Merge pull request #33 from Jylpah/stable-filters-update2

Backport stable fixes

analyze_wotb_replays.py

@@ -28,6 +28,8 @@ REPLAY_I = 0
 STAT_TANK_BATTLE_MIN = 100
 BATTLE_TIME_BUCKET = 3600*24*14
 
+OPT_EXPORT_REPLAY_DIR = 'export_replays'
+
 RE_SRC_IS_DB = re.compile(r'^DB:')
 
 class StatFunc:
@@ -98,7 +100,7 @@ replay_summary_flds = [
 	'protagonist',
 	'player_name',
 	'title', 
-	'mastery_badge'
+	'mastery_badge'	
 ]
 
 replay_details_flds = [
@@ -165,6 +167,7 @@ class BattleCategorizationList():
 		'Regular'				: 1,
 		'Training Room'			: 2,
 		'Tournament'			: 4,
+		'Quick Tournament'		: 5,  # maybe? 
 		'Rating'				: 7,
 		'Mad Games'				: 8,
 		'Realistic Battles'		: 22,
@@ -182,6 +185,7 @@ class BattleCategorizationList():
 		'battle_type'			: [ 'Battle Type', 		'category', 	[ 'Encounter', 'Supremacy']],
 		'room_type'				: [ 'Battle Mode', 		'category', 	_BATTLE_MODES ],
 		'top_tier'				: [ 'Tier', 			'category', 	[ 'Bottom tier', 'Top tier']],
+		'in_platoon'			: [ 'Platoon', 			'category', 	[ 'Solo', 'Platoon']],
 		'tank_tier'				: [ 'Tank Tier', 		'number' ],
 		'is_premium'			: [ 'Tank Type', 		'category', 	[ 'Tech tree', 'Premium']],
 		'tank_type'				: [ 'Tank Class', 		'category', 	WG.TANK_TYPE_STR],
@@ -201,7 +205,11 @@ class BattleCategorizationList():
 		'battle_duration'		: [ 'Battle Duration', 	'bucket', [ x*60 for x in range(0,8) ], 'int' ],
 		'distance_travelled'	: [ 'Distance Driven',	'bucket', [ x*250 for x in range(0,20) ], 'int' ],
 		'allies_wins'			: [ 'Allies WR', 		'bucket', [ 0, .35, .45, .50, .55, .65], '%' ],
+		'allies_battles'		: [ 'Player Battles',	'bucket', [ 0, 500, 1000, 2500, 5e3, 10e3, 15e3, 25e3], 'int' ],
+		'allies_damage_dealt'	: [ 'Player Avg Dmg',	'bucket', [ 0, 500, 1000, 1250, 1500, 1750, 2000, 2500], 'int' ],
 		'enemies_wins'			: [ 'Enemies WR', 		'bucket', [ 0, .35, .45, .50, .55, .65], '%' ],
+		'enemies_battles'		: [ 'Player Battles',	'bucket', [ 0, 500, 1000, 2500, 5e3, 10e3, 15e3, 25e3], 'int' ],
+		'enemies_damage_dealt'	: [ 'Player Avg Dmg',	'bucket', [ 0, 500, 1000, 1250, 1500, 1750, 2000, 2500], 'int' ],
 		'player_name'			: [ 'Player', 			'string', 25 ],
 		'protagonist'			: [ 'account_id', 		'number' ],
 		'tank_name'				: [ 'Tank', 			'string', 25],
@@ -219,6 +227,12 @@ class BattleCategorizationList():
 		]
 
 
+	_categorizations_stats = [
+		'player_wins',
+		'player_battles',
+		'player_battles'
+	]
+
 	@classmethod
 	def get_categorizations_all(cls):
 		return cls._categorizations.keys()
@@ -321,8 +335,7 @@ class BattleCategorizationList():
 
 	def __init__(self, cats: list):
 		self.urls 				= collections.OrderedDict()
-		self.url_title_max_len 	= 0
-		
+		self.url_title_max_len 	= 0		
 		
 		cats = list(set(cats))  	# remove duplicates
 		# ordering		
@@ -352,6 +365,11 @@ class BattleCategorizationList():
 				bu.error('BattleCategorizationList()', exception=err) 
 
 
+	def __iter__(self):
+		"""Returns the Iterator object"""
+		return BattleCategorizationListIterator(self)
+
+
 	def get_categorization(self, cat: str):	
 		"""Return a BattleCategorization for a category"""
 		try:
@@ -409,7 +427,7 @@ class BattleCategorizationList():
 
 	def get_filter_categories(self, cat: str, filters: list) -> dict:
 		try:
-			return self.categorizations[cat].get_filter_categories(filters)
+			return self.categorizations[str(cat)].get_filter_categories(filters)
 		except Exception as err:
 			bu.error(exception=err)
 		return None
@@ -455,6 +473,20 @@ class BattleCategorizationList():
 		return None
 
 
+class BattleCategorizationListIterator():
+	def __init__(self, btl_cats: BattleCategorizationList) -> None:
+		self._btl_cats = btl_cats
+		self._ndx = 0
+
+	def __next__(self):
+		if self._ndx < len(self._btl_cats.categorizations_list):
+			cat = self._btl_cats.categorizations_list[self._ndx]
+			res = self._btl_cats.categorizations[cat]
+			self._ndx += 1
+			return res
+		raise StopIteration
+
+
 class BattleCategorization():
 
 	RESULT_CAT_HEADER_LEN = 15
@@ -489,10 +521,13 @@ class BattleCategorization():
 
 
 	def get_filter_categories(self, filters: list) -> list:
-		if type(filters) != list:
-			filters = [ filters ]
-		return [ x for x in filters if x in self.get_categories() ]
-
+		try:
+			if type(filters) != list:
+				filters = [ filters ]
+			return [ x for x in filters if x in self.get_categories() ]
+		except Exception as err:
+			bu.error('Category: ' + str(self.category_key), exception=err)
+		return None
 
 	def help(self): 
 		"""Help"""
@@ -711,7 +746,10 @@ class BattleStringCategorization(BattleCategorization):
 
 	def get_filter_categories(self, filters: list) -> list:
 		try:
-			return filters
+			if type(filters) != list:
+				filters = [ filters ]
+			return [ x for x in filters if x in self.get_categories() ]
+			# return filters
 		except Exception as err:
 			bu.error('Category: ' + str(self.category_key), exception=err)
 		return None
@@ -747,11 +785,14 @@ class BattleNumberCategorization(BattleCategorization):
 
 	def get_filter_categories(self, filters: list) -> list:
 		try:
+			if type(filters) != list:
+				filters = [ filters ]
 			return list(map(str,filters))
 		except Exception as err:
 			bu.error('Category: ' + str(self.category_key), exception=err)
 		return None
 
