import collections import cProfile import io import itertools import os import pstats import psutil import random import re import sqlite3 import struct import subprocess import sys import threading import time import traceback import typing import yaml from hydrus.core import HydrusBoot from hydrus.core import HydrusConstants as HC from hydrus.core import HydrusExceptions from hydrus.core import HydrusGlobals as HG from hydrus.core import HydrusText def default_dict_list(): return collections.defaultdict( list ) def default_dict_set(): return collections.defaultdict( set ) def BuildKeyToListDict( pairs ): d = collections.defaultdict( list ) for ( key, value ) in pairs: d[ key ].append( value ) return d def BuildKeyToSetDict( pairs ): d = collections.defaultdict( set ) for ( key, value ) in pairs: d[ key ].add( value ) return d def BytesToNoneOrHex( b: typing.Optional[ bytes ] ): if b is None: return None else: return b.hex() def CalculateScoreFromRating( count, rating ): # https://www.evanmiller.org/how-not-to-sort-by-average-rating.html positive = count * rating negative = count * ( 1.0 - rating ) # positive + negative = count # I think I've parsed this correctly from the website! Not sure though! score = ( ( positive + 1.9208 ) / count - 1.96 * ( ( ( positive * negative ) / count + 0.9604 ) ** 0.5 ) / count ) / ( 1 + 3.8416 / count ) return score def CheckProgramIsNotShuttingDown(): if HG.model_shutdown: raise HydrusExceptions.ShutdownException( 'Application is shutting down!' ) def CleanRunningFile( db_path, instance ): # just to be careful path = os.path.join( db_path, instance + '_running' ) try: os.remove( path ) except: pass def ConvertFloatToPercentage( f ): return '{:.1f}%'.format( f * 100 ) def ConvertIntToPixels( i ): if i == 1: return 'pixels' elif i == 1000: return 'kilopixels' elif i == 1000000: return 'megapixels' else: return 'megapixels' def ConvertIndexToPrettyOrdinalString( index: int ): if index >= 0: return ConvertIntToPrettyOrdinalString( index + 1 ) else: return ConvertIntToPrettyOrdinalString( index ) def ConvertIntToPrettyOrdinalString( num: int ): if num == 0: return 'unknown position' tens = ( abs( num ) % 100 ) // 10 if tens == 1: ordinal = 'th' else: remainder = abs( num ) % 10 if remainder == 1: ordinal = 'st' elif remainder == 2: ordinal = 'nd' elif remainder == 3: ordinal = 'rd' else: ordinal = 'th' s = '{}{}'.format( ToHumanInt( abs( num ) ), ordinal ) if num < 0: if num == -1: s = 'last' else: s = '{} from last'.format( s ) return s def ConvertIntToUnit( unit ): if unit == 1: return 'B' elif unit == 1024: return 'KB' elif unit == 1048576: return 'MB' elif unit == 1073741824: return 'GB' def ConvertMillisecondsToPrettyTime( ms ): hours = ms // 3600000 if hours == 1: hours_result = '1 hour' else: hours_result = str( hours ) + ' hours' ms = ms % 3600000 minutes = ms // 60000 if minutes == 1: minutes_result = '1 minute' else: minutes_result = str( minutes ) + ' minutes' ms = ms % 60000 seconds = ms // 1000 if seconds == 1: seconds_result = '1 second' else: seconds_result = str( seconds ) + ' seconds' detailed_seconds = ms / 1000 detailed_seconds_result = '{:.1f} seconds'.format( detailed_seconds ) ms = ms % 1000 if hours > 0: return hours_result + ' ' + minutes_result if minutes > 0: return minutes_result + ' ' + seconds_result if seconds > 0: return detailed_seconds_result ms = int( ms ) if ms == 1: milliseconds_result = '1 millisecond' else: milliseconds_result = '{} milliseconds'.format( ms ) return milliseconds_result def ConvertNumericalRatingToPrettyString( lower, upper, rating, rounded_result = False, out_of = True ): rating_converted = ( rating * ( upper - lower ) ) + lower if rounded_result: rating_converted = round( rating_converted ) s = '{:.2f}'.format( rating_converted ) if out_of and lower in ( 0, 1 ): s += '/{:.2f}'.format( upper ) return s def ConvertPixelsToInt( unit ): if unit == 'pixels': return 1 elif unit == 'kilopixels': return 1000 elif unit == 'megapixels': return 1000000 def ConvertPrettyStringsToUglyNamespaces( pretty_strings ): result = { s for s in pretty_strings if s != 'no namespace' } if 'no namespace' in pretty_strings: result.add( '' ) return result def ConvertResolutionToPrettyString( resolution ): if resolution in HC.NICE_RESOLUTIONS: return HC.NICE_RESOLUTIONS[ resolution ] ( width, height ) = resolution return '{}x{}'.format( ToHumanInt( width ), ToHumanInt( height ) ) def ConvertStatusToPrefix( status ): if status == HC.CONTENT_STATUS_CURRENT: return '' elif status == HC.CONTENT_STATUS_PENDING: return '(+) ' elif status == HC.CONTENT_STATUS_PETITIONED: return '(-) ' elif status == HC.CONTENT_STATUS_DELETED: return '(X) ' def TimeDeltaToPrettyTimeDelta( seconds, show_seconds = True ): if seconds is None: return 'per month' if seconds == 0: return '0 seconds' if seconds < 0: seconds = abs( seconds ) if seconds >= 60: seconds = int( seconds ) MINUTE = 60 HOUR = 60 * MINUTE DAY = 24 * HOUR MONTH = 30 * DAY YEAR = 12 * MONTH lines = [] lines.append( ( 'year', YEAR ) ) lines.append( ( 'month', MONTH ) ) lines.append( ( 'day', DAY ) ) lines.append( ( 'hour', HOUR ) ) lines.append( ( 'minute', MINUTE ) ) if show_seconds: lines.append( ( 'second', 1 ) ) result_components = [] for ( time_string, duration ) in lines: time_quantity = seconds // duration seconds %= duration if time_quantity > 0: s = ToHumanInt( time_quantity ) + ' ' + time_string if time_quantity > 1: s += 's' result_components.append( s ) if len( result_components ) == 2: # we now have 1 month 2 days break else: if len( result_components ) > 0: # something like '1 year' -- in which case we do not care about the days and hours break result = ' '.join( result_components ) elif seconds > 1: if int( seconds ) == seconds: result = ToHumanInt( seconds ) + ' seconds' else: result = '{:.1f} seconds'.format( seconds ) elif seconds == 1: result = '1 second' elif seconds > 0.1: result = '{} milliseconds'.format( int( seconds * 1000 ) ) elif seconds > 0.01: result = '{:.1f} milliseconds'.format( int( seconds * 1000 ) ) elif seconds > 0.001: result = '{:.2f} milliseconds'.format( int( seconds * 1000 ) ) else: result = '{} microseconds'.format( int( seconds * 1000000 ) ) return result def ConvertTimestampToPrettyExpires( timestamp ): if timestamp is None: return 'does not expire' if timestamp == 0: return 'unknown expiration' try: time_delta_string = TimestampToPrettyTimeDelta( timestamp ) if TimeHasPassed( timestamp ): return 'expired ' + time_delta_string else: return 'expires ' + time_delta_string except: return 'unparseable time {}'.format( timestamp ) def ConvertTimestampToPrettyTime( timestamp, in_utc = False, include_24h_time = True ): if timestamp is None: return 'unknown time' if include_24h_time: phrase = '%Y-%m-%d %H:%M:%S' else: phrase = '%Y-%m-%d' try: if in_utc: struct_time = time.gmtime( timestamp ) phrase = phrase + ' UTC' else: struct_time = time.localtime( timestamp ) return time.strftime( phrase, struct_time ) except: return 'unparseable time {}'.format( timestamp ) def TimestampToPrettyTimeDelta( timestamp, just_now_string = 'now', just_now_threshold = 3, history_suffix = ' ago', show_seconds = True, no_prefix = False ): if timestamp is None: return 'at an unknown time' if not show_seconds: just_now_threshold = max( just_now_threshold, 60 ) try: time_delta = abs( timestamp - GetNow() ) if time_delta <= just_now_threshold: return just_now_string time_delta_string = TimeDeltaToPrettyTimeDelta( time_delta, show_seconds = show_seconds ) if TimeHasPassed( timestamp ): return '{}{}'.format( time_delta_string, history_suffix ) else: if no_prefix: return time_delta_string else: return 'in ' + time_delta_string