├── README.md ├── config.txt ├── launcher.py ├── screenshot.png ├── ui ├── ui.css ├── ui.html └── ui.js └── video2sub.py /README.md: -------------------------------------------------------------------------------- 1 | # video2sub 2 | 3 | ## 硬字幕OCR提取工具 4 | 5 | video2sub 提供图形界面，方便执行字幕 OCR 识别、编辑、校对等操作。 6 | 7 | video2sub 本身没有实现任何 OCR 识别算法，而是使用外部库进行 OCR 识别。目前支持 [chineseocr](https://github.com/chineseocr/chineseocr) 和[百度文字识别 API](https://ai.baidu.com/tech/ocr/general)。其中 chineseocr 支持离线运行，而百度文字识别 API 需要联网运行。建议先用 chineseocr 识别出哪些帧含有字幕，再调用百度文字识别 API 进行精确识别。 8 | 9 | ![屏幕截图](screenshot.png) 10 | 11 | ## 下载地址 12 | 13 | - [**百度网盘**](https://pan.baidu.com/s/1sLc22mr1PUh0X2HYXnAQzg) （提取码: **ydkn**） 14 | - [**GitHub**](https://github.com/zhangboyang/video2sub/releases) 15 | 16 | |文件名|说明| 17 | |-|-| 18 | |video2sub-20210331.7z|下载后解压并运行 **launcher.exe** 即可。| 19 | |chineseocr-gpu.7z|GPU 版 chineseocr 识别引擎，需要 NVIDIA 显卡。下载后解压并运行 **chineseocr-gpu.exe** 即可。 20 | |chineseocr-cpu.7z|CPU 版 chineseocr 识别引擎，无需显卡，但速度很慢，不建议使用。 21 | 22 | ## 系统需求 23 | 24 | - Windows 7 （64位）或更高版本 25 | - 8GB 以上内存（若不使用 GPU 则 4GB 即可） 26 | - NVIDIA 显卡：用于 chineseocr 加速（纯 CPU 也能运行，就是太慢） 27 | -------------------------------------------------------------------------------- /config.txt: -------------------------------------------------------------------------------- 1 | # 请按Python语法编辑此文件（代码使用ast.literal_eval读入此文件） 2 | 3 | { 4 | ## 通用设置 ###################################################### 5 | 6 | 'host': '127.0.0.1', 7 | 'port': [20000, 21000], # 前端与后端的通信端口范围（左闭右开区间） 8 | 9 | 'max_log': 1500, # (显示出来的)最大日志条目项数 10 | 'max_checkpoint': 1500, # (显示出来的)最大恢复点项数 11 | 12 | 'fix_ntsc_fps': True, # 修正NTSC帧率（如29.970改为30000/1001） 13 | 14 | 'export_suffix': '.chs', # 导出文件的后缀 15 | 'export_overwrite': False, # 导出时是否覆盖已存在文件 16 | 17 | ## UI 界面设置 ###################################################### 18 | 19 | 'thumbnail': { # 缩略图设置 20 | 'height': 150, # 高度 21 | 'npart': 10, # 同一文件中合并存储几张缩略图 22 | 'jpg_quality': 85 # JPG质量参数 23 | }, 24 | 25 | 'subthumb': { # 字幕预览设置 26 | 'width': 470, # 宽度（只对新建的字幕有效） 27 | 'jpg_quality': 90 # JPG质量参数 28 | }, 29 | 30 | 'default': { # 默认设置 31 | 'OCR': { 32 | 'engine': 'chineseocr:multi', # 默认OCR引擎名 33 | 'top': -1, # 默认字幕区域Y1 34 | 'bottom': -1, # 默认字幕区域Y2 35 | }, 36 | 'UI': { 37 | 'editorfontsize': 30, # 默认编辑框字体大小 38 | }, 39 | }, 40 | 41 | ## OCR 引擎设置 ###################################################### 42 | 43 | 'allengines': [ # 引擎名字 44 | ['chineseocr:multi', 'chineseocr(多行模式) -- 推荐“新OCR”使用'], 45 | ['chineseocr:single', 'chineseocr(单行模式)'], 46 | ['baiduocr:accurate', '百度OCR(高精度,批量) -- 推荐“重新OCR”使用'], 47 | ['baiduocr:accurate_basic', '百度OCR(高精度,单独) -- 推荐“空项OCR”使用'], 48 | ['baiduocr:general', '百度OCR(标准版,批量)'], 49 | ['baiduocr:general_basic', '百度OCR(标准版,单独)'], 50 | ['dummyocr', 'dummyocr(调试用)'], 51 | ], 52 | 53 | #### chineseocr设置 54 | 'chineseocr': { 55 | 'url': 'http://127.0.0.1:8080/ocr', # API地址 56 | }, 57 | 58 | #### 百度OCR设置 59 | 'baiduocr': { 60 | # API介绍 https://ai.baidu.com/tech/ocr/general 61 | # 账号申请方法 https://cloud.baidu.com/doc/OCR/s/dk3iqnq51 62 | # 请把下面的API_KEY、SECRET_KEY改为自己申请的账号 63 | 'API_KEY': 'GmhC18eVP1Fo1ECX911dtOzw', 64 | 'SECRET_KEY': 'PQ2ukO4Aec2PTsgQU9UkiEKYciavlZk8', 65 | 66 | 'qps_limit': 2, # 每秒调用次数限制 67 | 'batch_size': 20, # 批处理大小 68 | 'language_type': 'CHN_ENG', # 语言类型，参见 https://ai.baidu.com/ai-doc/OCR/1k3h7y3db 69 | 70 | 'TOKEN_URL': 'https://aip.baidubce.com/oauth/2.0/token', 71 | 'OCR_URL': 'https://aip.baidubce.com/rest/2.0/ocr/v1/', 72 | }, 73 | 74 | #### dummyocr(调试用) 75 | 'dummyocr': { 76 | 'always_error': 0, 77 | 'text': '测试', 78 | }, 79 | 80 | ## ASS 导出设置 ###################################################### 81 | 82 | 'ass_format': { 83 | 'header': ################ ASS文件头模板 84 | '''[Script Info] 85 | Title: {{文件名}} 86 | ScriptType: v4.00+ 87 | WrapStyle: 0 88 | ScaledBorderAndShadow: yes 89 | YCbCr Matrix: None 90 | PlayResX: {{视频宽度}} 91 | PlayResY: {{视频高度}} 92 | 93 | [V4+ Styles] 94 | Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding 95 | Style: Default,方正兰亭圆_GBK_准,{{字幕高度}},&H00FFFFFF,&H000000FF,&H00000000,&H00000000,0,0,0,0,100,100,0,0,1,2,0,2,10,10,{{下半屏底边距}},1 96 | Style: TopHalf,方正兰亭圆_GBK_准,{{字幕高度}},&H00FFFFFF,&H000000FF,&H00000000,&H00000000,0,0,0,0,100,100,0,0,1,2,0,8,10,10,{{上半屏顶边距}},1 97 | 98 | [Events] 99 | Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text 100 | ''', 101 | 'top_half': ################ ASS字幕行模板（上半屏） 102 | '''Dialogue: 0,{{开始时间}},{{结束时间}},TopHalf,,0,0,0,,{{字幕文本}} 103 | ''', 104 | 'bottom_half': ################ ASS字幕行模板（下半屏） 105 | '''Dialogue: 0,{{开始时间}},{{结束时间}},Default,,0,0,0,,{{字幕文本}} 106 | ''', 107 | 'time_alignment': { ################ ASS字幕时间对齐方式 108 | 'start': ('floor', +0.5, +1e-3, 0), 'end': ('floor', +0.5, +1e-3, 0), # 适用于VLC播放器 109 | #'start': ('floor', 0, +1e-3, 0), 'end': ('floor', 0, +1e-3, 0), # 适用于MPC-HC播放器 110 | }, 111 | }, 112 | 113 | } 114 | -------------------------------------------------------------------------------- /launcher.py: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env python3 2 | import ast 3 | import sys 4 | import os 5 | import subprocess 6 | import json 7 | import urllib.request 8 | import time 9 | import traceback 10 | import secrets 11 | import socket 12 | from threading import Thread 13 | from datetime import datetime 14 | if os.name == 'nt': 15 | import msvcrt 16 | import win32con 17 | import win32console 18 | import win32api 19 | import win32gui 20 | import win32file 21 | import winerror 22 | windows = True 23 | else: 24 | import signal 25 | import fcntl 26 | windows = False 27 | 28 | if windows: 29 | class LockedFileWriter: 30 | def __init__(self, path): 31 | while True: 32 | try: 33 | self.handle = win32file.CreateFile(path, win32file.GENERIC_READ | win32file.GENERIC_WRITE, win32file.FILE_SHARE_READ, None, win32file.TRUNCATE_EXISTING, 0, None) 34 | break 35 | except win32file.error as e: 36 | if e.winerror != winerror.ERROR_SHARING_VIOLATION: 37 | raise e 38 | time.sleep(0.1) 39 | def write(self, b): 40 | n = win32file.WriteFile(self.handle, b)[1] 41 | assert n == len(b) 42 | def close(self): 43 | self.handle.Close() 44 | def file_select(title, flt): 45 | try: 46 | ret = win32gui.GetOpenFileNameW( 47 | hwndOwner=win32console.GetConsoleWindow(), 48 | Title=title, 49 | MaxFile=1048576, 50 | Flags= 51 | win32con.OFN_ALLOWMULTISELECT | 52 | win32con.OFN_PATHMUSTEXIST | 53 | win32con.OFN_FILEMUSTEXIST | 54 | win32con.OFN_HIDEREADONLY | 55 | win32con.OFN_EXPLORER | 56 | win32con.OFN_DONTADDTORECENT | 57 | win32con.OFN_NOCHANGEDIR, 58 | Filter=flt) 59 | files = ret[0].split('\0') 60 | if len(files) > 1: 61 | files = [os.path.join(files[0], file) for file in files[1:]] 62 | return files 63 | except win32gui.error: 64 | return [] 65 | class ListSelectDialog: 66 | className = 'VIDEO2SUB_SELECTDLG' 67 | def __init__(self, title, msg, lst): 68 | self.title = title 69 | self.msg = msg 70 | self.lst = lst 71 | self.selitem = None 72 | wc = win32gui.WNDCLASS() 73 | wc.style = win32con.CS_VREDRAW | win32con.CS_HREDRAW 74 | wc.SetDialogProc() 75 | wc.cbWndExtra = win32con.DLGWINDOWEXTRA 76 | wc.hInstance = win32gui.dllhandle 77 | wc.hCursor = win32gui.LoadCursor(0, win32con.IDC_ARROW) 78 | wc.hbrBackground = win32con.COLOR_WINDOW + 1 79 | wc.lpszClassName = self.className 80 | try: 81 | win32gui.RegisterClass(wc) 82 | except win32gui.error as e: 83 | if e.winerror != winerror.ERROR_CLASS_ALREADY_EXISTS: 84 | raise e 85 | def DoModel(self): 86 | style = win32con.DS_SETFONT | win32con.DS_MODALFRAME | win32con.WS_POPUP | win32con.WS_SYSMENU | win32con.WS_VISIBLE | win32con.WS_CAPTION | win32con.CS_DBLCLKS 87 | s = win32con.WS_CHILD | win32con.WS_VISIBLE 88 | win32gui.DialogBoxIndirect(win32gui.dllhandle, [ 89 | [self.title, (0, 0, 180, 148), style, None, (12, "宋体"), None, self.className], 90 | [128, "确定", win32con.IDOK, (68, 127, 50, 14), s | win32con.WS_TABSTOP | win32con.BS_DEFPUSHBUTTON], 91 | [128, "取消", win32con.IDCANCEL, (123, 127, 50, 14), s | win32con.WS_TABSTOP | win32con.BS_PUSHBUTTON], 92 | [130, self.msg, -1, (7, 7, 166, 13), s | win32con.SS_LEFT], 93 | [131, None, 1000, (7, 22, 166, 98), s | win32con.WS_TABSTOP | win32con.LBS_NOINTEGRALHEIGHT | win32con.LBS_NOTIFY | win32con.WS_VSCROLL | win32con.WS_BORDER] 94 | ], win32console.GetConsoleWindow(), { 95 | win32con.WM_COMMAND: self.OnCommand, 96 | win32con.WM_INITDIALOG: self.OnInitDialog, 97 | }) 98 | def OnCommand(self, hwnd, msg, wparam, lparam): 99 | if wparam == win32con.IDCANCEL: 100 | self.selitem = None 101 | win32gui.EndDialog(hwnd, 0) 102 | elif wparam == win32con.IDOK or (win32api.LOWORD(wparam) == 1000 and win32api.HIWORD(wparam) == win32con.LBN_DBLCLK): 103 | listbox = win32gui.GetDlgItem(hwnd, 1000) 104 | sel = win32gui.SendMessage(listbox, win32con.LB_GETCURSEL, 0, 0) 105 | self.selitem = self.lst[sel] if 0 <= sel and sel < len(self.lst) else None 106 | if self.selitem is not None: 107 | win32gui.EndDialog(hwnd, 0) 108 | else: 109 | win32api.MessageBox(hwnd, '请从列表中选择一个项目', self.title, win32con.MB_ICONWARNING) 110 | def OnInitDialog(self, hwnd, msg, wparam, lparam): 111 | l, t, r, b = win32gui.GetWindowRect(hwnd) 112 | pl, pt, pr, pb = win32gui.GetWindowRect(win32gui.GetParent(hwnd) if win32gui.GetParent(hwnd) else win32gui.GetDesktopWindow()) 113 | xoff = ((pr - pl) - (r - l)) // 2 114 | yoff = ((pb - pt) - (b - t)) // 2 115 | win32gui.SetWindowPos(hwnd, win32con.HWND_TOP, pl + xoff, pt + yoff, 0, 0, win32con.SWP_NOSIZE) 116 | listbox = win32gui.GetDlgItem(hwnd, 1000) 117 | for item in self.lst: 118 | win32gui.SendMessage(listbox, win32con.LB_ADDSTRING, 0, item[1]) 119 | #win32gui.SendMessage(listbox, win32con.LB_SETCURSEL, 0, 0) 120 | def list_select(title, msg, lst): 121 | print('请在弹出的窗口中进行选择') 122 | dlg = ListSelectDialog(title, msg, lst) 123 | dlg.DoModel() 124 | return dlg.selitem 125 | else: 126 | class LockedFileWriter: 127 | def __init__(self, path): 128 | self.f = open(path, 'r+b') 129 | fcntl.flock(self.f.fileno(), fcntl.LOCK_EX) 130 | self.f.seek(0) 131 | self.f.truncate() 132 | def write(self, b): 133 | n = self.f.write(b) 134 | assert n == len(b) 135 | self.f.flush() 136 | def close(self): 137 | self.f.close() 138 | def list_select(title, msg, lst): 139 | print(msg + ':') 140 | for i, item in enumerate(lst): 141 | print(' %d. %s' % (i + 1, item[1])) 142 | while True: 143 | print('请输入编号: ', end='', flush=True) 144 | try: 145 | sel = int(input()) 146 | if 1 <= sel and sel <= len(lst): 147 | return lst[sel - 1] 148 | except ValueError: 149 | pass 150 | print('无效输入') 151 | 152 | sys.stdout.reconfigure(errors='replace') 153 | sys.stderr.reconfigure(errors='replace') 154 | 155 | if getattr(sys, 'frozen', False) and hasattr(sys, '_MEIPASS'): 156 | bundle_dir = os.path.abspath(sys._MEIPASS) 157 | else: 158 | bundle_dir = os.path.dirname(os.path.abspath(__file__)) 159 | 160 | EXTENSIONS_VIDEO = "*.3g2;*.3gp;*.3gp2;*.3gpp;*.amv;*.asf;*.avi;*.bik;*.bin;*.crf;*.dav;*.divx;*.drc;*.dv;*.dvr-ms;*.evo;*.f4v;*.flv;*.gvi;*.gxf;*.iso;*.m1v;*.m2v;" \ 161 | "*.m2t;*.m2ts;*.m4v;*.mkv;*.mov;*.mp2;*.mp2v;*.mp4;*.mp4v;*.mpe;*.mpeg;*.mpeg1;" \ 162 | "*.mpeg2;*.mpeg4;*.mpg;*.mpv2;*.mts;*.mtv;*.mxf;*.mxg;*.nsv;*.nuv;" \ 163 | "*.ogg;*.ogm;*.ogv;*.ogx;*.ps;" \ 164 | "*.rec;*.rm;*.rmvb;*.rpl;*.thp;*.tod;*.tp;*.ts;*.tts;*.txd;*.vob;*.vro;*.webm;*.wm;*.wmv;*.wtv;*.xesc" 165 | 166 | class BackendDiedError(Exception): 167 | pass 168 | 169 | waitkey = False 170 | fail = False 171 | 172 | gconfig = ast.literal_eval(open(os.path.join(bundle_dir, 'config.txt'), 'r', encoding='utf_8_sig').read()) 173 | url = None 174 | backend = None 175 | frontend = None 176 | session = None 177 | lastwatch = 0 178 | video = None 179 | backendlog = None 180 | backendlog_file = os.path.join(bundle_dir, 'backendlog_%s_%d.txt' % (datetime.now().strftime('%Y%m%d_%H%M%S'), os.getpid())) 181 | backendlog_preserve = False 182 | 183 | def alloc_hostport(): 184 | host = gconfig['host'] 185 | port_range = list(range(*gconfig['port'])) 186 | while len(port_range) > 0: 187 | port = secrets.choice(port_range) 188 | port_range = [p for p in port_range if p != port] 189 | with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s: 190 | try: 191 | s.bind((host, port)) 192 | return (host, port) 193 | except: 194 | pass 195 | raise Exception('无可用后端通信端口') 196 | 197 | def exe(path): 198 | if windows: 199 | root, ext = os.path.splitext(path) 200 | if ext == '.py': 201 | if getattr(sys, 'frozen', False) and hasattr(sys, '_MEIPASS'): 202 | ext = '.exe' 203 | else: 204 | return ['python', path] 205 | return [root + ext] 206 | else: 207 | if os.path.isabs(path): 208 | return [path] 209 | else: 210 | return [os.path.join('.', path)] 211 | 212 | def run_backend(file, silent=False, nosession=False): 213 | global backend 214 | global url 215 | global video 216 | global session 217 | global lastwatch 218 | global backendlog 219 | hostport = alloc_hostport() 220 | cmdline = exe('video2sub.py') + [hostport[0], str(hostport[1]), file] 221 | if silent: 222 | if backendlog is not None: 223 | backendlog.close() 224 | with open(backendlog_file, 'a', errors='replace') as f: 225 | f.write('>>>>> [%s] >>>>> %s\n' % (datetime.now().strftime('%Y-%m-%d %H:%M:%S'), json.dumps(cmdline, ensure_ascii=False))) 226 | backendlog = open(backendlog_file, 'ab') 227 | backend = subprocess.Popen(cmdline, stdout=backendlog if silent else None, stderr=subprocess.STDOUT if silent else None) 228 | video = os.path.basename(file) 229 | url = 'http://%s:%d' % hostport 230 | session = None 231 | lastwatch = 0 232 | if api('/getpid') != backend.pid: 233 | raise Exception("pid mismatch") 234 | if not nosession: 235 | session = api('/session') 236 | waitsec = 0.1 237 | while not api('/state')['loaded']: 238 | watch() 239 | time.sleep(waitsec) 240 | waitsec = min(waitsec * 2, 1) 241 | watch() 242 | 243 | def check_backend(): 244 | if backend.poll() is not None: 245 | raise BackendDiedError() 246 | 247 | def run_frontend(): 248 | global frontend 249 | apppath = 'APP-win32-x64' if windows else 'APP-linux-x64' 250 | appexe = 'APP' 251 | appconf = os.path.join(apppath, 'resources', 'app', 'nativefier.json') 252 | with open(appconf + '-template', 'r', encoding='utf-8') as f: 253 | conf = json.load(f) 254 | conf['targetUrl'] = url 255 | w = LockedFileWriter(appconf) 256 | w.write(json.dumps(conf, ensure_ascii=False, separators=(',', ':')).encode()) 257 | frontend = subprocess.Popen(exe(os.path.join(apppath, appexe)) + ['--disable-smooth-scrolling'], close_fds=True) 258 | waitsec = 0.1 259 | while not api('/havesession'): 260 | time.sleep(waitsec) 261 | waitsec = min(waitsec * 2, 1) 262 | w.close() 263 | 264 | def api(path, data=b'', retry=True): 265 | while True: 266 | try: 267 | req = urllib.request.Request(url + path) 268 | if session is not None: 269 | req.add_header('X-VIDEO2SUB-SESSION', session) 270 | if data is not None and type(data) is not bytes: 271 | data = json.dumps(data).encode('utf-8') 272 | req.add_header('Content-Type', 'application/json') 273 | req.data = data 274 | req = urllib.request.urlopen(req) 275 | rsp = req.read() 276 | try: 277 | return json.loads(rsp.decode('utf-8')) 278 | except: 279 | return rsp 280 | except urllib.error.HTTPError as e: 281 | raise e 282 | except Exception as e: 283 | check_backend() 284 | if not retry: 285 | return None 286 | time.sleep(0.1) 287 | 288 | def watch(): 289 | global lastwatch 290 | logs = api('/logs') 291 | for arr in logs['row']: 292 | row = dict(zip(logs['col'], arr)) 293 | if row['cursession'] and row['id'] > lastwatch: 294 | print(video, '#', row['date'], '#', row['level'], row['message']) 295 | lastwatch = row['id'] 296 | 297 | try: 298 | files = sys.argv[1:] 299 | if windows and len(files) == 0: 300 | print('请从弹出的窗口中选择要处理的文件') 301 | print(' 若只指定一个文件，则使用图形界面模式') 302 | print(' 若指定多个文件，则使用批处理模式') 303 | print() 304 | files = file_select('请选择要处理的文件', '视频文件 (%s)\0%s\0video2sub 数据库文件 (*.v2s)\0*.v2s\0所有文件 (*.*)\0*.*\0'%(EXTENSIONS_VIDEO, EXTENSIONS_VIDEO)) 305 | files = [os.path.abspath(p) for p in files] 306 | os.chdir(bundle_dir) 307 | if len(files) == 0: 308 | print('使用方法:') 309 | print(' %s [视频/数据库文件...]'