+
 	def help(self):
 		super().help()
 		print('INTEGER')
@@ -892,15 +933,23 @@ class BattleCategory():
 		'battles%'			: [ '% Battles', 'Share of Battles',								9, '{:9.0%}' ],
 		'win'				: [ 'WR', 'Win rate', 												6, '{:6.1%}' ],
 		'damage_made'		: [ 'DPB', 'Average Damage', 										5, '{:5.0f}' ],
+		'damage_received'	: [ 'DPBr', 'Average Damage  Received', 							5, '{:5.0f}' ],
 		'enemies_destroyed'	: [ 'KPB', 'Kills / Battle', 										4, '{:4.2f}' ],
 		'DR'				: [ 'DR', 'Damage Ratio', 											5, '{:5.1f}' ],
 		'KDR'				: [ 'KDR', 'Kills / Death', 										4, '{:4.1f}' ],
 		'enemies_spotted'	: [ 'Spot', 'Enemies spotted per battle', 							4, '{:4.1f}' ],
 		'hit_rate'			: [ 'Hit rate', 'Shots hit / all shots made', 						8, '{:8.1%}' ],
+		'shots_made'		: [ 'Shots', 'Shots made',			 								5, '{:5.1f}' ],
+		'hits_received'		: [ 'Hits R.', 'Hits reveived',		 								7, '{:7.1f}' ],
+		'hits_bounced'		: [ 'Bounced R.', 'Hits & bounced reveived',						10, '{:10.1f}' ],
+		'in_platoon'		: [ 'Plat%', 'Platoon rate',										6, '{:6.1%}' ],
 		'pen_rate'			: [ 'Pen rate', 'Shots pen / shots hit', 							8, '{:8.1%}' ],
+		'pen_rate_received'	: [ 'Penned%', 'Received hits pen / hits received',					7, '{:7.1%}' ],
+		'splash_rate'		: [ 'Splash%', 'HE splash shots / shots hit',						7, '{:7.1%}' ],
+		'splash_rate_received'	: [ 'Splashed%', 'HE splash shots received / hits hit',			9, '{:9.1%}' ],		
 		'survived'			: [ 'Surv%', 'Survival rate', 										6, '{:6.1%}' ],
 		'time_alive'		: [ 'T alive', 'Time being alive in a battle in secs', 				7, '{:7.0f}' ],
-		'alive'				: [ 'Share live', 'Percentage of time being alive in a battle', 		8, '{:8.0%}' ],
+		'alive'				: [ 'Share live', 'Percentage of time being alive in a battle', 	8, '{:8.0%}' ],
 		'battle_duration'	: [ 'Duration', 'Battle duration', 									8, '{:8.0f}' ], 
 		'distance_travelled': [ 'Distance', 'Distance travelled',								8, '{:8.0f}' ], 
 		'top_tier'			: [ 'Top tier', 'Share of games as top tier', 						8, '{:8.0%}' ],
@@ -910,9 +959,9 @@ class BattleCategory():
 		'player_battles'	: [ 'Player Btls', 'Average number battles of the player', 			11, '{:11.0f}' ],
 		'allies_battles'	: [ 'Allies Btls', 'Average number battles of the allies', 			11, '{:11.0f}' ],
 		'enemies_battles'	: [ 'Enemies Btls', 'Average number battles of the enemies', 		12, '{:12.0f}' ],
-		'player_damage_dealt'	: [ 'Player Avg Dmg', 'Player Average damage', 					11, '{:11.0f}' ],
-		'allies_damage_dealt'	: [ 'Allies Avg Dmg', 'Average damage of the allies', 			11, '{:11.0f}' ],
-		'enemies_damage_dealt'	: [ 'Enemies Avg Dmg', 'Average damage of the enemies', 		12, '{:12.0f}' ],
+		'player_damage_dealt'	: [ 'Player Avg Dmg', 'Player Average damage', 					14, '{:14.0f}' ],
+		'allies_damage_dealt'	: [ 'Allies Avg Dmg', 'Average damage of the allies', 			14, '{:14.0f}' ],
+		'enemies_damage_dealt'	: [ 'Enemies Avg Dmg', 'Average damage of the enemies', 		15, '{:14.0f}' ],
 		MISSING_STATS		: [ 'No stats', 'Players without stats avail', 						8, '{:8.1%}']		
 	}
 
@@ -921,13 +970,15 @@ class BattleCategory():
 	_result_fields_modes = {
 		'default'		: [ 'battles',	'win','damage_made', 'enemies_destroyed', 'enemies_spotted', 
 							'top_tier', 'DR', 'survived', 'allies_wins', 'enemies_wins'],
-		'team'			: [ 'battles',	'win','player_wins','allies_wins','enemies_wins'],
-		'team_extended'	: [ 'battles',	'battles%',	'win','player_wins','allies_wins','enemies_wins', 
+		'team'			: [ 'battles',	'win', 'in_platoon', 'player_wins', 'allies_wins','enemies_wins'],
+		'team_extended'	: [ 'battles',	'battles%',	'win', 'in_platoon', 'player_wins', 'allies_wins', 'enemies_wins', 
 							'player_damage_dealt', 'allies_damage_dealt', 'enemies_damage_dealt', 
 							'player_battles','allies_battles','enemies_battles' ],
-		'extended'		: [ 'battles',	'battles%',	'win','damage_made', 'enemies_destroyed', 'enemies_spotted', 
+		'extended'		: [ 'battles',	'battles%',	'win', 'in_platoon', 'damage_made', 'enemies_destroyed', 'enemies_spotted', 
 							'top_tier', 'DR', 'KDR', 'hit_rate', 'pen_rate', 'survived', 'alive', 
-							'player_wins', 'allies_wins', 'enemies_wins', MISSING_STATS ], 
+							'player_wins', 'allies_wins', 'enemies_wins', MISSING_STATS ],
+		'dmg_received'	: [ 'battles',	'win','damage_made', 'damage_received', 'enemies_destroyed', 'top_tier', 'DR', 
+							'hits_received', 'hits_bounced', 'pen_rate_received', 'splash_rate_received'], 
 		'all'			: [ cat for cat in _result_fields.keys() ]
 	}
 
@@ -940,11 +991,14 @@ class BattleCategory():
 	]
 
 	_result_ratios = {
-		'KDR'				: [ 'enemies_destroyed', 'destroyed' ],
-		'DR'				: [ 'damage_made', 'damage_received' ],
-		'hit_rate'			: [ 'shots_hit', 'shots_made' ],
-		'pen_rate'			: [ 'shots_pen', 'shots_hit' ],
-		'dmg_block_rate' 	: [ 'damage_blocked', 'damage_received' ]
+		'KDR'					: [ 'enemies_destroyed', 'destroyed' ],
+		'DR'					: [ 'damage_made', 'damage_received' ],
+		'hit_rate'				: [ 'shots_hit', 'shots_made' ],
+		'splash_rate'			: [ 'shots_splash', 'shots_hit' ],
+		'splash_rate_received'	: [ 'hits_splash', 'hits_received' ],
+		'pen_rate'				: [ 'shots_pen', 'shots_hit' ],
+		'pen_rate_received'		: [ 'hits_pen', 'hits_received' ],		
+		'dmg_block_rate' 		: [ 'damage_blocked', 'damage_received' ]
 	}
 
 	# _extended_stats = False
@@ -1014,12 +1068,21 @@ class BattleCategory():
 
 
 	@classmethod
-	def get_field_name(cls, field: str):
+	def get_field_name(cls, field: str) -> str:
 		try:
 			return cls._result_fields[field][0]
 		except Exception as err:
 			bu.error(exception=err)
 
+	
+	@classmethod
+	def get_field_description(cls, field: str) -> str:
+		try:
+			return cls._result_fields[field][1]
+		except Exception as err:
+			bu.error(exception=err)
+
+	
 	@classmethod
 	def get_field_width(cls, field: str):
 		try:
@@ -1081,6 +1144,17 @@ class BattleCategory():
 		return None
 
 
+	@classmethod
+	def help(cls):
+		try:
+			for field in cls.get_result_fields_all():
+				print('{:24s} : {}'.format(field, cls.get_field_description(field)))
+		except Exception as err:
+			bu.error(exception=err) 
+		return None
+
+
+
 	def __init__(self):
 		try:
 			
@@ -1302,7 +1376,6 @@ class PlayerHistogram():
 		return None
 
 
-
 class ErrorCatchingArgumentParser(argparse.ArgumentParser):
 	def exit(self, status=0, message=None):
 		if status:
@@ -1358,14 +1431,14 @@ async def main(argv):
 
 			try:
 				if 'ANALYZER' in config.sections():
-					configAnalyzer	= config['ANALYZER']
-					OPT_MODE_DEFAULT= configAnalyzer.getboolean('mode', OPT_MODE_DEFAULT)
-					OPT_HIST		= configAnalyzer.getboolean('histograms', OPT_HIST)
-					OPT_STAT_FUNC	= configAnalyzer.get('stat_func', fallback=OPT_STAT_FUNC)
-					OPT_WORKERS_N 	= configAnalyzer.getint('workers', OPT_WORKERS_N)
-					OPT_EXPORT_CSV_FILE  = configAnalyzer.get('csv_file', OPT_EXPORT_CSV_FILE)
-					OPT_EXPORT_JSON_FILE = configAnalyzer.get('json_file', OPT_EXPORT_JSON_FILE)
-					res_categorizations  = configAnalyzer.get('categorizations', None)
+					configAnalyzer		= config['ANALYZER']
+					OPT_MODE_DEFAULT	= configAnalyzer.getboolean('mode', OPT_MODE_DEFAULT)
+					OPT_HIST			= configAnalyzer.getboolean('histograms', OPT_HIST)
+					OPT_STAT_FUNC		= configAnalyzer.get('stat_func', fallback=OPT_STAT_FUNC)
+					OPT_WORKERS_N 		= configAnalyzer.getint('workers', OPT_WORKERS_N)
+					OPT_EXPORT_CSV_FILE = configAnalyzer.get('csv_file', OPT_EXPORT_CSV_FILE)
+					OPT_EXPORT_JSON_FILE= configAnalyzer.get('json_file', OPT_EXPORT_JSON_FILE)
+					res_categorizations = configAnalyzer.get('categorizations', None)
 					if res_categorizations != None:
 						res_categorizations.replace(' ','')
 						BattleCategorizationList.set_categorizations_default(res_categorizations.split(','))
@@ -1420,9 +1493,10 @@ async def main(argv):
 		parser.add_argument('--mapfile', type=str, default='maps.json', help='JSON file to read Blitz map names from. Default is "maps.json"')
 		parser.add_argument('--json', action='store_true', default=False, help='Export data in JSON format')
 		parser.add_argument('--csv', action='store_true', default=False, help='Export data in CSV format')
+		parser.add_argument('--replays', action='store_true', default=False, help='Export replay info in JSON format')
 		parser.add_argument('-o','--outfile', type=str, default='-', metavar="OUTPUT", help='File to write results. Default STDOUT')
 		parser.add_argument('--db', action='store_true', default=OPT_DB, help='Use DB - You are unlikely to have it')
-		parser.add_argument('--filters', type=str, default=None, help='Filter replays based on categories. Filters given in JSON format.\nUse array "[]" for multiple filters/values. see --mode help.\nExample: : [ {"tier" : [8,9,10] }, { "player_wins" : 5 }]')
+		parser.add_argument('--filters', type=str, default=None, help='Filter replays based on categories. Filters given in JSON format.\nUse array "[]" for multiple filters/values. see --mode help.\nExample: : [ {"tank_tier" : [8,9,10] }, { "player_wins" : 5 }]')
 		parser.add_argument('--filters_db', type=str, default=None, help='[only for DB setup] Filter replays in DB based on categories. Filters given in MongoDB JSON format. See --mode help')
 		parser.add_argument('--min', type=int, default=None, help='Only select replays from players with minimum number of replays in the dataset')
 		parser.add_argument('-d', '--debug', action='store_true', default=False, help='Debug mode')
@@ -1441,7 +1515,7 @@ async def main(argv):
 		# res_categories = BattleCategorization.get_categorizations(OPT_CATEGORIZATIONS, args)
 