except: return 'unparseable time {}'.format( timestamp ) def ConvertUglyNamespaceToPrettyString( namespace ): if namespace is None or namespace == '': return 'no namespace' else: return namespace def ConvertUglyNamespacesToPrettyStrings( namespaces ): namespaces = sorted( namespaces ) result = [ ConvertUglyNamespaceToPrettyString( namespace ) for namespace in namespaces ] return result def ConvertUnitToInt( unit ): if unit == 'B': return 1 elif unit == 'KB': return 1024 elif unit == 'MB': return 1048576 elif unit == 'GB': return 1073741824 def ConvertValueRangeToBytes( value, range ): return ToHumanBytes( value ) + '/' + ToHumanBytes( range ) def ConvertValueRangeToPrettyString( value, range ): return ToHumanInt( value ) + '/' + ToHumanInt( range ) def ConvertValueRangeToScanbarTimestampsMS( value_ms, range_ms ): value_ms = int( round( value_ms ) ) range_hours = range_ms // 3600000 value_hours = value_ms // 3600000 range_minutes = ( range_ms % 3600000 ) // 60000 value_minutes = ( value_ms % 3600000 ) // 60000 range_seconds = ( range_ms % 60000 ) // 1000 value_seconds = ( value_ms % 60000 ) // 1000 range_ms = range_ms % 1000 value_ms = value_ms % 1000 if range_hours > 0: # 0:01:23.033/1:12:57.067 time_phrase = '{}:{:0>2}:{:0>2}.{:0>3}' args = ( value_hours, value_minutes, value_seconds, value_ms, range_hours, range_minutes, range_seconds, range_ms ) elif range_minutes > 0: # 01:23.033/12:57.067 or 0:23.033/1:57.067 if range_minutes > 9: time_phrase = '{:0>2}:{:0>2}.{:0>3}' else: time_phrase = '{:0>1}:{:0>2}.{:0>3}' args = ( value_minutes, value_seconds, value_ms, range_minutes, range_seconds, range_ms ) else: # 23.033/57.067 or 3.033/7.067 or 0.033/0.067 if range_seconds > 9: time_phrase = '{:0>2}.{:0>3}' else: time_phrase = '{:0>1}.{:0>3}' args = ( value_seconds, value_ms, range_seconds, range_ms ) full_phrase = '{}/{}'.format( time_phrase, time_phrase ) result = full_phrase.format( *args ) return result def DebugPrint( debug_info ): Print( debug_info ) sys.stdout.flush() sys.stderr.flush() def DedupeList( xs ): xs_seen = set() xs_return = [] for x in xs: if x in xs_seen: continue xs_return.append( x ) xs_seen.add( x ) return xs_return def GenerateKey(): return os.urandom( HC.HYDRUS_KEY_LENGTH ) def Get64BitHammingDistance( phash1, phash2 ): # old way of doing this was: #while xor > 0: # # distance += 1 # xor &= xor - 1 # # convert to unsigned long long, then xor # then through the power of stackexchange magic, we get number of bits in record time # Here it is: https://stackoverflow.com/questions/9829578/fast-way-of-counting-non-zero-bits-in-positive-integer/9830282#9830282 n = struct.unpack( '!Q', phash1 )[0] ^ struct.unpack( '!Q', phash2 )[0] n = ( n & 0x5555555555555555 ) + ( ( n & 0xAAAAAAAAAAAAAAAA ) >> 1 ) # 10101010, 01010101 n = ( n & 0x3333333333333333 ) + ( ( n & 0xCCCCCCCCCCCCCCCC ) >> 2 ) # 11001100, 00110011 n = ( n & 0x0F0F0F0F0F0F0F0F ) + ( ( n & 0xF0F0F0F0F0F0F0F0 ) >> 4 ) # 11110000, 00001111 n = ( n & 0x00FF00FF00FF00FF ) + ( ( n & 0xFF00FF00FF00FF00 ) >> 8 ) # etc... n = ( n & 0x0000FFFF0000FFFF ) + ( ( n & 0xFFFF0000FFFF0000 ) >> 16 ) n = ( n & 0x00000000FFFFFFFF ) + ( n >> 32 ) # you technically are going n & 0xFFFFFFFF00000000 at the end, but that's a no-op with the >> 32 afterwards, so can be omitted return n def GetEmptyDataDict(): data = collections.defaultdict( default_dict_list ) return data def GetNonDupeName( original_name, disallowed_names ): i = 1 non_dupe_name = original_name while non_dupe_name in disallowed_names: non_dupe_name = original_name + ' (' + str( i ) + ')' i += 1 return non_dupe_name def GetNow(): return int( time.time() ) def GetNowFloat(): return time.time() def GetNowPrecise(): return time.perf_counter() def GetSiblingProcessPorts( db_path, instance ): path = os.path.join( db_path, instance + '_running' ) if