diff --git a/.gitignore b/.gitignore index a945659..e98820f 100644 --- a/.gitignore +++ b/.gitignore @@ -1,6 +1,7 @@ # mystuff /.vscode - +*.mid +*.midi # Byte-compiled / optimized / DLL files __pycache__/ diff --git a/example_convert_linear_mcpack.py b/example_convert_linear_mcpack.py new file mode 100644 index 0000000..7f3202f --- /dev/null +++ b/example_convert_linear_mcpack.py @@ -0,0 +1,22 @@ +# THIS PROGRAM IS ONLY A TEST EXAMPLE + + +from msctPkgver.main import * + +convertion = midiConvert() +convertion.convert(input('请输入midi文件路径：'), input('请输入输出路径：')) +convertion.tomcpack( + 1, + bool(int(input('是否自动重置计分板(1|0)：'))), + bool(int(input('是否开启进度条(1|0)：'))), + input('请输入计分板名称：'), + float(input('请输入音量（0-1）：')), + float(input('请输入速度倍率：')), +) + +# for the test +# if __name__ == '__main__': +# convertion = midiConvert() +# convertion.convert(r"C:\Users\lc\Documents\MuseScore3\乐谱\乐谱\victory.mid", ".") +# convertion.tomcpack( +# 1, True, True, "scb", 1, 1) diff --git a/msctPkgver/exceptions.py b/msctPkgver/exceptions.py index 1cd9028..23ec562 100644 --- a/msctPkgver/exceptions.py +++ b/msctPkgver/exceptions.py @@ -62,8 +62,10 @@ class MidiDestroyedError(MSCTBaseException): pass class ChannelOverFlowError(MSCTBaseException): - '''一个midi中含有过多的通道（应≤16）''' + '''一个midi中含有过多的通道（数量应≤16）''' pass - +class NotDefineProgramError(MSCTBaseException): + '''没有Program设定导致没有乐器可以选择的错误''' + pass diff --git a/msctPkgver/main.py b/msctPkgver/main.py index 63047f9..e8a4a30 100644 --- a/msctPkgver/main.py +++ b/msctPkgver/main.py @@ -63,10 +63,87 @@ def makeZip(sourceDir, outFilename, compression=8, exceptFile=None): zipf.close() +class SingleNote: + def __init__(self, inst: int, pitch: int, velocity, startTime, lastTime): + '''用于存储单个音符的类 + :param inst 乐器编号 + :param pitch 音符编号 + :param velocity 力度/响度 + :param startTime 开始之时(ms) + 注：此处的时间是用从乐曲开始到当前的毫秒数 + :param lastTime 音符延续时间(ms)''' + self.inst = inst + self.note = pitch + self.velocity = velocity + self.startTime = startTime + self.lastTime = lastTime + @property + def inst(self): + return self.inst + + @property + def note(self): + return self.note + + @property + def velocity(self): + return self.velocity + + @property + def startTime(self): + return self.startTime + + @property + def lastTime(self): + return self.lastTime + + def __str__(self): + return f"Note(inst = {self.inst}, pitch = {self.note}, velocity = {self.velocity}, startTime = {self.startTime}, lastTime = {self.lastTime}, )" + + def __tuple__(self): + return (self.inst, self.note, self.velocity, self.startTime, self.lastTime) + + def __dict__(self): + return { + "inst": self.inst, + "pitch": self.note, + "velocity": self.velocity, + "startTime": self.startTime, + "lastTime": self.lastTime, + } + + +''' +学习笔记： +tempo: microseconds per quarter note 毫秒每四分音符，换句话说就是一拍占多少毫秒 +tick: midi帧 +ticks_per_beat: 帧每拍，即一拍多少帧 + +那么： + +tick / ticks_per_beat => amount_of_beats 拍数(四分音符数) + +tempo * amount_of_beats => 毫秒数 + +所以： + +tempo * tick / ticks_per_beat => 毫秒数 + +''' + + class midiConvert: def __init__(self, debug: bool = False): """简单的midi转换类，将midi文件转换为我的世界结构或者包""" self.debugMode = debug + self.methods = [ + self._toCmdList_m1, + self._toCmdList_m2, + self._toCmdList_m3, + ] + self.methods_byDelay = [ + self._toCmdList_withDelay_m1, + ] def convert(self, midiFile: str, outputPath: str): """转换前需要先运行此函数来获取基本信息""" @@ -446,7 +523,7 @@ class midiConvert: / ((self.midi.ticks_per_beat * float(speed)) * 50000) ) except NameError: - raise NotDefineTempoError('计算当前分数时出错 未定义参量') + raise NotDefineTempoError('计算当前分数时出错 未定义参量 Tempo') maxscore = max(maxscore, nowscore) soundID, _X = self.__Inst2soundIDwithX(instrumentID) singleTrack.append( @@ -463,68 +540,302 @@ class midiConvert: return [tracks, commands, maxscore] - # 值得注意的是，我这里没有修改多少 + # 这与上面的算法几乎没有差别 甚至更慢了一点 但是是为了线性插值做准备 def _toCmdList_m2( - self, scoreboardname: str = "mscplay", MaxVolume: float = 1.0, speed: float = 1.0 + self, + scoreboardname: str = "mscplay", + MaxVolume: float = 1.0, + speed: float = 1.0, ) -> list: """ - 使用金羿的转换思路，将midi转换为我的世界命令列表，使用线性方法调整音量 + 使用金羿的转换思路，将midi转换为我的世界命令列表 :param scoreboardname: 我的世界的计分板名称 - :param volume: 音量，注意：这里的音量范围为(0,1]，如果超出将被处理为正确值，其原理为在距离玩家 (1 / volume -1) 的地方播放音频 + :param MaxVolume: 音量，注意：这里的音量范围为(0,1]，如果超出将被处理为正确值，其原理为在距离玩家 (1 / volume -1) 的地方播放音频 :param speed: 速度，注意：这里的速度指的是播放倍率，其原理为在播放音频的时候，每个音符的播放时间除以 speed :return: tuple(命令列表, 命令个数, 计分板最大值) """ - tracks = [] + if MaxVolume > 1: MaxVolume = 1 if MaxVolume <= 0: MaxVolume = 0.001 - commands = 0 - maxscore = 0 + # 一个midi中仅有16通道 我们通过通道来识别而不是音轨 + channels = [[], [], [], [], [], [], [], [], [], [], [], [], [], [], [], []] + + + # 我们来用通道统计音乐信息 + for i, track in enumerate(self.midi.tracks): + + microseconds = 0 + + for msg in track: + + if msg.time != 0: + try: + microseconds += msg.time * tempo / self.midi.ticks_per_beat + except NameError: + raise NotDefineTempoError('计算当前分数时出错 未定义参量 Tempo') + + + if msg.is_meta: + if msg.type == "set_tempo": + tempo = msg.tempo + else: + + try: + msg.channel + channelMsg = True + except: + channelMsg = False + if channelMsg: + if msg.channel > 15: + raise ChannelOverFlowError(f"当前消息 {msg} 的通道超限(≤15)") + + if msg.type == "program_change": + channels[msg.channel].append( + ("PgmC", msg.program, microseconds) + ) + + elif msg.type == 'note_on' and msg.velocity != 0: + channels[msg.channel].append( + ("NoteS", msg.note, msg.velocity, microseconds) + ) + + elif (msg.type == 'note_on' and msg.velocity == 0) or ( + msg.type == 'note_off' + ): + channels[msg.channel].append( + ("NoteE", msg.note, microseconds) + ) + + + '''整合后的音乐通道格式 + 每个通道包括若干消息元素其中逃不过这三种： + + 1 切换乐器消息 + + ("PgmC", 切换后的乐器ID: int, 距离演奏开始的毫秒) + + 2 音符开始消息 + + ("NoteS", 开始的音符ID, 力度（响度）, 距离演奏开始的毫秒) + + 3 音符结束消息 + + ("NoteS", 结束的音符ID, 距离演奏开始的毫秒)''' + + tracks = [] + cmdAmount = 0 + maxScore = 0 + CheckFirstChannel = False + + # 此处 我们把通道视为音轨 + for track in channels: + # 如果当前通道为空 则跳过 + if not track: + continue + + + if channels.index(track) == 0: + CheckFirstChannel = True + else: + CheckFirstChannel = False + + nowTrack = [] + + for msg in track: + + + if msg[0] == "PgmC": + InstID = msg[1] + + elif msg[0] == "NoteS": + + soundID, _X = self.