 		bu.set_log_level(args.silent, args.verbose, args.debug)
-		bu.set_progress_step(250)  						# Set the frequency of the progress dots. 
+		# bu.set_progress_step(250)  						# Set the frequency of the progress dots. 
 
 		#### Connect to MongoDB. You are unlikely to have this set up... 
 		bu.debug('DB_SERVER: ' + DB_SERVER)
@@ -1491,7 +1565,7 @@ async def main(argv):
 			# Make replay Queue
 
 			scanner_task = asyncio.create_task(mk_replayQ(replayQ, args, db))
-			bu.debug('Replay scanner started')
+			bu.set_counter('Reading replays', 10)
 			# Start tasks to process the Queue
 			for i in range(OPT_WORKERS_N):
 				reader_tasks.append(asyncio.create_task(replay_reader(replayQ, i, args)))
@@ -1499,7 +1573,7 @@ async def main(argv):
 
 			bu.debug('Waiting for the replay scanner to finish')
 			await asyncio.wait([scanner_task])
-			
+			bu.finish_progress_bar()
 			# bu.debug('Scanner finished. Waiting for replay readers to finish the queue')
 			await replayQ.join()
 			await asyncio.sleep(0.1)
@@ -1509,17 +1583,28 @@ async def main(argv):
 				await asyncio.sleep(0.1)	
 			results = []
 			players = set()
+			replays	= dict()
 			for res in await asyncio.gather(*reader_tasks):
 				results.extend(res[0])
 				players.update(res[1])
+				replays.update(res[2])
+			
 			if len(players) == 0:
 				raise Exception("No players found to fetch stats for. No replays found?")
 
 			if args.min != None:
 				results = filter_min_replays_by_player(results, args.min)
 				players = get_players(results)
+				replays = filter_replays(replays, results)				
+
+			# Filter based on non-stats filters
+			results = filter_results(results, args.filters, stats_filters=False)	
+
+      (player_stats, stat_id_map) = await process_player_stats(players, OPT_WORKERS_N, args, db)
+      
+      # Filter based on stats filters
+			results = filter_results(results, args.filters, stats_filters=True)
 
-			(player_stats, stat_id_map) = await process_player_stats(players, OPT_WORKERS_N, args, db)
 			bu.debug('Number of player stats: ' + str(len(player_stats)))
 			teamresults = calc_team_stats(results, player_stats, stat_id_map, args)
 
@@ -1564,6 +1649,9 @@ async def main(argv):
 					csvwriter.writerows(rows)
 					bu.verbose_std('Results exported to: ' + export_file)			
 			
+			if args.replays:
+				await export_replays(replays)
+
 			bu.debug('Finished. Cleaning up..................')
 		except Exception as err:
 			bu.error(exception=err)
@@ -1588,11 +1676,33 @@ async def help_extended(db : motor.motor_asyncio.AsyncIOMotorDatabase = None, pa
 		parser.print_help()
 	# Result category options
 
+	# Results fields
+	print('\n-------------------------------------------------------------------')
+	print('| Result fields                                                   |')
+	print('-------------------------------------------------------------------\n')
+	
+	BattleCategory.help()
+
+	print('\n-------------------------------------------------------------------')
+	print('| Filters                                                         |')
+	print('-------------------------------------------------------------------\n')
+	print("\t--filters '{ \"categorization\" : category }'")
+	print('\tThese filters let you filter only battles matching the filter. The filter has been proper JSON dict.')
+	print("\tExample: --filter '{ \"room_type\" :1 }'")
+	print('')
+	BattleCategorizationList.help()
+	print('\n-------------------------------------------------------------------')
+	print('| Rounding errors                                                 |')
+	print('-------------------------------------------------------------------\n')
+	print('Yes, the numbers are rounded UP because reasons.')
+	print('If interested why please read this: ')
+	print('\thttps://realpython.com/python-rounding/#pythons-built-in-round-function')
+	print('No, I am not planning to fix these for now.')	
+
 	# Filter usage
 	print('\n-------------------------------------------------------------------')
 	print('| DB Filter usage                                                 |')
-	print('-------------------------------------------------------------------')
-	print('')
+	print('-------------------------------------------------------------------\n')
 	print('You are unlikely to have DB setup required for these filters. This is "Jylpah\'s special"')
 	print('')
 	print("Syntax: --filters_db '{ \"replay.param\": VALUE, \"replay.param2\": { \"$MONGO_OPERATOR\" : VALUE }, ... }'")
@@ -1611,23 +1721,24 @@ async def help_extended(db : motor.motor_asyncio.AsyncIOMotorDatabase = None, pa
 		except Exception as err:
 			bu.error(exception=err)
 
-	print('\n-------------------------------------------------------------------')
-	print('| Filters                                                         |')
-	print('-------------------------------------------------------------------')
-	print('')
-	print("\t--filters '{ \"categorization\" : category }'")
-	print('\tThese filters let you filter only battles matching the filter. The filter has been proper JSON dict.')
-	print("\tExample: --filter '{ \"room_type\" :1 }'")
-	print('')
-	BattleCategorizationList.help()
-	print('\n-------------------------------------------------------------------')
-	print('| Rounding errors                                                 |')
-	print('-------------------------------------------------------------------')
-	print('')
-	print('Yes, the numbers are rounded UP because reasons.')
-	print('If interested why please read this: ')
-	print('\thttps://realpython.com/python-rounding/#pythons-built-in-round-function')
-	print('No, I am not planning to fix these for now.')	
+
+async def export_replays(replays: dict):
+	"""Export replays"""
+	try:
+		replay_dir = OPT_EXPORT_REPLAY_DIR +  '_' + bu.get_date_str()
+		bu.verbose_std('Exporting replays to ' + replay_dir)
+		if not os.path.isdir(replay_dir):
+			os.mkdir(replay_dir)
+
+		i = 0
+		for replay in replays.values():
+			filename = wg.get_replay_filename(replay)
+			async with aiofiles.open(os.path.join(replay_dir, filename), 'w', encoding="utf8") as f:
+				await f.write(json.dumps(replay, indent=4))
+			i += 1
+		bu.verbose_std( str(i) + ' replays exported')
+	except Exception as err:
+		bu.error(exception=err)
 
 
 ## move the class PlayerHistogram?
@@ -1669,7 +1780,7 @@ def filter_min_replays_by_player(results: list, min_replays: int) -> list:
 		for res in results:
 			if res['protagonist'] in players_enough_replays:
 				res_ret.append(res)
-
+		bu.verbose_std('Replays after filtering: ' + str(len(res_ret)))	
 		return res_ret
 
 	except Exception as err:
@@ -1712,15 +1823,15 @@ def print_player_dist(histograms: dict):
 		print('')
 		hist.print()
 
+
 def process_battle_results(results: dict, args : argparse.Namespace):
 	"""Process replay battle results""" 
 	try:
 		url 		= args.url
 		cats 		= BattleCategorizationList.get_categorizations(args)
+		blt_cat_list= None
 		
-		
-		if args.filters != None:
-			results = filter_replays(results, args.filters)
+		# remove if? 
 		if len(results) > 0:
 			blt_cat_list = BattleCategorizationList(cats)
 			for result in results:
@@ -1731,7 +1842,7 @@ def process_battle_results(results: dict, args : argparse.Namespace):
 			blt_cat_list.calc_results()
 		else:
 			warning_txt = 'No replays to process '
-			if args.filter != None: 
+			if args.filters != None: 
 				warning_txt = warning_txt + 'after filtering'
 			bu.warning(warning_txt)
 
