sub_helper.go 23 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619
  1. package sub_helper
  2. import (
  3. "errors"
  4. "github.com/allanpk716/ChineseSubFinder/internal/common"
  5. "github.com/allanpk716/ChineseSubFinder/internal/pkg/archive_helper"
  6. "github.com/allanpk716/ChineseSubFinder/internal/pkg/decode"
  7. "github.com/allanpk716/ChineseSubFinder/internal/pkg/language"
  8. "github.com/allanpk716/ChineseSubFinder/internal/pkg/log_helper"
  9. "github.com/allanpk716/ChineseSubFinder/internal/pkg/my_folder"
  10. "github.com/allanpk716/ChineseSubFinder/internal/pkg/my_util"
  11. "github.com/allanpk716/ChineseSubFinder/internal/pkg/regex_things"
  12. "github.com/allanpk716/ChineseSubFinder/internal/pkg/sub_parser_hub"
  13. "github.com/allanpk716/ChineseSubFinder/internal/pkg/vad"
  14. "github.com/allanpk716/ChineseSubFinder/internal/types/subparser"
  15. "github.com/allanpk716/ChineseSubFinder/internal/types/supplier"
  16. "github.com/go-rod/rod/lib/utils"
  17. "io/ioutil"
  18. "math"
  19. "os"
  20. "path/filepath"
  21. "strconv"
  22. "strings"
  23. "time"
  24. )
  25. // OrganizeDlSubFiles 需要从汇总来是网站字幕中,解压对应的压缩包中的字幕出来
  26. func OrganizeDlSubFiles(tmpFolderName string, subInfos []supplier.SubInfo) (map[string][]string, error) {
  27. // 缓存列表,整理后的字幕列表
  28. // SxEx - []string 字幕的路径
  29. var siteSubInfoDict = make(map[string][]string)
  30. tmpFolderFullPath, err := my_folder.GetTmpFolderByName(tmpFolderName)
  31. if err != nil {
  32. return nil, err
  33. }
  34. // 把后缀名给改好
  35. ChangeVideoExt2SubExt(subInfos)
  36. // 第三方的解压库,首先不支持 io.Reader 的操作,也就是得缓存到本地硬盘再读取解压
  37. // 且使用 walk 会无法解压 rar,得指定具体的实例,太麻烦了,直接用通用的接口得了,就是得都缓存下来再判断
  38. // 基于以上两点,写了一堆啰嗦的逻辑···
  39. for i := range subInfos {
  40. // 先存下来,保存是时候需要前缀,前缀就是从那个网站下载来的
  41. nowFileSaveFullPath := filepath.Join(tmpFolderFullPath, GetFrontNameAndOrgName(&subInfos[i]))
  42. err = utils.OutputFile(nowFileSaveFullPath, subInfos[i].Data)
  43. if err != nil {
  44. log_helper.GetLogger().Errorln("getFrontNameAndOrgName - OutputFile", subInfos[i].FromWhere, subInfos[i].Name, subInfos[i].TopN, err)
  45. continue
  46. }
  47. nowExt := strings.ToLower(subInfos[i].Ext)
  48. epsKey := my_util.GetEpisodeKeyName(subInfos[i].Season, subInfos[i].Episode)
  49. _, ok := siteSubInfoDict[epsKey]
  50. if ok == false {
  51. // 不存在则实例化
  52. siteSubInfoDict[epsKey] = make([]string, 0)
  53. }
  54. if nowExt != ".zip" && nowExt != ".tar" && nowExt != ".rar" && nowExt != ".7z" {
  55. // 是否是受支持的字幕类型
  56. if sub_parser_hub.IsSubExtWanted(nowExt) == false {
  57. continue
  58. }
  59. // 加入缓存列表
  60. siteSubInfoDict[epsKey] = append(siteSubInfoDict[epsKey], nowFileSaveFullPath)
  61. } else {