__Inst2soundIDwithX(InstID) + score_now = round(msg[-1]/ float(speed) / 50000) + maxScore = max(maxScore,score_now) + + nowTrack.append( + "execute @a[scores={" + + str(scoreboardname) + + "=" + + str(score_now) + + "}" + + f"] ~ ~ ~ playsound {soundID} @s ~ ~{1 / MaxVolume - 1} ~ {msg[2] * (0.7 if CheckFirstChannel else 0.9)} {2 ** ((msg[1] - 60 - _X) / 12)}" + ) + + cmdAmount += 1 + + tracks.append(nowTrack) + + return [tracks, cmdAmount, maxScore] + + # 这才是加了线性插值的算法 + def _toCmdList_m3( + self, + scoreboardname: str = "mscplay", + MaxVolume: float = 1.0, + speed: float = 1.0, + ) -> list: + """ + 使用金羿的转换思路，将midi转换为我的世界命令列表，并使用线性插值算法优化音量，此处的插值是固定斜率的方法 + :param scoreboardname: 我的世界的计分板名称 + :param MaxVolume: 音量，注意：这里的音量范围为(0,1]，如果超出将被处理为正确值，其原理为在距离玩家 (1 / volume -1) 的地方播放音频 + :param speed: 速度，注意：这里的速度指的是播放倍率，其原理为在播放音频的时候，每个音符的播放时间除以 speed + :return: tuple(命令列表, 命令个数, 计分板最大值) + """ + + if MaxVolume > 1: + MaxVolume = 1 + if MaxVolume <= 0: + MaxVolume = 0.001 # 一个midi中仅有16通道 我们通过通道来识别而不是音轨 channels = [[], [], [], [], [], [], [], [], [], [], [], [], [], [], [], []] + # 我们来用通道统计音乐信息 for i, track in enumerate(self.midi.tracks): - ticks = 0 - instrumentID = 0 - singleTrack = [] - noteOn = [] + microseconds = 0 for msg in track: - ticks += msg.time - # print(msg) + + if msg.time != 0: + try: + microseconds += msg.time * tempo / self.midi.ticks_per_beat + except NameError: + raise NotDefineTempoError('计算当前分数时出错 未定义参量 Tempo') + + if msg.is_meta: if msg.type == "set_tempo": tempo = msg.tempo else: - if msg.type == "program_change": - # print("TT") - instrumentID = msg.program - if msg.type == 'note_on' and msg.velocity != 0: - - nowscore = round( - (ticks * tempo) - / ((self.midi.ticks_per_beat * float(speed)) * 50000) - ) - maxscore = max(maxscore, nowscore) - soundID, _X = self.__Inst2soundIDwithX(instrumentID) - singleTrack.append( - "execute @a[scores={" - + str(scoreboardname) - + "=" - + str(nowscore) - + "}" - + f"] ~ ~ ~ playsound {soundID} @s ~ ~{1 / MaxVolume - 1} ~ {msg.velocity * (0.7 if msg.channel == 0 else 0.9)} {2 ** ((msg.note - 60 - _X) / 12)}" - ) - commands += 1 - if len(singleTrack) != 0: - tracks.append(singleTrack) - return [tracks, commands, maxscore] + try: + msg.channel + channelMsg = True + except: + channelMsg = False + if channelMsg: + if msg.channel > 15: + raise ChannelOverFlowError(f"当前消息 {msg} 的通道超限(≤15)") + + if msg.type == "program_change": + channels[msg.channel].append( + ("PgmC", msg.program, microseconds) + ) + + elif msg.type == 'note_on' and msg.velocity != 0: + channels[msg.channel].append( + ("NoteS", msg.note, msg.velocity, microseconds) + ) + + elif (msg.type == 'note_on' and msg.velocity == 0) or ( + msg.type == 'note_off' + ): + channels[msg.channel].append( + ("NoteE", msg.note, microseconds) + ) + + + '''整合后的音乐通道格式 + 每个通道包括若干消息元素其中逃不过这三种： + + 1 切换乐器消息 + + ("PgmC", 切换后的乐器ID: int, 距离演奏开始的毫秒) + + 2 音符开始消息 + + ("NoteS", 