rework downloading of variants: store progress of download in database to resume downloads
This commit is contained in:
parent
0ffb323b7f
commit
5eac42e5c0
GPG key ID:
BCC5A619923C0BA5
2 changed files with 155 additions and 89 deletions
|
|
@ -131,12 +131,17 @@ CREATE INDEX variants_name_idx ON variants (name);
|
||||||
*/
|
*/
|
||||||
DROP TABLE IF EXISTS variant_suits CASCADE;
|
DROP TABLE IF EXISTS variant_suits CASCADE;
|
||||||
CREATE TABLE variant_suits (
|
CREATE TABLE variant_suits (
|
||||||
variant_id INT NOT NULL,
|
variant_id INT NOT NULL REFERENCES variants ON DELETE CASCADE,
|
||||||
suit_id INT NOT NULL,
|
suit_id INT NOT NULL REFERENCES suits ON DELETE RESTRICT,
|
||||||
index SMALLINT NOT NULL,
|
index SMALLINT NOT NULL,
|
||||||
FOREIGN KEY (variant_id) REFERENCES variants (id) ON DELETE CASCADE,
|
|
||||||
FOREIGN KEY (suit_id) REFERENCES suits (id) ON DELETE CASCADE,
|
|
||||||
UNIQUE (variant_id, suit_id),
|
UNIQUE (variant_id, suit_id),
|
||||||
UNIQUE (variant_id, index)
|
UNIQUE (variant_id, index)
|
||||||
);
|
);
|
||||||
CREATE INDEX variant_suits_index ON variant_suits (variant_id, index);
|
CREATE INDEX variant_suits_index ON variant_suits (variant_id, index);
|
||||||
|
|
||||||
|
|
||||||
|
DROP TABLE IF EXISTS variant_game_downloads CASCADE;
|
||||||
|
CREATE TABLE variant_game_downloads (
|
||||||
|
variant_id INT PRIMARY KEY REFERENCES variants ON DELETE CASCADE,
|
||||||
|
last_game_id INT NOT NULL
|
||||||
|
)
|
||||||
|
|
|
||||||
229
download_data.py
229
download_data.py
|
|
@ -1,100 +1,161 @@
|
||||||
import json
|
import json
|
||||||
|
from typing import Dict, Optional
|
||||||
|
|
||||||
from site_api import get, api, replay
|
from site_api import get, api, replay
|
||||||
from database import Game, store, load, commit, conn
|
from database.database import Game, store, load, commit, conn, cur
|
||||||
from compress import compress_deck, compress_actions, DeckCard, Action
|
from compress import compress_deck, compress_actions, DeckCard, Action, InvalidFormatError
|
||||||
from variants import variant_id
|
from variants import variant_id, variant_name
|
||||||
from hanabi import HanabiInstance, GameState, Action
|
from hanab_live import HanabLiveInstance, HanabLiveGameState
|
||||||
|
|
||||||
with open('variants.json') as f:
|
with open('variants.json') as f:
|
||||||
variants = json.loads(f.read())
|
variants = json.loads(f.read())
|
||||||
|
|
||||||
def download_games(variant_id, name=None):
|
|
||||||
url = "variants/{}".format(variant_id)
|
|
||||||
r = api(url)
|
|
||||||
if not r:
|
|
||||||
print("Not a valid variant: {}".format(variant_id))
|
|
||||||
return
|
|
||||||
num_entries = r['total_rows']
|
|
||||||
print("Downloading {} entries for variant {} ({})".format(num_entries, variant_id, name))
|
|
||||||
num_pages = (num_entries + 99) // 100
|
|
||||||
for page in range(0, num_pages):
|
|
||||||
print("Downloading page {} of {}".format(page + 1, num_pages), end = '\r')
|
|
||||||
r = api(url + "?page={}".format(page))
|
|
||||||
for row in r['rows']:
|
|
||||||
row.pop('users')
|
|
||||||
row.pop('datetime')
|
|
||||||
g = Game(row)
|
|
||||||
g.variant_id = variant_id
|
|
||||||
store(g)
|
|
||||||
print()
|
|
||||||
print('Downloaded and stored {} entries for variant {} ({})'.format(num_entries, variant_id, name))
|
|
||||||
commit()
|
|
||||||
|
|
||||||
# requires seed AND game to already have an entry in database
|
#
|
||||||
# return: (successfully exported game, game without cheat options, null if not exported)
|
def detailed_export_game(game_id: int, score: Optional[int] = None, seed_exists: bool = False) -> None:
|
||||||
def export_game(game_id) -> [bool, bool]:
|
"""
|
||||||
with conn.cursor() as cur:
|
Downloads full details of game, inserts seed and game into DB
|
||||||
cur.execute("SELECT deck_plays, one_extra_card, one_less_card, all_or_nothing, actions FROM games WHERE id = (%s)", (game_id,))
|
If seed is already present, it is left as is
|
||||||
res = cur.fetchall()
|
If game is already present, game details will be updated
|
||||||
if len(res) == 1:
|
|
||||||
print(res)
|
|
||||||
return
|
|
||||||
else:
|
|
||||||
print('game is completely new')
|
|
||||||
# return
|
|
||||||
|
|
||||||
r = get("export/{}".format(game_id))
|
:param game_id:
|
||||||
if r is None:
|
:param score: If given, this will be inserted as score of the game. If not given, score is calculated
|
||||||
print("Failed to export game id {}".format(game_id))
|
:param seed_exists: If specified and true, assumes that the seed is already present in database.