  62. // 那么就是需要解压的文件了
  63. // 解压,给一个单独的文件夹
  64. unzipTmpFolder := filepath.Join(tmpFolderFullPath, subInfos[i].FromWhere)
  65. err = os.MkdirAll(unzipTmpFolder, os.ModePerm)
  66. if err != nil {
  67. return nil, err
  68. }
  69. err = archive_helper.UnArchiveFile(nowFileSaveFullPath, unzipTmpFolder)
  70. // 解压完成后,遍历受支持的字幕列表,加入缓存列表
  71. if err != nil {
  72. log_helper.GetLogger().Errorln("archiver.UnArchive", subInfos[i].FromWhere, subInfos[i].Name, subInfos[i].TopN, err)
  73. continue
  74. }
  75. // 搜索这个目录下的所有符合字幕格式的文件
  76. subFileFullPaths, err := SearchMatchedSubFileByDir(unzipTmpFolder)
  77. if err != nil {
  78. log_helper.GetLogger().Errorln("searchMatchedSubFile", subInfos[i].FromWhere, subInfos[i].Name, subInfos[i].TopN, err)
  79. continue
  80. }
  81. // 这里需要给这些下载到的文件进行改名,加是从那个网站来的前缀,后续好查找
  82. for _, fileFullPath := range subFileFullPaths {
  83. newSubName := AddFrontName(subInfos[i], filepath.Base(fileFullPath))
  84. newSubNameFullPath := filepath.Join(tmpFolderFullPath, newSubName)
  85. // 改名
  86. err = os.Rename(fileFullPath, newSubNameFullPath)
  87. if err != nil {
  88. log_helper.GetLogger().Errorln("os.Rename", subInfos[i].FromWhere, subInfos[i].Name, subInfos[i].TopN, err)
  89. continue
  90. }
  91. // 加入缓存列表
  92. siteSubInfoDict[epsKey] = append(siteSubInfoDict[epsKey], newSubNameFullPath)
  93. }
  94. }
  95. }
  96. return siteSubInfoDict, nil
  97. }
  98. // ChangeVideoExt2SubExt 检测 Name,如果是视频的后缀名就改为字幕的后缀名
  99. func ChangeVideoExt2SubExt(subInfos []supplier.SubInfo) {
  100. for x, info := range subInfos {
  101. tmpSubFileName := info.Name
  102. // 如果后缀名是下载字幕目标的后缀名 或者 是压缩包格式的,则跳过
  103. if strings.Contains(tmpSubFileName, info.Ext) == true || archive_helper.IsWantedArchiveExtName(tmpSubFileName) == true {
  104. } else {
  105. subInfos[x].Name = tmpSubFileName + info.Ext
  106. }
  107. }
  108. }
  109. // SelectChineseBestBilingualSubtitle 找到合适的双语中文字幕,简体->繁体,以及 字幕类型的优先级选择
  110. func SelectChineseBestBilingualSubtitle(subs []subparser.FileInfo, subTypePriority int) *subparser.FileInfo {
  111. // 先傻一点实现优先双语的,之前的写法有 bug
  112. for _, info := range subs {
  113. // 找到了中文字幕
  114. if language.HasChineseLang(info.Lang) == true {
  115. // 字幕的优先级 0 - 原样, 1 - srt , 2 - ass/ssa
  116. if subTypePriority == 1 {
  117. // 1 - srt
  118. if strings.ToLower(info.Ext) == common.SubExtSRT {
  119. // 优先双语
  120. if language.IsBilingualSubtitle(info.Lang) == true {
  121. return &info
  122. }
  123. }
  124. } else if subTypePriority == 2 {
  125. // 2 - ass/ssa
  126. if strings.ToLower(info.Ext) == common.SubExtASS || strings.ToLower(info.Ext) == common.SubExtSSA {
  127. // 优先双语
  128. if language.IsBilingualSubtitle(info.Lang) == true {