@@ -1740,25 +1851,25 @@ def process_battle_results(results: dict, args : argparse.Namespace):
 	return blt_cat_list
 
 
-def filter_replays(results: list, filter_json : str) -> bool:
+def filter_results(results: list, filter_json : str, stats_filters=False) -> bool:
 	"""Filter replays based on battle category filters"""
 	try:
 		filters = json.loads(filter_json)
 		bu.debug(str(filters))
-		for filter in filters:
-			if type(filters[filter]) != list:
-				filters[filter] = [ filters[filter]]
-
 		if type(filters) != dict:
-			bu.error('invalid --filter arguments: ' + str(filters))
+			bu.error('invalid --filter arguments: ' + str(filter))
 			sys.exit(1)
 		
 		filter_cats = BattleCategorizationList(filters.keys())
-		for cat in filters:
+		bu.verbose('Filters applied:')
+		for category in filter_cats:
+			cat = category.category_key
 			categorization = filter_cats.get_categorization(cat)
+			bu.debug('Category: ' + cat + ' filters type: ' + str(type(filters)) + ' filters: ' + str(filters))
 			filters[cat] = categorization.get_filter_categories(filters[cat])
-			bu.verbose('Filter: ' + cat + ': ' + ','.join(filters[cat]))		
-		
+			bu.verbose('  ' + cat + ' = ' + ','.join(filters[cat]))		
+		bu.verbose('')
+
 		res = list()
 		for result in results:
 			try:
@@ -1774,13 +1885,27 @@ def filter_replays(results: list, filter_json : str) -> bool:
 					res.append(result)
 			except Exception as err:
 				bu.error(exception=err)
-		bu.verbose('Replays after filtering: ' + str(len(res)))	
+		bu.verbose_std('Replays after filtering: ' + str(len(res)))	
 		return res
 	except Exception as err:
 		bu.error(exception=err)
 	return None
 
 
+def filter_replays(replays_in: dict, results: list) -> list:
+	"""Filter source replays based on (filtered) results"""
+	replays = dict()
+	for res in results:
+		try:
+			# bu.debug('res[_id] : ' + res['_id'], force=True)			
+			_id = res['_id']
+			replays[_id] = replays_in[_id]
+		except Exception as err:
+			bu.error(exception=err)	
+	return replays
+	
+	
+
 async def process_player_stats(players, N_workers: int, args : argparse.Namespace, db : motor.motor_asyncio.AsyncIOMotorDatabase) -> dict:
 	"""Start stat_workers to retrieve and store players' stats"""
 	## Create queue of player-tank pairs to find stats for
@@ -1934,42 +2059,38 @@ async def stat_worker(queue : asyncio.Queue, workerID: int, args : argparse.Name
 			try:
 				bu.debug('Stat_id: ' + stat_id, id=workerID)
 				bu.print_progress()
-				# Analysing player performance based on their stats on the tier tanks they are playing 
-
+				stats_tmp = None
+	
 				# Try cache first
-				if (stat_id not in stats):
-					stats_tmp = None
-					pruned_stat_id = prune_stat_id(stat_id)
-					if (pruned_stat_id not in stats):
-						stats_tmp = await stat_wg_func(pruned_stat_id, cache_only = True)
-					else:
-						stat_id_remap[stat_id] = pruned_stat_id
-						queue.task_done()
-						continue
+				pruned_stat_id = prune_stat_id(stat_id)
+				if (pruned_stat_id not in stats):
+					stats_tmp = await stat_wg_func(pruned_stat_id, cache_only = True)
+				else:
+					stat_id_remap[stat_id] = pruned_stat_id
+					continue
 
-					if (stats_tmp != None):
-						stats[pruned_stat_id] = stats_tmp
-						stat_id_remap[stat_id] = pruned_stat_id
-						queue.task_done()
-						continue
-					
-					# try DB
-					stats[stat_id] = await stat_db_func(db, stat_id)				
-					bu.debug('get_db_' + args.stat_func + '_stats returned: account_id=' + str(get_account_id_f_stat_id(stat_id)) + ': ' + str(stats[stat_id]), workerID)
+				if (stats_tmp != None):
+					stats[pruned_stat_id] = stats_tmp
+					stat_id_remap[stat_id] = pruned_stat_id					
+					continue
+				
+				# try DB
+				stats[stat_id] = await stat_db_func(db, stat_id)				
+				bu.debug('get_db_' + args.stat_func + '_stats returned: account_id=' + str(get_account_id_f_stat_id(stat_id)) + ': ' + str(stats[stat_id]), workerID)
 
-					# no DB stats found, trying WG AP
-					if (stats[stat_id] == None):							
-						stats[pruned_stat_id] = await stat_wg_func(pruned_stat_id)
-						stat_id_remap[stat_id] = pruned_stat_id						
-						del stats[stat_id]
-						bu.debug('get_wg_' + args.stat_func + '_stats returned: account_id=' + str(get_account_id_f_stat_id(pruned_stat_id)) + ': ' + str(stats[pruned_stat_id]), workerID)
+				# no DB stats found, trying WG AP
+				if (stats[stat_id] == None):							
+					stats[pruned_stat_id] = await stat_wg_func(pruned_stat_id)
+					stat_id_remap[stat_id] = pruned_stat_id						
+					del stats[stat_id]
+					bu.debug('get_wg_' + args.stat_func + '_stats returned: account_id=' + str(get_account_id_f_stat_id(pruned_stat_id)) + ': ' + str(stats[pruned_stat_id]), workerID)
 
-					
 			except KeyError as err:
 				bu.error('Key not found', err, id=workerID)
 			except Exception as err:
 				bu.error('Unexpected error', err, id=workerID)
-			queue.task_done()
+			finally:
+				queue.task_done()
 
 	except (asyncio.CancelledError, concurrent.futures.CancelledError):
 		bu.debug('Stats queue is empty', id=workerID)		
@@ -2003,7 +2124,7 @@ async def get_wg_tank_stats(stat_id_str: str, cache_only = False) -> dict:
 		player_stats = await wg.get_player_tank_stats(account_id, [ tank_id ], hist_stats, cache_only = cache_only)
 		#bu.debug('account_id: ' + str(account_id) + ' ' + str(player_stats))
 
-		return await tank_stats_helper(player_stats)
+		return tank_stats_helper(player_stats)
 
 	except KeyError as err:
 		bu.error('account_id: ' + str(account_id) + ' tank_id:' + str(tank_id) +' : Key not found', err)
@@ -2029,8 +2150,8 @@ async def get_db_tank_stats(db : motor.motor_asyncio.AsyncIOMotorDatabase, stat_
 		for stat in hist_stats:			
 			project[stat] = True
 		project['_id'] = False
-		
-		cursor = dbc.find({ '$and': [ { 'account_id': account_id }, { 'last_battle_time': { '$gte': battletime - time_buffer }}, { 'tank_id' : tank_id } ] }, projection=project).sort('last_battle_time',-1).limit(1)
+		bu.debug('find(): tank_id={} account_id={} last_battle_time>={}'.format(tank_id, account_id, battletime - time_buffer))
+		cursor = dbc.find({ '$and': [ { 'tank_id' : tank_id }, { 'account_id': account_id }, { 'last_battle_time': { '$gte': battletime - time_buffer }} ] }, projection=project).sort('last_battle_time',-1).limit(1)
 
 		# pipeline = 	[ { '$match': { '$and': [ { 'account_id': account_id }, { 'last_battle_time': { '$lte': battletime + time_buffer }}, { 'tank_id' : tank_id } ]}}, 
 		# 		{ '$sort': { 'last_battle_time': -1 }}, 
@@ -2041,7 +2162,11 @@ async def get_db_tank_stats(db : motor.motor_asyncio.AsyncIOMotorDatabase, stat_
 		# cursor = dbc.aggregate(pipeline, allowDiskUse=True)
 		#cursor = dbc.aggregate(pipeline)
 
-		return await tank_stats_helper(await cursor.to_list(1)) 
+		stats = await cursor.to_list(1)
+		# store cache
+		# await wg.save_tank_stats(account_id, [tank_id], stats)
+		
+		return tank_stats_helper(stats) 	
 				
 	except Exception as err:
 		bu.error('account_id: ' + str(account_id) + ' Error', err)
@@ -2062,7 +2187,7 @@ async def get_wg_tank_tier_stats(stat_id_str: str, cache_only = False) -> dict:
 		player_stats = await wg.get_player_tank_stats(account_id, tier_tanks, hist_stats, cache_only = cache_only)
 		#bu.debug('account_id: ' + str(account_id) + ' ' + str(player_stats))
 