|
||||||
return False, None
|
If this is not the case, call will raise a DB insertion error
|
||||||
assert(r['id'] == game_id)
|
"""
|
||||||
# print(r)
|
|
||||||
|
|
||||||
try:
|
game_json = get("export/{}".format(game_id))
|
||||||
num_players = len(r['players'])
|
assert game_json.get('id') == game_id, "Invalid response format from hanab.live"
|
||||||
seed = r['seed']
|
|
||||||
options = r.get('options', {})
|
|
||||||
var_id = variant_id(options.get('variant', 'No Variant'))
|
|
||||||
deck_plays = options.get('deckPlays', False)
|
|
||||||
one_extra_card = options.get('oneExtraCard', False)
|
|
||||||
one_less_card = options.get('oneLessCard', False)
|
|
||||||
all_or_nothing = options.get('allOrNothing', False)
|
|
||||||
actions = [Action.from_json(action) for action in r['actions']]
|
|
||||||
deck = [DeckCard.from_json(card) for card in r['deck']]
|
|
||||||
except KeyError:
|
|
||||||
print('Error parsing JSON when exporting game {}'.format(game_id))
|
|
||||||
raise
|
|
||||||
|
|
||||||
|
players = game_json.get('players', [])
|
||||||
|
num_players = len(players)
|
||||||
|
seed = game_json.get('seed', None)
|
||||||
|
options = game_json.get('options', {})
|
||||||
|
var_id = variant_id(options.get('variant', 'No Variant'))
|
||||||
|
deck_plays = options.get('deckPlays', False)
|
||||||
|
one_extra_card = options.get('oneExtraCard', False)
|
||||||
|
one_less_card = options.get('oneLessCard', False)
|
||||||
|
all_or_nothing = options.get('allOrNothing', False)
|
||||||
|
actions = [Action.from_json(action) for action in game_json.get('actions', [])]
|
||||||
|
deck = [DeckCard.from_json(card) for card in game_json.get('deck', None)]
|
||||||
|
|
||||||
|
assert (players != [])
|
||||||
|
assert (seed is not None)
|
||||||
|
|
||||||
|
if score is None:
|
||||||
# need to play through the game once to find out its score
|
# need to play through the game once to find out its score
|
||||||
game = GameState(HanabiInstance(deck, num_players))
|
game = HanabLiveGameState(HanabLiveInstance(deck, num_players, var_id))
|
||||||
for action in actions:
|
for action in actions:
|
||||||
game.make_action(action)
|
game.make_action(action)
|
||||||
|
score = game.score
|
||||||
|
|
||||||
try:
|
try:
|
||||||
compressed_deck = compress_deck(deck)
|
compressed_deck = compress_deck(deck)
|
||||||
except:
|
except InvalidFormatError:
|
||||||
print("Failed to compress deck while exporting game {}".format(game_id))
|
print("Failed to compress deck while exporting game {}: {}".format(game_id, deck))
|
||||||
raise
|
raise
|
||||||
try:
|
try:
|
||||||
compressed_actions = compress_actions(actions)
|
compressed_actions = compress_actions(actions)
|
||||||
except:
|
except InvalidFormatError:
|
||||||
print("Failed to compress actions while exporting game {}".format(game_id))
|
print("Failed to compress actions while exporting game {}".format(game_id))
|
||||||
raise
|
raise
|
||||||
|
|
||||||
with conn.cursor() as cur:
|
if not seed_exists:
|
||||||
# cur.execute("UPDATE seeds SET deck=(%s) WHERE seed=(%s);", (deck, seed))
|
cur.execute(
|
||||||
cur.execute(
|
"INSERT INTO seeds (seed, num_players, variant_id, deck)"
|
||||||
"INSERT INTO seeds (seed, num_players, variant_id, deck)"
|
"VALUES (%s, %s, %s, %s)"
|
||||||
"VALUES (%s, %s, %s, %s)"
|
"ON CONFLICT (seed) DO NOTHING",
|
||||||
"ON CONFLICT (seed) DO NOTHING",
|
(seed, num_players, var_id, compressed_deck)
|
||||||
(seed, num_players, var_id, compressed_deck)
|
)
|
||||||
)
|
|
||||||
cur.execute(
|
|
||||||
"INSERT INTO games (id, num_players, score, seed, variant_id, deck_plays, one_extra_card, one_less_card, all_or_nothing, actions)"
|
|
||||||
"VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s)",
|
|
||||||
(game_id, num_players, game.score, seed, var_id, deck_plays, one_extra_card, one_less_card, all_or_nothing, compressed_actions))
|
|
||||||
conn.commit()
|
|
||||||
return True, not any([deck_plays, one_extra_card, one_less_card, all_or_nothing])