  129. return &info
  130. }
  131. }
  132. } else {
  133. // 优先双语
  134. if language.IsBilingualSubtitle(info.Lang) == true {
  135. return &info
  136. }
  137. }
  138. }
  139. }
  140. return nil
  141. }
  142. // SelectChineseBestSubtitle 找到合适的中文字幕,简体->繁体,以及 字幕类型的优先级选择
  143. func SelectChineseBestSubtitle(subs []subparser.FileInfo, subTypePriority int) *subparser.FileInfo {
  144. // 先傻一点实现优先双语的,之前的写法有 bug
  145. for _, info := range subs {
  146. // 找到了中文字幕
  147. if language.HasChineseLang(info.Lang) == true {
  148. // 字幕的优先级 0 - 原样, 1 - srt , 2 - ass/ssa
  149. if subTypePriority == 1 {
  150. // 1 - srt
  151. if strings.ToLower(info.Ext) == common.SubExtSRT {
  152. return &info
  153. }
  154. } else if subTypePriority == 2 {
  155. // 2 - ass/ssa
  156. if strings.ToLower(info.Ext) == common.SubExtASS || strings.ToLower(info.Ext) == common.SubExtSSA {
  157. return &info
  158. }
  159. } else {
  160. return &info
  161. }
  162. }
  163. }
  164. return nil
  165. }
  166. // GetFrontNameAndOrgName 返回的名称包含,那个网站下载的,这个网站中排名第几,文件名
  167. func GetFrontNameAndOrgName(info *supplier.SubInfo) string {
  168. infoName := ""
  169. fileName, err := decode.GetVideoInfoFromFileName(info.Name)
  170. if err != nil {
  171. log_helper.GetLogger().Warnln("", err)
  172. infoName = info.Name
  173. } else {
  174. infoName = fileName.Title + "_S" + strconv.Itoa(fileName.Season) + "E" + strconv.Itoa(fileName.Episode) + filepath.Ext(info.Name)
  175. }
  176. info.Name = infoName
  177. return "[" + info.FromWhere + "]_" + strconv.FormatInt(info.TopN, 10) + "_" + infoName
  178. }
  179. // AddFrontName 添加文件的前缀
  180. func AddFrontName(info supplier.SubInfo, orgName string) string {
  181. return "[" + info.FromWhere + "]_" + strconv.FormatInt(info.TopN, 10) + "_" + orgName
  182. }
  183. // SearchMatchedSubFileByDir 搜索符合后缀名的视频文件,排除 Sub_SxE0 这样的文件夹中的文件
  184. func SearchMatchedSubFileByDir(dir string) ([]string, error) {
  185. // 这里有个梗,会出现 __MACOSX 这类文件夹,那么里面会有一样的文件,需要用文件大小排除一下,至少大于 1 kb 吧
  186. var fileFullPathList = make([]string, 0)
  187. pathSep := string(os.PathSeparator)
  188. files, err := ioutil.ReadDir(dir)
  189. if err != nil {
  190. return nil, err
  191. }
  192. for _, curFile := range files {
  193. fullPath := dir + pathSep + curFile.Name()
  194. if curFile.IsDir() {
  195. // 需要排除 Sub_S1E0、Sub_S2E0 这样的整季的字幕文件夹,这里仅仅是缓存,不会被加载的
  196. matched := regex_things.RegOneSeasonSubFolderNameMatch.FindAllStringSubmatch(curFile.Name(), -1)
  197. if matched == nil || len(matched) > 0 {
  198. continue
  199. }
  200. // 内层的错误就无视了
  201. oneList, _ := SearchMatchedSubFileByDir(fullPath)
  202. if oneList != nil {
  203. fileFullPathList = append(fileFullPathList, oneList...)