%os.path.basename(sys.argv[0])) 310 | print('若只指定一个文件，则使用图形界面模式') 311 | print('若指定多个文件，则使用批处理模式') 312 | elif len(files) == 1: 313 | print('图形界面模式') 314 | print('=====') 315 | run_backend(files[0], nosession=True) 316 | run_frontend() 317 | frontend.wait() 318 | elif len(files) > 1: 319 | print('批处理模式') 320 | print('=====') 321 | print('文件列表:\n '+'\n '.join(map(os.path.basename, files))) 322 | print('=====') 323 | op = list_select('批处理模式', '已选择%d个文件，请选择要执行的操作'%len(files), [ 324 | ('ocr', '批量OCR'), 325 | ('exportass', '批量导出ASS'), 326 | ('exportcsv', '批量导出CSV'), 327 | ('gui', '同时打开多个图形界面进行操作'), 328 | ]) 329 | if op is not None: 330 | print('操作:', op[1]) 331 | op = op[0] 332 | else: 333 | print('没有选择操作') 334 | print('=====') 335 | start_time = time.time() 336 | success = 0 337 | if op is None: 338 | success = -1 339 | elif op == 'ocr': 340 | print('正从第一个文件“%s”中读取OCR设置'%os.path.basename(files[0])) 341 | run_backend(files[0], silent=True) 342 | ocrconfig = api('/loadconfig', {'key':'OCR'}) 343 | backend.kill() 344 | backend.wait() 345 | print('将要使用的OCR设置:', ocrconfig) 346 | print('=====') 347 | for file in files: 348 | print('正处理:', os.path.basename(file)) 349 | run_backend(file, silent=True) 350 | file_start_time = time.time() 351 | for i in range(9999999): 352 | api('/saveconfig', {'key':'OCR', 'value':ocrconfig, 'msg':'设定批处理OCR设置: '+str(ocrconfig)}) 353 | state = api('/state') 354 | if state['lastjob'] > 0 or state['curarea'] == 0: 355 | if i > 10: 356 | print('迭代次数过多，已放弃') 357 | break 358 | if i > 0: 359 | print('迭代（第%d次）'%i) 360 | 361 | if state['lastjob'] > 0: 362 | print('数据库中有未处理完毕的项目，执行“继续OCR”操作') 363 | ret = api('/continueocr', {'item_range': None, 'restarttype': ''}) 364 | elif state['curarea'] == 0: 365 | print('数据库中无该OCR区域的项目，执行“新OCR”操作') 366 | ret = api('/startocr', {'frame_range': None}) 367 | else: 368 | assert False 369 | 370 | if ret == b'ok': 371 | while api('/state')['ocractive']: 372 | watch() 373 | time.sleep(1) 374 | else: 375 | watch() 376 | print('无法启动OCR任务，请检查第一个文件的OCR设置是否正确') 377 | break 378 | watch() 379 | else: 380 | print('没有要做的任务' if i == 0 else '任务已完成') 381 | success += 1 382 | break 383 | backend.kill() 384 | backend.wait() 385 | print('耗时%.2f秒'%(time.time() - file_start_time)) 386 | print('=====') 387 | elif op.startswith('export'): 388 | for file in files: 389 | print('正处理:', os.path.basename(file)) 390 | run_backend(file, silent=True) 391 | if api('/' + op) == b'ok': 392 | success += 1 393 | watch() 394 | backend.kill() 395 | backend.wait() 396 | print('=====') 397 | elif op == 'gui': 398 | success = -1 399 | children = [] 400 | waiters = [] 401 | def do_wait(i): 402 | children[i].wait() 403 | print('已退出:', os.path.basename(files[i])) 404 | def do_join(): 405 | for waiter in waiters: 406 | waiter.join() 407 | try: 408 | for i, file in enumerate(files): 409 | if i > 0: 410 | time.sleep(1) 411 | print('正打开:', os.path.basename(file)) 412 | children.append(subprocess.Popen(exe('launcher.py') + [file], close_fds=True)) 413 | waiters.append(Thread(target=do_wait, args=(i,))) 414 | waiters[i].start() 415 | if windows: 416 | joiner = Thread(target=do_join) 417 | joiner.start() 418 | while joiner.is_alive(): # avoid join() which can't be interrupted 419 | time.sleep(1) 420 | else: 421 | do_join() 422 | except: 423 | traceback.print_exc() 424 | if not windows: 425 | for child in children: 426 | child.send_signal(signal.SIGINT) 427 | do_join() 428 | else: 429 | assert False 430 | if success >= 0: 431 | print('共%d个文件，成功%d个，失败%d个，耗时%.2f秒'%(len(files), success, len(files)-success, time.time()-start_time)) 432 | if success != len(files): 433 | backendlog_preserve = True 434 | if windows: 435 | waitkey = True 436 | except KeyboardInterrupt: 437 | print('手动中断') 438 | except Exception as e: 439 | if isinstance(e, BackendDiedError): 440 | print('（后端异常退出）') 441 | else: 442 | print('===== 异常 =====') 443 | traceback.print_exc() 444 | if backendlog is not None: 445 | backendlog.close() 446 | with open(backendlog_file, 'r', errors='replace') as f: 447 | lines = f.readlines() 448 | for i in range(len(lines)): 449 | if lines[-i - 1].startswith('>>>>>'): 450 | lines = lines[-i:] if i > 0 else [] 451 | break 452 | maxshow = 50 453 | if len(lines) > maxshow: 454 | print('===== 后端日志的最后 %d 行 ====='%maxshow) 455 | lines = lines[-maxshow:] 456 | else: 457 | print('===== 后端日志 =====') 458 | print(''.join(lines), end='') 459 | fail = True 460 | except: 461 | traceback.print_exc() 462 | fail = True 463 | 464 | if backend: 465 | backend.kill() 466 | backend.wait() 467 | if frontend: 468 | frontend.kill() 469 | frontend.wait() 470 | 471 | if backendlog is not None: 472 | backendlog.close() 473 | if not fail and not backendlog_preserve and os.path.exists(backendlog_file): 474 | os.remove(backendlog_file) 475 | else: 476 | print('（后端日志已存储为: %s）'%backendlog_file) 477 | 478 | exitmsg = [] 479 | exitcode = 0 480 | if fail: 481 | print('=====') 482 | exitmsg.append('遇到致命错误') 483 | exitcode = 1 484 | if windows: 485 | waitkey = True 486 | if waitkey: 487 | if windows: 488 | exitmsg.append('按任意键退出程序') 489 | else: 490 | exitmsg.append('按回车退出程序') 491 | if len(exitmsg) > 0: 492 | print('，'.join(exitmsg)) 493 | if waitkey: 494 | if windows: 495 | msvcrt.getch() 496 | else: 497 | input() 498 | sys.exit(exitcode) 499 | -------------------------------------------------------------------------------- /screenshot.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/zhangboyang/video2sub/5964e919d3e024cfb4272fa3b90c4943adeb85dc/screenshot.png -------------------------------------------------------------------------------- /ui/ui.css: -------------------------------------------------------------------------------- 1 | html, body { 2 | margin: 0; 3 | padding: 0; 4 | width: 100%; 5 | height: 100%; 6 | font-size: 16px; 7 | line-height: 20px; 8 | background-color: white; 9 | font-family: Tahoma, '宋体', SimSun, sans-serif; 10 | overflow-x: hidden; 11 | overflow-y: auto; 12 | } 13 | input, select { 14 | font-family: Tahoma, '宋体', SimSun, sans-serif; 15 | font-size: 14px; 16 | } 17 | input:focus, select:focus { 18 | outline: none; 19 | } 20 | a { 21 | color: blue; 22 | text-decoration: underline; 23 | cursor: pointer; 24 | user-select: none; 25 | } 26 | a.button { 27 | display: inline-block; 28 | height: 20px; 29 | vertical-align: middle; 30 | position: relative; 31 | background-color: white; 32 | border: 1px solid gray; 33 | padding: 2px 3px; 34 | text-align: center; 35 | } 36 | a.button:hover { 37 | border-color: black; 38 | } 39 | a.button > div { 40 | display: none; 41 | } 42 | a.button:hover > div { 43 | text-align: left; 44 | position: absolute; 45 | width: max-content; 46 | display: inline-block; 47 | background-color: white; 48 | border: 1px solid black; 49 | z-index: 100; 50 | padding: 2px; 51 | color: black; 52 | pointer-events: none; 53 | } 54 | a.button:hover > div.bl { /* bottom-left tooltip */ 55 | top: 26px; 56 | left: -1px; 57 | } 58 | a.button:hover > div.br { /* bottom-right tooltip*/ 59 | top: 26px; 60 | right: -1px; 61 | } 62 | a.button:hover > div.tr { /* top-right tooltip*/ 63 | bottom: 26px; 64 | right: -1px; 65 | } 66 | [v-cloak] { 67 | display: none; 68 | } 69 | #app { 70 | width: 100%; 71 | height: 100%; 72 | position: relative; 73 | } 74 | #loading { 75 | position: absolute; 76 | left: 0; 77 | top: 0; 78 | height: 100%; 79 | width: 100%; 80 | background-color: white; 81 | padding: 20px; 82 | box-sizing: border-box; 83 | } 84 | #apparea { 85 | width: 100%; 86 | height: 100%; 87 | display: flex; 88 | flex-direction: column; 89 | } 90 | #apparea > div { 91 | position: relative; 92 | } 93 | #apparea > div:not(:first-child) { 94 | border-top: 1px black solid; 95 | } 96 | .dialogbackground { 97 | position: absolute; 98 | left: 0; 99 | top: 0; 100 | width: 100%; 101 | height: 100%; 102 | background-color: rgba(0, 0, 0, 0.5); 103 | } 104 | #title { 105 | text-align: center; 106 | } 107 | 108 | div.imgdb { 109 | position: relative; 110 | overflow: hidden; 111 | } 112 | div.imgdb > img { 113 | position: absolute; 114 | } 115 | #video.editorup { 116 | flex-direction: column-reverse; 117 | } 118 | #video { 119 | display: flex; 120 | user-select: none; 121 | flex-direction: column; 122 | } 123 | #framearea { 124 | display: flex; 125 | margin-bottom: 5px; 126 | } 127 | #framebox { 128 | display: flex; 129 | flex-direction: column; 130 | margin: 0 10px; 131 | } 132 | .editorup #framebox { 133 | flex-direction: column-reverse; 134 | } 135 | #frametitle { 136 | border: 1px solid black; 137 | border-bottom: 0; 138 | } 139 | .editorup #frametitle { 140 | border: 1px solid black; 141 | border-bottom: 1px solid black; 142 | border-top: 0; 143 | } 144 | #frame { 145 | border: 1px solid black; 146 | position: relative; 147 | background-color: cadetblue; 148 | } 149 | #frame > #frameevent { 150 | left: 0; 151 | top: 0; 152 | position: absolute; 153 | width: 100%; 154 | height: 100%; 155 | } 156 | #frame > img { 157 | width: 100%; 158 | height: 100%; 159 | } 160 | #frameleft { 161 | flex: 1; 162 | overflow-wrap: anywhere; 163 | writing-mode: vertical-rl; 164 | } 165 | #frameright { 166 | flex: 1; 167 | overflow-wrap: anywhere; 168 | writing-mode: vertical-lr; 169 | } 170 | #frameleft > *, #frameright > * { 171 | writing-mode: horizontal-tb; 172 | margin: 0 5px; 173 | } 174 | div.toolbox { 175 | border: 1px solid gray; 176 | } 177 | div.toolbox > strong { 178 | box-sizing: border-box; 179 | display: block; 180 | width: 100%; 181 | padding: 2px 5px; 182 | text-align: center; 183 | } 184 | div.toolbox > div:not(:first-child) { 185 | border-top: 1px solid gray; 186 | } 187 | div.toolbox > div.text { 188 | padding: 2px 5px; 189 | } 190 | 191 | #frame > #ocrarea { 192 | width: calc(100% + 2px); 193 | position: absolute; 194 | border-style: solid; 195 | border-width: 0px 9px; 196 | left: -10px; 197 | } 198 | .neighborframe { 199 | float: right; 200 | display: flex; 201 | flex-direction: column; 202 | } 203 | .neighborframe > .imgdb { 204 | cursor: pointer; 205 | } 206 | .editorup .neighborframe { 207 | flex-direction: column-reverse; 208 | } 209 | .editorup div.neighborframe.toolbox > div:not(:first-child) { 210 | border-top: 0; 211 | border-bottom: 1px solid gray; 212 | } 213 | 214 | #dispsize, #jumpframe, #ocrinfo { 215 | margin-top: 10px; 216 | float: left; 217 | } 218 | #jumpframe td { 219 | text-align: right; 220 | cursor: pointer; 221 | color: blue; 222 | text-decoration: underline; 223 | } 224 | #ocrinfo { 225 | max-width: 140px; 226 | } 227 | 228 | #editor { 229 | margin-bottom: 5px; 230 | display: flex; 231 | } 232 | #editor > div { 233 | flex: 1; 234 | margin: 0 5px; 235 | position: relative; 236 | } 237 | #editor > div:first-child > div { 238 | right: 0; 239 | } 240 | #editor > div > div { 241 | min-width: max-content; 242 | position: absolute; 243 | bottom: 0; 244 | } 245 | #editor > input { 246 | display: block; 247 | margin: 0 auto; 248 | padding: 2px; 249 | text-align: center; 250 | border: 1px solid black; 251 | font-family: Arial, '黑体', SimHei; 252 | } 253 | #editor > input:read-only { 254 | background-color: lightgray; 255 | } 256 | 257 | #thumbnail { 258 | position: absolute; 259 | border: 1px solid #00ff00; 260 | bottom: 0; 261 | } 262 | #timebar { 263 | background-color: magenta; 264 | height: 30px; 265 | min-height: 30px; 266 | user-select: none; 267 | overflow: hidden; 268 | } 269 | #timebar > div { 270 | position: absolute; 271 | left: 0; 272 | top: 0; 273 | height: 100%; 274 | } 275 | #timebar > div#barbg { 276 | width: 100%; 277 | } 278 | #timebar > div#posmarker, #timebar > div#mousemarker { 279 | width: 0; 280 | height: 0; 281 | margin: 0 -5px; 282 | border-style: solid; 283 | border-width: 0 5px 10px 5px; 284 | border-color: transparent transparent #00ff00 transparent; 285 | z-index: 2; 286 | } 287 | #timebar > #mouselabel { 288 | margin-left: 10px; 289 | color: black; 290 | z-index: 2; 291 | } 292 | #timebar > canvas { 293 | height: 100%; 294 | position: absolute; 295 | left: 0; 296 | top: 0; 297 | z-index: 1; 298 | } 299 | #thumbbar { 300 | background-color:lightgray; 301 | } 302 | 303 | 304 | #op input[type='text'] { 305 | width: 70px; 306 | height: 26px; 307 | border: 1px solid black; 308 | box-sizing: border-box; 309 | vertical-align: middle; 310 | } 311 | #op input[type='file'] { 312 | display: none; 313 | } 314 | #op select { 315 | width: 150px; 316 | height: 26px; 317 | box-sizing: border-box; 318 | border: 1px solid black; 319 | vertical-align: middle; 320 | background-color: greenyellow; 321 | } 322 | #op { 323 | background-color: #e8ffcd; 324 | user-select: none; 325 | display: flex; 326 | } 327 | #op > div:nth-child(2) { 328 | flex: 1 329 | } 330 | #op > div > div { 331 | display: block; 332 | padding: 2px 10px; 333 | height: 26px; 334 | } 335 | #op > div > div > span { 336 | line-height: 26px; 337 | display: inline-block; 338 | vertical-align: top; 339 | } 340 | #op > div:not(:first-child) > div { 341 | border-left: 1px solid black; 342 | } 343 | #op > div > div:not(:first-child) { 344 | border-top: 1px solid black; 345 | } 346 | #op #export { 347 | background-color: white; 348 | } 349 | #op #export a.button { 350 | color:rgb(0, 80, 0); 351 | } 352 | 353 | #codeeditor { 354 | border: 1px solid black; 355 | background-color: white; 356 | width: 80%; 357 | height: 80%; 358 | position: absolute; 359 | top: 10%; 360 | left: 10%; 361 | z-index: 20; 362 | display: flex; 363 | flex-direction: column; 364 | user-select: none; 365 | } 366 | #codeeditor > div:not(:first-child) { 367 | border-top: 1px solid black; 368 | } 369 | #codeeditor > #codeeditortitle { 370 | text-align: center; 371 | line-height: 24px; 372 | position: relative; 373 | } 374 | #codeeditor > #codeeditortitle > a { 375 | position: absolute; 376 | width: 24px; 377 | border-left: 1px solid black; 378 | right: 0; 379 | top: 0; 380 | } 381 | #codeeditor > #editordiv { 382 | flex: 1; 383 | overflow: hidden; 384 | } 385 | #codeeditor > #scriptselect { 386 | padding: 2px 5px; 387 | } 388 | #codeeditor > #scriptselect > select { 389 | border: 1px solid black; 390 | background-color: azure; 391 | height: 26px; 392 | width: 300px; 393 | } 394 | #codeeditor > #editordiv > .CodeMirror { 395 | height: 100%; 396 | width: 100%; 397 | } 398 | #codeeditor > #codeeditorop { 399 | padding: 2px 5px; 400 | } 401 | #codeeditor > #codeeditorop > .button { 402 | float: right; 403 | min-width: 60px; 404 | min-height: 26px; 405 | line-height: 26px; 406 | margin-right: 10px; 407 | } 408 | 409 | table { 410 | border-collapse: collapse; 411 | } 412 | table td, table th { 413 | border: 1px gray solid; 414 | padding: 2px 5px; 415 | vertical-align: top; 416 | } 417 | table tr:first-child th, table tr:first-child td { 418 | border-top: 0; 419 | } 420 | table td:first-child, table th:first-child { 421 | border-left: 0; 422 | } 423 | table td:last-child, table th:last-child { 424 | border-right: 0; 425 | } 426 | table tr:last-child td { 427 | border-bottom: 0; 428 | } 429 | 430 | #tblarea { 431 | flex: 1; 432 | display: flex; 433 | flex-direction: column; 434 | min-height: 5px; 435 | } 436 | #tblsummary { 437 | width: 350px; 438 | text-align: center; 439 | } 440 | #tblbottom { 441 | border-top: 1px solid gray; 442 | flex: 1; 443 | color: gray; 444 | user-select: none; 445 | } 446 | #tblbottom > div { 447 | padding: 2px 10px; 448 | } 449 | #tblnoresult { 450 | text-align: center; 451 | } 452 | #tblhint > span { 453 | color: rgb(64, 128, 64); 454 | } 455 | #tblop { 456 | display: flex; 457 | border-bottom: 1px solid black; 458 | user-select: none; 459 | } 460 | #tblop > div { 461 | padding: 2px 4px; 462 | } 463 | #tblop > div:not(:first-child) { 464 | border-left: 1px solid black; 465 | } 466 | #tblop > div > span { 467 | line-height: 24px; 468 | vertical-align: top; 469 | display: inline-block; 470 | } 471 | #tblop select { 472 | border: 1px solid black; 473 | background-color: azure; 474 | padding: 2px; 475 | width: 100px; 476 | } 477 | #tblfltcustom { 478 | flex: 1; 479 | margin: 2px; 480 | padding: 2px; 481 | border: 1px solid black; 482 | font-family: monospace; 483 | } 484 | #text { 485 | flex: 1; 486 | overflow-y: scroll; 487 | display: flex; 488 | flex-direction: column; 489 | } 490 | #text > table { 491 | width: 100%; 492 | } 493 | #text > table th { 494 | white-space: nowrap; 495 | } 496 | #text > table td:not(:nth-child(4)) { 497 | white-space: nowrap; 498 | } 499 | #text > table td:nth-child(4) { 500 | width: 100%; 501 | white-space: pre-wrap; 502 | } 503 | #text > table tr { 504 | scroll-margin-bottom: 30px; 505 | } 506 | #text > table tr[data-state='waitocr'] { 507 | background-color: lightgray; 508 | } 509 | #text > table tr[data-state='error'] { 510 | background-color: lightpink; 511 | } 512 | #text > table tr[data-state='merged'] { 513 | background-color: lightgray; 514 | } 515 | #text > table tbody tr:hover { 516 | background-color: lemonchiffon !important; 517 | } 518 | #text > table tr td:nth-child(1) { 519 | color: gray; 520 | } 521 | #text > table tr[data-tag~='primary'] td:nth-child(1) { 