|
|
||||||
|
|
||||||
if __name__ == "__main__":
|
cur.execute(
|
||||||
export_game(961092)
|
"INSERT INTO games ("
|
||||||
|
"id, num_players, score, seed, variant_id, deck_plays, one_extra_card, one_less_card,"
|
||||||
|
"all_or_nothing, actions"
|
||||||
|
")"
|
||||||
|
"VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s)"
|
||||||
|
"ON CONFLICT (id) DO UPDATE SET ("
|
||||||
|
"deck_plays, one_extra_card, one_less_card, all_or_nothing, actions"
|
||||||
|
") = ("
|
||||||
|
"EXCLUDED.deck_plays, EXCLUDED.one_extra_card, EXCLUDED.one_less_card, EXCLUDED.all_or_nothing,"
|
||||||
|
"EXCLUDED.actions"
|
||||||
|
")",
|
||||||
|
(
|
||||||
|
game_id, num_players, score, seed, var_id, deck_plays, one_extra_card, one_less_card,
|
||||||
|
all_or_nothing, compressed_actions
|
||||||
|
)
|
||||||
|
)
|
||||||
|
|
||||||
|
|
||||||
|
def process_game_row(game: Dict, var_id):
|
||||||
|
game_id = game.get('id', None)
|
||||||
|
seed = game.get('seed', None)
|
||||||
|
num_players = game.get('num_players', None)
|
||||||
|
score = game.get('score', None)
|
||||||
|
|
||||||
|
if any(v is None for v in [game_id, seed, num_players, score]):
|
||||||
|
raise ValueError("Unknown response format on hanab.live")
|
||||||
|
|
||||||
|
cur.execute("SELECT seed FROM seeds WHERE seed = %s", (seed,))
|
||||||
|
seed_exists = cur.fetchone()
|
||||||
|
if seed_exists is not None:
|
||||||
|
cur.execute(
|
||||||
|
"INSERT INTO games (id, seed, num_players, score, variant_id)"
|
||||||
|
"VALUES"
|
||||||
|
"(%s, %s ,%s ,%s ,%s)"
|
||||||
|
"ON CONFLICT (id) DO NOTHING",
|
||||||
|
(game_id, seed, num_players, score, var_id)
|
||||||
|
)
|
||||||
|
else:
|
||||||
|
detailed_export_game(game_id, score)
|
||||||
|
|
||||||
|
|
||||||
|
def download_games(var_id):
|
||||||
|
name = variant_name(var_id)
|
||||||
|
page_size = 100
|
||||||
|
if name is None:
|
||||||
|
raise ValueError("{} is not a known variant_id.".format(var_id))
|
||||||
|
|
||||||
|
url = "variants/{}".format(var_id)
|
||||||
|
r = api(url)
|
||||||
|
if not r:
|
||||||
|
raise RuntimeError("Failed to download request from hanab.live")
|
||||||
|
|
||||||
|
num_entries = r.get('total_rows', None)
|
||||||
|
if num_entries is None:
|
||||||
|
raise ValueError("Unknown response format on hanab.live")
|
||||||
|
|
||||||
|
cur.execute(
|
||||||
|
"SELECT COUNT(*) FROM games WHERE variant_id = %s AND id <= "
|
||||||
|
"(SELECT COALESCE (last_game_id, 0) FROM variant_game_downloads WHERE variant_id = %s)",
|
||||||
|
(var_id, var_id)
|
||||||
|
)
|
||||||
|
num_already_downloaded_games = cur.fetchone()[0]
|
||||||
|
next_page = num_already_downloaded_games // page_size
|
||||||
|
last_page = (num_entries - 1) // page_size
|
||||||
|
|
||||||
|
if num_already_downloaded_games == num_entries:
|
||||||
|
print("Already downloaded all games for variant {} [{}]".format(var_id, name))
|
||||||
|
return
|
||||||
|
|
||||||
|
print(
|
||||||
|
"Downloading remaining {} (total {}) entries for variant {} [{}]".format(
|
||||||
|
num_entries - num_already_downloaded_games, num_entries, var_id, name
|
||||||
|
)
|
||||||
|
)
|
||||||
|
|
||||||
|
for page in range(next_page, last_page + 1):
|
||||||
|
r = api(url + "?col[0]=0&page={}".format(page))
|
||||||
|
rows = r.get('rows', [])
|
||||||
|
assert page == last_page or len(rows) == page_size, \
|
||||||
|
"Received unexpected row count ({}) when querying page {}".format(len(rows), page)
|
||||||
|
for row in rows:
|
||||||
|
process_game_row(row, var_id)
|
||||||
|
cur.execute(
|
||||||
|
"INSERT INTO variant_game_downloads (variant_id, last_game_id) VALUES"
|
||||||
|
"(%s, %s)"
|
||||||
|
"ON CONFLICT (variant_id) DO UPDATE SET last_game_id = EXCLUDED.last_game_id",
|
||||||
|
(var_id, r['rows'][-1]['id'])
|
||||||
|
)
|
||||||
|
conn.commit()
|
||||||
|
print('Downloaded and processed page {}'.format(page))
|
||||||
|
|
|
||||||
Loading…
Reference in a new issue