os.path.exists( path ): with open( path, 'r', encoding = 'utf-8' ) as f: file_text = f.read() try: ( pid, create_time ) = HydrusText.DeserialiseNewlinedTexts( file_text ) pid = int( pid ) create_time = float( create_time ) except ValueError: return None try: if psutil.pid_exists( pid ): ports = [] p = psutil.Process( pid ) for conn in p.connections(): if conn.status == 'LISTEN': ports.append( int( conn.laddr[1] ) ) return ports except psutil.Error: return None return None def GetSubprocessEnv(): if HG.subprocess_report_mode: env = os.environ.copy() ShowText( 'Your unmodified env is: {}'.format( env ) ) env = os.environ.copy() if HydrusBoot.ORIGINAL_PATH is not None: env[ 'PATH' ] = HydrusBoot.ORIGINAL_PATH if HC.RUNNING_FROM_FROZEN_BUILD: # let's make a proper env for subprocess that doesn't have pyinstaller woo woo in it changes_made = False orig_swaperoo_strings = [ 'LD_LIBRARY_PATH', 'XDG_DATA_DIRS' ] ok_to_remove_absent_orig = [ 'LD_LIBRARY_PATH' ] for key in orig_swaperoo_strings: orig_key = '{}_ORIG'.format( key ) if orig_key in env: env[ key ] = env[ orig_key ] changes_made = True elif key in env and key in ok_to_remove_absent_orig: del env[ key ] changes_made = True remove_if_hydrus_base_dir = [ 'QT_PLUGIN_PATH', 'QML2_IMPORT_PATH', 'SSL_CERT_FILE' ] hydrus_base_dir = HG.controller.GetDBDir() for key in remove_if_hydrus_base_dir: if key in env and env[ key ].startswith( hydrus_base_dir ): del env[ key ] changes_made = True if ( HC.PLATFORM_LINUX or HC.PLATFORM_MACOS ): if 'PATH' in env: # fix for pyinstaller, which drops this stuff for some reason and hence breaks ffmpeg path = env[ 'PATH' ] path_locations = set( path.split( ':' ) ) desired_path_locations = [ '/usr/bin', '/usr/local/bin' ] for desired_path_location in desired_path_locations: if desired_path_location not in path_locations: path = desired_path_location + ':' + path env[ 'PATH' ] = path changes_made = True if 'XDG_DATA_DIRS' in env: xdg_data_dirs = env[ 'XDG_DATA_DIRS' ] # pyinstaller can just replace this nice usually long str with multiple paths with base_dir/share # absent the _orig above to rescue this, we'll populate with basic if ':' not in xdg_data_dirs and HC.BASE_DIR in xdg_data_dirs: xdg_data_dirs = '/usr/local/share:/usr/share' changes_made = True if not changes_made: env = None else: env = None return env def GetSubprocessHideTerminalStartupInfo(): if HC.PLATFORM_WINDOWS: # This suppresses the terminal window that tends to pop up when calling ffmpeg or whatever startupinfo = subprocess.STARTUPINFO() startupinfo.dwFlags |= subprocess.STARTF_USESHOWWINDOW else: startupinfo = None return startupinfo def GetSubprocessKWArgs( hide_terminal = True, text = False ): sbp_kwargs = {} sbp_kwargs[ 'env' ] = GetSubprocessEnv() if text: # probably need to override the stdXXX pipes with i/o encoding wrappers in the case of 3.5 here if sys.version_info.minor >= 6: sbp_kwargs[ 'encoding' ] = 'utf-8' if sys.version_info.minor >= 7: sbp_kwargs[ 'text' ] = True else: sbp_kwargs[ 'universal_newlines' ] = True if hide_terminal: sbp_kwargs[ 'startupinfo' ] = GetSubprocessHideTerminalStartupInfo() if HG.subprocess_report_mode: message = 'KWargs are: {}'.format( sbp_kwargs ) ShowText( message ) return sbp_kwargs def GetTimeDeltaSinceTime( timestamp ): time_since = timestamp - GetNow() result = min( time_since, 0 ) return - result def GetTimeDeltaUntilTime( timestamp ): time_remaining = timestamp - GetNow() return max( time_remaining, 0 ) def GetTimeDeltaUntilTimeFloat( timestamp ): time_remaining = timestamp - GetNowFloat() return max( time_remaining, 0.0 ) def GetTimeDeltaUntilTimePrecise( t ): time_remaining = t - GetNowPrecise() return max( time_remaining, 0.0 ) def GetTypeName( obj_type ): if hasattr( obj_type, '__name__' ): return obj_type.