-		return await tank_stats_helper(player_stats)
+		return tank_stats_helper(player_stats)
 
 	except KeyError as err:
 		bu.error('account_id: ' + str(account_id) + ' Key not found', err)
@@ -2085,7 +2210,7 @@ async def get_wg_tier_x_stats(stat_id_str: str, cache_only = False) -> dict:
 		player_stats = await wg.get_player_tank_stats(account_id, tier_tanks, hist_stats, cache_only = cache_only)
 		#bu.debug('account_id: ' + str(account_id) + ' ' + str(player_stats))
 
-		return await tank_stats_helper(player_stats)
+		return tank_stats_helper(player_stats)
 
 	except KeyError as err:
 		bu.error('account_id: ' + str(account_id) + ' Key not found', err)
@@ -2113,8 +2238,10 @@ async def get_db_tank_tier_stats(db : motor.motor_asyncio.AsyncIOMotorDatabase,
 		# cursor = dbc.aggregate(pipeline, allowDiskUse=True)
 		cursor = dbc.aggregate(pipeline)
 
-		return await tank_stats_helper(await cursor.to_list(1000)) 
-				
+		stats = await cursor.to_list(1000)
+		# await wg.save_tank_stats(account_id, tier_tanks, stats)
+		return tank_stats_helper(stats) 
+		
 	except Exception as err:
 		bu.error('account_id: ' + str(account_id) + ' Error', err)
 	return None
@@ -2124,23 +2251,9 @@ async def get_db_tier_x_stats(db : motor.motor_asyncio.AsyncIOMotorDatabase, sta
 	"""Get player stats from MongoDB (you are very unlikely to have it, unless you have set it up)"""
 	if db == None:
 		return None
-	try:
-		dbc = db[DB_C_TANK_STATS]
+	try:		
 		( account_id, battletime ) = str2ints(stat_id_str)
-		tier_tanks = wg.get_tanks_by_tier(10)
-		time_buffer = 2*7*24*3600
-
-		pipeline = 	[ { '$match': { '$and': [ { 'account_id': account_id }, { 'last_battle_time': { '$lte': battletime + time_buffer }}, { 'tank_id' : {'$in': tier_tanks} } ]}}, 
-				{ '$sort': { 'last_battle_time': -1 }}, 
-				{ '$group': { '_id': '$tank_id', 'doc': { '$first': '$$ROOT' }}}, 
-				{ '$replaceRoot': { 'newRoot': '$doc' }}, 
-				{ '$project': { '_id': 0 }} ]
-
-		# cursor = dbc.aggregate(pipeline, allowDiskUse=True)
-		cursor = dbc.aggregate(pipeline)
-
-		return await tank_stats_helper(await cursor.to_list(1000)) 
-				
+		return await get_db_tank_tier_stats(db,get_stat_id(account_id, 10, battletime) )
 	except Exception as err:
 		bu.error('account_id: ' + str(account_id) + ' Error', err)
 	return None
@@ -2158,14 +2271,16 @@ async def get_db_player_stats(db : motor.motor_asyncio.AsyncIOMotorDatabase, sta
 		
 		pipeline = 	[ { '$match': { '$and': [ { 'account_id': account_id }, { 'last_battle_time': { '$lte': battletime + time_buffer }} ]}}, 
 				{ '$sort': { 'last_battle_time': -1 }}, 
-				{ '$group': { '_id': '$tank_id', 'doc': { '$first': '$$ROOT' }}}, 
+				{ '$group': { '_id': { 'tank_id':'$tank_id'}, 'doc': { '$first': '$$ROOT' }}}, 
 				{ '$replaceRoot': { 'newRoot': '$doc' }}, 
 				{ '$project': { '_id': 0 }} ]
 
 		# cursor = dbc.aggregate(pipeline, allowDiskUse=True)
 		cursor = dbc.aggregate(pipeline)
 
-		return await tank_stats_helper(await cursor.to_list(1000)) 
+		stats = await cursor.to_list(1000)
+		# await wg.save_tank_stats(account_id, [], stats)
+		return tank_stats_helper(stats) 
 				
 	except Exception as err:
 		bu.error('account_id: ' + str(account_id) + ' Error', err)
@@ -2193,20 +2308,17 @@ async def get_wg_player_stats(stat_id_str: str, cache_only = False) -> dict:
 	return None
 
 
-async def tank_stats_helper(stat_list: list):
+def tank_stats_helper(stat_list: list):
 	"""Helpher func got get_db_tank_tier_stats() and get_wg_tank_tier_stats()"""
 	try:
 		if stat_list == None: 
 			return None
-		stats = dict()
+		stats = collections.defaultdict(def_value_zero)
 		hist_fields = histogram_fields.keys()
 		if 'battles' not in hist_fields:
 			bu.error('\'battles\' must be defined in \'histogram_fields\'')
 			return None
 
-		for field in hist_fields:
-			stats[field] = 0
-		
 		for tank_stat in stat_list: 
 			tank_stat = tank_stat['all']
 			for field in hist_fields:
@@ -2276,7 +2388,7 @@ async def mk_replayQ(queue : asyncio.Queue, args : argparse.Namespace, db : moto
 			async for replay_json in cursor:
 				_id = replay_json['_id']
 				#del(replay_json['_id'])
-				await queue.put(await mk_readerQ_item(replay_json, 'DB: _id = ' + _id))
+				await queue.put(await mk_readerQ_item(replay_json, _id=_id))
 				Nreplays += 1
 			bu.debug('All the matching replays have been read from the DB')
 		except Exception as err:
@@ -2292,7 +2404,7 @@ async def mk_replayQ(queue : asyncio.Queue, args : argparse.Namespace, db : moto
 				
 				if (p_replayfile.match(line) != None):
 					replay_json = await bu.open_JSON(line, wi.chk_JSON_replay)						
-					await queue.put(await mk_readerQ_item(replay_json, line))
+					await queue.put(await mk_readerQ_item(replay_json, filename=line))
 			except Exception as err:
 				bu.error(exception=err)
 	else:
@@ -2303,7 +2415,7 @@ async def mk_replayQ(queue : asyncio.Queue, args : argparse.Namespace, db : moto
 					fn = fn[:-1] 
 				if os.path.isfile(fn) and (p_replayfile.match(fn) != None):
 					replay_json = await bu.open_JSON(fn, wi.chk_JSON_replay)
-					await queue.put(await mk_readerQ_item(replay_json, fn))
+					await queue.put(await mk_readerQ_item(replay_json, filename=fn))
 					bu.debug('File added to queue: ' + fn)
 					Nreplays += 1
 				elif os.path.isdir(fn):
@@ -2312,34 +2424,35 @@ async def mk_replayQ(queue : asyncio.Queue, args : argparse.Namespace, db : moto
 							if entry.is_file() and (p_replayfile.match(entry.name) != None): 
 								bu.debug(entry.name)
 								replay_json = await bu.open_JSON(entry.path, wi.chk_JSON_replay)
-								await queue.put(await mk_readerQ_item(replay_json, entry.name))
+								await queue.put(await mk_readerQ_item(replay_json, filename=entry.name))
 								bu.debug('File added to queue: ' + entry.path)
 								Nreplays += 1
 			except Exception as err:
 				bu.error(exception=err)					
-	bu.verbose('Finished scanning replays: ' + str(Nreplays)  + ' replays to process') 
 	return Nreplays
 
 
-async def mk_readerQ_item(replay_json, filename : str = None) -> list:
+async def mk_readerQ_item(replay_json, filename : str = None, _id: str = None) -> list:
 	"""Make an item to replay queue"""
 	global REPLAY_N
 	REPLAY_N +=1
 	try:
 		if wi.chk_JSON_replay(replay_json): 
 			if not '_id' in replay_json:
-				id = wi.read_replay_id(replay_json)
-				replay_json['_id'] = id	
+				if _id  != None:			
+					_id = wi.read_replay_id(replay_json)
+				replay_json['_id'] = _id
 		else:
 			replay_json = None # mark error
 	except Exception as err:
 		bu.error(exception=err)
-	if filename == None:
-		return [replay_json, REPLAY_N, '' ]
-	elif RE_SRC_IS_DB.match(filename) != None:
-		return [replay_json, REPLAY_N, filename ]
-	else:
+	
+	if filename != None:
 		return [replay_json, REPLAY_N, os.path.basename(filename) ]
+	elif _id != None: 
+		return [replay_json, REPLAY_N, 'DB: ' + _id ]
+	else:
+		return None
 
 
 async def replay_reader(queue: asyncio.Queue, readerID: int, args : argparse.Namespace):
@@ -2347,6 +2460,7 @@ async def replay_reader(queue: asyncio.Queue, readerID: int, args : argparse.Nam
 	#global SKIPPED_N
 	results = []
 	playerstanks = set()
+	replays		 = dict()
 	try:
 		while True:
 			item = await queue.get()
@@ -2373,8 +2487,10 @@ async def replay_reader(queue: asyncio.Queue, readerID: int, args : argparse.Nam
 				playerstanks.update(set(result['allies']))
 				playerstanks.update(set(result['enemies']))	
 				playerstanks.update(set([result['player']]))
-				
+				_id = wi.read_replay_id(replay_json)
+				replays[_id] = replay_json
 				results.append(result)
+
 				bu.debug('Marking replay [' + str(replayID) + '] done', id=readerID)
 						
 			except Exception as err:
@@ -2382,7 +2498,7 @@ async def replay_reader(queue: asyncio.Queue, readerID: int, args : argparse.Nam
 			queue.task_done()
 	except (asyncio.CancelledError, concurrent.futures.CancelledError):		
 		bu.debug( str(len(results)) + ' replays, ' + str(len(playerstanks)) + ' player/tanks', readerID)
-		return results, playerstanks
+		return results, playerstanks, replays
 	return None
 