  204. }
  205. } else {
  206. // 这里就是文件了
  207. if curFile.Size() < 1000 {
  208. continue
  209. }
  210. if sub_parser_hub.IsSubExtWanted(filepath.Ext(curFile.Name())) == true {
  211. fileFullPathList = append(fileFullPathList, fullPath)
  212. }
  213. }
  214. }
  215. return fileFullPathList, nil
  216. }
  217. // SearchMatchedSubFileByOneVideo 搜索这个视频当前目录下匹配的字幕
  218. func SearchMatchedSubFileByOneVideo(oneVideoFullPath string) ([]string, error) {
  219. dir := filepath.Dir(oneVideoFullPath)
  220. fileName := filepath.Base(oneVideoFullPath)
  221. fileName = strings.ToLower(fileName)
  222. fileName = strings.ReplaceAll(fileName, filepath.Ext(fileName), "")
  223. pathSep := string(os.PathSeparator)
  224. files, err := ioutil.ReadDir(dir)
  225. if err != nil {
  226. return nil, err
  227. }
  228. var matchedSubs = make([]string, 0)
  229. for _, curFile := range files {
  230. if curFile.IsDir() {
  231. continue
  232. }
  233. // 这里就是文件了
  234. if curFile.Size() < 1000 {
  235. continue
  236. }
  237. // 判断的时候用小写的,后续重命名的时候用原有的名称
  238. nowFileName := strings.ToLower(curFile.Name())
  239. // 后缀名得对
  240. if sub_parser_hub.IsSubExtWanted(filepath.Ext(nowFileName)) == false {
  241. continue
  242. }
  243. // 字幕文件名应该包含 视频文件名(无后缀)
  244. if strings.Contains(nowFileName, fileName) == false {
  245. continue
  246. }
  247. oldPath := dir + pathSep + curFile.Name()
  248. matchedSubs = append(matchedSubs, oldPath)
  249. }
  250. return matchedSubs, nil
  251. }
  252. // SearchVideoMatchSubFileAndRemoveExtMark 找到找个视频目录下相匹配的字幕,同时去除这些字幕中 .default 或者 .forced 的标记。注意这两个标记不应该同时出现,否则无法正确去除
  253. func SearchVideoMatchSubFileAndRemoveExtMark(oneVideoFullPath string) error {
  254. dir := filepath.Dir(oneVideoFullPath)
  255. fileName := filepath.Base(oneVideoFullPath)
  256. fileName = strings.ToLower(fileName)
  257. fileName = strings.ReplaceAll(fileName, filepath.Ext(fileName), "")
  258. pathSep := string(os.PathSeparator)
  259. files, err := ioutil.ReadDir(dir)
  260. if err != nil {
  261. return err
  262. }
  263. for _, curFile := range files {
  264. if curFile.IsDir() {
  265. continue
  266. } else {
  267. // 这里就是文件了
  268. if curFile.Size() < 1000 {
  269. continue
  270. }
  271. // 判断的时候用小写的,后续重命名的时候用原有的名称
  272. nowFileName := strings.ToLower(curFile.Name())
  273. // 后缀名得对
  274. if sub_parser_hub.IsSubExtWanted(filepath.Ext(nowFileName)) == false {
  275. continue
  276. }
  277. // 字幕文件名应该包含 视频文件名(无后缀)
  278. if strings.Contains(nowFileName, fileName) == false {
  279. continue
  280. }
  281. // 得包含 .default. 找个关键词
  282. if strings.Contains(nowFileName, subparser.Sub_Ext_Mark_Default+".") == true {
  283. oldPath := dir + pathSep + curFile.Name()
  284. newPath := dir + pathSep + strings.ReplaceAll(curFile.Name(), subparser.Sub_Ext_Mark_Default+".", ".")
  285. err = os.Rename(oldPath, newPath)
  286. if err != nil {
  287. return err
  288. }
  289. } else if strings.Contains(nowFileName, subparser.Sub_Ext_Mark_Forced+".") == true {
  290. // 得包含 .forced. 找个关键词
  291. oldPath := dir + pathSep + curFile.Name()
  292. newPath := dir + pathSep + strings.ReplaceAll(curFile.Name(), subparser.Sub_Ext_Mark_Forced+".", ".")