__name__ else: return repr( obj_type ) def GenerateHumanTextSortKey(): """Solves the 19, 20, 200, 21, 22 issue when sorting 'Page 21.jpg' type strings. Breaks the string into groups of text and int (i.e. ( "Page ", 21, ".jpg" ) ).""" int_convert = lambda t: int( t ) if t.isdecimal() else t split_alphanum = lambda t: tuple( ( int_convert( sub_t ) for sub_t in re.split( '([0-9]+)', t.lower() ) ) ) return split_alphanum HumanTextSortKey = GenerateHumanTextSortKey() def HumanTextSort( texts ): texts.sort( key = HumanTextSortKey ) def IntelligentMassIntersect( sets_to_reduce ): answer = None for set_to_reduce in sets_to_reduce: if len( set_to_reduce ) == 0: return set() if answer is None: answer = set( set_to_reduce ) else: if len( answer ) == 0: return set() else: answer.intersection_update( set_to_reduce ) if answer is None: return set() else: return answer def IsAlreadyRunning( db_path, instance ): path = os.path.join( db_path, instance + '_running' ) if os.path.exists( path ): with open( path, 'r', encoding = 'utf-8' ) as f: file_text = f.read() try: ( pid, create_time ) = HydrusText.DeserialiseNewlinedTexts( file_text ) pid = int( pid ) create_time = float( create_time ) except ValueError: return False try: me = psutil.Process() if me.pid == pid and me.create_time() == create_time: # this is me! there is no conflict, lol! # this happens when a linux process restarts with os.execl(), for instance (unlike Windows, it keeps its pid) return False if psutil.pid_exists( pid ): p = psutil.Process( pid ) if p.create_time() == create_time and p.is_running(): return True except psutil.Error: return False return False def IterateHexPrefixes(): hex_chars = '0123456789abcdef' for ( one, two ) in itertools.product( hex_chars, hex_chars ): prefix = one + two yield prefix def LastShutdownWasBad( db_path, instance ): path = os.path.join( db_path, instance + '_running' ) if os.path.exists( path ): return True else: return False def MassUnion( lists ): return { item for item in itertools.chain.from_iterable( lists ) } def MedianPop( population ): # assume it has at least one and comes sorted median_index = len( population ) // 2 row = population.pop( median_index ) return row def MergeKeyToListDicts( key_to_list_dicts ): result = collections.defaultdict( list ) for key_to_list_dict in key_to_list_dicts: for ( key, value ) in list(key_to_list_dict.items()): result[ key ].extend( value ) return result def PartitionIterator( pred: typing.Callable[ [ object ], bool ], stream: typing.Iterable[ object ] ): ( t1, t2 ) = itertools.tee( stream ) return ( itertools.filterfalse( pred, t1 ), filter( pred, t2 ) ) def PartitionIteratorIntoLists( pred: typing.Callable[ [ object ], bool ], stream: typing.Iterable[ object ] ): ( a, b ) = PartitionIterator( pred, stream ) return ( list( a ), list( b ) ) def ParseHashesFromRawHexText( hash_type, hex_hashes_raw ): hash_type_to_hex_length = { 'md5' : 32, 'sha1' : 40, 'sha256' : 64, 'sha512' : 128 } hex_hashes = HydrusText.DeserialiseNewlinedTexts( hex_hashes_raw ) # convert md5:abcd to abcd hex_hashes = [ hex_hash.split( ':' )[-1] for hex_hash in hex_hashes ] hex_hashes = [ HydrusText.HexFilter( hex_hash ) for hex_hash in hex_hashes ] expected_hex_length = hash_type_to_hex_length[ hash_type ] bad_hex_hashes = [ hex_hash for hex_hash in hex_hashes if len( hex_hash ) != expected_hex_length ] if len( bad_hex_hashes ): m = 'Sorry, {} hashes should have {} hex characters! These did not:'.format( hash_type, expected_hex_length ) m += os.linesep * 2 m += os.linesep.join( ( '{} ({} characters)'.format( bad_hex_hash, len( bad_hex_hash ) ) for bad_hex_hash in bad_hex_hashes ) ) raise Exception( m ) hex_hashes = [ hex_hash for hex_hash in hex_hashes if len( hex_hash ) % 2 == 0 ] hex_hashes = DedupeList( hex_hashes ) hashes = tuple( [ bytes.fromhex( hex_hash ) for hex_hash in hex_hashes ] ) return hashes def Print( text ): try: print( str( text ) ) except: print( repr( text ) ) ShowText = Print def PrintException( e, do_wait = True ): ( etype, value, tb ) = sys.exc_info() PrintExceptionTuple( etype, value, tb, do_wait = do_wait ) def PrintExceptionTuple( etype, value, tb, do_wait = True ): if etype is None: etype = HydrusExceptions.UnknownException if etype == HydrusExceptions.ShutdownException: return if value is None: value = 'Unknown error' if tb is None: trace = 'No error trace--here is the stack:' + os.linesep + ''.join( traceback.format_stack() ) else: trace = ''.join( traceback.format_exception( etype, value, tb ) ) stack_list = traceback.format_stack() stack = ''.join( stack_list ) message = str( etype.__name__ ) + ': ' + str( value ) + os.linesep + trace + os.linesep + stack Print( '' ) Print( 'Exception:' ) DebugPrint( message ) if do_wait: time.sleep( 1 ) ShowException = PrintException ShowExceptionTuple = PrintExceptionTuple def Profile( summary, code, g, l, min_duration_ms = 20, show_summary = False ): profile = cProfile.Profile() started = GetNowPrecise() profile.runctx( code, g, l ) finished = GetNowPrecise() time_took = finished - started time_took_ms = int( time_took * 1000.0 ) if time_took_ms > min_duration_ms: output = io.StringIO() stats = pstats.Stats( profile, stream = output ) stats.strip_dirs() stats.sort_stats( 'tottime' ) output.write( 'Stats' ) output.write( os.linesep * 2 ) stats.print_stats() output.write( 'Callers' ) output.write( os.linesep * 2 ) stats.print_callers() output.seek( 0 ) details = output.read() with HG.profile_counter_lock: HG.profile_slow_count += 1 if show_summary: ShowText( summary ) HG.controller.PrintProfile( summary, details ) else: with HG.profile_counter_lock: HG.profile_fast_count += 1 if show_summary: HG.controller.PrintProfile( summary, details ) def PullNFromIterator( iterator, n ): chunk = [] for item in iterator: chunk.append( item ) if len( chunk ) == n: return chunk return chunk def RandomPop( population ): random_index = random.randint( 0, len( population ) - 1 ) row = population.pop( random_index ) return row def RecordRunningStart( db_path, instance ): path = os.path.join( db_path, instance + '_running' ) record_string = '' try: me = psutil.Process() record_string += str( me.pid ) record_string += os.linesep record_string += str( me.create_time() ) except psutil.Error: return with open( path, 'w', encoding = 'utf-8' ) as f: f.write( record_string ) def RestartProcess(): time.sleep( 1 ) # time for ports to unmap exe = sys.executable me = sys.argv[0] if HC.RUNNING_FROM_SOURCE: # exe is python's exe, me is the script args = [ sys.executable ] + sys.argv else: # we are running a frozen release--both exe and me are the built exe # wrap it in quotes because pyinstaller passes it on as raw text, breaking any path with spaces :/ if not me.startswith( '"' ): me = '"{}"'.format( me ) args = [ me ] + sys.argv[1:] os.execv( exe, args ) def SampleSetByGettingFirst( s: set, n ): # sampling from a big set can be slow, so if we don't care about super random, let's just rip off the front and let __hash__ be our random n = min( len( s ), n ) sample = set() if n == 0: return sample for ( i, obj ) in enumerate( s ): sample.add( obj ) if i >= n - 1: break return sample def SetsIntersect( a, b ): # not a.isdisjoint( b ) if not isinstance( a, set ): a = set( a ) if not isinstance( b, set ): b = set( b ) if len( a ) > len( b ): ( a, b ) = ( b, a ) return True in ( i in b for i in a ) def SmoothOutMappingIterator( xs, n ): # de-spikifies mappings, so if there is ( tag, 20k files ), it breaks that up into manageable chunks chunk_weight = 0 chunk = [] for ( tag_item, hash_items ) in xs: for chunk_of_hash_items in SplitIteratorIntoChunks( hash_items, n ): yield ( tag_item, chunk_of_hash_items ) def SplayListForDB( xs ): return '(' + ','.join( ( str( x ) for x