 
@@ -2449,20 +2565,20 @@ async def read_replay_JSON(replay_json: dict, args : argparse.Namespace) -> dict
 			player_tank_tier = wg.get_tank_data(player['vehicle_descr'], 'tier')
 			btl_tier = max(btl_tier, player_tank_tier)
 
-			if (protagonist != None) and (player['dbid'] == protagonist):
-				protagonist_tank = player['vehicle_descr']
-
 			if player['dbid'] == account_id:
 				# player itself is not put in results['allies']
 				tmp = dict()
 				tmp['account_id'] 	= account_id
-				tmp['tank_id'] 		= player['vehicle_descr']
+				tank_id 			= player['vehicle_descr']
+				tmp['tank_id'] 		= tank_id
 				tmp['tank_tier'] 	= player_tank_tier
 				tmp['tank_name']	= wg.get_tank_name(tmp['tank_id'])
 				tmp['tank_type']	= wg.get_tank_type_id(tmp['tank_id'])
 				tmp['tank_nation']	= wg.get_tank_nation_id(tmp['tank_id'])
 				tmp['is_premium']	= int(wg.is_premium(tmp['tank_id']))
-				tmp['squad_index'] 	= player['squad_index']
+				platoon_ndx 		= player['squad_index']
+				tmp['squad_index'] 	= platoon_ndx
+				tmp['in_platoon']	= 0 if platoon_ndx == None else 1
 				for key in replay_details_flds:
 					tmp[key] = player[key]
 				if player['hitpoints_left'] == 0:
@@ -2495,7 +2611,7 @@ async def read_replay_JSON(replay_json: dict, args : argparse.Namespace) -> dict
 
 		result['battle_duration'] = btl_duration
 		## Rather use 'player' than incomprehensible 'protagonist'...		
-		result['player'] = get_stat_id(protagonist, protagonist_tank, result['battle_start_timestamp'])
+		result['player'] = get_stat_id(account_id, tank_id, result['battle_start_timestamp'])
 		bu.debug('Player stats_id: ' + result['player'])
 		# remove platoon buddy from stats 			
 		if result['squad_index'] != None:
@@ -2540,8 +2656,8 @@ def prune_stat_id(stat_id_str: str) -> str:
 	return ':'.join(stat_id)
 
 
-def get_stat_id(account_id: int, tank_id: int, battletime: int) -> str:
-	return ':'.join([ str(account_id), str(tank_id), str(battletime)])
+def get_stat_id(account_id: int, tank_id: int, battle_time: int) -> str:
+	return ':'.join(map(str, [account_id, tank_id, battle_time ]))
 
 
 def get_stat_id_tank_tier(stat_id_str: str) -> str:
@@ -2551,7 +2667,7 @@ def get_stat_id_tank_tier(stat_id_str: str) -> str:
 		account_id	= stat_id[0]
 		tank_tier 	= wg.get_tank_tier(stat_id[1])
 		battle_time = (stat_id[2] // BATTLE_TIME_BUCKET) * BATTLE_TIME_BUCKET
-		return ':'.join(map(str, [account_id, tank_tier, battle_time ]))
+		return get_stat_id(account_id, tank_tier, battle_time)
 	except Exception as err:
 		bu.error('Stats_id: ' + stat_id_str, exception=err)
 	return None
@@ -2564,7 +2680,7 @@ def get_stat_id_tank(stat_id_str: str) -> str:
 		account_id	= stat_id[0]
 		tank_id 	= stat_id[1]
 		battle_time = (stat_id[2] // BATTLE_TIME_BUCKET) * BATTLE_TIME_BUCKET
-		return ':'.join(map(str, [account_id, tank_id, battle_time ]))
+		return get_stat_id(account_id, tank_id, battle_time)		
 	except Exception as err:
 		bu.error('Stats_id: ' + stat_id_str, exception=err)
 	return None

blitzutils.py

@@ -22,6 +22,11 @@ _log_level = NORMAL
 LOG = False
 LOGGER = None
 
+MODE_TANK_STATS         = 'tank_stats'
+MODE_PLAYER_STATS       = 'player_stats'
+MODE_PLAYER_ACHIEVEMENTS= 'player_achievements'
+MODE_TANKOPEDIA			= 'tankopedia'
+
 # Progress display
 _progress_N = 100
 _progress_i = 0
@@ -418,12 +423,14 @@ def set_progress_bar(heading: str, max_value: int, step: int = None, slow: bool
     return
 
 
-def set_counter(heading: str):
+def set_counter(heading: str, step = None):
     global _progress_obj, _progress_i
     _progress_i = 0
     if _progress_obj != None:
         finish_progress_bar()
-    _progress_obj = Counter(heading)
+    _progress_obj = Counter(heading + ': ')
+    if step != None:
+        set_progress_step(step)
     return 
 
 
@@ -479,6 +486,16 @@ def NOW() -> int:
     return int(time.time())
 
 
+def get_date_str(timestamp:int = NOW(), date_format = '%Y%m%d_%H%M%S'):
+    """Return YYYYMMDD_HHmm date string"""
+    try:
+        ts = datetime.fromtimestamp(timestamp)
+        return ts.strftime(date_format)
+    except Exception as err:
+        error(exception=err)
+    return None
+
+
 def rebase_file_args(current_dir, files):
     """REbase file command line params after moving working dir to the script's dir""" 
     if isinstance(files, list):    
@@ -626,6 +643,14 @@ def get_JSON_value(json, key : str = None, keys : list = None, keypath = None):
     raise KeyError('Key not found: ' + get_JSON_keypath(keypath, keys[0]))
 
 
+def sort_dict(d: dict, number: bool = False) -> dict:
+    """Sort a dict by keys"""
+    if number:
+        return dict(sorted(d.items(), key=lambda item: int(item[0])))
+    else:
+        return dict(sorted(d.items(), key=lambda item: item[0]))
+
+
 # -----------------------------------------------------------
 # Class SlowBar 
 # -----------------------------------------------------------
@@ -649,8 +674,6 @@ class StatsNotFound(Exception):
     pass
 
 
-
-
 # -----------------------------------------------------------
 # Class WG 
 # -----------------------------------------------------------
@@ -686,7 +709,7 @@ class WG:
     #                                 WHERE account_id = {} AND stat = {} 
     #                                 AND date >= {} LIMIT 1;"""
           
-    
+    # TANK_STATS
     SQL_TANK_STATS_TBL          = 'tank_stats'
 
     SQL_TANK_STATS_CREATE_TBL   = 'CREATE TABLE IF NOT EXISTS ' + SQL_TANK_STATS_TBL + \
@@ -700,6 +723,7 @@ class WG:
 
     SQL_TANK_STATS_UPDATE       = 'REPLACE INTO ' + SQL_TANK_STATS_TBL + '(account_id, tank_id, update_time, stats) VALUES(?,?,?,?)'
 
+    # PLAYER_STATS
     SQL_PLAYER_STATS_TBL        = 'player_stats'
 
     SQL_PLAYER_STATS_CREATE_TBL = 'CREATE TABLE IF NOT EXISTS ' + SQL_PLAYER_STATS_TBL + \
@@ -713,6 +737,7 @@ class WG:
 
     SQL_PLAYER_STATS_CACHED     = 'SELECT * FROM ' +  SQL_PLAYER_STATS_TBL + ' WHERE account_id = ? AND update_time > ?'
 