522 | color: black; 523 | } 524 | #text > table tr[data-state='done'][data-tag~='sel'] { 525 | background-color: rgb(130, 215, 255); 526 | } 527 | #text > table tr[data-state='done'][data-tag~='cur'] { 528 | background-color: yellowgreen; 529 | } 530 | #text > table tr[data-tag~='cur'] td:nth-child(2) { 531 | background-color: yellowgreen; 532 | } 533 | #text > table tr[data-tag~='cur'] td:first-child { 534 | background-color: yellowgreen; 535 | } 536 | #text > table tr[data-tag~='sel'] div.img { 537 | outline: 2px solid blue !important; 538 | } 539 | #text > table tr[data-tag~='sel'] td:first-child { 540 | background-color: rgb(130, 215, 255) !important; 541 | } 542 | #text > table tr[data-lock='locked'] { 543 | text-decoration: line-through; 544 | } 545 | #text > table tr:not([data-position="2"]) td:nth-child(2) { 546 | color: gray; 547 | } 548 | #text > table tr[data-empty='empty'] td:nth-child(4) { 549 | color: gray; 550 | } 551 | 552 | #text > table tr:hover td:nth-child(4) { 553 | text-decoration: underline; 554 | } 555 | #text > table tr div.img { 556 | outline: 1px solid white; 557 | } 558 | #text > table tr:hover div.img { 559 | outline: 1px dashed red; 560 | } 561 | 562 | 563 | #logarea { 564 | overflow: hidden; 565 | } 566 | #logarea > div.buttonbar { 567 | display: inline-block; 568 | position: absolute; 569 | right: 32px; 570 | bottom: 5px; 571 | text-align: right; 572 | } 573 | #logarea > div.buttonbar .button { 574 | margin: 2px 0; 575 | } 576 | #logs { 577 | width: 100%; 578 | height: 100%; 579 | overflow-y: scroll; 580 | } 581 | #logs > table { 582 | width: 100%; 583 | } 584 | #logs > table tr[data-level='I'] { 585 | background-color: azure; 586 | } 587 | #logs > table tr[data-level='E'] { 588 | background-color: lightpink; 589 | } 590 | #logs > table tr[data-level='S'] { 591 | background-color: lightgreen; 592 | } 593 | #logs > table tr[data-level='W'] { 594 | background-color: lightyellow; 595 | } 596 | #logs > table tr[data-level='C'] { 597 | background-color: gold; 598 | } 599 | #logs > table td:not(:last-child) { 600 | white-space: nowrap; 601 | } 602 | #logs > table td:last-child { 603 | width: 100%; 604 | white-space: pre-wrap; 605 | } 606 | 607 | #promptbg { 608 | display: flex; 609 | align-items: center; 610 | justify-content: center; 611 | z-index: 20; 612 | } 613 | #promptdialog { 614 | border: 1px solid black; 615 | background-color: white; 616 | width: 350px; 617 | height: 160px; 618 | position: absolute; 619 | margin: auto; 620 | z-index: 20; 621 | display: flex; 622 | flex-direction: column; 623 | user-select: none; 624 | padding: 10px; 625 | } 626 | #promptdialog > div:nth-child(1) { 627 | padding: 10px; 628 | white-space: pre-wrap; 629 | } 630 | #promptdialog > div:nth-child(2) { 631 | flex: 1; 632 | } 633 | #promptdialog > div:nth-child(2) > input { 634 | width: 90%; 635 | margin: auto; 636 | display: block; 637 | border: 1px solid black; 638 | padding: 5px; 639 | font-size: 16px; 640 | } 641 | #promptdialog > div:nth-child(3) { 642 | text-align: right; 643 | } 644 | #promptdialog > div:nth-child(3) > a.button { 645 | padding: 10px 20px; 646 | } 647 | 648 | #adveditorbg { 649 | display: flex; 650 | align-items: center; 651 | justify-content: center; 652 | z-index: 20; 653 | } 654 | #adveditor { 655 | border: 1px solid black; 656 | background-color: white; 657 | width: 600px; 658 | height: 600px; 659 | position: absolute; 660 | margin: auto; 661 | z-index: 20; 662 | display: flex; 663 | flex-direction: column; 664 | user-select: none; 665 | padding: 10px; 666 | } 667 | #adveditor > div:nth-child(1) { 668 | flex: 1; 669 | margin-bottom: 10px; 670 | overflow: hidden; 671 | border: 1px solid gray; 672 | } 673 | #adveditor > div:nth-child(1) > .CodeMirror { 674 | height: 100%; 675 | width: 100%; 676 | } 677 | #adveditor > div:nth-child(2) { 678 | text-align: right; 679 | } 680 | #adveditor > div:nth-child(2) > a.button { 681 | padding: 10px 20px; 682 | } -------------------------------------------------------------------------------- /ui/ui.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 硬字幕OCR提取工具 11 | 12 | 13 |
14 |
15 |
{{info.file}} ({{info.width}}x{{info.height}}, {{info.nframes}}帧, {{info.fps}}fps)
16 |
17 |
18 |
19 |
20 | 上一帧 ({{pos-1}}) 21 | 22 |
无
23 |
24 |
25 | 显示大小 26 |
27 | 缩略图 28 | 小 29 | 大 30 |
31 |
32 |
33 | 跳转 34 |
35 | 36 | 37 | 40 | 41 |
38 | {{sgn>0?'+':'-'}}{{n}} 39 |
42 |
43 |
44 | 手工输入 45 |
46 |
47 |
48 |
49 |
当前帧 ({{pos}})
50 |
51 | 52 | 53 |
54 |
55 |
56 |
57 |
58 |
59 | 下一帧 ({{pos+1}}) 60 | 61 |
无
62 |
63 |
64 | {{ocrselmode?'选中的字幕':'新生成字幕'}}
OCR区域 65 |
66 | 请在图片上用左键拖动来指定字幕的范围 67 | y1={{ocrsel[0]}}
y2={{ocrsel[1]}} 68 |
69 |
70 | 手工输入 71 |
72 |
73 | x={{mousexy[0]}}
74 | y={{mousexy[1]}} 75 |
76 |
77 |
78 |
79 | 100 | 101 |
102 |
103 |
104 |
105 | 106 |
107 |
108 |
{{mousepos}}
109 |
110 |
111 |
112 |
{{tblinfo}}{{tblinfo2}}
113 |
114 | 排序 115 | 118 |
119 |
120 | 筛选 121 | 124 |
125 | 126 |
127 |
128 | 129 | 130 | 131 | 132 | 133 | 134 | 135 | 136 | 137 | 138 | 139 |
帧范围OCR区域引擎文本预览
140 |
141 |
（无结果）
142 |
单击:选择 / Ctrl+单击:多选 / Shift+单击:选择范围
143 |
144 |
145 |
146 | 191 |
192 |
193 | {{checkpointonly?'显示所有日志':'只显示恢复点'}}
{{checkpointonly?'在日志栏中显示所有类型日志':'在日志栏中只显示恢复点'}}
 194 | 创建恢复点
对全部字幕项“存档”，以便需要时“读档”
 195 | 显示{{morelog?'更少':'更多'}}日志
{{morelog?'减小':'增大'}}日志栏高度
 196 |
197 |
198 | 199 | 200 | 201 | 202 | 203 | 204 | 205 |
{{log.date}}#{{log.checkpoint_id}} {{log.message}} {{log.level=='C'?'撤销至此':'撤销至此操作之前'}}
处理中…
与后端通信失败，继续操作数据可能会丢失，请重新启动本程序！！！
206 |
207 |
208 |
209 |
210 |
211 |
212 | 执行脚本 213 | X 214 |
215 |
216 | 219 | 重命名 220 | 克隆 221 | 删除 222 | 新建 223 | 该脚本已锁定，不可更改 224 | 提示：按 Ctrl+Shift+I 可打开控制台。 225 |
226 |
227 |
228 | 执行 229 | 取消 230 |
231 |
232 |
233 |
234 |
235 |
236 | 取消 237 | 确定 238 |
239 |
240 |
241 |
242 |
243 |
{{myprompt.message}}
244 |
245 |
246 | 取消 247 | 确定 248 |
249 |
250 |
251 |
加载中，请稍候…

若数据较多，加载速度可能会比较慢
252 |
253 | 254 | 255 | -------------------------------------------------------------------------------- /ui/ui.js: -------------------------------------------------------------------------------- 1 | "use strict"; 2 | 3 | // use 'var' instead of 'let' for global variables 4 | // or compile_expr() can't capture them!!! 5 | 6 | var geval = eval; 7 | function compile_expr(gvars, param, expr) { 8 | /* NOT SAFE BUT EFFECTIVE */ 9 | return geval('(function(){"use strict";let ' + Object.keys(window).filter(i => gvars.indexOf(i) < 0).join(';let ') + ';return function('+param.join(',')+'){return ('+expr+');};})()'); 10 | } 11 | function compile_func(gvars, func) { 12 | return compile_expr(gvars, [], '('+func+')')(); 13 | } 14 | function strcmp(a, b) { 15 | if (a < b) return -1; 16 | if (a > b) return 1; 17 | return 0; 18 | } 19 | function ziprow(col, arr) { 20 | let obj = {}; 21 | col.forEach((key, i) => obj[key] = arr[i]); 22 | return obj; 23 | } 24 | function selectcol(sel, arr) { 25 | return arr.map((item) => { 26 | let obj = {}; 27 | sel.forEach((key) => obj[key] = item[key]); 28 | return obj; 29 | }); 30 | } 31 | function array_equal(a, b) { 32 | let n = a.length; 33 | for (let i = 0; i < n; i++) { 34 | if (a[i] !== b[i]) { 35 | return false; 36 | } 37 | } 38 | return true; 39 | } 40 | function array_cmp(a, b) { 41 | let n = a.length; 42 | for (let i = 0; i < n; i++) { 43 | if (a[i] !== b[i]) { 44 | if (typeof a[i] === 'number') { 45 | return a[i] - b[i]; 46 | } else { 47 | return strcmp(a[i], b[i]); 48 | } 49 | } 50 | } 51 | return 0; 52 | } 53 | function sec2str(sec) { 54 | sec = Math.floor(sec); 55 | let s = sec % 60; 56 | let m = Math.floor(sec / 60) % 60; 57 | let h = Math.floor(sec / 60 / 60); 58 | return h.toString() +':'+ ('0'+m.toString()).slice(-2) +':'+ ('0'+s.toString()).slice(-2); 59 | } 60 | 61 | /////////////////////////////////////////////////////////////////////////////// 62 | var curedit = null; 63 | function savecur() { 64 | if (curedit) { 65 | let upd_item = id2item.get(curedit[colmap['id']]); 66 | if (upd_item && !upd_item.locked) { 67 | let upd_data = ziprow(ocrresult.col, upd_item.r); 68 | let upd_text = app.geteditboxval(); 69 | if (upd_data.state == 'done' && upd_data.ocrtext != upd_text) { 70 | upd_data.ocrtext = curedit[colmap['ocrtext']] = upd_text; 71 | tbllock(upd_item); 72 | app.refreshafter(axios.post('/updateresult', { 73 | changes: selectcol(['id','ocrtext'], [upd_data]), 74 | checkpoint: '“手动修改：'+upd_data.ocrtext+'”之前', 75 | message: '手动修改已保存：'+upd_data.ocrtext, 76 | compatlog: true, 77 | })); 78 | } 79 | } 80 | } 81 | } 82 | function tbledit(item, scroll) { 83 | let newedit = item ? JSON.parse(JSON.stringify(item.r)) : null; 84 | 85 | let change_item = false; 86 | if (newedit) { 87 | if (!curedit || curedit[colmap['id']] != newedit[colmap['id']]) { 88 | change_item = true; 89 | app.setframepos(newedit[colmap['frame_start']]); 90 | } 91 | } 92 | if (newedit && newedit[colmap['state']] == 'done') { 93 | if (curedit && change_item) { 94 | savecur(); 95 | } 96 | app.$refs.editbox.readOnly = item.locked; 97 | if (item.locked || change_item || (curedit && array_cmp(curedit, newedit) != 0)) { 98 | app.seteditboxval(newedit[colmap['ocrtext']]); 99 | app.$refs.editbox.focus(); 100 | app.setocrsel(1, [newedit[colmap['top']], newedit[colmap['bottom']]]); 101 | if (scroll) { 102 | item.ele.scrollIntoView({block: "nearest", inline: "nearest"}); 103 | } 104 | } 105 | } else { 106 | if (curedit) { 107 | savecur(); 108 | } 109 | app.seteditboxval(''); 110 | app.$refs.editbox.readOnly = true; 111 | app.setocrsel(0); 112 | } 113 | curedit = newedit; 114 | if (item && !tblselect.has(item)) { 115 | tblsel([item]); 116 | } 117 | updatetblstyle(); 118 | } 119 | function updatetbledit() { 120 | if (curedit) { 121 | let item = id2item.get(curedit[colmap['id']]); 122 | if (item && tblselect.has(item)) { 123 | tbledit(item); 124 | } else if (tblselect.size > 0) { 125 | tbledit(tblitem.find((item) => tblselect.has(item))); 126 | } else { 127 | tbledit(null); 128 | } 129 | } 130 | } 131 | 132 | /////////////////////////////////////////////////////////////////////////////// 133 | 134 | var coldesc = [ 135 | ['id', '字幕唯一内部ID(不可修改)'], 136 | ['state', '字幕状态：waitocr, error, done, merged'], 137 | ['frame_start', '字幕起始帧(含)'], 138 | ['frame_end', '字幕结束帧(含)'], 139 | ['engine', '使用的OCR引擎名'], 140 | ['top', '字幕区域Y1(含)'], 141 | ['bottom', '字幕区域Y2(含)'], 142 | ['position', '字幕位置(2下半屏, 8上半屏)'], 143 | ['ocrtext', '字幕文本'], 144 | ]; 145 | 146 | var lastresult = ''; 147 | var ocrresult = {col:[],row:[]}; 148 | var colmap = {}; 149 | var id2item = new Map(); // id => { ele: , sortkey: [], r: [c1, c2...] } 150 | 151 | var tableele = null; 152 | var tblview = new Set(); 153 | var tblitem = []; 154 | var viewopt = { 155 | filter: null, 156 | sortkey: null, 157 | }; 158 | var tblsortfunc = [ 159 | ['上/下半屏','[-position,frame_start,frame_end,top,bottom,engine,id]'], 160 | ['帧范围','[frame_start,frame_end,-position,top,bottom,engine,id]'], 161 | ['帧OCR区域','[top,bottom,-position,frame_start,frame_end,engine,id]'], 162 | ['引擎名称','[engine,-position,frame_start,frame_end,top,bottom,id]'], 163 | ]; 164 | var tblfltfunc = [ 165 | ['全部','true'], 166 | ['所有待处理',"!(state == 'merged' || (state == 'done' && ocrtext == ''))"], 167 | ['所有空项',"state == 'done' && ocrtext == ''"], 168 | ['=====',null], 169 | ['上半屏','position == 8'], 170 | ['上半屏待处理',"position == 8 && !(state == 'merged' || (state == 'done' && ocrtext == ''))"], 171 | ['上半屏空项',"position == 8 && state == 'done' && ocrtext == ''"], 172 | ['=====',null], 173 | ['下半屏','position == 2'], 174 | ['下半屏待处理',"position == 2 && !(state == 'merged' || (state == 'done' && ocrtext == ''))"], 175 | ['下半屏空项',"position == 2 && state == 'done' && ocrtext == ''"], 176 | ['=====',null], 177 | ['多行',"ocrtext.includes('\\n')"], 178 | ['待清理',"state == 'merged' || (state == 'done' && ocrtext == '')"], 179 | ['等待OCR',"state == 'waitocr'"], 180 | ['OCR失败',"state == 'error'"], 181 | ['无','false'], 182 | ]; 183 | function setviewopt(filter, sortkey) { 184 | if (filter !== null) { 185 | let fn = null; 186 | app.tblflterr = false; 187 | try { 188 | fn = compile_expr([], ocrresult.col, filter); 189 | } catch(e) { 190 | app.tblflterr = true; 191 | } 192 | viewopt.filter = function (item) { 193 | try { 194 | return fn.apply(this, item.r); 195 | } catch(e) { 196 | app.tblflterr = true; 197 | return false; 198 | } 199 | }; 200 | } 201 | if (sortkey !== null) { 202 | let fn = compile_expr([], ocrresult.col, sortkey); 203 | viewopt.sortkey = function (item) { 204 | return fn.apply(this, item.r); 205 | }; 206 | } 207 | if (filter !== null || sortkey !== null) { 208 | updateview(); 209 | if (curedit) { 210 | let item = id2item.get(curedit[colmap['id']]); 211 | if (item) { 212 | item.ele.scrollIntoView({block: "nearest", inline: "nearest"}); 213 | } 214 | } 215 | } 216 | } 217 | function sortedview(expr) { 218 | if (expr === undefined) { 219 | expr = '[position,frame_start,frame_end,top,bottom,engine,id]'; 220 | } 221 | let fn = compile_expr([], ocrresult.col, expr); 222 | let getsortkey = function (item) { 223 | return fn.apply(this, item.r); 224 | }; 225 | return Array.from(tblview).sort((a, b) => array_cmp(getsortkey(a),getsortkey(b))); 226 | } 227 | 228 | var tblselect = new Set(); 229 | var tblselect0 = new Set(); 230 | var tbllastsel = null; 231 | 232 | function tbllock(item) { 233 | if (item) { 234 | item.locked = true; 235 | item.ele.dataset.lock = 'locked'; 236 | if (curedit && curedit[colmap['id']] == item.r[colmap['id']]) { 237 | app.$refs.editbox.readOnly = true; 238 | } 239 | } 240 | } 241 | 242 | function updatetblstyle() { 243 | let lastitem = null; 244 | tblitem.forEach((item) => { 245 | let tag = []; 246 | let state = item.r[colmap['state']]; 247 | let frame_start = item.r[colmap['frame_start']]; 248 | if (state == 'done') { 249 | if (!lastitem || frame_start != lastitem.r[colmap['frame_end']] + 1) { 250 | tag.push('primary'); 251 | } 252 | lastitem = item; 253 | } 254 | if (tblselect.has(item)) { 255 | tag.push('sel'); 256 | } 257 | if (curedit && item.r[colmap['id']] == curedit[colmap['id']]) { 258 | tag.push('cur'); 259 | } 260 | let tagstr = tag.join(' '); 261 | if (item.ele.dataset.tag != tagstr) { 262 | item.ele.dataset.tag = tagstr; 263 | } 264 | let lockstr = item.locked ? 'locked' : ''; 265 | if (item.ele.dataset.lock != lockstr) { 266 | item.ele.dataset.lock = lockstr; 267 | } 268 | }); 269 | } 270 | function updateselect() { 271 | tblselect.forEach((item) => { 272 | if (!tblview.has(item)) { 273 | tblselect.delete(item); 274 | } 275 | }); 276 | tblselect0.forEach((item) => { 277 | if (!tblview.has(item)) { 278 | tblselect0.delete(item); 279 | } 280 | }); 281 | if (!tblview.has(tbllastsel)) { 282 | tbllastsel = null; 283 | } 284 | 285 | updatetbledit(); 286 | updatetblstyle(); 287 | 288 | if (tblselect.size == 0) { 289 | app.selinfo = ''; 290 | app.selinfo2 = '空'; 291 | } else if (tblselect.size == id2item.size) { 292 | app.selinfo = '全部'; 293 | app.selinfo2 = ' (共'+id2item.size+'个)'; 294 | } else { 295 | app.selinfo = tblselect.size.toString(); 296 | app.selinfo2 = ' / ' + id2item.size.toString(); 297 | } 298 | if (id2item.size != tblview.size) { 299 | app.tblinfo = '筛选后显示了'+tblview.size+'条'; 300 | app.tblinfo2 = '（共'+id2item.size+'条字幕）'; 301 | } else { 302 | app.tblinfo = '共'+id2item.size+'条字幕'; 303 | app.tblinfo2 = ''; 304 | } 305 | document.getElementById('tblnoresult').style.display = tblview.size ? 'none' : ''; 306 | 307 | app.redrawtimebar(); 308 | } 309 | function tblclick(e) { 310 | let item = this.myitem; 311 | e.stopPropagation(); 312 | e.preventDefault(); 313 | 314 | if (e.shiftKey && tbllastsel) { 315 | tblselect = new Set(tblselect0); 316 | let flag = 0; 317 | tblitem.forEach((curitem) => { 318 | if (curitem === tbllastsel) flag ^= 1; 319 | if (curitem === item) flag ^= 1; 320 | if (flag || curitem === tbllastsel || curitem === item) { 321 | tblselect.add(curitem); 322 | } 323 | }); 324 | } else { 325 | if (e.ctrlKey) { 326 | if (tblselect.has(item)) { 327 | tblselect.delete(item); 328 | } else { 329 | tblselect.add(item); 330 | } 331 | } else { 332 | tblselect.clear(); 333 | tblselect.add(item); 334 | } 335 | tbllastsel = item; 336 | tblselect0 = new Set(tblselect); 337 | } 338 | updateselect(); 339 | if (tblselect.has(item)) { 340 | tbledit(item); 341 | } 342 | } 343 | function tblsel(list) { 344 | tblselect = new Set(list); 345 | tblselect0 = new Set(list); 346 | tbllastsel = list.length == 1 ? list[0] : null; 347 | updateselect(); 348 | } 349 | function tblselnone() { 350 | tblsel([]); 351 | } 352 | function tblselall() { 353 | tblsel(tblitem); 354 | } 355 | function removeelement(item) { 356 | if (item.ele) { 357 | item.ele.remove(); 358 | item.ele = null; 359 | item.tblidx = -1; 360 | } 361 | } 362 | function updateelement(item, lastele, rebuildtable) { 363 | if (item.ele === null) { 364 | item.ele = document.createElement('tr'); 365 | item.ele.myitem = item; 366 | for (let i = 0; i < 5; i++) { 367 | item.ele.appendChild(document.createElement('td')); 368 | } 369 | item.ele.addEventListener('click', tblclick); 370 | item.ele.addEventListener('mousedown', function (e) { 371 | if (e.shiftKey) { 372 | e.preventDefault(); 373 | } 374 | }); 375 | item.ele.dataset.sel = ''; 376 | if (!rebuildtable) { 377 | if (lastele === null) { 378 | tableele.tBodies[0].prepend(item.ele); 379 | } else { 380 | lastele.after(item.ele); 381 | } 382 | } 383 | } 384 | let row = ziprow(ocrresult.col, item.r); 385 | item.ele.dataset.state = row.state; 386 | item.ele.dataset.empty = row.ocrtext == '' ? 