in xs ) ) + ')' def SplitIteratorIntoChunks( iterator, n ): chunk = [] for item in iterator: chunk.append( item ) if len( chunk ) == n: yield chunk chunk = [] if len( chunk ) > 0: yield chunk def SplitIteratorIntoAutothrottledChunks( iterator, starting_n, precise_time_to_stop ): n = starting_n chunk = PullNFromIterator( iterator, n ) while len( chunk ) > 0: time_work_started = GetNowPrecise() yield chunk work_time = GetNowPrecise() - time_work_started items_per_second = n / work_time time_remaining = precise_time_to_stop - GetNowPrecise() if TimeHasPassedPrecise( precise_time_to_stop ): n = 1 else: expected_items_in_remaining_time = max( 1, int( time_remaining * items_per_second ) ) quad_speed = n * 4 n = min( quad_speed, expected_items_in_remaining_time ) chunk = PullNFromIterator( iterator, n ) def SplitListIntoChunks( xs, n ): if isinstance( xs, set ): xs = list( xs ) for i in range( 0, len( xs ), n ): yield xs[ i : i + n ] def SplitMappingIteratorIntoAutothrottledChunks( iterator, starting_n, precise_time_to_stop ): n = starting_n chunk_weight = 0 chunk = [] for ( tag_item, hash_items ) in iterator: chunk.append( ( tag_item, hash_items ) ) chunk_weight += len( hash_items ) if chunk_weight >= n: time_work_started = GetNowPrecise() yield chunk work_time = GetNowPrecise() - time_work_started chunk_weight = 0 chunk = [] items_per_second = n / work_time time_remaining = precise_time_to_stop - GetNowPrecise() if TimeHasPassedPrecise( precise_time_to_stop ): n = 1 else: expected_items_in_remaining_time = max( 1, int( time_remaining * items_per_second ) ) quad_speed = n * 4 n = min( quad_speed, expected_items_in_remaining_time ) if len( chunk ) > 0: yield chunk def TimeHasPassed( timestamp ): if timestamp is None: return False return GetNow() > timestamp def TimeHasPassedFloat( timestamp ): return GetNowFloat() > timestamp def TimeHasPassedPrecise( precise_timestamp ): return GetNowPrecise() > precise_timestamp def TimeUntil( timestamp ): return timestamp - GetNow() def ToHumanBytes( size ): if size is None: return 'unknown size' if size < 1024: return ToHumanInt( size ) + 'B' suffixes = ( '', 'K', 'M', 'G', 'T', 'P' ) suffix_index = 0 while size >= 1024: size = size / 1024 suffix_index += 1 suffix = suffixes[ suffix_index ] if size < 10.0: # 3.1MB return '{:.1f}{}B'.format( size, suffix ) else: # 23MB return '{:.0f}{}B'.format( size, suffix ) def ToHumanInt( num ): num = int( num ) # this got stomped on by mpv, which resets locale #text = locale.format_string( '%d', num, grouping = True ) text = '{:,}'.format( num ) return text def WaitForProcessToFinish( p, timeout ): started = GetNow() while p.poll() is None: if TimeHasPassed( started + timeout ): p.kill() raise Exception( 'Process did not finish within ' + ToHumanInt( timeout ) + ' seconds!' ) time.sleep( 2 ) class HydrusYAMLBase( yaml.YAMLObject ): yaml_loader = yaml.SafeLoader yaml_dumper = yaml.SafeDumper class BigJobPauser( object ): def __init__( self, period = 10, wait_time = 0.1 ): self._period = period self._wait_time = wait_time self._next_pause = GetNow() + self._period def Pause( self ): if TimeHasPassed( self._next_pause ): time.sleep( self._wait_time ) self._next_pause = GetNow() + self._period class Call( object ): def __init__( self, func, *args, **kwargs ): self._label = None self._func = func self._args = args self._kwargs = kwargs def __call__( self ): self._func( *self._args, **self._kwargs ) def __repr__( self ): label = self._GetLabel() return 'Call: {}'.format( label ) def _GetLabel( self ) -> str: if self._label is None: # this can actually cause an error with Qt objects that are dead or from the wrong thread, wew! label = '{}( {}, {} )'.format( self._func, self._args, self._kwargs ) else: label = self._label return label def GetLabel( self ) -> str: return self._GetLabel() def SetLabel( self, label: str ): self._label = label class ContentUpdate( object ): def __init__( self, data_type, action, row, reason = None ): self._data_type = data_type self._action = action self._row = row self._reason = reason def __eq__( self, other ): if isinstance( other, ContentUpdate ): return self.