+    # PLAYER_ACHIEVEMENTS
     SQL_PLAYER_ACHIEVEMENTS_TBL  = 'player_achievements'
 
     SQL_PLAYER_ACHIEVEMENTS_CREATE_TBL = 'CREATE TABLE IF NOT EXISTS ' + SQL_PLAYER_ACHIEVEMENTS_TBL + \
@@ -720,12 +745,13 @@ class WG:
                                     update_time INTEGER NOT NULL, 
                                     stats TEXT)"""
 
+    SQL_PLAYER_ACHIEVEMENTS_COUNT      = 'SELECT COUNT(*) FROM ' + SQL_PLAYER_ACHIEVEMENTS_TBL
+    
     SQL_PLAYER_ACHIEVEMENTS_CACHED     = 'SELECT * FROM ' +  SQL_PLAYER_ACHIEVEMENTS_TBL + ' WHERE account_id = ? AND update_time > ?'
 
     SQL_PLAYER_ACHIEVEMENTS_UPDATE     = 'REPLACE INTO ' + SQL_PLAYER_ACHIEVEMENTS_TBL + '(account_id, update_time, stats) VALUES(?,?,?)'
 
-    SQL_PLAYER_ACHIEVEMENTS_COUNT      = 'SELECT COUNT(*) FROM ' + SQL_PLAYER_ACHIEVEMENTS_TBL
-
+    
     SQL_TABLES                  = [ SQL_PLAYER_STATS_TBL, SQL_TANK_STATS_TBL, SQL_PLAYER_ACHIEVEMENTS_TBL ]
 
     SQL_CHECK_TABLE_EXITS       = """SELECT name FROM sqlite_master WHERE type='table' AND name=?"""
@@ -908,33 +934,18 @@ class WG:
 
 
     @classmethod
-    def get_map(cls, map_str: str) -> str:
+    def get_map(self, map_str: str) -> str:
         """Return map name from short map string in replays"""
         try:
-            return cls.maps[map_str]
+            return self.maps[map_str]
         except:
             error('Map ' + map_str + ' not found')
         return None
     
 
     @classmethod
-    def get_map_user_strs(cls) -> str:
-        return cls.maps.keys()
-
-
-    @classmethod
-    def tank_str2name(cls, tank_str: str) -> str:
-        """Return tank name from short tank string in replays"""
-        try:
-            return cls.tanks["userStr"][tank_str]
-        except:
-            error('Tank ' + tank_str + ' not found')
-        return tank_str
-
-
-    @classmethod
-    def get_tank_user_strs(cls) -> str:
-        return cls.tanks["userStr"].keys()
+    def get_tank_user_strs(self) -> str:
+        return self.tanks["userStr"].keys()
 
 
     @classmethod
@@ -1070,11 +1081,15 @@ class WG:
     def load_tanks(self, tankopedia_fn: str):
         """Load tanks from tankopedia JSON"""
         if tankopedia_fn == None:
-            return False 
-
+            return False
         try:
             with open(tankopedia_fn, 'rt', encoding='utf8') as f:
                 self.tanks = json.loads(f.read())
+                self.tanks['tankStr'] = dict()
+                p = re.compile('.+_short$')
+                for usr_str in self.tanks['userStr']:
+                    if p.match(usr_str) == None:
+                        self.tanks['tankStr'][self.tanks['userStr'][usr_str]] = usr_str
                 self.tanks_by_tier = dict()
                 for tier in range(1,11):
                     self.tanks_by_tier[str(tier)] = list()
@@ -1084,7 +1099,44 @@ class WG:
         except Exception as err:
             error('Could not read tankopedia: ' + tankopedia_fn, err) 
         return False     
-     
+    
+
+    def get_replay_filename(self, replay: dict):
+        try:
+            summary = replay['data']['summary']
+            timestamp   = int(summary['battle_start_timestamp'])
+            timestamp    = get_date_str(timestamp, date_format='%Y%m%d_%H%M')
+            player      = summary['player_name']
+            arena_id    = int(summary['arena_unique_id'])
+            vehicle     = self.name2tank_str(summary['vehicle'])
+            vehicle.replace(' ','_')
+
+            return '{:s}_{:s}_{:s}_{:d}.wotbreplay.json'.format(timestamp, player, vehicle, arena_id)
+        except Exception as err:
+            error('Unexpected Exception', err) 
+            return None 
+
+    def get_map_user_strs(self) -> str:
+        return self.maps.keys()
+
+
+    def tank_str2name(self, tank_str: str) -> str:
+        """Return tank name from short tank string in replays"""
+        try:
+            return self.tanks["userStr"][tank_str]
+        except:
+            error('Tank ' + tank_str + ' not found')
+        return tank_str
+
+    
+    def name2tank_str(self, name: str) -> str:
+        """Return shorttank string from tank string in replays"""
+        try:
+            return self.tanks["tankStr"][name]
+        except:
+            debug('Tank ' + name + ' not found')
+        return name
+
 
     def get_tanks_by_tier(self, tier: int) -> list():
         """Returns tank_ids by tier"""
@@ -1261,7 +1313,7 @@ class WG:
                 # debug('JSON Response received: ' + str(json_data))
                 stats = json_data['data'][str(account_id)]
                 if cache:
-                    await self.put_2_statsQ('tank_stats', [account_id, tank_ids], stats)
+                    await self.save_tank_stats(account_id, tank_ids, stats)
                 return stats
         except Exception as err:
             error(exception=err)
@@ -1294,7 +1346,7 @@ class WG:
                 # debug('JSON Response received: ' + str(json_data))
                 stats = json_data['data'][str(account_id)]
                 if cache:
-                    await self.put_2_statsQ('player_stats', [account_id], stats)
+                    await self.save_player_stats(account_id, stats)
                 return stats
         except Exception as err:
             error(exception=err)
@@ -1333,7 +1385,7 @@ class WG:
                 for account_id in json_data['data'].keys():
                     stats[account_id] = json_data['data'][account_id]
                     if cache:
-                        await self.put_2_statsQ('player_achievements', [int(account_id)], json_data['data'][account_id])
+                        await self.save_player_achievements(int(account_id), json_data['data'][account_id])
                 return stats
         except Exception as err:
             error(exception=err)
@@ -1440,7 +1492,7 @@ class WG:
         return None
 
 
-    async def put_2_statsQ(self, statsType: str, key: list, stats: list):
+    async def save_stats(self, statsType: str, key: list, stats: list):
         """Save stats to a async queue to be saved by the stat_saver -task"""
         if self.statsQ == None:
             return False
@@ -1449,6 +1501,30 @@ class WG:
             return True
 
 
+    async def save_player_stats(self, account_id: int, stats: dict):
+        try:
+            await self.save_stats(MODE_PLAYER_STATS, [int(account_id)], stats)
+        except Exception as err:
+            error(exception=err)
+        return None
+
+
+    async def save_player_achievements(self, account_id: int, stats: dict):
+        try:
+            await self.save_stats(MODE_PLAYER_ACHIEVEMENTS, [int(account_id)], stats)
+        except Exception as err:
+            error(exception=err)
+        return None
+
+
+    async def save_tank_stats(self, account_id: int, tank_ids: list, stats: dict):
+        try:
+            await self.save_stats(MODE_TANK_STATS, [int(account_id), tank_ids], stats)
+        except Exception as err:
+            error(exception=err)
+        return None
+
+
     async def stat_saver(self): 
         """Async task for saving stats into cache in background"""
 
@@ -1464,8 +1540,13 @@ class WG:
 
             await self.cache.commit()
             
-            async with self.cache.execute(WG.SQL_TANK_STATS_COUNT) as cursor:
-                debug('Cache contains: ' + str((await cursor.fetchone())[0]) + ' cached player tank stat records' )
+            if is_debug():
+                async with self.cache.execute(WG.SQL_PLAYER_STATS_COUNT) as cursor:
+                    debug('Cache contains: ' + str((await cursor.fetchone())[0]) + ' player stat records' )
+                async with self.cache.execute(WG.SQL_PLAYER_ACHIEVEMENTS_COUNT) as cursor:
+                    debug('Cache contains: ' + str((await cursor.fetchone())[0]) + ' player achievement records' )
+                async with self.cache.execute(WG.SQL_TANK_STATS_COUNT) as cursor:
+                    debug('Cache contains: ' + str((await cursor.fetchone())[0]) + ' player tank stat records' )
         except Exception as err:
             error(exception=err)
             sys.exit(1)
@@ -1474,19 +1555,20 @@ class WG:
             try:
                 stats = await self.statsQ.get()
             
-                stats_type  = stats[0]
+                stat_type  = stats[0]
                 key         = stats[1]
                 stats_data  = stats[2]
                 update_time = stats[3]
 