'empty' : ''; 387 | item.ele.dataset.position = row.position.toString(); 388 | item.ele.children[0].title = sec2str(row.frame_start / app.info.fps); 389 | item.ele.children[0].innerText = row.frame_start + '-' + row.frame_end; 390 | item.ele.children[1].innerText = row.top + '-' + row.bottom; 391 | item.ele.children[2].innerText = row.engine; 392 | item.ele.children[3].textContent = row.state == 'done' ? (row.ocrtext == '' ? '(空)' : row.ocrtext.replaceAll('\n','\\n\n')) : 393 | row.state == 'waitocr' ? '等待OCR' : 394 | row.state == 'error' ? 'OCR失败：' + row.ocrtext : 395 | row.state == 'merged' ? row.comment : row.state; 396 | if (row.file_id) { 397 | let img = document.createElement('div'); 398 | img.classList.add('img'); 399 | img.style.backgroundImage = 'url(/file?id=' + row.file_id + ')'; 400 | img.style.height = row.imgdb_h+'px'; 401 | img.style.width = row.imgdb_w+'px'; 402 | img.style.backgroundPositionX = '-'+row.imgdb_l+'px'; 403 | img.style.backgroundPositionY = '-'+row.imgdb_t+'px'; 404 | item.ele.children[4].innerHTML = ''; 405 | item.ele.children[4].appendChild(img); 406 | } 407 | } 408 | function updateview(needdelete, needupdate) { 409 | let fullupdate = (needdelete === undefined && needupdate === undefined); 410 | if (fullupdate) { 411 | tblview.forEach((item) => removeelement(item)); 412 | tblview.clear(); 413 | needdelete = new Set(); 414 | needupdate = new Set(id2item.values()); 415 | app.tblflterr = false; 416 | } 417 | let tbllast = tblview.size; 418 | let tblchange = 0; 419 | needupdate.forEach((item) => { 420 | if (viewopt.filter(item)) { 421 | tblview.add(item); 422 | item.sortkey = viewopt.sortkey(item); 423 | if (!item.ele) { 424 | tblchange++; 425 | } 426 | } else { 427 | needdelete.add(item); 428 | } 429 | }); 430 | needdelete.forEach((item) => { 431 | tblview.delete(item); 432 | needupdate.delete(item); 433 | if (item.ele) { 434 | tblchange++; 435 | } 436 | removeelement(item); 437 | }); 438 | 439 | tblitem = Array.from(tblview.values()); 440 | tblitem.sort((a, b) => array_cmp(a.sortkey, b.sortkey)); 441 | 442 | let rebuildtable = false; 443 | let lasttblidx = -1; 444 | for (let item of tblitem) { 445 | if (item.tblidx >= 0) { 446 | if (item.tblidx < lasttblidx) { 447 | rebuildtable = true; 448 | break; 449 | } 450 | lasttblidx = item.tblidx; 451 | } 452 | } 453 | 454 | if (navigator.userAgent.toLowerCase().indexOf('firefox') > -1) { 455 | rebuildtable = rebuildtable || tblchange > Math.sqrt(tbllast); 456 | } 457 | 458 | tableele = document.getElementById('ocrresult'); 459 | 460 | let lastele = null; 461 | tblitem.forEach((item, i) => { 462 | item.tblidx = i; 463 | if (needupdate.has(item)) { 464 | updateelement(item, lastele, rebuildtable); 465 | } 466 | lastele = item.ele; 467 | }); 468 | 469 | if (rebuildtable) { 470 | tableele.tBodies[0].innerHTML = ''; 471 | tblitem.forEach((item) => tableele.tBodies[0].appendChild(item.ele)); 472 | } 473 | 474 | updateselect(); 475 | } 476 | function updateresult(jsonstr) { 477 | let needdelete = new Set(); 478 | let needupdate = new Set(); 479 | 480 | if (lastresult !== jsonstr) { 481 | ocrresult = JSON.parse(lastresult = jsonstr); 482 | 483 | ocrresult.col.forEach((k, i) => colmap[k] = i); 484 | if (viewopt.filter === null || viewopt.sortkey === null) { 485 | setviewopt(app.tblflt, app.tblsort); 486 | } 487 | 488 | let delmap = new Map(id2item); 489 | ocrresult.row.forEach((r, i) => delmap.delete(r[colmap['id']])); 490 | delmap.forEach((item, id) => id2item.delete(id)); 491 | needdelete = new Set(delmap.values()); 492 | 493 | for (let r of ocrresult.row) { 494 | let id = r[colmap['id']]; 495 | let item = id2item.get(id); 496 | if (item === undefined) { 497 | item = { ele: null, sortkey: null, tblidx: -1, locked: false, mergedto: null, r: r }; 498 | id2item.set(id, item); 499 | } else { 500 | item.locked = false; 501 | if (array_equal(item.r, r)) { 502 | continue; 503 | } 504 | item.mergedto = null; 505 | item.r = r; 506 | } 507 | needupdate.add(item); 508 | } 509 | } else { 510 | for (let r of ocrresult.row) { 511 | let id = r[colmap['id']]; 512 | let item = id2item.get(id); 513 | if (item !== undefined) { 514 | item.locked = false; 515 | } 516 | } 517 | } 518 | 519 | let changed = needdelete.size > 0 || needupdate.size > 0; 520 | 521 | updateview(needdelete, needupdate); // may modify needdelete, needupdate 522 | 523 | return changed; 524 | } 525 | 526 | 527 | /////////////////////////////////////////////////////////////////////////////// 528 | 529 | Vue.component('imgdb', { 530 | props: ['src'], 531 | data() { 532 | return { 533 | imgloaded: false, 534 | }; 535 | }, 536 | template: 537 | `
538 | 539 |
`, 540 | watch: { 541 | src: { 542 | deep: true, 543 | handler(newsrc, oldsrc) { 544 | if (newsrc.hasOwnProperty('file_id') && oldsrc.hasOwnProperty('file_id')) { 545 | if (newsrc.file_id === oldsrc.file_id) { 546 | if (this.imgloaded) { 547 | this.$emit('my-load'); 548 | return; 549 | } 550 | } 551 | } 552 | this.imgloaded = false; 553 | }, 554 | }, 555 | }, 556 | methods: { 557 | imgload() { 558 | if (!this.imgloaded) { 559 | this.imgloaded = true; 560 | this.$emit('my-load'); 561 | } 562 | }, 563 | }, 564 | }) 565 | 566 | var app; 567 | app = new Vue({ 568 | el: '#app', 569 | data: { 570 | loaded: 0, 571 | info: {file: '', width: 100, height: 100, thumb_w: 100, thumb_h: 100, fps: 0, nframes: 100}, 572 | thumbnail: {col:[],row:[]}, 573 | 574 | mousexy: [-1, -1], // frame image 575 | mousexydown: 0, 576 | ocrsel_origin: -1, 577 | ocrsel: [-1, -1], 578 | ocrselsv: [-1, -1], 579 | ocrselmode: 0, 580 | framescale: 1, 581 | 582 | pos: -1, 583 | postitle: '', 584 | mousepos: -1, mouseposdown: 0, // timebar 585 | pixeldata: null, 586 | 587 | editorup: 0, 588 | editboxundoinfo: null, 589 | editboxredoinfo: null, 590 | editorfontsize: 20, 591 | adveditor: 0, 592 | adveditcm: null, 593 | 594 | tblsortfunc: tblsortfunc, 595 | tblsort: tblsortfunc[0][1], 596 | tblfltfunc: tblfltfunc, 597 | tblflt: tblfltfunc[0][1], 598 | tblfltedit: tblfltfunc[0][1], 599 | tblflterr: false, 600 | selinfo: '', 601 | selinfo2: '', 602 | tblinfo: '', 603 | tblinfo2: '', 604 | 605 | logs: {col:[],row:[]}, 606 | checkpointonly: 0, 607 | morelog: 0, 608 | waitstatus: 0, 609 | noautoscrollstatus: 0, 610 | pageerror: 0, 611 | statetimer: null, 612 | rerefresh: false, 613 | 614 | ocrconfig: { engine: '', top: -1, bottom: -1 }, 615 | engines: new Map(), 616 | 617 | codeeditor: 0, 618 | codeedit: null, 619 | myscript: { lastid: 0, scripts: [] }, 620 | scriptuploadtimer: null, 621 | scriptsel: null, 622 | scriptsel2: null, 623 | scripttemplate: `// 自定义脚本 624 | function (items) { 625 | for (let item of items) { 626 | ` + coldesc.map((d) => ' // item.' + d[0] + ' '.repeat(15 - d[0].length) + d[1]).join('\n') + 627 | ` 628 | } 629 | //return '处理完毕'; 630 | } 631 | `, 632 | defaultscripts: [ 633 | { name: '合并相同', locked: true, value: `// 合并相同：合并相邻且文字完全相同的字幕 634 | function (items) { 635 | if (items.length < 2) { 636 | alert('请选择多条字幕'); 637 | return '没有合并字幕'; 638 | } 639 | let nmerged = 0; 640 | let lastitem = null; 641 | for (let item of items) { 642 | if (lastitem != null) { 643 | if (item.state == 'done' && lastitem.state == 'done') { // 都是已OCR字幕 644 | if (item.frame_start - 1 == lastitem.frame_end) { // 相邻 645 | if (item.ocrtext == lastitem.ocrtext) { // 文字完全相同 646 | lastitem.frame_end = item.frame_end; 647 | item.state = 'merged'; 648 | item.comment = '已合并到上一字幕'; 649 | nmerged++; 650 | continue; 651 | } 652 | } 653 | } 654 | } 655 | if (item.state != 'merged') { 656 | lastitem = item; 657 | } 658 | } 659 | return '合并了' + nmerged + '条字幕'; 660 | } 661 | ` }, 662 | { name: '强制合并相邻', locked: true, value: `// 强制合并相邻：合并相邻字幕，不论文字是否相同 663 | function (items) { 664 | if (items.length < 2) { 665 | alert('请选择多条字幕'); 666 | return '没有合并字幕'; 667 | } 668 | if (confirm('确定要强制合并相邻字幕吗？')) { 669 | let nmerged = 0; 670 | let lastitem = null; 671 | for (let item of items) { 672 | if (lastitem != null) { 673 | if (item.frame_start - 1 == lastitem.frame_end) { // 相邻 674 | lastitem.frame_end = item.frame_end; 675 | item.state = 'merged'; 676 | item.comment = '已合并到上一字幕'; 677 | nmerged++; 678 | continue; 679 | } 680 | } 681 | if (item.state == 'done') { 682 | lastitem = item; 683 | } 684 | } 685 | return '强制合并了' + nmerged + '条相邻字幕'; 686 | } else { 687 | return '没有合并字幕'; 688 | } 689 | } 690 | ` }, 691 | { name: '强制合并', locked: true, value: `// 强制合并：合并字幕，不论文字是否相同，也不论是否相邻 692 | function (items) { 693 | if (items.length < 2) { 694 | alert('请选择多条字幕'); 695 | return '没有合并字幕'; 696 | } 697 | if (confirm('确定要强制合并吗？')) { 698 | let nmerged = 0; 699 | let lastitem = null; 700 | for (let item of items) { 701 | if (lastitem != null) { 702 | lastitem.frame_end = item.frame_end; 703 | item.state = 'merged'; 704 | item.comment = '已合并到上一字幕'; 705 | nmerged++; 706 | continue; 707 | } 708 | if (item.state == 'done') { 709 | lastitem = item; 710 | } 711 | } 712 | return '强制合并了' + nmerged + '条字幕'; 713 | } else { 714 | return '没有合并字幕'; 715 | } 716 | } 717 | ` }, 718 | { name: '清理', locked: true, value: `// 清理：删除空字幕、被合并字幕 719 | function (items) { 720 | if (confirm('确定要删除空字幕、被合并字幕吗？\\n提示：是否已检查“假空项”（有文字但被误识别为无文字）？')) { 721 | let nempty = 0, nmerged = 0; 722 | for (let item of items) { 723 | // 删除空字幕 724 | if (item.state == 'done' && item.ocrtext == '') { 725 | item.state = 'delete'; 726 | nempty++; 727 | } 728 | // 删除被合并字幕 729 | if (item.state == 'merged') { 730 | item.state = 'delete'; 731 | nmerged++; 732 | } 733 | } 734 | return '清理了' + nempty + '条空字幕，' + nmerged + '条被合并字幕'; 735 | } 736 | return '没有清理字幕'; 737 | } 738 | ` }, 739 | { name: '删除', locked: true, value: `// 删除：删除操作范围内的所有字幕 740 | function (items) { 741 | let n = items.length; 742 | if (confirm('确定要删除'+ n +'条字幕吗？')) { 743 | for (let item of items) { 744 | item.state = 'delete'; 745 | } 746 | return '删除了' + n + '条字幕'; 747 | } 748 | return '没有删除字幕'; 749 | } 750 | ` }, 751 | { name: '克隆', locked: true, value: `// 克隆：克隆一份选中的字幕 752 | function (items) { 753 | let n = items.length; 754 | if (confirm('确定要克隆'+ n +'条字幕吗？')) { 755 | for (let i = 0; i < n; i++) { 756 | let newitem = JSON.parse(JSON.stringify(items[i])); 757 | newitem.id = 0; 758 | items.push(newitem); 759 | } 760 | return '克隆了'+ n + '条字幕'; 761 | } 762 | return '没有克隆字幕'; 763 | } 764 | ` }, 765 | ], 766 | 767 | myprompt: null, 768 | }, 769 | watch: { 770 | pos(newpos) { 771 | //updateselect(); 772 | [this.$refs.prevframe, this.$refs.curframe, this.$refs.nextframe].forEach((ref) => { 773 | ref.style.color = 'gray'; 774 | ref.style.textDecoration = 'line-through'; 775 | }); 776 | this.postitle = sec2str(newpos / this.info.fps); 777 | }, 778 | editorfontsize() { this.saveui(); }, 779 | editorup() { this.saveui(); }, 780 | tblsort(newsort, oldsort) { 781 | if (newsort !== oldsort) { 782 | setviewopt(null, newsort); 783 | } 784 | }, 785 | tblflt(newflt, oldflt) { 786 | if (newflt != oldflt) { 787 | this.tblfltedit = newflt; 788 | setviewopt(newflt, null); 789 | } 790 | }, 791 | }, 792 | async mounted() { 793 | try { 794 | let session = (await axios.post('/session')).data; 795 | axios.defaults.headers.common['X-VIDEO2SUB-SESSION'] = session; 796 | this.info = (await axios.post('/info')).data; 797 | document.title += '：' + this.info.file; 798 | this.ocrconfig = await this.loadconfig('OCR') 799 | this.setocrsel(0); 800 | this.engines = new Map((await axios.post('/allengines')).data); 801 | await this.loadui(); 802 | this.myscript = await this.loadconfig('SCRIPT', { lastid: 0, scripts: [] }); 803 | this.myscript.scripts = this.defaultscripts.concat(this.myscript.scripts.filter((s)=>!s.locked)); 804 | this.scriptsel = this.myscript.scripts.slice(-1)[0]; 805 | this.pos = 0; 806 | tbledit(null); 807 | window.onmouseup = (e) => this.appmouseup(e); 808 | window.onmousemove = (e) => this.appmousemove(e); 809 | window.addEventListener("resize", this.appresize); 810 | this.codeedit = CodeMirror(this.$refs.codemirror, { 811 | value: '', 812 | mode: 'javascript', 813 | lineNumbers: true, 814 | tabSize: 2, 815 | }); 816 | this.adveditcm = CodeMirror(this.$refs.adveditbox, { 817 | value: '', 818 | mode: 'javascript', 819 | tabSize: 2, 820 | }); 821 | this.codeedit.on('change', () => this.savescript()); 822 | this.refresh(); 823 | } catch (e) { 824 | console.log(e); 825 | this.$refs.loading.innerHTML = '加载时发生错误，请重新启动本程序！！！'; 826 | } 827 | }, 828 | methods: { 829 | ziprowidx(dbresult, idx) { 830 | return ziprow(dbresult.col, dbresult.row[idx]); 831 | }, 832 | ziprowarr(dbresult) { 833 | return dbresult.row.map((r) => ziprow(dbresult.col, r)); 834 | }, 835 | 836 | exportass() { 837 | if (Array.from(id2item.values()).some((item) => item.r[colmap['state']] != 'done' || item.r[colmap['ocrtext']] == '')) { 838 | if (!confirm('有待清理或待处理条目，确认要继续导出吗？')) { 839 | return; 840 | } 841 | } 842 | this.setwaitstatus(); 843 | this.refreshafter(axios.post('/exportass')); 844 | }, 845 | exportcsv() { 846 | this.setwaitstatus(); 847 | this.refreshafter(axios.post('/exportcsv')); 848 | }, 849 | importcsv(asnew) { 850 | let fileele = [this.$refs.csvfile0, this.$refs.csvfile1][asnew]; 851 | let fromdata = new FormData(); 852 | if (fileele.files.length > 0) { 853 | fromdata.append("csv", fileele.files[0]); 854 | fromdata.append('asnew', asnew); 855 | fromdata.append('checkpoint', asnew ? '执行“添加CSV”之前' : '执行“导入CSV”之前'); 856 | this.setwaitstatus(); 857 | this.refreshafter(axios.post('/importcsv', fromdata, { 858 | headers: { 859 | 'Content-Type': 'multipart/form-data' 860 | } 861 | })); 862 | fileele.value = ''; 863 | } 864 | }, 865 | 866 | setframescale(h) { this.framescale = h / this.info.height; this.saveui(); }, 867 | getdisph() { return Math.round(this.info.height*this.framescale); }, 868 | getdispw() { return Math.round(this.info.width*this.framescale); }, 869 | 870 | jumpframe(expr) { 871 | this.showmyprompt('请输入表达式：\n如“100”，“n+10”，“n-10”，“n+fps*10”', expr !== undefined ? expr : this.pos.toString(), (val) => { 872 | if (val !== null) { 873 | let next; 874 | try { 875 | next = Math.round(compile_expr([], ['n','fps'], val)(this.pos, this.info.fps)); 876 | if (isNaN(next)) { 877 | throw 'nan'; 878 | } 879 | } catch (e) { 880 | alert('求值时发生错误！'); 881 | this.$nextTick(function() { 882 | this.jumpframe(val); 883 | }); 884 | return; 885 | } 886 | this.setframepos(next, 1); 887 | } 888 | }); 889 | }, 890 | 891 | y2percent(y) { 892 | return y / this.info.height * 100 + '%'; 893 | }, 894 | framemouse(e) { 895 | let rect = this.$refs.frameevent.getBoundingClientRect(); 896 | return [ 897 | Math.max(0, Math.min(this.info.width - 1, Math.round((e.clientX - rect.left) / (rect.right - rect.left) * this.info.width))), 898 | Math.max(0, Math.min(this.info.height - 1, Math.round((e.clientY - rect.top) / (rect.bottom - rect.top) * this.info.height))) 899 | ]; 900 | }, 901 | framemousemove(e) { 902 | this.mousexy = this.framemouse(e); 903 | }, 904 | videomousemove(e) { 905 | if (this.mousexydown) { 906 | let y = this.framemouse(e)[1]; 907 | this.ocrsel = [this.ocrsel_origin, y].sort((a, b) => a - b); 908 | //console.log(this.ocrsel); 909 | } 910 | }, 911 | framemousedown(e) { 912 | this.mousexydown = 1; 913 | let y = this.framemouse(e)[1]; 914 | this.ocrsel_origin = y; 915 | this.ocrsel = [y, y]; 916 | }, 917 | framemouseup(e) { 918 | if (this.mousexydown) { 919 | this.mousexydown = 0; 920 | this.framemousemove(e); 921 | this.videomousemove(e); 922 | if (confirm(this.ocrselmode?'确定要修改选中的'+tblselect.size+'条字幕的OCR区域吗？':'确定要修改新项目的OCR区域吗？')) { 923 | this.setocrsel(-1, this.ocrsel); 924 | this.saveocrsel(); 925 | } else { 926 | this.setocrsel(-1); 927 | } 928 | } 929 | }, 930 | inputocrsel() { 931 | this.showmyprompt((this.ocrselmode?'