  293. err = os.Rename(oldPath, newPath)
  294. if err != nil {
  295. return err
  296. }
  297. } else {
  298. continue
  299. }
  300. }
  301. }
  302. return nil
  303. }
  304. // DeleteOneSeasonSubCacheFolder 删除一个连续剧中的所有一季字幕的缓存文件夹
  305. func DeleteOneSeasonSubCacheFolder(seriesDir string) error {
  306. files, err := ioutil.ReadDir(seriesDir)
  307. if err != nil {
  308. return err
  309. }
  310. pathSep := string(os.PathSeparator)
  311. for _, curFile := range files {
  312. if curFile.IsDir() == true {
  313. matched := regex_things.RegOneSeasonSubFolderNameMatch.FindAllStringSubmatch(curFile.Name(), -1)
  314. if matched == nil || len(matched) < 1 {
  315. continue
  316. }
  317. fullPath := seriesDir + pathSep + curFile.Name()
  318. err = os.RemoveAll(fullPath)
  319. if err != nil {
  320. return err
  321. }
  322. }
  323. }
  324. return nil
  325. }
  326. /*
  327. 只针对英文字幕进行合并分散的 Dialogues
  328. 会遇到这样的字幕,如下0
  329. 2line-The Card Counter (2021) WEBDL-1080p.chinese(inside).ass
  330. 它的对白一句话分了两个 dialogue 去做。这样做后续字幕时间轴校正就会遇到问题,因为只有一半,匹配占比会很低
  331. (每一个 Dialogue 的首字母需要分析,大写和小写的占比是多少,统计一下,正常的,和上述特殊的)
  332. 那么,就需要额外的逻辑去对 DialoguesEx 进行额外的推断
  333. 暂时考虑的方案是,英文对白每一句的开头应该是英文大写字幕,如果是小写字幕,就应该与上语句合并,且每一句的字符长度有大于一定才触发
  334. */
  335. func MergeMultiDialogue4EngSubtitle(inSubParser *subparser.FileInfo) {
  336. merger := NewDialogueMerger()
  337. for _, dialogueEx := range inSubParser.DialoguesEx {
  338. merger.Add(dialogueEx)
  339. }
  340. inSubParser.DialoguesEx = merger.Get()
  341. }
  342. // GetVADInfoFeatureFromSub 跟下面的 GetVADInfoFeatureFromSubNeedOffsetTimeWillInsert 函数功能一致
  343. func GetVADInfoFeatureFromSub(fileInfo *subparser.FileInfo, frontAndEndPer float64, subUnitMaxCount int, insert bool) ([]SubUnit, error) {
  344. return GetVADInfoFeatureFromSubNeedOffsetTimeWillInsert(fileInfo, frontAndEndPer, subUnitMaxCount, 0, insert)
  345. }
  346. /*
  347. GetVADInfoFeatureFromSubNeedOffsetTimeWillInsert 只不过这里可以加一个每一句话固定的偏移时间
  348. 这里的字幕要求是完整的一个字幕
  349. 1. 抽取字幕的时间片段的时候,暂定,前 15% 和后 15% 要避开,前奏、主题曲、结尾曲
  350. 2. 将整个字幕,抽取连续 5 句对话为一个单元,提取时间片段信息
  351. 3. 这里抽取的是特征,也就有额外的逻辑去找这个特征(本程序内会描述为“钥匙”)
  352. */
  353. func GetVADInfoFeatureFromSubNeedOffsetTimeWillInsert(fileInfo *subparser.FileInfo, SkipFrontAndEndPer float64, subUnitMaxCount int, offsetTime float64, insert bool) ([]SubUnit, error) {
  354. if subUnitMaxCount < 0 {
  355. subUnitMaxCount = 0
  356. }
  357. srcSubUnitList := make([]SubUnit, 0)
  358. srcSubDialogueList := make([]subparser.OneDialogueEx, 0)
  359. srcOneSubUnit := NewSubUnit()
  360. // 最后一个对话的结束时间
  361. lastDialogueExTimeEnd, err := fileInfo.ParseTime(fileInfo.DialoguesEx[len(fileInfo.DialoguesEx)-1].EndTime)
  362. if err != nil {
  363. return nil, err
  364. }
  365. // 相当于总时长
  366. fullDuration := my_util.Time2SecondNumber(lastDialogueExTimeEnd)
  367. // 最低的起始时间,因为可能需要裁剪范围
  368. startRangeTimeMin := fullDuration * SkipFrontAndEndPer
  369. endRangeTimeMax := fullDuration * (1.0 - SkipFrontAndEndPer)
  370. println(startRangeTimeMin)
  371. println(endRangeTimeMax)
  372. for i := 0; i < len(fileInfo.DialoguesEx); i++ {
  373. oneDialogueExTimeStart, err := fileInfo.ParseTime(fileInfo.DialoguesEx[i].StartTime)
  374. if err != nil {
  375. return nil, err
  376. }
  377. oneDialogueExTimeEnd, err := fileInfo.ParseTime(fileInfo.DialoguesEx[i].EndTime)
  378. if err != nil {
  379. return nil, err
  380. }
  381. oneStart := my_util.Time2SecondNumber(oneDialogueExTimeStart)
  382. if SkipFrontAndEndPer > 0 {
  383. if fullDuration*SkipFrontAndEndPer > oneStart || fullDuration*(1.0-SkipFrontAndEndPer) < oneStart {
  384. continue
  385. }
  386. }
  387. // 如果当前的这一句话,为空,或者进过正则表达式剔除特殊字符后为空,则跳过
  388. if my_util.ReplaceSpecString(fileInfo.GetDialogueExContent(i), "") == "" {
  389. continue
  390. }
  391. // 低于 5句对白,则添加
  392. if srcOneSubUnit.GetDialogueCount() < subUnitMaxCount {
  393. // 算上偏移
  394. offsetTimeDuration := time.Duration(offsetTime * math.Pow10(9))
  395. oneDialogueExTimeStart = oneDialogueExTimeStart.Add(offsetTimeDuration)
  396. oneDialogueExTimeEnd = oneDialogueExTimeEnd.Add(offsetTimeDuration)
  397. // 如果没有偏移就是 0
  398. if insert == true {
  399. srcOneSubUnit.AddAndInsert(oneDialogueExTimeStart, oneDialogueExTimeEnd)
  400. } else {
  401. srcOneSubUnit.Add(oneDialogueExTimeStart, oneDialogueExTimeEnd)
  402. }
  403. // 这一个单元的 Dialogue 需要合并起来,才能判断是否符合“钥匙”的要求
  404. srcSubDialogueList = append(srcSubDialogueList, fileInfo.DialoguesEx[i])
  405. } else {
  406. // 用完清空
  407. srcSubDialogueList = make([]subparser.OneDialogueEx, 0)
  408. // 将拼凑起来的对话组成一个单元进行存储起来
  409. srcSubUnitList = append(srcSubUnitList, *srcOneSubUnit)
  410. // 然后重置
  411. srcOneSubUnit = NewSubUnit()
  412. }
  413. }
  414. if srcOneSubUnit.GetDialogueCount() > 0 {
  415. srcSubUnitList = append(srcSubUnitList, *srcOneSubUnit)
  416. }
  417. return srcSubUnitList, nil
  418. }
  419. /*
  420. GetVADInfoFeatureFromSubNew 将 Sub 文件转换为 VAD List 信息
  421. */
  422. func GetVADInfoFeatureFromSubNew(fileInfo *subparser.FileInfo, SkipFrontAndEndPer float64) (*SubUnit, error) {
  423. outSubUnits := NewSubUnit()
  424. if len(fileInfo.DialoguesEx) <= 0 {
  425. return nil, errors.New("GetVADInfoFeatureFromSubNew fileInfo Dialogue Length is 0")
  426. }
  427. /*
  428. 先拼凑出完整的一个 VAD List
  429. 因为 VAD 的窗口是 10ms,那么需要多每一句话按 10 ms 的单位进行取整
  430. 每一句话开始、结束的时间,需要向下取整
  431. */
  432. subStartTimeFloor, subEndTimeFloor, err := ReadSubStartAndEndTime(fileInfo)
  433. if err != nil {
  434. return nil, err
  435. }
  436. subFullSecondTimeFloor := subEndTimeFloor - subStartTimeFloor
  437. // 根据这个时长就能够得到一个完整的 VAD List,然后再通过每一句对白进行 VAD 值的调整即可,这样就能够保证
  438. // 相同的一个字幕因为使用 ffmpeg 导出 srt 和 ass 后的,可能存在总体时间轴不一致的问题
  439. // 123.450 - > 12345
  440. vadLen := int(subFullSecondTimeFloor * 100)
  441. subVADs := make([]vad.VADInfo, vadLen)
  442. subStartTimeFloor10ms := subStartTimeFloor * 100
  443. for i := 0; i < vadLen; i++ {
  444. subVADs[i] = *vad.NewVADInfoBase(false, time.Duration((subStartTimeFloor10ms+float64(i))*math.Pow10(7)))
  445. }
  446. // 计算出需要截取的片段,起始和结束
  447. skipLen := int(float64(vadLen) * SkipFrontAndEndPer)
  448. skipStartIndex := skipLen
  449. skipEndIndex := vadLen - skipLen
  450. // 现在需要从 fileInfo 的每一句对白也就对应一段连续的 VAD active = true 来进行改写,记得向下取整
  451. lastDialogueIndex := 0
  452. for index, dialogueEx := range fileInfo.DialoguesEx {
  453. // 如果当前的这一句话,为空,或者进过正则表达式剔除特殊字符后为空,则跳过
  454. if my_util.ReplaceSpecString(fileInfo.GetDialogueExContent(index), "") == "" {
  455. continue
  456. }
  457. // 字幕的开始时间
  458. oneDialogueStartTime, err := fileInfo.ParseTime(dialogueEx.StartTime)
  459. if err != nil {
  460. return nil, err
  461. }
  462. // 字幕的结束时间
  463. oneDialogueEndTime, err := fileInfo.ParseTime(dialogueEx.EndTime)
  464. if err != nil {
  465. return nil, err
  466. }
  467. // 字幕的时长,对时间进行向下取整
  468. oneDialogueStartTimeFloor := my_util.MakeCeil10msMultipleFromFloat(my_util.Time2SecondNumber(oneDialogueStartTime))
  469. oneDialogueEndTimeFloor := my_util.MakeFloor10msMultipleFromFloat(my_util.Time2SecondNumber(oneDialogueEndTime))
  470. // 得到一句对白的时长
  471. changeVADStartIndex := int(oneDialogueStartTimeFloor * 100)
  472. changeVADEndIndex := int(oneDialogueEndTimeFloor * 100)
  473. // 不能超过 最后一句话的时常
  474. if changeVADStartIndex > int(subEndTimeFloor*100) {
  475. continue
  476. }
  477. // 也不能比起始的第一句话时间轴更低
  478. if changeVADStartIndex < int(subStartTimeFloor10ms) {
  479. continue
  480. }
  481. // 当前这句话的开始和结束信息
  482. changerStartIndex := changeVADStartIndex - int(subStartTimeFloor10ms)
  483. if changerStartIndex < 0 {
  484. continue
  485. }
  486. changerEndIndex := changeVADEndIndex - int(subStartTimeFloor10ms)
  487. if changerEndIndex < 0 {
  488. continue
  489. }
  490. // 如果上一个对白的最后一个 OffsetIndex 连接着当前这一句的索引的 VAD 信息 active 是 true 就设置为 false
  491. if lastDialogueIndex == changerStartIndex {
  492. for i := 1; i <= 10; i++ {
  493. if lastDialogueIndex-i >= 0 && subVADs[lastDialogueIndex-i].Active == true {
  494. subVADs[lastDialogueIndex-i].Active = false
  495. }
  496. }
  497. }
  498. // 开始根据当前这句话进行 VAD 信息的设置
  499. // 调整之前做好的整体 VAD 的信息,符合 VAD active = true
  500. if changerEndIndex >= vadLen {
  501. changerEndIndex = vadLen - 1
  502. }
  503. for i := changerStartIndex; i <= changerEndIndex; i++ {
  504. subVADs[i].Active = true
  505. }
  506. lastDialogueIndex = changerEndIndex
  507. }
  508. // 截取出来当前这一段
  509. tmpVADList := subVADs[skipStartIndex:skipEndIndex]
  510. outSubUnits.VADList = tmpVADList
  511. tmpStartTime := time.Time{}
  512. tmpStartTime = tmpStartTime.Add(tmpVADList[0].Time)
  513. tmpEndTime := time.Time{}
  514. tmpEndTime = tmpEndTime.Add(tmpVADList[len(tmpVADList)-1].Time)
  515. outSubUnits.SetBaseTime(tmpStartTime)
  516. outSubUnits.SetOffsetStartTime(tmpStartTime)
  517. outSubUnits.SetOffsetEndTime(tmpEndTime)
  518. return outSubUnits, nil
  519. }
  520. func ReadSubStartAndEndTime(fileInfo *subparser.FileInfo) (float64, float64, error) {
  521. /*
  522. 因为是先构建完整的时间轴 VAD ,然后再用每一句话去修改对应的 VAD 段
  523. 那么,如果字幕第一句话时间轴有问题,就会出问题,所以这里返回的时候需要判断是否 StartTime 正确
  524. 因为可能会有一种情况,读取到的字幕是经过 V1 校正时间的,那么第一句和前几句话,可能时间是 Dialogue: 0,23:59:31.32,23:59:33.23
  525. 明显时间过大,导致减出来的值是负值,会越界访问
  526. */
  527. getTimeFunc := func(fileInfo *subparser.FileInfo, startIndex int) (bool, float64, float64, error) {
  528. // 字幕的开始时间
  529. subStartTime, err := fileInfo.ParseTime(fileInfo.DialoguesEx[startIndex].StartTime)
  530. if err != nil {
  531. return false, 0, 0, err
  532. }
  533. // 字幕的结束时间
  534. subEndTime, err := fileInfo.ParseTime(fileInfo.DialoguesEx[len(fileInfo.DialoguesEx)-1].EndTime)
  535. if err != nil {
  536. return false, 0, 0, err
  537. }
  538. // 字幕的时长,对时间进行向下取整
  539. subStartTimeFloor := my_util.MakeFloor10msMultipleFromFloat(my_util.Time2SecondNumber(subStartTime))
  540. subEndTimeFloor := my_util.MakeFloor10msMultipleFromFloat(my_util.Time2SecondNumber(subEndTime))
  541. if subEndTimeFloor-subStartTimeFloor < 0 {
  542. // 说明 StartTime 的数值太大,不正常,超过 EndTime 了,startIndex 需要累加
  543. return false, 0, 0, nil
  544. }
  545. return true, subStartTimeFloor, subEndTimeFloor, nil
  546. }
  547. startIndex := 0
  548. var err error
  549. var subStartTimeFloor, subEndTimeFloor float64
  550. bok := false
  551. for bok == false {
  552. bok, subStartTimeFloor, subEndTimeFloor, err = getTimeFunc(fileInfo, startIndex)
  553. if err != nil {
  554. return 0, 0, err
  555. }
  556. if bok == true {
  557. break
  558. }
  559. startIndex++
  560. }
  561. return subStartTimeFloor, subEndTimeFloor, err
  562. }