", ".join(card.get("types", [])) |
| 162 | + |
| 163 | + flat["attacks_json"] = json.dumps(card.get("attacks", []), ensure_ascii=False) |
| 164 | + |
| 165 | + attacks = card.get("attacks", []) |
| 166 | + for i, atk in enumerate(attacks): |
| 167 | + prefix = f"attack_{i+1}" |
| 168 | + flat[f"{prefix}_name"] = atk.get("name") |
| 169 | + flat[f"{prefix}_damage"] = atk.get("damage") |
| 170 | + flat[f"{prefix}_effect"] = atk.get("effect") |
| 171 | + flat[f"{prefix}_cost"] = ", ".join(atk.get("cost", [])) |
| 172 | + |
| 173 | + all_flat_cards.append(flat) |
| 174 | + |
| 175 | + df = pl.DataFrame(all_flat_cards) |
| 176 | + |
| 177 | + return df |
0 commit comments