修改选中的'+tblselect.size+'条字幕的OCR区域':'修改新项目的OCR区域') + '\n请输入OCR区域信息，格式为“Y1-Y2”', (this.ocrsel[0]>=0 ? this.ocrsel[0]+'-'+this.ocrsel[1] : ''), (val) => { 932 | if (val !== null) { 933 | let newsel = val.split('-').map((v) => parseInt(v, 10)); 934 | if (newsel.length == 2 && newsel[0] >= 0 && newsel[1] >= 0) { 935 | newsel.sort((a, b) => a - b); 936 | if (newsel[1] < this.info.height) { 937 | this.setocrsel(-1, newsel); 938 | this.saveocrsel(); 939 | return; 940 | } 941 | } 942 | alert('无效值'); 943 | } 944 | }) 945 | }, 946 | saveocrsel() { 947 | if (this.ocrselmode == 0) { 948 | this.ocrconfig.top = this.ocrsel[0]; 949 | this.ocrconfig.bottom = this.ocrsel[1]; 950 | this.saveconfig('OCR', this.ocrconfig, 'OCR区域已设定 (Y1='+this.ocrconfig.top+', Y2='+this.ocrconfig.bottom+')'); 951 | } else if (this.ocrselmode == 1) { 952 | let selrange = tblitem.filter((item) => tblselect.has(item)); 953 | let changes = selrange.map((item) => ziprow(ocrresult.col, item.r)); 954 | let top = this.ocrsel[0], bottom = this.ocrsel[1]; 955 | changes.forEach((item) => (item.top = top, item.bottom = bottom)); 956 | if (curedit) { 957 | curedit[colmap['top']] = top; 958 | curedit[colmap['bottom']] = bottom; 959 | } 960 | app.refreshafter(axios.post('/updateresult', { 961 | changes: selectcol(['id','top','bottom'], changes), 962 | checkpoint: '“修改'+selrange.length+'条字幕的OCR范围”之前', 963 | message: '已修改'+selrange.length+'条字幕的OCR范围', 964 | compatlog: true, 965 | })); 966 | selrange.forEach((item) => tbllock(item)); 967 | } 968 | }, 969 | setocrsel(mode, newsel) { 970 | if (mode == 1) { 971 | this.ocrsel = [newsel[0], newsel[1]]; 972 | this.ocrselmode = 1; 973 | } else if (mode == 0) { 974 | this.ocrsel = [this.ocrconfig.top, this.ocrconfig.bottom]; 975 | this.ocrselmode = 0; 976 | } else { 977 | if (newsel) { 978 | this.ocrsel = [newsel[0], newsel[1]]; 979 | } else { 980 | this.ocrsel = [this.ocrselsv[0], this.ocrselsv[1]]; 981 | } 982 | } 983 | this.ocrselsv = [this.ocrsel[0], this.ocrsel[1]]; 984 | }, 985 | 986 | 987 | imgload(ref) { 988 | ref.style.color = 'black'; 989 | ref.style.textDecoration = ''; 990 | }, 991 | 992 | seteditboxvalinternal(s) { 993 | this.$refs.editbox.value = ''; 994 | this.$refs.editbox.value = s; 995 | }, 996 | geteditboxval() { 997 | return this.$refs.editbox.value.replaceAll('\\n', '\n'); 998 | }, 999 | seteditboxval(s) { 1000 | this.seteditboxvalinternal(s.replaceAll('\n', '\\n')); 1001 | this.$refs.editbox.setSelectionRange(0, 0); 1002 | this.editboxsavehistory(-1, -1); 1003 | }, 1004 | editboxselinfo() { 1005 | let v = this.$refs.editbox.value; 1006 | let st = this.$refs.editbox.selectionStart; 1007 | let ed = this.$refs.editbox.selectionEnd; 1008 | return [v, st, ed]; 1009 | }, 1010 | editboxsavehistory(undo, redo) { 1011 | this.editboxundoinfo = undo > 0 ? this.editboxselinfo() : (undo < 0 ? null : [null,null,null]); 1012 | this.editboxredoinfo = redo > 0 ? this.editboxselinfo() : (redo < 0 ? null : [null,null,null]); 1013 | }, 1014 | editboxundo(e) { 1015 | if (this.editboxundoinfo) { 1016 | let [v, st, ed] = this.editboxundoinfo; 1017 | if (v !== null) { 1018 | this.editboxsavehistory(0, 1); 1019 | this.seteditboxvalinternal(v); 1020 | this.$refs.editbox.selectionStart = st; 1021 | this.$refs.editbox.selectionEnd = ed; 1022 | } 1023 | e.preventDefault(); 1024 | } 1025 | }, 1026 | editboxredo(e) { 1027 | if (this.editboxredoinfo) { 1028 | let [v, st, ed] = this.editboxredoinfo; 1029 | if (v !== null) { 1030 | this.editboxsavehistory(1, 0); 1031 | this.seteditboxvalinternal(v); 1032 | this.$refs.editbox.selectionStart = st; 1033 | this.$refs.editbox.selectionEnd = ed; 1034 | } 1035 | e.preventDefault(); 1036 | } 1037 | }, 1038 | editboxinput(e) { 1039 | if (!e.isComposing) { 1040 | this.editboxsavehistory(-1, -1); 1041 | } 1042 | }, 1043 | editboxcut(e) { 1044 | let [v, st, ed] = this.editboxselinfo(); 1045 | if (v.substring(st, ed).length > 0) { 1046 | this.editboxsavehistory(1, 0); 1047 | e.clipboardData.setData('text/plain', v.substring(st, ed).replaceAll('\\n', '\n')); 1048 | this.seteditboxvalinternal(v.substring(0, st) + v.substring(ed, v.length)); 1049 | this.$refs.editbox.selectionStart = this.$refs.editbox.selectionEnd = st; 1050 | } 1051 | e.preventDefault(); 1052 | }, 1053 | editboxcopy(e) { 1054 | let [v, st, ed] = this.editboxselinfo(); 1055 | if (v.substring(st, ed).length > 0) { 1056 | e.clipboardData.setData('text/plain', v.substring(st, ed).replaceAll('\\n', '\n')); 1057 | } 1058 | e.preventDefault(); 1059 | }, 1060 | editboxpaste(e) { 1061 | let s = e.clipboardData.getData('text/plain').replaceAll('\n', '\\n'); 1062 | if (s.length > 0) { 1063 | this.editboxsavehistory(1, 0); 1064 | let [v, st, ed] = this.editboxselinfo(); 1065 | this.seteditboxvalinternal(v.substring(0, st) + s + v.substring(ed, v.length)); 1066 | this.$refs.editbox.selectionStart = this.$refs.editbox.selectionEnd = st + s.length; 1067 | } 1068 | e.preventDefault(); 1069 | }, 1070 | 1071 | findneighbor(curid) { 1072 | if (curid === undefined) { 1073 | if (!curedit) return [null, null, null]; 1074 | curid = curedit[colmap['id']]; 1075 | } 1076 | let curr = null; 1077 | let prev = null; 1078 | let next = null; 1079 | let multisel = tblselect.size > 1; 1080 | for (let i = 0; i < tblitem.length; i++) { 1081 | let item = tblitem[i]; 1082 | if (multisel && !tblselect.has(item)) { 1083 | continue; 1084 | } 1085 | if (item.r[colmap['id']] == curid) { 1086 | curr = item; 1087 | } 1088 | if (item.r[colmap['state']] == 'done') { 1089 | if (curr === null) { 1090 | prev = item; 1091 | } else if (curr !== item) { 1092 | next = item; 1093 | break; 1094 | } 1095 | } 1096 | } 1097 | return [prev, curr, next]; 1098 | }, 1099 | mergeneighbor(prev, curr, replace, nextcursor) { 1100 | if (prev && curr && !curr.locked) { 1101 | let item = ziprow(ocrresult.col, curr.r); 1102 | if (item.state != 'done') { 1103 | return; 1104 | } 1105 | let trueprev = prev; 1106 | while (trueprev.mergedto !== null) { 1107 | trueprev = trueprev.mergedto; 1108 | } 1109 | if (item.frame_start - 1 == prev.r[colmap['frame_end']]) { // 相邻 1110 | let lastitem = ziprow(ocrresult.col, trueprev.r); 1111 | lastitem.frame_end = item.frame_end; 1112 | if (replace) { 1113 | lastitem.ocrtext = this.geteditboxval(); 1114 | } 1115 | item.state = 'merged'; 1116 | item.comment = '已合并到上一字幕'; 1117 | if (nextcursor) tbledit(nextcursor, 1); 1118 | tbllock(prev); 1119 | tbllock(trueprev); 1120 | tbllock(curr, true); 1121 | curr.mergedto = trueprev; 1122 | updatetbledit(); 1123 | this.refreshafter(axios.post('/updateresult', { 1124 | changes: selectcol(['id','frame_end'].concat(replace ? ['ocrtext'] : []), [lastitem]).concat(selectcol(['id','state','comment'], [item])), 1125 | checkpoint: '“与上条合并：'+lastitem.ocrtext+'”之前', 1126 | message: '已合并到上一字幕：'+lastitem.ocrtext, 1127 | compatlog: true, 1128 | })); 1129 | //console.log(ziprow(ocrresult.col,trueprev.r), ziprow(ocrresult.col,prev.r), ziprow(ocrresult.col,curr.r)); 1130 | } else { 1131 | alert('起始帧与上条字幕结束帧不相邻，拒绝合并'); 1132 | } 1133 | } 1134 | }, 1135 | mergeprev() { 1136 | let [prev, curr, next] = this.findneighbor(); 1137 | this.mergeneighbor(prev, curr, false, next); 1138 | }, 1139 | replaceprev() { 1140 | let [prev, curr, next] = this.findneighbor(); 1141 | this.mergeneighbor(prev, curr, true, next); 1142 | }, 1143 | jumpprev() { 1144 | if (!curedit) { 1145 | tbledit(tblitem.find((item) => tblselect.has(item)), 1); 1146 | return; 1147 | } 1148 | let prev = this.findneighbor()[0]; 1149 | if (prev) { 1150 | tbledit(prev, 1); 1151 | } else { 1152 | savecur(); 1153 | } 1154 | }, 1155 | jumpnext() { 1156 | if (!curedit) { 1157 | tbledit(tblitem.find((item) => tblselect.has(item)), 1); 1158 | return; 1159 | } 1160 | let next = this.findneighbor()[2]; 1161 | if (next) { 1162 | tbledit(next, 1); 1163 | } else { 1164 | savecur(); 1165 | } 1166 | }, 1167 | editorkeydown(e) { 1168 | if (e.keyCode == 90 && e.ctrlKey) { 1169 | if (e.shiftKey) { 1170 | this.editboxredo(e); 1171 | } else { 1172 | this.editboxundo(e); 1173 | } 1174 | return; 1175 | } 1176 | if (e.keyCode == 38) { 1177 | this.jumpprev(); 1178 | } else if (e.keyCode == 40 || (e.keyCode == 13 && !e.ctrlKey)) { 1179 | this.jumpnext(); 1180 | } else if (e.keyCode == 13 && e.ctrlKey) { 1181 | if (e.shiftKey) { 1182 | this.replaceprev(); 1183 | } else { 1184 | this.mergeprev(); 1185 | } 1186 | } else if (e.key === "Escape") { 1187 | curedit = null; 1188 | tbledit(null); 1189 | } else { 1190 | return; 1191 | } 1192 | e.preventDefault(); 1193 | }, 1194 | editorwheel(e) { 1195 | if (e.deltaY > 0) { 1196 | this.jumpnext(); 1197 | } else if (e.deltaY < 0) { 1198 | this.jumpprev(); 1199 | } else { 1200 | return; 1201 | } 1202 | e.preventDefault(); 1203 | }, 1204 | jumppos(t) { 1205 | if (curedit) { 1206 | this.pos = [curedit[colmap['frame_start']], curedit[colmap['frame_end']]][t]; 1207 | } 1208 | }, 1209 | 1210 | advedit() { 1211 | if (curedit) { 1212 | //console.log(ziprow(ocrresult.col, curedit)); 1213 | this.adveditor = 1; 1214 | let lines = []; 1215 | for (let pair of coldesc) { 1216 | if (pair[0] == 'id') continue; 1217 | lines.push('') 1218 | lines.push(' // ' + pair[1]) 1219 | lines.push(' "' + pair[0] + '": ' + JSON.stringify(curedit[colmap[pair[0]]]) + ','); 1220 | } 1221 | this.$nextTick(function() { 1222 | this.adveditcm.refresh(); 1223 | }); 1224 | this.adveditcm.getDoc().setValue('{' + lines.join('\n') + '\n}\n'); 1225 | this.adveditcm.scrollTo(0, 0); 1226 | } 1227 | }, 1228 | adveditok() { 1229 | let changes; 1230 | try { 1231 | changes = compile_expr([], [], this.adveditcm.getDoc().getValue())(); 1232 | } catch (e) { 1233 | console.log(e); 1234 | alert('错误: ' + ('lineNumber' in e ? '('+e.lineNumber+'行 '+e.columnNumber+'列)' : '') + '\n' + e); 1235 | return; 1236 | } 1237 | 1238 | let newitem = { id: curedit[colmap['id']] }; 1239 | for (let [key, value] of Object.entries(changes)) { 1240 | if (key == 'id') continue; 1241 | newitem[key] = value; 1242 | } 1243 | this.refreshafter(axios.post('/updateresult', { 1244 | changes: [newitem], 1245 | checkpoint: '“高级修改：'+JSON.stringify(changes)+'”之前', 1246 | message: '已保存高级修改：'+JSON.stringify(changes), 1247 | compatlog: true, 1248 | })); 1249 | tbllock(id2item.get(newitem.id)); 1250 | this.adveditor = 0; 1251 | }, 1252 | adveditcancel() { 1253 | this.adveditor = 0; 1254 | }, 1255 | 1256 | 1257 | barpos(e) { 1258 | return Math.max(0, Math.min(this.info.nframes - 1, 1259 | Math.floor((e.clientX - this.$refs.barbg.getBoundingClientRect().left) / this.$refs.barbg.clientWidth * this.info.nframes))); 1260 | }, 1261 | setframepos(n, scrolltype) { 1262 | this.pos = Math.max(0, Math.min(this.info.nframes - 1, n)); 1263 | if (tblview.size > 0 && scrolltype) { 1264 | let tblsorted = sortedview(); 1265 | let target = tblsorted.findIndex((item) => item.r[colmap['frame_start']] >= n); 1266 | let item; 1267 | if (target < 0) { 1268 | item = tblsorted.slice(-1)[0]; 1269 | } else { 1270 | if (tblsorted[target].r[colmap['frame_start']] > n) { 1271 | target--; 1272 | } 1273 | item = target >= 0 ? tblsorted[target] : tblsorted[0]; 1274 | } 1275 | 1276 | if (scrolltype === 2) { 1277 | tbledit(null); 1278 | item.ele.scrollIntoView(true); 1279 | } else { 1280 | item.ele.scrollIntoView({block: "nearest", inline: "nearest"}); 1281 | } 1282 | } 1283 | }, 1284 | setpos(e) { 1285 | this.setframepos(this.barpos(e), 2); 1286 | }, 1287 | setmousepos(e) { 1288 | this.mousepos = this.barpos(e); 1289 | if (this.mouseposdown) this.setpos(e); 1290 | }, 1291 | timebarmouseup(e) { 1292 | if (this.mouseposdown) { 1293 | this.mouseposdown = 0; 1294 | let rect = this.$refs.barbg.getBoundingClientRect(); 1295 | if (e.clientY < rect.top || e.clientY >= rect.bottom) { 1296 | this.mousepos = -1; 1297 | } 1298 | } 1299 | }, 1300 | timebarmousemove(e) { 1301 | if (this.mouseposdown) { 1302 | this.mousepos = this.barpos(e); 1303 | this.setpos(e); 1304 | } 1305 | }, 1306 | pos2percent(n, leftmargin) { 1307 | let w = Math.floor(document.body.clientWidth); 1308 | let p = n / this.info.nframes * 100 + '%'; 1309 | if (leftmargin) { 1310 | return 'min('+p+','+(w-leftmargin)+'px)'; 1311 | } else { 1312 | return p; 1313 | } 1314 | }, 1315 | redrawtimebar() { 1316 | let w = Math.floor(document.body.clientWidth); 1317 | if (this.$refs.canvas.width != w) { 1318 | this.$refs.canvas.width = w; 1319 | this.$refs.canvas.style.width = w+'px'; 1320 | } 1321 | if (this.$refs.canvas.height != 1) { 1322 | this.$refs.canvas.height = 1; 1323 | } 1324 | let ctx = this.$refs.canvas.getContext('2d'); 1325 | if (this.pixeldata === null || this.pixeldata.width != w) { 1326 | this.pixeldata = ctx.createImageData(w, 1); 1327 | } 1328 | let buffer = this.pixeldata.data; 1329 | buffer.fill(0); 1330 | 1331 | let state_color = [ 1332 | null, 1333 | [255,255,255], // 1: not in view 1334 | [176,94,0], // 2: waitocr 1335 | [84,168,0], // 3: done (ocrtext != '') 1336 | [255,0,0], // 4: error 1337 | [0,123,255], // 5: selected 1338 | ]; 1339 | let f = w / this.info.nframes; 1340 | id2item.forEach((item) => { 1341 | let p = 0; 1342 | switch (item.r[colmap['state']]) { 1343 | case 'waitocr': p = 2; break; 1344 | case 'done': p = item.r[colmap['ocrtext']] != '' ? 3 : 0; break; 1345 | case 'error': p = 4; break; 1346 | } 1347 | if (p > 0 && !tblview.has(item)) { 1348 | p = 1; 1349 | } 1350 | if (tblselect.has(item)) { 1351 | p = 5; 1352 | } 1353 | if (p > 0) { 1354 | for (let i = item.r[colmap['frame_start']]; i <= item.r[colmap['frame_end']]; i++) { 1355 | let j = Math.floor(i * f); 1356 | buffer[j * 4] = Math.max(buffer[j * 4], p); 1357 | } 1358 | } 1359 | }); 1360 | for (let i = 0; i < w * 4; i += 4) { 1361 | let p = buffer[i]; 1362 | if (p) { 1363 | buffer[i] = state_color[p][0]; 1364 | buffer[i + 1] = state_color[p][1]; 1365 | buffer[i + 2] = state_color[p][2]; 1366 | buffer[i + 3] = 255; 1367 | } 1368 | } 1369 | ctx.putImageData(this.pixeldata, 0, 0); 1370 | }, 1371 | 1372 | tblfltkeydown(e) { 1373 | if (e.keyCode == 13) { 1374 | e.preventDefault(); 1375 | this.tblflt = this.tblfltedit; 1376 | } 1377 | }, 1378 | 1379 | tblselall() { 1380 | tblselall(); 1381 | }, 1382 | tblselnone() { tblselnone(); }, 1383 | 1384 | showcodeedit(show) { 1385 | this.codeeditor = show; 1386 | if (show) { 1387 | if (this.myscript.scripts.every((x) => x.locked)) { 1388 | this.newscript(); 1389 | } 1390 | this.changescript(); 1391 | this.$nextTick(function() { 1392 | this.codeedit.refresh(); 1393 | }); 1394 | } else { 1395 | this.savescript(true); 1396 | } 1397 | }, 1398 | changescript() { 1399 | if (this.scriptsel != null) { 1400 | this.codeedit.getDoc().setValue(this.scriptsel.value); 1401 | this.scriptsel2 = this.scriptsel; 1402 | this.codeedit.scrollTo(0, 0); 1403 | this.codeedit.setOption("readOnly", this.scriptsel.locked); 1404 | this.uploadscript(true); 1405 | } 1406 | }, 1407 | renamescript() { 1408 | if (this.scriptsel.locked) { 1409 | alert('不能修改此脚本！'); 1410 | return; 1411 | } 1412 | this.showmyprompt('请输入新的名字', this.scriptsel.name, (newname) => { 1413 | if (newname !== null) { 1414 | this.scriptsel.name = newname; 1415 | } 1416 | }); 1417 | }, 1418 | clonescript() { 1419 | if (!confirm('确定要克隆一份吗？')) { 1420 | return; 1421 | } 1422 | this.myscript.scripts.push(JSON.parse(JSON.stringify(this.scriptsel))); 1423 | this.scriptsel = this.myscript.scripts.slice(-1)[0]; 1424 | this.scriptsel.name += ' (副本)'; 1425 | this.scriptsel.locked = false; 1426 | this.changescript(); 1427 | }, 1428 | delscript() { 1429 | if (this.scriptsel.locked) { 1430 | alert('不能删除此脚本！'); 1431 | return; 1432 | } 1433 | if (!confirm('确定要删除此脚本吗？')) { 1434 | return; 1435 | } 1436 | let idx = this.myscript.scripts.indexOf(this.scriptsel); 1437 | this.myscript.scripts = this.myscript.scripts.filter((s) => s !== this.scriptsel); 1438 | this.scriptsel = this.myscript.scripts[idx] ? this.myscript.scripts[idx] : this.myscript.scripts.slice(-1)[0]; 1439 | this.changescript(); 1440 | }, 1441 | newscript() { 1442 | this.myscript.scripts.push({ 1443 | name: '我的脚本' + (++this.myscript.lastid), 1444 | locked: false, 1445 | value: this.scripttemplate, 1446 | }); 1447 | this.scriptsel = this.myscript.scripts.slice(-1)[0]; 1448 | this.changescript(); 1449 | }, 1450 | savescript(uploadnow) { 1451 | if (this.scriptsel2 === this.scriptsel) { 1452 | if (!this.scriptsel.locked) { 1453 | this.scriptsel.value = this.codeedit.getDoc().getValue(); 1454 | } 1455 | } 1456 | if (this.scriptuploadtimer !