__hash__() == other.__hash__() return NotImplemented def __hash__( self ): return hash( ( self._data_type, self._action, repr( self._row ) ) ) def __repr__( self ): return 'Content Update: ' + str( ( self._data_type, self._action, self._row, self._reason ) ) def GetAction( self ): return self._action def GetDataType( self ): return self._data_type def GetHashes( self ): hashes = set() if self._data_type == HC.CONTENT_TYPE_FILES: if self._action == HC.CONTENT_UPDATE_ADVANCED: ( sub_action, possible_hashes ) = self._row if possible_hashes is None: hashes = set() else: hashes = possible_hashes elif self._action == HC.CONTENT_UPDATE_ADD: ( file_info_manager, timestamp ) = self._row hashes = { file_info_manager.hash } else: hashes = self._row elif self._data_type == HC.CONTENT_TYPE_DIRECTORIES: hashes = set() elif self._data_type == HC.CONTENT_TYPE_URLS: ( urls, hashes ) = self._row elif self._data_type == HC.CONTENT_TYPE_MAPPINGS: if self._action == HC.CONTENT_UPDATE_ADVANCED: hashes = set() else: ( tag, hashes ) = self._row elif self._data_type in ( HC.CONTENT_TYPE_TAG_PARENTS, HC.CONTENT_TYPE_TAG_SIBLINGS ): hashes = set() elif self._data_type == HC.CONTENT_TYPE_RATINGS: if self._action == HC.CONTENT_UPDATE_ADD: ( rating, hashes ) = self._row elif self._data_type == HC.CONTENT_TYPE_NOTES: if self._action == HC.CONTENT_UPDATE_SET: ( hash, name, note ) = self._row hashes = { hash } elif self._action == HC.CONTENT_UPDATE_DELETE: ( hash, name ) = self._row hashes = { hash } elif self._data_type == HC.CONTENT_TYPE_FILE_VIEWING_STATS: if self._action == HC.CONTENT_UPDATE_ADD: ( hash, preview_views_delta, preview_viewtime_delta, media_views_delta, media_viewtime_delta ) = self._row hashes = { hash } elif self._action == HC.CONTENT_UPDATE_DELETE: hashes = self._row if not isinstance( hashes, set ): hashes = set( hashes ) return hashes def GetReason( self ): if self._reason is None: return 'No reason given.' else: return self._reason def GetRow( self ): return self._row def GetWeight( self ): return len( self.GetHashes() ) def IsInboxRelated( self ): return self._action in ( HC.CONTENT_UPDATE_ARCHIVE, HC.CONTENT_UPDATE_INBOX ) def SetRow( self, row ): self._row = row def ToTuple( self ): return ( self._data_type, self._action, self._row ) class JobDatabase( object ): def __init__( self, job_type, synchronous, action, *args, **kwargs ): self._type = job_type self._synchronous = synchronous self._action = action self._args = args self._kwargs = kwargs self._result_ready = threading.Event() def __str__( self ): return 'DB Job: {}'.format( self.ToString() ) def _DoDelayedResultRelief( self ): pass def GetCallableTuple( self ): return ( self._action, self._args, self._kwargs ) def GetResult( self ): time.sleep( 0.00001 ) # this one neat trick can save hassle on superquick jobs as event.wait can be laggy while True: if self._result_ready.wait( 2 ) == True: break elif HG.model_shutdown: raise HydrusExceptions.ShutdownException( 'Application quit before db could serve result!' ) self._DoDelayedResultRelief() if isinstance( self._result, Exception ): e = self._result raise e else: return self._result def GetType( self ): return self._type def IsSynchronous( self ): return self._synchronous def PutResult( self, result ): self._result = result self._result_ready.set() def ToString( self ): return '{} {}'.format( self._type, self._action ) class ServiceUpdate( object ): def __init__( self, action, row = None ): self._action = action self._row = row def ToTuple( self ): return ( self._action, self._row )