开始的音符ID, 力度（响度）, 距离演奏开始的毫秒) + + 3 音符结束消息 + + ("NoteS", 结束的音符ID, 距离演奏开始的毫秒)''' + + + note_channels = [[], [], [], [], [], [], [], [], [], [], [], [], [], [], [], []] + + # 此处 我们把通道视为音轨 + for i in range(len(channels)): + # 如果当前通道为空 则跳过 + + noteMsgs = [] + + for msg in channels[i]: + + if msg[0] == "PgmC": + InstID = msg[1] + + elif msg[0] == "NoteS": + noteMsgs.append(msg[1:]) + + elif msg[0] == "NoteE": + record_pop_ids = [] + for j in range(len(noteMsgs)): + if noteMsgs[j][0] == msg[1]: + note_channels[i].append(SingleNote(InstID, msg[1],noteMsgs[j][1],noteMsgs[j][2],msg[-1]-noteMsgs[j][2])) + record_pop_ids.append(j) + for j in record_pop_ids: + noteMsgs.pop(j) + + tracks = [] + + + cmdAmount = 0 + maxScore = 0 + CheckFirstChannel = False + + # 此处 我们把通道视为音轨 + for track in note_channels: + # 如果当前通道为空 则跳过 + if not track: + continue + + + if channels.index(track) == 0: + CheckFirstChannel = True + else: + CheckFirstChannel = False + + nowTrack = [] + + for note in track: + + for i in range(note.lastTime / 500): + # 实在没时间写了 + # 用半秒做一个时间单位 + # 然后计算即可 + # 我找时间写完 + + + soundID, _X = self.__Inst2soundIDwithX(InstID) + score_now = round(msg[-1]/ float(speed) / 50000) + maxScore = max(maxScore,score_now) + + nowTrack.append( + "execute @a[scores={" + + str(scoreboardname) + + "=" + + str(score_now) + + "}" + + f"] ~ ~ ~ playsound {soundID} @s ~ ~{1 / MaxVolume - 1} ~ {msg[2] * (0.7 if CheckFirstChannel else 0.9)} {2 ** ((msg[1] - 60 - _X) / 12)}" + ) + + cmdAmount += 1 + + tracks.append(nowTrack) + + + return [tracks, cmdAmount, maxScore] + + + return [tracks, cmdAmount, maxScore] + + def _toCmdList_withDelay_m1( self, @@ -722,9 +1033,9 @@ class midiConvert: :param speed: 速度，注意：这里的速度指的是播放倍率，其原理为在播放音频的时候，每个音符的播放时间除以 speed :return 成功与否，成功返回(True,True)，失败返回(False,str失败原因) """ - if method == 1: - cmdlist, _a, maxscore = self._toCmdList_m1(scoreboardname, volume, speed) - else: + try: + cmdlist, _a, maxscore = self.methods[method-1](scoreboardname, volume, speed) + except: return (False, f"无法找到算法ID{method}对应的转换算法") del _a @@ -858,13 +1169,12 @@ class midiConvert: :param isAutoReset: 是否自动重置计分板 :return 成功与否，成功返回(True,未经过压缩的源,结构占用大小)，失败返回(False,str失败原因) """ + try: + cmdlist, totalcount, maxScore = self.methods[method-1](scoreboardname, volume, speed) + except: + return (False, f"无法找到算法ID{method}对应的转换算法") + - if method == 1: - cmdlist, totalcount, maxScore = self._toCmdList_m1( - scoreboardname, volume, speed - ) - else: - return (False, f"无法找到算法ID {method} 对应的转换算法") if not os.path.exists(self.outputPath): os.makedirs(self.outputPath) @@ -995,10 +1305,10 @@ class midiConvert: :return 成功与否，成功返回(True,未经过压缩的源,结构占用大小)，失败返回(False,str失败原因) """ - if method == 1: - cmdlist = self._toCmdList_withDelay_m1(volume, speed, player, progressbar) - else: - return (False, f"无法找到算法ID {method} 对应的转换算法") + try: + cmdlist = self.methods_byDelay[method-1](volume, speed, player, progressbar) + except: + return (False, f"无法找到算法ID{method}对应的转换算法") if not os.path.exists(self.outputPath): os.makedirs(self.outputPath)