== null) { 1457 | clearTimeout(this.scriptuploadtimer); 1458 | this.scriptuploadtimer = null; 1459 | } 1460 | if (uploadnow) { 1461 | this.uploadscript(); 1462 | } else { 1463 | this.scriptuploadtimer = setTimeout(this.uploadscript, 1000); 1464 | } 1465 | }, 1466 | uploadscript() { 1467 | this.scriptuploadtimer = null; 1468 | this.saveconfig('SCRIPT', { 1469 | lastid: this.myscript.lastid, 1470 | scripts: this.myscript.scripts.filter((s)=>!s.locked), 1471 | }, ''); 1472 | }, 1473 | checklock() { 1474 | return Array.from(tblselect).every((item) => !item.locked); 1475 | }, 1476 | checkselect(permissive) { 1477 | if (!this.checklock()) { 1478 | return false; 1479 | } 1480 | if (!tblitem.some((item) => tblselect.has(item))) { 1481 | if (permissive) { 1482 | if (!confirm('没有选择要处理的字幕，继续吗？')) { 1483 | return false; 1484 | } 1485 | } else { 1486 | alert('请先选中要处理的字幕'); 1487 | return false; 1488 | } 1489 | } 1490 | return true; 1491 | }, 1492 | runscript(s) { 1493 | if (!this.checklock()) { 1494 | return false; 1495 | } 1496 | let f; 1497 | let message; 1498 | try { 1499 | f = compile_func(['alert', 'confirm'], s.value); 1500 | } catch(e) { 1501 | console.log(e); 1502 | alert('语法错误: ' + ('lineNumber' in e ? '('+e.lineNumber+'行 '+e.columnNumber+'列)' : '') + '\n' + e); 1503 | return false; 1504 | } 1505 | let items = sortedview().filter((item) => tblselect.has(item)).map((item) => ziprow(ocrresult.col, item.r)); 1506 | let old = JSON.stringify(items, ocrresult.col); 1507 | try { 1508 | message = f(items); 1509 | } catch(e) { 1510 | console.log(e); 1511 | alert('运行时错误:\n' + e); 1512 | return false; 1513 | } 1514 | //console.log(old); 1515 | let changed = JSON.stringify(items, ocrresult.col) != old; 1516 | if (message !== '') this.setwaitstatus(); 1517 | this.refreshafter(axios.post('/updateresult', { 1518 | changes: changed ? items : [], 1519 | checkpoint: changed ? '运行脚本“'+s.name+'”之前' : '', 1520 | message: typeof message === 'string' ? message : ('执行脚本“'+s.name+'”成功' + (changed ? '' : '，但没有任何条目被修改')), 1521 | compatlog: s.locked&&changed, 1522 | })); 1523 | if (changed) { 1524 | items.forEach((item) => tbllock(id2item.get(item.id))); 1525 | } 1526 | updatetbledit(); 1527 | return true; 1528 | }, 1529 | 1530 | startocr() { 1531 | if (ocrresult.row.length == 0) { 1532 | this.setwaitstatus(); 1533 | this.refreshafter(axios.post('/startocr', { frame_range: Array.from(Array(this.info.nframes).keys()) })); 1534 | return; 1535 | } 1536 | this.showmyprompt('请输入OCR帧范围，格式为“起-终”', 0+'-'+(this.info.nframes-1), (val) => { 1537 | if (val !== null) { 1538 | let timesel = val.split('-').map((v) => parseInt(v, 10)); 1539 | if (timesel.length == 2 && timesel[0] >= 0 && timesel[1] >= 0) { 1540 | timesel.sort((a, b) => a - b); 1541 | if (timesel[0] >= 0 && timesel[1] < this.info.nframes) { 1542 | this.setwaitstatus(); 1543 | let frame_range = []; 1544 | for (let i = timesel[0]; i <= timesel[1]; i++) { 1545 | frame_range.push(i); 1546 | } 1547 | this.refreshafter(axios.post('/startocr', { frame_range: frame_range })); 1548 | return; 1549 | } 1550 | } 1551 | alert('无效值'); 1552 | } 1553 | }); 1554 | }, 1555 | continueocr(restarttype) { 1556 | let msg = '确定要对'; 1557 | let item_range = Array.from(tblselect).map((item) => item.r[colmap['id']]); 1558 | if (item_range.length == 0) { 1559 | msg += '上次未完成的任务'; 1560 | } else { 1561 | msg += '选中的' + item_range.length + '条字幕'; 1562 | } 1563 | msg += '执行“'; 1564 | msg += ({ '': '继续OCR', 'all': '重新OCR', 'empty': '空项OCR' })[restarttype]; 1565 | msg += '”操作吗？'; 1566 | if (!confirm(msg)) { 1567 | return; 1568 | } 1569 | this.refreshafter(axios.post('/continueocr', { item_range: item_range, restarttype: restarttype })); 1570 | }, 1571 | async stopocr() { 1572 | this.setwaitstatus(); 1573 | this.refreshafter(axios.post('/stopocr')); 1574 | }, 1575 | 1576 | createcheckpoint() { 1577 | this.showmyprompt('请输入备注（可选）：', '手动创建的恢复点', (msg) => { 1578 | if (msg !== null) { 1579 | this.checkpoint(msg); 1580 | } 1581 | }); 1582 | }, 1583 | async checkpoint(msg) { 1584 | this.setwaitstatus(); 1585 | this.refreshafter(axios.post('/checkpoint', { msg: msg })); 1586 | }, 1587 | async rollback(checkpoint_id) { 1588 | if (confirm('确定要恢复到 #'+checkpoint_id+' 吗？\n字幕数据将被替换为恢复点中的数据。')) { 1589 | curedit = null; 1590 | tbledit(null); 1591 | tblitem.forEach((item) => tbllock(item)); 1592 | this.checkpointonly = 0; 1593 | this.morelog = 0; 1594 | this.setwaitstatus(); 1595 | this.refreshafter(axios.post('/rollback', { checkpoint_id: checkpoint_id })); 1596 | } 1597 | }, 1598 | 1599 | async saveui() { 1600 | await this.saveconfig('UI', { 1601 | disp_h: this.getdisph(), 1602 | editorfontsize: this.editorfontsize, 1603 | editorup: this.editorup, 1604 | }, ''); 1605 | }, 1606 | async loadui() { 1607 | let uiconfig = await this.loadconfig('UI') 1608 | this.setframescale(uiconfig.disp_h !== undefined ? uiconfig.disp_h : 360); 1609 | this.editorfontsize = uiconfig.editorfontsize; 1610 | if (this.editorup !== undefined) this.editorup = uiconfig.editorup; 1611 | await this.saveui(); 1612 | }, 1613 | 1614 | appmouseup(e) { 1615 | this.framemouseup(e); 1616 | this.timebarmouseup(e); 1617 | }, 1618 | appmousemove(e) { 1619 | this.videomousemove(e); 1620 | this.timebarmousemove(e); 1621 | }, 1622 | appresize(e) { 1623 | this.redrawtimebar(); 1624 | }, 1625 | scrollstatus() { 1626 | if (!this.noautoscrollstatus) { 1627 | this.$nextTick(function() { 1628 | this.$refs.logs.scrollTop = this.$refs.logs.scrollHeight; 1629 | }); 1630 | } 1631 | }, 1632 | setwaitstatus() { 1633 | this.waitstatus = 1; 1634 | this.scrollstatus(); 1635 | }, 1636 | 1637 | async loadconfig(key, default_value) { 1638 | return (await axios.post('/loadconfig', {key:key, default_value:default_value})).data; 1639 | }, 1640 | async saveconfig(key, value, msg) { 1641 | if (msg != '') { 1642 | this.setwaitstatus(); 1643 | } 1644 | await axios.post('/saveconfig', {key:key, value:value, msg:msg}); 1645 | if (msg != '') { 1646 | this.refreshnow(); 1647 | } 1648 | }, 1649 | refreshafter(p) { 1650 | p.then(() => this.refreshnow()); 1651 | this.rerefresh = true; 1652 | }, 1653 | refreshnow() { 1654 | if (this.statetimer === null) { 1655 | this.rerefresh = true; 1656 | } else { 1657 | clearTimeout(this.statetimer); 1658 | this.refresh(); 1659 | } 1660 | }, 1661 | async refresh() { 1662 | this.rerefresh = false; 1663 | this.statetimer = null; 1664 | try { 1665 | if (this.thumbnail.row.length < this.info.nframes) { 1666 | this.thumbnail = Object.freeze((await axios.post('/thumbnail')).data); 1667 | //console.log(this.thumbnail.row.length, this.info.nframes); 1668 | } 1669 | 1670 | } catch (err) { 1671 | console.log(err); 1672 | } 1673 | try { 1674 | let newresultdata = (await axios.post('/loadresult', null, { transformResponse: []})).data; 1675 | if (!this.rerefresh) { 1676 | if (updateresult(newresultdata)) { 1677 | this.scrollstatus(); 1678 | } 1679 | } 1680 | 1681 | let lastline = JSON.stringify(this.logs.row.slice(-1)); 1682 | this.logs = Object.freeze((await axios.post('/logs')).data); 1683 | if (lastline != JSON.stringify(this.logs.row.slice(-1))) { 1684 | this.scrollstatus(); 1685 | } 1686 | this.waitstatus = 0; 1687 | } catch (err) { 1688 | console.log(err); 1689 | this.pageerror = 1; 1690 | this.scrollstatus(); 1691 | } 1692 | this.loaded = 1; 1693 | if (this.rerefresh) { 1694 | await this.refresh(); 1695 | return; 1696 | } 1697 | this.statetimer = setTimeout(this.refresh, 1000); 1698 | }, 1699 | 1700 | showmyprompt(message, value, callback) { 1701 | this.myprompt = { 1702 | message: message, 1703 | value: value, 1704 | callback: callback, 1705 | }; 1706 | this.$nextTick(function() { 1707 | this.$refs.promptbox.focus(); 1708 | this.$refs.promptbox.select(); 1709 | }); 1710 | }, 1711 | mypromptok() { 1712 | this.myprompt.callback(this.myprompt.value); 1713 | this.myprompt = null; 1714 | }, 1715 | mypromptcancel() { 1716 | this.myprompt.callback(null); 1717 | this.myprompt = null; 1718 | }, 1719 | mypromptkeydown(e) { 1720 | if (e.keyCode == 13) { 1721 | this.mypromptok(); 1722 | } else if (e.key === "Escape") { 1723 | this.mypromptcancel(); 1724 | } else { 1725 | return; 1726 | } 1727 | e.preventDefault(); 1728 | }, 1729 | }, 1730 | }); -------------------------------------------------------------------------------- /video2sub.py: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env python3 2 | import ast 3 | import flask 4 | from flask.json import jsonify 5 | import werkzeug 6 | import sys 7 | import sqlite3 8 | import json 9 | import os 10 | import time 11 | import base64 12 | import urllib.parse 13 | import urllib.request 14 | import traceback 15 | import zlib 16 | import secrets 17 | import functools 18 | import csv 19 | import io 20 | import math 21 | from threading import Thread 22 | from cv2 import cv2 # make VSCode happy 23 | import numpy as np 24 | if os.name == 'nt': 25 | import win32file 26 | class FileLock: 27 | def __init__(self, path): 28 | self.path = path 29 | self.handle = None 30 | self.shared() 31 | def exclusive(self): 32 | self.close() 33 | self.handle = win32file.CreateFile(self.path, win32file.GENERIC_READ, 0, None, win32file.OPEN_EXISTING, 0, None) 34 | def shared(self): 35 | self.close() 36 | self.handle = win32file.CreateFile(self.path, win32file.GENERIC_READ, win32file.FILE_SHARE_READ, None, win32file.OPEN_EXISTING, 0, None) 37 | def close(self): 38 | if self.handle is not None: 39 | self.handle.Close() 40 | self.handle = None 41 | else: 42 | import fcntl 43 | class FileLock: 44 | def __init__(self, path): 45 | self.fd = os.open(path, os.O_RDONLY) 46 | def exclusive(self): 47 | fcntl.flock(self.fd, fcntl.LOCK_EX | fcntl.LOCK_NB) 48 | def shared(self): 49 | fcntl.flock(self.fd, fcntl.LOCK_SH | fcntl.LOCK_NB) 50 | def close(self): 51 | os.close(self.fd) 52 | 53 | sys.stdout.reconfigure(errors='replace') 54 | sys.stderr.reconfigure(errors='replace') 55 | 56 | version = '1.0' 57 | dbver = 1 58 | 59 | if len(sys.argv) != 4: 60 | print('请使用launcher启动器') 61 | print('若要手动启动后端：') 62 | print(' %s [主机名] [端口名] [视频/数据库文件名]'%os.path.basename(sys.argv[0])) 63 | sys.exit(1) 64 | host = sys.argv[1] 65 | port = sys.argv[2] 66 | if sys.argv[3].lower().endswith('.v2s'): 67 | dbfile = sys.argv[3] 68 | if not os.path.isfile(dbfile): 69 | print('无法打开数据库文件') 70 | sys.exit(1) 71 | with sqlite3.connect(dbfile) as conn: 72 | video = os.path.join(os.path.dirname(dbfile), conn.cursor().execute("SELECT value FROM config WHERE key = 'file'").fetchone()[0]) 73 | else: 74 | video = sys.argv[3] 75 | dbfile = os.path.splitext(video)[0] + '.v2s' 76 | 77 | try: 78 | lock = FileLock(video) 79 | except Exception: 80 | print('无法打开文件:', video) 81 | sys.exit(1) 82 | try: 83 | lock.exclusive() 84 | lock.shared() 85 | except Exception: 86 | print('无法取得文件锁（文件被占用？）') 87 | sys.exit(1) 88 | 89 | gconfig = ast.literal_eval(open('config.txt', 'r', encoding='utf_8_sig').read()) 90 | 91 | mime = { 92 | '.jpg': 'image/jpeg', 93 | '.bmp': 'image/bmp', 94 | } 95 | 96 | class VideoReader: 97 | def __init__(self): 98 | self.file = video 99 | self.cap = cv2.VideoCapture(self.file) 100 | if not self.cap.isOpened(): 101 | raise Exception('无法打开视频：%s'%self.file) 102 | self.width = int(self.cap.get(cv2.CAP_PROP_FRAME_WIDTH)) 103 | self.height = int(self.cap.get(cv2.CAP_PROP_FRAME_HEIGHT)) 104 | self.fps = self.cap.get(cv2.CAP_PROP_FPS) 105 | self.nframes = int(self.cap.get(cv2.CAP_PROP_FRAME_COUNT)) 106 | self.logs = [] 107 | def read(self, frame): 108 | if self.cap.get(cv2.CAP_PROP_POS_FRAMES) != frame: 109 | self.cap.set(cv2.CAP_PROP_POS_FRAMES, frame) 110 | succ, img = self.cap.read() 111 | if succ: 112 | return img 113 | self.cap.release() 114 | self.cap = cv2.VideoCapture(self.file) 115 | if self.cap.isOpened(): 116 | self.cap.set(cv2.CAP_PROP_POS_FRAMES, frame) 117 | succ, img = self.cap.read() 118 | if succ: 119 | self.logs.append(('读取视频帧%d/%d时发生错误，已尝试恢复；请考虑先转码一遍视频'%(frame, self.nframes), 'E')) 120 | return img 121 | self.logs.append(('读取视频帧%d/%d时发生错误；请考虑先转码一遍视频'%(frame, self.nframes), 'E')) 122 | return np.zeros((self.height, self.width, 3), np.uint8) 123 | def close(self): 124 | self.cap.release() 125 | def writelog(self, db): 126 | if len(self.logs) > 0: 127 | for str, level in self.logs: 128 | log(str, level, db=db) 129 | self.logs = [] 130 | return True 131 | else: 132 | return False 133 | 134 | 135 | 136 | def connect_db(): 137 | return sqlite3.connect(dbfile, timeout=999999) 138 | 139 | cap = VideoReader() 140 | width = cap.width 141 | height = cap.height 142 | fps = cap.fps 143 | nframes = cap.nframes 144 | 145 | def fpsfix(fps): 146 | fpsdict = { 147 | '23.976': (24000,1001), 148 | '29.970': (30000,1001), 149 | '59.940': (60000,1001), 150 | } 151 | key = '%.3f'%fps 152 | if gconfig['fix_ntsc_fps'] and key in fpsdict: 153 | return fpsdict[key] 154 | else: 155 | return (fps, 1) 156 | 157 | def frame2sec(n, fps): 158 | a, b = fpsfix(fps) 159 | return n * b / a 160 | 161 | ############################################################################## 162 | 163 | conn = connect_db() 164 | c = conn.cursor() 165 | c.execute('CREATE TABLE IF NOT EXISTS config (key TEXT PRIMARY KEY, value)') 166 | 167 | c.execute("INSERT OR IGNORE INTO config VALUES ('create_date', datetime('now','localtime'))") 168 | c.execute("INSERT OR IGNORE INTO config VALUES ('create_ver', ?)", (version,)) 169 | 170 | c.execute("INSERT OR IGNORE INTO config VALUES ('db_ver', ?)", (dbver,)) 171 | file_dbver = c.execute("SELECT value FROM config WHERE key = 'db_ver'").fetchone()[0] 172 | if dbver < file_dbver: 173 | print("版本过低，请升级") 174 | sys.exit(1) 175 | if file_dbver < dbver: 176 | print("正在升级数据库") 177 | if file_dbver == 1: 178 | assert False 179 | 180 | c.execute("INSERT OR REPLACE INTO config VALUES ('lastopen_date',datetime('now','localtime'))") 181 | c.execute("INSERT OR REPLACE INTO config VALUES ('lastopen_ver', ?)", (version,)) 182 | 183 | c.execute("INSERT OR REPLACE INTO config VALUES ('file', ?)", (os.path.basename(video),)) 184 | 185 | c.executemany('INSERT OR IGNORE INTO config VALUES (?,?)', [(k, json.dumps(v)) for k, v in gconfig['default'].items()]) 186 | 187 | 188 | c.execute('CREATE TABLE IF NOT EXISTS filedb (file_id INTEGER PRIMARY KEY AUTOINCREMENT, format TEXT, data BLOB)') 189 | c.execute(''' 190 | CREATE TABLE IF NOT EXISTS imgdb ( 191 | imgdb_id INTEGER PRIMARY KEY AUTOINCREMENT, 192 | file_id INTEGER, 193 | imgdb_l INT, imgdb_t INT, imgdb_w INT, imgdb_h INT, 194 | FOREIGN KEY(file_id) REFERENCES filedb(file_id) 195 | )''') 196 | 197 | c.execute('CREATE TABLE IF NOT EXISTS thumbnail (frame_id INT UNIQUE, imgdb_id INT)') 198 | c.execute(''' 199 | CREATE TABLE IF NOT EXISTS ocrresult ( 200 | id INTEGER PRIMARY KEY AUTOINCREMENT, 201 | state TEXT, 202 | date TEXT, 203 | frame_start INT, 204 | frame_end INT, 205 | imgdb_id INT, 206 | engine TEXT, 207 | left INT, 208 | top INT, 209 | right INT, 210 | bottom INT, 211 | position INT, 212 | ocrtext TEXT, 213 | comment TEXT 214 | )''') 215 | c.execute('CREATE INDEX IF NOT EXISTS itemstate ON ocrresult (state)') 216 | 217 | c.execute('CREATE TABLE IF NOT EXISTS jobrange (id INTEGER PRIMARY KEY)') 218 | 219 | c.execute('CREATE TABLE IF NOT EXISTS logs (date TEXT, level CHAR(1), message TEXT, checkpoint_id INTEGER)') 220 | c.execute('CREATE TABLE IF NOT EXISTS checkpoint (checkpoint_id INTEGER PRIMARY KEY AUTOINCREMENT, dbver INT, date TEXT, message TEXT, data BLOB)') 221 | 222 | if file_dbver < dbver: 223 | if file_dbver == 1: 224 | assert False 225 | c.execute("INSERT OR REPLACE INTO config VALUES ('db_ver', ?)", (dbver,)) 226 | print("数据库升级完成") 227 | 228 | conn.commit() 229 | 230 | logid_start = c.execute('SELECT MAX(rowid) FROM logs').fetchone()[0] 231 | logid_start = logid_start if logid_start is not None else 0 232 | 233 | def getconfig(db, key): 234 | val = db.cursor().execute('SELECT value FROM config WHERE key = ?', (key,)).fetchone() 235 | return val[0] if val is not None else None 236 | def putconfig(db, key, val): 237 | db.cursor().execute('INSERT OR REPLACE INTO config VALUES (?, ?)', (key, val)) 238 | 239 | #levels: [E]rror [W]arning [I]nfo [S]uccess [C]heckpoint 240 | def log(str, level='I', checkpoint_id=None, db=None): 241 | #db = None 242 | if db != None: 243 | db.cursor().execute("INSERT INTO logs VALUES (datetime('now','localtime'), ?, ?, ?)", (level, str, checkpoint_id)) 244 | else: 245 | print(level, str) 246 | 247 | def thread_yield(): 248 | time.sleep(0.001) # 好像没用 249 | 250 | def addfile(fmt, blob, db): 251 | c = db.cursor() 252 | c.execute('INSERT INTO filedb (format, data) VALUES (?, ?)', (fmt, blob)) 253 | return c.lastrowid 254 | 255 | def addimages(imglist, db, fmt='.jpg', fmtparam=None): 256 | succ, blob = cv2.imencode(fmt, cv2.vconcat(imglist), fmtparam) 257 | assert succ 258 | file_id = addfile(fmt, blob, db) 259 | c = db.cursor() 260 | result = [] 261 | t = 0 262 | for img in imglist: 263 | l = 0 264 | h = img.shape[0] 265 | w = img.shape[1] 266 | c.execute('INSERT INTO imgdb (file_id, imgdb_l, imgdb_t, imgdb_w, imgdb_h) VALUES (?,?,?,?,?)', (file_id, l, t, w, h)) 267 | result.append(c.lastrowid) 268 | t += h 269 | return result 270 | 271 | def db2json(db, sql, *param): 272 | c = db.cursor() 273 | row = c.execute(sql, *param).fetchall() 274 | col = [x[0] for x in c.description] 275 | return flask.jsonify({ 276 | 'col': col, 277 | 'row': row, 278 | }) 279 | 280 | def checkpoint(message, generate_log=True): 281 | dump = zlib.compress( 282 | json.dumps( 283 | c.execute('SELECT * FROM ocrresult').fetchall(), 284 | ensure_ascii=False, separators=(',', ':')).encode(), 285 | level=1) 286 | c.execute("INSERT INTO checkpoint (dbver, date, message, data) VALUES (?, datetime('now','localtime'), ?, ?)", (dbver, message, dump)) 287 | checkpoint_id = c.lastrowid 288 | if generate_log: 289 | log('已创建恢复点 #%d %s'%(checkpoint_id, message), 'C', checkpoint_id=checkpoint_id, db=conn) 290 | print('checkpoint #%d (%s) is %.2fKB (zlib)' % (checkpoint_id, message, len(dump)/1024)) 291 | return checkpoint_id 292 | 293 | def rollback(checkpoint_id): 294 | checkpoint_dbver, msg, dump = c.execute('SELECT dbver, message, data FROM checkpoint WHERE checkpoint_id = ?', (checkpoint_id,)).fetchone() 295 | if checkpoint_dbver != dbver: 296 | return '版本不符', msg 297 | c.execute('DELETE FROM ocrresult') 298 | rows = json.loads(zlib.decompress(dump).decode()) 299 | if len(rows): 300 | c.executemany('INSERT INTO ocrresult VALUES (' + ','.join(['?']*len(rows[0])) + ')', rows) 301 | return None, msg 302 | 303 | ############################################################################## 304 | 305 | thumb_h = gconfig['thumbnail']['height'] 306 | thumb_w = int(width / height * thumb_h) 307 | thumb_npart = gconfig['thumbnail']['npart'] 308 | thumb_fmt = '.jpg' 309 | thumb_fmtparam = [int(cv2.IMWRITE_JPEG_QUALITY), gconfig['thumbnail']['jpg_quality']] 310 | 311 | subthumb_w = gconfig['subthumb']['width'] 312 | subthumb_fmt = '.jpg' 313 | subthumb_fmtparam = [int(cv2.IMWRITE_JPEG_QUALITY), gconfig['subthumb']['jpg_quality']] 314 | 315 | frame_fmt = '.bmp' 316 | frame_fmtparam = None 317 | 318 | 319 | ############################################################################## 320 | 321 | class ImageVConcat: 322 | def __init__(self, imglist): 323 | self.h, self.w, _ = imglist[0].shape 324 | pad = np.zeros((self.h, self.w, 3), np.uint8) 325 | self.hdict = {} 326 | self.n = len(imglist) 327 | cat = [] 328 | cath = 0 329 | for img in imglist: 330 | if len(cat): 331 | cat.append(pad) 332 | cath += self.h 333 | cat.append(img) 334 | for i in range(cath, cath + self.h): 335 | self.hdict[i] = [] 336 | cath += self.h 337 | self.img = cv2.vconcat(cat) 338 | assert self.img.shape[0] == cath 339 | #print(self.hdict) 340 | def addresult(self, left, top, right, bottom, ocrtext, comment): 341 | left, top, right, bottom = int(left), int(top), int(right), int(bottom) 342 | #print(left, top, right, bottom, ocrtext, comment) 343 | for i in range(top, bottom): 344 | if i in self.hdict: 345 | self.hdict[i].append((top, bottom, left, right, ocrtext, comment)) 346 | def getresult(self): 347 | result = [] 348 | for h0 in range(0, 2*self.h*self.n, 2*self.h): 349 | s = set() 350 | for i in range(h0, h0 + self.h): 351 | print(h0, self.hdict[i]) 352 | s.update(self.hdict[i]) 353 | ocrtext = '\n'.join([ocrtext for top, bottom, left, right, ocrtext, comment in sorted(s)]) 354 | comment = '#'.join([comment for top, bottom, left, right, ocrtext, comment in sorted(s)]) 355 | result.append(('done', ocrtext, comment)) 356 | return result 357 | 358 | ocr_stop = False 359 | 360 | 361 | class BaiduOcr: 362 | token = None 363 | lastreq = 0 364 | def __init__(self, arg): 365 | self.service = arg 366 | self.use_batch = not arg.endswith('_basic') 367 | self.ocr_batch = gconfig['baiduocr']['batch_size'] if self.use_batch else 1 368 | self.ninvoke = 0 369 | 370 | def fetch_token(self): 371 | if BaiduOcr.token is not None: 372 | return 373 | params = {'grant_type': 'client_credentials', 374 | 'client_id': gconfig['baiduocr']['API_KEY'], 375 | 'client_secret': gconfig['baiduocr']['SECRET_KEY']} 376 | post_data = urllib.parse.urlencode(params) 377 | post_data = post_data.encode('utf-8') 378 | req = urllib.request.Request(gconfig['baiduocr']['TOKEN_URL'], post_data) 379 | f = urllib.request.urlopen(req, timeout=5) 380 | result_str = f.read() 381 | 382 | result_str = result_str.decode() 383 | result = json.loads(result_str) 384 | if ('access_token' in result.keys() and 'scope' in result.keys()): 385 | if not 'brain_all_scope' in result['scope'].split(' '): 386 | raise Exception('please ensure has check the ability') 387 | BaiduOcr.token = result['access_token'] 388 | else: 389 | raise Exception('please overwrite the correct API_KEY and SECRET_KEY') 390 | def request(self, url, data): 391 | req = urllib.request.Request(url, data.encode('utf-8')) 392 | f = urllib.request.urlopen(req) 393 | result_str = f.read() 394 | result_str = result_str.decode() 395 | return result_str 396 | def run(self, imglist): 397 | mintimediff = 1 / gconfig['baiduocr']['qps_limit'] 398 | if self.use_batch: 399 | vcat = ImageVConcat(imglist) 400 | imglist = [vcat.img] 401 | else: 402 | results = [] 403 | for img in imglist: 404 | errmsg = None 405 | try: 406 | if time.time() - BaiduOcr.lastreq < mintimediff: 407 | time.sleep(mintimediff - (time.time() - BaiduOcr.lastreq)) 408 | self.fetch_token() 409 | url = gconfig['baiduocr']['OCR_URL'] + self.service + "?access_token=" + BaiduOcr.token 410 | #print(url) 411 | succ, blob = cv2.imencode('.jpg', img, [int(cv2.IMWRITE_JPEG_QUALITY), 99]) 412 | assert succ 413 | file_content = blob.tobytes() 414 | #with open('dump.jpg', 'wb') as f: 415 | # f.write(file_content) 416 | self.ninvoke += 1 417 | result = self.request(url, urllib.parse.urlencode({ 418 | 'image': base64.b64encode(file_content), 419 | 'language_type': gconfig['baiduocr']['language_type'], 420 | 'probability': 'true', 421 | })) 422 | result_json = json.loads(result) 423 | if 'error_msg' in result_json: 424 | errmsg = result_json['error_msg'] 425 | if "words_result" not in result_json: 426 | raise Exception(str(result_json)) 427 | print(str(result_json)) 428 | 429 | if self.use_batch: 430 | for words_result in result_json["words_result"]: 431 | left = words_result['location']['left'] 432 | top = words_result['location']['top'] 433 | right = left + words_result['location']['width'] 434 | bottom = top + words_result['location']['height'] 435 | vcat.addresult(left, top, right, bottom, words_result["words"], str(words_result)) 436 | results = vcat.getresult() 437 | else: 438 | results.append(('done', '\n'.join([words_result["words"] for words_result in result_json["words_result"]]), str(result_json))) 439 | except Exception as e: 440 | traceback.print_exc() 441 | if errmsg is None: 442 | errmsg = str(e) 443 | if self.use_batch: 444 | results = [('error', errmsg)] * vcat.n 445 | else: 446 | results.append(('error', errmsg)) 447 | BaiduOcr.lastreq = time.time() 448 | if ocr_stop: 449 | break 450 | return results 451 | def done(self): 452 | return '调用了%d次百度文字识别API(%s)'%(self.ninvoke,self.service) 453 | 454 | class ChineseOcr: 455 | ocr_batch = 5 456 | global ocr_stop 457 | def __init__(self, arg): 458 | if arg == 'multi': 459 | self.textLine = b'false' 460 | if arg == 'single': 461 | self.textLine = b'true' 462 | def run(self, imglist): 463 | results = [] 464 | for img in imglist: 465 | format = '.bmp' 466 | succ, blob = cv2.imencode(format, img) 467 | assert succ 468 | #with open('dump.bmp', 'wb') as f: 469 | # f.write(blob.tobytes()) 470 | try: 471 | data = b'{"imgString":"data:%s;base64,%s","textAngle":false,"textLine":%s}' % (mime[format].encode(), base64.b64encode(blob.tobytes()), self.textLine) 472 | req = urllib.request.urlopen(gconfig['chineseocr']['url'], data, timeout=5) 473 | rsp = json.load(req) 474 | result = ('done', '\n'.join([item['text'] for item in rsp['res']])) 475 | except Exception as e: 476 | traceback.print_exc() 477 | result = ('error', str(e)) 478 | results.append(result) 479 | if ocr_stop: 480 | break 481 | return results 482 | def done(self): 483 | return None 484 | 485 | class DummyOcr: 486 | ocr_batch = 30 487 | def run(self, imglist): 488 | if gconfig['dummyocr']['always_error']: 489 | return [('error', None)] * len(imglist) 490 | else: 491 | return [('done', gconfig['dummyocr']['text'])] * len(imglist) 492 | def done(self): 493 | return None 494 | 495 | def createocr(engine, h): 496 | if engine == 'dummyocr': 497 | return DummyOcr() 498 | if engine.startswith('chineseocr:'): 499 | return ChineseOcr(engine[len('chineseocr:'):]) 500 | if engine.startswith('baiduocr:'): 501 | return BaiduOcr(engine[len('baiduocr:'):]) 502 | assert False, "无效OCR引擎名"+engine 503 | 504 | def run_ocrjob(): 505 | global ocr_stop 506 | cap = VideoReader() 507 | conn = connect_db() 508 | c = conn.cursor() 509 | 510 | total = c.execute("SELECT COUNT(id) FROM ocrresult WHERE EXISTS (SELECT * FROM jobrange WHERE ocrresult.id = jobrange.id)").fetchone()[0] 511 | progress = c.execute("SELECT COUNT(id) FROM ocrresult WHERE EXISTS (SELECT * FROM jobrange WHERE ocrresult.id = jobrange.id) AND state != 'waitocr' AND state != 'error'").fetchone()[0] 512 | emptycnt = c.execute("SELECT COUNT(id) FROM ocrresult WHERE EXISTS (SELECT * FROM jobrange WHERE ocrresult.id = jobrange.id) AND state != 'waitocr' AND state != 'error' AND ocrtext == ''").fetchone()[0] 513 | errcnt = c.execute("SELECT COUNT(id) FROM ocrresult WHERE EXISTS (SELECT * FROM jobrange WHERE ocrresult.id = jobrange.id) AND state == 'error'").fetchone()[0] 514 | 515 | period_frames = 0 516 | period_start = 0 517 | 518 | curengine = '' 519 | ocr = None 520 | 521 | while True: 522 | lines = c.execute("SELECT id, engine, top, bottom, frame_start FROM ocrresult WHERE EXISTS (SELECT * FROM jobrange WHERE ocrresult.id = jobrange.id) AND state = 'waitocr' ORDER BY engine, top, frame_start").fetchall() 523 | if len(lines) == 0: 524 | break 525 | lines.reverse() 526 | while True: 527 | if ocr_stop: 528 | cap.writelog(conn) 529 | msg = ocr.done() 530 | if msg: 531 | log(msg, 'I', db=conn) 532 | log('OCR任务已暂停', 'W', db=conn) 533 | conn.commit() 534 | conn.close() 535 | cap.close() 536 | return 537 | # show status 538 | if len(lines) == 0 or (time.time() - period_start) > 5: 539 | speed = period_frames / (time.time() - period_start) 540 | log('OCR任务进度: %.1f%% [总共%d, 完成%d(空项%d), 错误%d, 剩余%d, fps=%.1f]' % (int((progress+errcnt)/total*1000)/10, total, progress, emptycnt, errcnt, total-progress-errcnt, speed), db=conn) 541 | conn.commit() 542 | period_frames = 0 543 | period_start = time.time() 544 | 545 | if len(lines) == 0: 546 | break 547 | 548 | # collect batch 549 | batch = [] 550 | _, engine0, top0, bottom0, _ = lines[-1] 551 | if engine0 != curengine: 552 | curengine = engine0 553 | if ocr: 554 | msg = ocr.done() 555 | if msg: 556 | log(msg, 'I', db=conn) 557 | ocr = createocr(engine0, bottom0+1 - top0) 558 | while len(lines) > 0 and len(batch) < ocr.ocr_batch: 559 | _, engine, top, bottom, _ = lines[-1] 560 | if (engine, top, bottom) == (engine0, top0, bottom0): 561 | batch.append(lines.pop()) 562 | else: 563 | break 564 | print(batch) 565 | 566 | # process batch 567 | parts = [] 568 | subthumbs = [] 569 | for id, engine, top, bottom, frame_id in batch: 570 | img = cap.read(frame_id) 571 | img = img[top:bottom+1] 572 | parts.append(img) 573 | h = int(subthumb_w * (bottom+1 - top) / width) 574 | img = cv2.resize(img, (subthumb_w, h), interpolation=cv2.INTER_AREA) 575 | subthumbs.append(img) 576 | 577 | result = ocr.run(parts) # [ (state, ocrtext, comment) 578 | print('\n'.join(['OCR[%d]: %s'%(batch[i][0], result[i] if i < len(result) else None) for i in range(len(batch))])) 579 | imgdb_ids = addimages(subthumbs, conn, subthumb_fmt, subthumb_fmtparam) 580 | for i, r in enumerate(result): 581 | assert type(r) is tuple 582 | c.execute('UPDATE ocrresult SET %s, imgdb_id = ? WHERE id = ?' % 583 | ', '.join(['state = ?', 'ocrtext = ?', 'comment = ?'][:len(r)]), 584 | r + (imgdb_ids[i], batch[i][0])) 585 | if r[0] == 'error': 586 | errcnt += 1 587 | else: 588 | if r[1] == '': 589 | emptycnt += 1 590 | progress += 1 591 | period_frames += len(result) 592 | conn.commit() 593 | thread_yield() 594 | 595 | cap.writelog(conn) 596 | msg = ocr.done() 597 | if msg: 598 | log(msg, 'I', db=conn) 599 | msg = 'OCR任务已完成(空项数%d)'%emptycnt if errcnt == 0 else 'OCR任务已完成(空项数%d)，但有%d个错误发生，请使用“继续OCR”功能来重试错误项'%(emptycnt,errcnt) 600 | log(msg, 'S', db=conn) 601 | if emptycnt: 602 | log('OCR结果中有空项，请注意处理', 'W', db=conn) 603 | conn.commit() 604 | conn.close() 605 | cap.close() 606 | 607 | ocr_thread = None 608 | def startocr(): 609 | global ocr_stop 610 | global ocr_thread 611 | if ocr_thread is not None: 612 | ocr_thread.join() 613 | ocr_stop = False 614 | ocr_thread = Thread(target=run_ocrjob) 615 | ocr_thread.start() 616 | 617 | ############################################################################## 618 | 619 | 620 | 621 | def make_thumbnail(): 622 | cap = VideoReader() 623 | conn = connect_db() 624 | c = conn.cursor() 625 | if getconfig(conn, 'thumb_h') != thumb_h: 626 | putconfig(conn, 'thumb_h', thumb_h) 627 | c.execute('DELETE FROM thumbnail') 628 | start = c.execute('SELECT MAX(frame_id) + 1 FROM thumbnail').fetchone()[0] 629 | start = 0 if start is None else start 630 | if start < nframes - 1: 631 | for part_start in range(start, nframes, thumb_npart): 632 | parts = [] 633 | frame_ids = [] 634 | for i in range(part_start, min(part_start + thumb_npart, nframes)): 635 | img = cap.read(i) 636 | img = cv2.resize(img, (thumb_w, thumb_h), interpolation=cv2.INTER_AREA) 637 | parts.append(img) 638 | frame_ids.append(i) 639 | imgdb_ids = addimages(parts, conn, thumb_fmt, thumb_fmtparam) 640 | c.executemany('INSERT INTO thumbnail (frame_id, imgdb_id) VALUES (?, ?)', list(zip(frame_ids, imgdb_ids))) 641 | if int(part_start / nframes * 100) != int((part_start - thumb_npart) / nframes * 100): 642 | log('生成缩略图: %d%%' % int(part_start / nframes * 100), db=conn) 643 | conn.commit() 644 | thread_yield() 645 | cap.writelog(conn) 646 | log('生成缩略图已完成', 'I', db=conn) 647 | conn.commit() 648 | conn.close() 649 | cap.close() 650 | 651 | def checkwaitocr(db): 652 | if db.cursor().execute("SELECT COUNT(id) FROM ocrresult WHERE state == 'waitocr'").fetchone()[0]: 653 | log('有未完成的OCR任务，可使用“继续OCR”功能继续上次的进度', 'W', db=db) 654 | def do_init(): 655 | conn = connect_db() 656 | c = conn.cursor() 657 | t = Thread(target=make_thumbnail) 658 | t.start() 659 | t.join() 660 | log('后端已启动', 'S', db=conn) 661 | checkwaitocr(db=conn) 662 | conn.commit() 663 | conn.close() 664 | 665 | init_thread = Thread(target=do_init) 666 | init_thread.start() 667 | 668 | 669 | 670 | ############################################################################## 671 | 672 | 673 | app = flask.Flask(__name__) 674 | session = None 675 | 676 | def session_header_required(f): 677 | @functools.wraps(f) 678 | def wrapper(*args, **kwds): 679 | if 'X-VIDEO2SUB-SESSION' not in flask.request.headers or flask.request.headers['X-VIDEO2SUB-SESSION'] != session: 680 | return flask.make_response('', 403) 681 | return f(*args, **kwds) 682 | return wrapper 683 | 684 | @app.route('/') 685 | def serve_home(): 686 | return flask.redirect('/ui/ui.html') 687 | 688 | @app.route('/ui/') 689 | def serve_ui(filename): 690 | return flask.send_from_directory('ui', filename) 691 | 692 | @app.route('/getpid', methods=['POST']) 693 | def serve_getpid(): 694 | return flask.jsonify(os.getpid()) 695 | 696 | @app.route('/session', methods=['POST']) 697 | def serve_session(): 698 | global session 699 | session = secrets.token_hex() 700 | return flask.jsonify(session) 701 | 702 | @app.route('/havesession', methods=['POST']) 703 | def serve_havesession(): 704 | global session 705 | return flask.jsonify(session is not None) 706 | 707 | @app.route('/info', methods=['POST']) 708 | @session_header_required 709 | def serve_info(): 710 | return flask.jsonify({ 711 | 'file': os.path.basename(video), 712 | 'width': width, 713 | 'height': height, 714 | 'fps': fps, 715 | 'nframes': nframes, 716 | 'thumb_w': thumb_w, 717 | 'thumb_h': thumb_h, 718 | }) 719 | 720 | @app.route('/file') 721 | def serve_img(): 722 | file_id = flask.request.args.get('id', type=int) 723 | row = c.execute('SELECT format, data FROM filedb WHERE file_id = ?', (file_id,)).fetchone() 724 | return flask.Response(row[1], mimetype=mime[row[0]]) 725 | 726 | @app.route('/frame') 727 | def serve_frame(): 728 | frame_id = flask.request.args.get('id', type=int) 729 | frame = cap.read(frame_id) 730 | if cap.writelog(conn): 731 | conn.commit() 732 | succ, blob = cv2.imencode(frame_fmt, frame, frame_fmtparam) 733 | assert succ 734 | return flask.Response(blob.tobytes(), mimetype=mime[frame_fmt]) 735 | 736 | @app.route('/thumbnail', methods=['POST']) 737 | @session_header_required 738 | def serve_thumbnail(): 739 | return db2json(conn, 'SELECT frame_id, imgdb.* FROM thumbnail JOIN imgdb USING (imgdb_id) ORDER BY frame_id') 740 | 741 | @app.route('/logs', methods=['POST']) 742 | @session_header_required 743 | def serve_logs(): 744 | return db2json(conn, ''' 745 | SELECT id, id > ? AS cursession, date, level, message, checkpoint_id FROM ( 746 | SELECT * FROM (SELECT ROWID AS id, * FROM logs WHERE checkpoint_id IS NULL ORDER BY ROWID DESC LIMIT ?) 747 | UNION ALL 748 | SELECT * FROM (SELECT ROWID AS id, * FROM logs WHERE checkpoint_id IS NOT NULL ORDER BY ROWID DESC LIMIT ?) 749 | ) ORDER BY id''', (logid_start, gconfig['max_log'], gconfig['max_checkpoint'])) 750 | 751 | @app.route('/state', methods=['POST']) 752 | @session_header_required 753 | def serve_state(): 754 | loaded = not init_thread.is_alive() 755 | ocractive = ocr_thread is not None and ocr_thread.is_alive() 756 | lastjob = c.execute("SELECT COUNT(id) FROM ocrresult WHERE state = 'waitocr' or state = 'error'").fetchone()[0] 757 | 758 | ocrconf = json.loads(getconfig(conn, 'OCR')) 759 | ocrtop, ocrbottom = ocrconf['top'], ocrconf['bottom'] 760 | if ocrtop >= 0 and ocrbottom >= 0: 761 | curarea = c.execute('SELECT COUNT(id) FROM ocrresult WHERE top = ? AND bottom = ?', (ocrtop, ocrbottom)).fetchone()[0] 762 | else: 763 | curarea = 0 764 | 765 | return flask.jsonify({ 766 | 'loaded': loaded, 767 | 'ocractive': ocractive, 768 | 'lastjob': lastjob, 769 | 'curarea': curarea, 770 | }) 771 | 772 | @app.route('/exportass', methods=['POST']) 773 | @session_header_required 774 | def serve_exportass(): 775 | row = c.execute("SELECT frame_start, frame_end, ocrtext, top, bottom, position FROM ocrresult WHERE state = 'done' AND ocrtext != '' ORDER BY frame_start, frame_end").fetchall() 776 | if len(row) == 0: 777 | log('无字幕数据', 'I', db=conn) 778 | conn.commit() 779 | return '' 780 | outfile = os.path.splitext(video)[0] + gconfig['export_suffix'] + '.ass' 781 | if os.path.exists(outfile) and not gconfig['export_overwrite']: 782 | log('输出文件已存在，请先删除：%s'%outfile, 'E', db=conn) 783 | conn.commit() 784 | return '' 785 | def frame2timestr(frame, fps, alignment): 786 | fn, shiftframe, eps, shift100 = alignment 787 | sec100 = getattr(math, fn)(frame2sec(frame + shiftframe, fps) * 100 + eps) + shift100 788 | sec100 = sec100 if sec100 >= 0 else 0 789 | fs = sec100 % 100 790 | s = sec100 // 100 % 60 791 | m = sec100 // 100 // 60 % 60 792 | h = sec100 // 100 // 60 // 60 793 | return '%d:%02d:%02d.%02d'%(h,m,s,fs) 794 | best_top = c.execute("SELECT top, COUNT(top) cnt FROM ocrresult WHERE state = 'done' AND ocrtext != '' AND position = 8 GROUP BY top ORDER BY cnt DESC LIMIT 1").fetchone() 795 | best_bottom = c.execute("SELECT bottom, COUNT(bottom) cnt FROM ocrresult WHERE state = 'done' AND ocrtext != '' AND position = 2 GROUP BY bottom ORDER BY cnt DESC LIMIT 1").fetchone() 796 | best_height = c.execute("SELECT bottom+1-top, COUNT(bottom+1-top) cnt FROM ocrresult WHERE state = 'done' AND ocrtext != '' AND position = 2 GROUP BY bottom+1-top ORDER BY cnt DESC LIMIT 1").fetchone() 797 | best_top = best_top[0] if best_top is not None else 10 798 | best_bottom = best_bottom[0] if best_bottom is not None else 10 799 | best_height = best_height[0] if best_height is not None else c.execute("SELECT bottom+1-top, COUNT(bottom+1-top) cnt FROM ocrresult WHERE state = 'done' AND ocrtext != '' GROUP BY bottom+1-top ORDER BY cnt DESC LIMIT 1").fetchone()[0] 800 | f = open(outfile, 'w', encoding='utf_8_sig', newline='\r\n') 801 | s = gconfig['ass_format']['header'] 802 | s = s.replace('{{文件名}}', os.path.basename(video)) 803 | s = s.replace('{{视频宽度}}', str(width)) 804 | s = s.replace('{{视频高度}}', str(height)) 805 | s = s.replace('{{字幕高度}}', str(best_height)) 806 | s = s.replace('{{上半屏顶边距}}', str(best_top)) 807 | s = s.replace('{{下半屏底边距}}', str(height-best_bottom-1)) 808 | f.write(s) 809 | for frame_start, frame_end, ocrtext, top, bottom, position in row: 810 | s = gconfig['ass_format']['bottom_half' if position == 2 else 'top_half'] 811 | start_time = frame2timestr(frame_start, fps, gconfig['ass_format']['time_alignment']['start']) 812 | end_time = frame2timestr(frame_end+1, fps, gconfig['ass_format']['time_alignment']['end']) 813 | s = s.replace('{{开始时间}}', start_time) 814 | s = s.replace('{{结束时间}}', end_time) 815 | s = s.replace('{{字幕文本}}', ocrtext.replace('\n', '\\N')) 816 | f.write(s) 817 | f.close() 818 | log('已导出至：%s'%outfile, 'S', db=conn) 819 | conn.commit() 820 | return 'ok' 821 | 822 | @app.route('/exportcsv', methods=['POST']) 823 | @session_header_required 824 | def serve_exportcsv(): 825 | outfile = os.path.splitext(video)[0] + gconfig['export_suffix'] + '.csv' 826 | output = io.StringIO() 827 | row = c.execute('SELECT * FROM ocrresult ORDER BY frame_start,frame_end,top,bottom,engine,id').fetchall() 828 | col = [x[0] for x in c.description] 829 | w = csv.writer(output) 830 | w.writerow(col) 831 | for r in row: 832 | w.writerow(map(lambda x: x if x is not None else 'SQLITE_NULL', r)) 833 | if os.path.exists(outfile) and not gconfig['export_overwrite']: 834 | log('输出文件已存在，请先删除：%s'%outfile, 'E', db=conn) 835 | output.close() 836 | conn.commit() 837 | return '' 838 | with open(outfile, 'wb') as f: 839 | f.write(output.getvalue().encode('utf_8_sig')) 840 | log('已导出至：%s'%outfile, 'S', db=conn) 841 | output.close() 842 | conn.commit() 843 | return 'ok' 844 | 845 | @app.route('/importcsv', methods=['POST']) 846 | @session_header_required 847 | def serve_importcsv(): 848 | try: 849 | with conn: 850 | csvfile = flask.request.files['csv'].read() 851 | encoding = 'utf_8_sig' if csvfile.startswith(b'\xEF\xBB\xBF') else None 852 | r = csv.reader(io.TextIOWrapper(io.BytesIO(csvfile), encoding=encoding)) 853 | col = next(r) 854 | colmap = dict([(c, i) for i, c in enumerate(col)]) 855 | coltype = dict(c.execute("SELECT name,type FROM pragma_table_info('ocrresult')").fetchall()) 856 | def convertvalue(v, c): 857 | if v == 'SQLITE_NULL': 858 | return None 859 | if coltype[c].startswith('INT'): 860 | return int(v) 861 | return v 862 | ins = [] 863 | upd = [] 864 | asnew = int(flask.request.form['asnew']) 865 | for row in r: 866 | id = row[colmap['id']] 867 | if id == '': 868 | continue 869 | if int(id) == 0 or asnew: 870 | ins.append(tuple([convertvalue(v, c) for v, c in zip(row, col) if c != 'id'])) 871 | else: 872 | upd.append(tuple([convertvalue(v, c) for v, c in zip(row, col) if c != 'id'] + [int(id)])) 873 | 874 | checkpoint(flask.request.form['checkpoint']) 875 | # FIXME: sql inject 876 | c.executemany('UPDATE ocrresult SET ' + ','.join([c + ' = ?' for c in col if c != 'id']) + ' WHERE id = ?', upd) 877 | updcnt = c.rowcount 878 | c.executemany('INSERT INTO ocrresult (' + ','.join([c for c in col if c != 'id']) + ') VALUES (' + ','.join(['?' for c in col if c != 'id']) + ')', ins) 879 | inscnt = c.rowcount 880 | log('导入CSV文件成功，修改了%d条字幕，新增了%d条字幕'%(updcnt,inscnt), 'S', db=conn) 881 | except Exception: 882 | traceback.print_exc() 883 | log('导入CSV文件失败，请查看控制台中的错误详细信息', 'E', db=conn) 884 | conn.commit() 885 | return '' 886 | 887 | @app.route('/checkpoint', methods=['POST']) 888 | @session_header_required 889 | def serve_checkpoint(): 890 | data = flask.request.get_json() 891 | msg = data['msg'] 892 | checkpoint(msg) 893 | conn.commit() 894 | return '' 895 | 896 | @app.route('/rollback', methods=['POST']) 897 | @session_header_required 898 | def serve_rollback(): 899 | if init_thread.is_alive(): 900 | log('请等待后端启动完成', 'E', db=conn) 901 | conn.commit() 902 | return '' 903 | if ocr_thread is not None and ocr_thread.is_alive(): 904 | log('请先暂停OCR任务', 'E', db=conn) 905 | conn.commit() 906 | return '' 907 | data = flask.request.get_json() 908 | checkpoint_id = data['checkpoint_id'] 909 | err, msg = rollback(checkpoint_id) 910 | if err is None: 911 | log('已还原到恢复点 #%d %s'%(checkpoint_id, msg), 'S', db=conn) 912 | checkwaitocr(db=conn) 913 | else: 914 | log('无法还原到恢复点 #%d (%s)'%(checkpoint_id, err), 'E', db=conn) 915 | conn.commit() 916 | return '' 917 | 918 | @app.route('/loadconfig', methods=['POST']) 919 | @session_header_required 920 | def serve_loadconfig(): 921 | data = flask.request.get_json() 922 | value = getconfig(conn, data['key']) 923 | if value: 924 | return value 925 | default_value = json.dumps(data['default_value']) 926 | c.execute('INSERT INTO config VALUES (?, ?)', (data['key'], default_value)) 927 | conn.commit() 928 | return default_value 929 | 930 | @app.route('/saveconfig', methods=['POST']) 931 | @session_header_required 932 | def serve_saveconfig(): 933 | data = flask.request.get_json() 934 | putconfig(conn, data['key'], json.dumps(data['value'])) 935 | msg = data['msg'] if 'msg' in data else ('%s设置已保存' % data['key']) 936 | if len(msg): 937 | log(msg, 'I', db=conn) 938 | conn.commit() 939 | return '' 940 | 941 | @app.route('/allengines', methods=['POST']) 942 | @session_header_required 943 | def serve_allengines(): 944 | return flask.jsonify(gconfig['allengines']) 945 | 946 | @app.route('/updateresult', methods=['POST']) 947 | @session_header_required 948 | def serve_updateresult(): 949 | data = flask.request.get_json() 950 | checkpoint_id = None 951 | if data['checkpoint']: 952 | checkpoint_id = checkpoint(data['checkpoint'], not data['compatlog']) 953 | cols = [x[0] for x in c.execute("SELECT name FROM pragma_table_info('ocrresult')").fetchall() if x[0] != 'id'] 954 | upd = [] 955 | ins = [] 956 | for item in data['changes']: 957 | if 'id' in item and item['id'] > 0: 958 | if all(col in item for col in cols): 959 | upd.append(tuple([item[col] for col in cols] + [item['id']])) 960 | else: 961 | curcols = [col for col in cols if col in item] 962 | c.execute('UPDATE ocrresult SET ' + ','.join([c + ' = ?' for c in curcols]) + ' WHERE id = ?', tuple(item[c] for c in curcols) + (item['id'],)) 963 | else: 964 | ins.append(tuple([item[col] for col in cols])) 965 | c.executemany('UPDATE ocrresult SET ' + ','.join([c + ' = ?' for c in cols]) + ' WHERE id = ?', upd) 966 | updcnt = c.rowcount 967 | c.executemany('INSERT INTO ocrresult (' + ','.join([c for c in cols]) + ') VALUES (' + ','.join(['?' for c in cols]) + ')', ins) 968 | inscnt = c.rowcount 969 | c.execute("DELETE FROM ocrresult WHERE state = 'delete'") 970 | delcnt = c.rowcount 971 | if data['message']: 972 | log(data['message'], 'S', db=conn, checkpoint_id=checkpoint_id if data['compatlog'] else None) 973 | conn.commit() 974 | return flask.jsonify({ 975 | 'updcnt': updcnt, 976 | 'inscnt': inscnt, 977 | 'delcnt': delcnt, 978 | }) 979 | 980 | @app.route('/loadresult', methods=['POST']) 981 | @session_header_required 982 | def serve_loadresult(): 983 | return db2json(conn, 'SELECT * FROM ocrresult LEFT JOIN imgdb USING (imgdb_id)') 984 | 985 | @app.route('/startocr', methods=['POST']) 986 | @session_header_required 987 | def serve_startocr(): 988 | if init_thread.is_alive(): 989 | log('请等待后端启动完成', 'E', db=conn) 990 | conn.commit() 991 | return '' 992 | ocrconf = json.loads(getconfig(conn, 'OCR')) 993 | ocrengine, ocrtop, ocrbottom = ocrconf['engine'], ocrconf['top'], ocrconf['bottom'] 994 | ocrleft, ocrright = 0, width-1 995 | position = 2 if ocrbottom >= height // 2 else 8 996 | if ocrtop < 0 or ocrbottom < 0: 997 | log('请先指定字幕在屏幕上的范围', 'E', db=conn) 998 | conn.commit() 999 | return '' 1000 | if ocr_thread is not None and ocr_thread.is_alive(): 1001 | log('已有OCR任务运行中，无法启动新任务', 'E', db=conn) 1002 | conn.commit() 1003 | return '' 1004 | checkpoint('“新OCR”之前 [%s]'%ocrengine) 1005 | data = flask.request.get_json() 1006 | frame_range = data['frame_range'] 1007 | if frame_range is None: 1008 | frame_range = list(range(0, nframes)) 1009 | c.execute('DELETE FROM jobrange') 1010 | for frame_id in sorted(frame_range): 1011 | c.execute("INSERT INTO ocrresult (date, state, frame_start, frame_end, engine, left, top, right, bottom, position) VALUES (datetime('now','localtime'), ?, ?, ?, ?, ?, ?, ?, ?, ?)", ('waitocr', frame_id, frame_id, ocrengine, ocrleft, ocrtop, ocrright, ocrbottom, position)) 1012 | c.execute('INSERT INTO jobrange VALUES (?)', (c.lastrowid,)) 1013 | log('OCR任务已提交 [%s]'%ocrengine, db=conn) 1014 | conn.commit() 1015 | startocr() 1016 | return 'ok' 1017 | 1018 | @app.route('/stopocr', methods=['POST']) 1019 | @session_header_required 1020 | def serve_stopocr(): 1021 | global ocr_stop 1022 | if ocr_thread is None or not ocr_thread.is_alive(): 1023 | log('无运行中的OCR任务', 'I', db=conn) 1024 | conn.commit() 1025 | return '' 1026 | ocr_stop = True 1027 | log('暂停请求已提交', 'I', db=conn) 1028 | conn.commit() 1029 | return '' 1030 | 1031 | @app.route('/continueocr', methods=['POST']) 1032 | @session_header_required 1033 | def serve_continueocr(): 1034 | if init_thread.is_alive(): 1035 | log('请等待后端启动完成', 'E', db=conn) 1036 | conn.commit() 1037 | return '' 1038 | if ocr_thread is not None and ocr_thread.is_alive(): 1039 | log('已有OCR任务运行中，无法启动新任务', 'E', db=conn) 1040 | conn.commit() 1041 | return '' 1042 | data = flask.request.get_json() 1043 | if data['item_range'] is not None: 1044 | if len(data['item_range']) > 0: 1045 | c.execute('DELETE FROM jobrange') 1046 | c.executemany('INSERT INTO jobrange VALUES (?)', [(id,) for id in sorted(data['item_range'])]) 1047 | else: 1048 | c.execute("INSERT OR IGNORE INTO jobrange SELECT id FROM ocrresult WHERE (state = 'waitocr' OR state = 'error')") 1049 | restarttype = data['restarttype'] 1050 | new_engine = json.loads(getconfig(conn, 'OCR'))['engine'] 1051 | if restarttype == '': 1052 | errcnt = c.execute(""" 1053 | SELECT COUNT(id) FROM ocrresult WHERE EXISTS (SELECT * FROM jobrange WHERE ocrresult.id = jobrange.id) AND 1054 | (state = 'error' OR state = 'waitocr')""").fetchone()[0] 1055 | if errcnt == 0: 1056 | log('没有任务要做', db=conn) 1057 | conn.commit() 1058 | return '' 1059 | checkpoint('“继续OCR”之前 [%s]'%new_engine) 1060 | c.execute("UPDATE ocrresult SET state = 'waitocr', engine = ? WHERE (state = 'waitocr' OR state = 'error') AND EXISTS (SELECT * FROM jobrange WHERE ocrresult.id = jobrange.id)", (new_engine,)) 1061 | elif restarttype == 'all': 1062 | donecnt = c.execute(""" 1063 | SELECT COUNT(id) FROM ocrresult WHERE EXISTS (SELECT * FROM jobrange WHERE ocrresult.id = jobrange.id) AND 1064 | (state = 'waitocr' OR state = 'error' OR state = 'waitocr' OR state = 'done')""").fetchone()[0] 1065 | if donecnt == 0: 1066 | log('没有任务要做', db=conn) 1067 | conn.commit() 1068 | return '' 1069 | checkpoint('“重新OCR”之前 [%s]'%new_engine) 1070 | c.execute("UPDATE ocrresult SET state = 'waitocr', engine = ? WHERE (state = 'waitocr' OR state = 'error' OR state = 'waitocr' OR state = 'done') AND EXISTS (SELECT * FROM jobrange WHERE ocrresult.id = jobrange.id)", (new_engine,)) 1071 | elif restarttype == 'empty': 1072 | emptycnt = c.execute(""" 1073 | SELECT COUNT(id) FROM ocrresult WHERE EXISTS (SELECT * FROM jobrange WHERE ocrresult.id = jobrange.id) AND 1074 | (state = 'done' AND ocrtext = '')""").fetchone()[0] 1075 | if emptycnt == 0: 1076 | log('没有任务要做', db=conn) 1077 | conn.commit() 1078 | return '' 1079 | checkpoint('“空项OCR”之前 [%s]'%new_engine) 1080 | c.execute("UPDATE ocrresult SET state = 'waitocr', engine = ? WHERE (state = 'done' AND ocrtext = '') AND EXISTS (SELECT * FROM jobrange WHERE ocrresult.id = jobrange.id)", (new_engine,)) 1081 | else: 1082 | assert False 1083 | log('OCR任务已提交 [%s]'%new_engine, db=conn) 1084 | conn.commit() 1085 | startocr() 1086 | return 'ok' 1087 | 1088 | @app.after_request 1089 | def add_header(response): 1090 | response.cache_control.no_cache = True 1091 | response.cache_control.no_store = True 1092 | response.cache_control.must_revalidate = True 1093 | response.cache_control.max_age = 0 1094 | response.expires = werkzeug.http.parse_date('Thu, 19 Nov 1981 08:52:00 GMT') 1095 | response.headers['Pragma'] = 'no-cache' 1096 | response.close_connection = True 1097 | return response 1098 | 1099 | app.run(threaded=False, host=host, port=port) 1100 | --------------------------------------------------------------------------------