-                if stats_type == 'tank_stats':
+                #debug('stat_type: ' + stat_type + ', key: ' + str(key))
+                if stat_type == MODE_TANK_STATS:
                     await self.store_tank_stats(key, stats_data, update_time)
-                elif stats_type == 'player_stats':
+                elif stat_type == MODE_PLAYER_STATS:
                     await self.store_player_stats(key, stats_data, update_time)
-                elif stats_type == 'player_achievements':
+                elif stat_type == MODE_PLAYER_ACHIEVEMENTS:
                     await self.store_player_achievements(key, stats_data, update_time)
                 else: 
-                    error('Function to saves stats type \'' + stats_type + '\' is not implemented yet')
+                    error('Function to saves stats type \'' + stat_type + '\' is not implemented yet')
             
             except asyncio.CancelledError:
                 # this is an eternal loop that will wait until cancelled	
@@ -1521,7 +1603,7 @@ class WG:
                 for stat in stats_data:
                     tank_id = stat['tank_id']
                     await self.cache.execute(WG.SQL_TANK_STATS_UPDATE, (account_id, tank_id, update_time, json.dumps(stat)))
-                    tank_ids.remove(tank_id)
+                    tank_ids.discard(tank_id)
             # no stats found => Add None to mark that
             for tank_id in tank_ids:
                 await self.cache.execute(WG.SQL_TANK_STATS_UPDATE, (account_id, tank_id, update_time, None))
@@ -1793,6 +1875,8 @@ class WoTinspector:
             except Exception as err:
                 debug(exception=err, id=N)
             await asyncio.sleep(SLEEP)
+            _id  = self.read_replay_id(json_resp)
+            json_resp['_id']  =  _id
             
         debug(' Could not post replay: ' + title, id=N)
         return json_resp

extract_tankopedia.py

@@ -79,8 +79,8 @@ async def main(argv):
         tankopedia = collections.OrderedDict()
         tankopedia['status'] = 'ok'
         tankopedia['meta'] = { "count":  len(tanks) }
-        tankopedia['data'] = tanks
-        tankopedia['userStr'] = userStrs
+        tankopedia['data'] = bu.sort_dict(tanks, number=True)
+        tankopedia['userStr'] = bu.sort_dict(userStrs)
         bu.verbose_std('New tankopedia \'' + args.tanks + '\' contains ' + str(len(tanks)) + ' tanks')
         bu.verbose_std('New tankopedia \'' + args.tanks + '\' contains ' + str(len(userStrs)) + ' tank strings')        
         await outfile.write(json.dumps(tankopedia, ensure_ascii=False, indent=4, sort_keys=False))
@@ -115,13 +115,12 @@ async def extract_tanks(blitz_app_base : str, nation: str):
             for data in tankList['root'].keys():
                 tank_xml = tankList['root'][data]
                 tank = dict()
-                tank['tank_id'] = await get_tank_id(nation, int(tank_xml['id']))
-                tank['userStr'] = tank_xml['userString']
-                tank['nation']  = nation
-                tank['tier']    = int(tank_xml['level'])
-                #debug(tank_xml['price'])
                 tank['is_premium'] = issubclass(type(tank_xml['price']), dict)
+                tank['nation']  = nation
+                tank['tank_id'] = await get_tank_id(nation, int(tank_xml['id']))
+                tank['tier']    = int(tank_xml['level'])
                 tank['type'] = await get_tank_type(tank_xml['tags'])
+                tank['userStr'] = tank_xml['userString']
                 #bu.debug('Reading tank string: ' + tank['userStr'], force=True)
                 tanks.append(tank)
         except Exception as err:
@@ -165,7 +164,10 @@ async def convert_tank_names(tanklist : list, tank_strs: dict) -> dict:
             tank['name'] = tank_strs[tank['userStr']]
             #userStrs[tank['userStr'].split(':')[1]] = tank['name']
             tank.pop('userStr', None)
-            tankopedia[str(tank['tank_id'])] = tank
+            tank_tmp = collections.OrderedDict()
+            for key in sorted(tank.keys()):
+                tank_tmp[key] = tank[key]
+            tankopedia[str(tank['tank_id'])] = tank_tmp
 
         for tank_str in tank_strs:
             skip = False

get_tankopedia.py

@@ -1,4 +1,4 @@
-#!/usr/local/bin/python3.7
+#!/usr/bin/env python3
 
 # Get Tankpedia from WoTinspector.com
 

maps.json

@@ -43,13 +43,17 @@
     "neptune_02": "Normandy - Beach",
     "north": "North",
     "ordeal": "Trial by Fire",
+    "plant": "Ghost Factory",
     "pliego": "Castilla",
     "port": "Port Bay",
     "rift": "Hellas",
     "rock": "Mayan Ruins",
     "rudniki": "Mines",
+    "rudniki_01": "Mines - Hill",
+    "rudniki_02": "Mines - Village",
+    "rudniki_03": "Mines - Center",
     "savanna": "Oasis Palms",
     "skit": "Naval Frontier",
     "test": "World of Ducks",
     "tutorial": "Proving Grounds"
-}
+}

releases.md

@@ -2,6 +2,13 @@
 
 | Release | Date     | Notes       | Inaccurate? |
 |:-------:|:--------:|-------------|-----|
+| 8.4 | 2021-10-20 | Polish heavy tanks |  |
+| 8.3 | 2021-09-16 | FV4005, FV4022, VK 91.01 (P) and WZ-121 legendary camos, -5 deg gun depression to WZ-120 |  |
+| 8.2 | 2021-08-18 | Ghost Factory map, XP/credits nerf for inactive players, new store UI |  |
+| 8.1 | 2021-07-15 | Armor highlighting updated |  |
+| 8.0 | 2021-06-08 | American experimental heavies (Yoh-line), PBR camos |  |
+| 7.9 | 2021-04-27 | Mines rework, new tutorial, colorblind mode, JgPz E100 legendary camo |  |
+| 7.8 | 2021-03-24 | Mid-tier balance changes, Grille line new consumables  |  |
 | 7.7 | 2021-02-24 | Czechoslovakian medium tanks  |  |
 | 7.6 | 2021-01-27 | AMC 13 90, TOG II buffs, free replay camera  |  |
 | 7.5 | 2020-12-09 | ATGM removal, Batchat, Kpz70, T-2020 buff |  |

upload_wotb_replays.py

@@ -195,7 +195,7 @@ async def replayWorker(queue: asyncio.Queue, workerID: int, account_id: int, pri
 						bu.verbose_std(msg_str + title + ' has already been posted. Skipping.' )
 					else:
 						os.remove(replay_json_fn)
-						bu.debug(msg_str + "Replay JSON not valid: Deleting " + replay_json_fn, id=workerID)
+						bu.debug(msg_str + "Replay JSON not valid/complete: Deleting " + replay_json_fn, id=workerID)
 					SKIPPED_N += 1						
 					queue.task_done()						
 					continue
@@ -216,13 +216,13 @@ async def replayWorker(queue: asyncio.Queue, workerID: int, account_id: int, pri
 							if not bu.debug(msg_str + 'Replay saved OK: ' + filename):
 								bu.verbose_std(msg_str + title + ' posted')
 						else:
-							bu.warning(msg_str +'Replay file is not valid: ' + filename)
+							bu.warning(msg_str +'Replay file is not valid/complete: ' + filename)
 							ERROR_N += 1
 					else:
 						bu.error(msg_str + 'Error saving replay: ' + filename)
 						ERROR_N += 1
 				else:
-					bu.error(msg_str + 'Replay file is not valid: ' + filename)
+					bu.error(msg_str + 'Replay file is not valid/complete: ' + filename)
 					ERROR_N += 1	
 		except Exception as err:
 			bu.error(msg_str + 'Unexpected Exception: ' + str(type(err)) + ' : ' + str(err) )
@@ -286,14 +286,14 @@ def getTitle(replayfile: str, title: str = None, i : int = 0) -> str:
 					# bu.debug('Replay file\'s metadata: opened')
 					try:
 						metadata_json = json.load(meta)
-						#player = metadata_json['playerName']
-						tank = wg.tank_str2name(metadata_json['playerVehicleName'])
+						#player = metadata_json['playerName']						
+						tank_name = wg.get_tank_name(metadata_json['vehicleCompDescriptor'])
 						map_name = wg.get_map(metadata_json['mapName'])
-						bu.debug('Tank: ' + tank +  ' Map: ' + map_name)
+						bu.debug('Tank: ' + tank_name if tank_name != None else "NOT FOUND IN tanks.json" +  ' Map: ' + map_name)
 					except Exception as err:
 						bu.error(exception = err)
-			if (tank != None) and  (map_name != None):
-				title = tank + ' @ ' + map_name
+			if (tank_name != None) and  (map_name != None):
+				title = tank_name + ' @ ' + map_name
 			else:
 				title = re.sub('\\.wotbreplay$', '', filename)
 		except Exception as err: