bind4spider.go 4.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172
  1. // 爬虫调试绑定
  2. package main
  3. import (
  4. qu "jygit.jydev.jianyu360.cn/data_processing/common_utils"
  5. "sort"
  6. be "spider_creator/backend"
  7. bdb "spider_creator/backend/db"
  8. bew "spider_creator/backend/webservice"
  9. )
  10. // DebugSpider 调试爬虫
  11. func (a *App) DebugSpider(url string, proxyServe string, maxPages int, listDealy int64, trunPageDelay int64, contentDelay int64, headless bool,
  12. showImage bool, threads int, cssMark map[string]interface{}) {
  13. exitCh = make(chan bool, 1)
  14. qu.Debug(url, proxyServe, maxPages, listDealy, trunPageDelay, contentDelay, headless, showImage, threads)
  15. qu.Debug("cssMark---", cssMark)
  16. if maxPages == 1 && threads == 1 {
  17. vm.RunSpider(url, maxPages, listDealy, contentDelay, headless, showImage, proxyServe, exitCh, cssMark)
  18. } else { //多页下载强制使用多线程模式
  19. vm.RunSpiderMulThreads(url, maxPages, listDealy, trunPageDelay, contentDelay, headless, showImage, proxyServe, threads, exitCh, cssMark)
  20. }
  21. }
  22. // VerifySpiderConfig 验证
  23. func (a *App) VerifySpiderConfig(cssMark map[string]interface{}) map[string]interface{} {
  24. var errtype int
  25. qu.Debug(cssMark)
  26. sc, err := be.NewSpiderConfig(cssMark)
  27. if err != nil {
  28. qu.Debug("CssMark Marshal Error:", err)
  29. return map[string]interface{}{
  30. "ret": nil,
  31. "err": errtype,
  32. "msg": "标注信息传输失败!",
  33. }
  34. }
  35. ret, err := vm.VerifySpiderConfig(sc) //验证
  36. if err != nil {
  37. qu.Debug("验证脚本配置失败::", err.Error())
  38. return map[string]interface{}{
  39. "ret": ret,
  40. "err": errtype,
  41. "msg": "验证脚本配置失败!",
  42. }
  43. }
  44. be.VerifyResults[sc.Code] = ret
  45. qu.Debug("验证结果---", *ret)
  46. return map[string]interface{}{
  47. "ret": ret,
  48. "err": 1,
  49. "msg": "验证脚本配置成功",
  50. }
  51. }
  52. // StopDebugSpider 停止调试
  53. func (a *App) StopDebugSpider() string {
  54. defer func() {
  55. if err := recover(); err != nil {
  56. qu.Debug(err)
  57. }
  58. }()
  59. exitCh <- true
  60. return "ok"
  61. }
  62. // ViewResultItemAll 查看所有结果,只显示最近的50条
  63. func (a *App) ViewResultItemAll(code string) be.ResultItems {
  64. ret := make(be.ResultItems, 0)
  65. index := 0
  66. if dataResult := be.DataResults[code]; dataResult != nil {
  67. for el := dataResult.Back(); el != nil; el = el.Prev() {
  68. if index > 50 {
  69. break
  70. }
  71. index += 1
  72. v, _ := el.Value.(*be.ResultItem)
  73. ret = append(ret, v)
  74. }
  75. }
  76. return ret
  77. }
  78. // ServerActionCurrentOpenTab 记录当前编辑code的cssmark,高亮回显使用
  79. func (a *App) ServerActionCurrentOpenTab(cssMark map[string]interface{}) {
  80. qu.Debug("----------------")
  81. sc, err := be.NewSpiderConfig(cssMark)
  82. if err != nil {
  83. qu.Debug("标注信息传输失败!")
  84. }
  85. bew.SetCurrentTabCssMark(sc)
  86. }
  87. // ExportEpubFile
  88. func (a *App) ExportEpubFile(bookname, filepath string) string {
  89. qu.Debug("EPUB 文件存储:", bookname, filepath)
  90. db.ExportEpubFile(bookname, filepath, currentResults)
  91. return "ok"
  92. }
  93. // ImportSpiderConfigByExcelFile 通过excel文件导入爬虫配置
  94. func (a *App) ImportSpiderConfigByExcelFile(filepath string) string {
  95. db.BatchImport(filepath)
  96. return "ok"
  97. }
  98. // CountYestodayArts
  99. func (a *App) CountYestodayArts(url string, listDealy int64, trunPageDelay int64, headless bool, showImage bool) {
  100. exitCh = make(chan bool, 1)
  101. vm.CountYestodayArts(url, listDealy, trunPageDelay, headless, showImage, exitCh, currentSpiderConfig)
  102. }
  103. // ExportExcelFile
  104. func (a *App) ExportExcelFile(filepath string) string {
  105. if err := db.ExportExcelFile(filepath, currentSpiderConfig.Site, currentSpiderConfig.Channel, currentResults); err == nil {
  106. return "ok"
  107. } else {
  108. return err.Error()
  109. }
  110. }
  111. // LoadAllJobs 加载我的所有作业
  112. func (a *App) LoadAllJobs() be.Jobs {
  113. rs, err := bdb.LoadEntities[be.Job]("jobs")
  114. if err != nil {
  115. return make(be.Jobs, 0)
  116. }
  117. jobs := be.Jobs(rs)
  118. sort.Sort(jobs)
  119. return jobs
  120. }
  121. // SaveJob
  122. func (a *App) SaveJob(job *be.Job) string {
  123. err := bdb.SaveEntity[be.Job]("jobs", job.Code, job)
  124. if err != nil {
  125. return err.Error()
  126. }
  127. return "ok"
  128. }
  129. // DeleteJob
  130. func (a *App) DeleteJob(code string) string {
  131. err := bdb.DeleteEntity[be.Job]("jobs", code)
  132. if err != nil {
  133. return err.Error()
  134. }
  135. return "ok"
  136. }
  137. // LoadJob
  138. func (a *App) LoadJob(code string) *be.Job {
  139. job, _ := bdb.LoadEntity[be.Job]("jobs", code)
  140. return job
  141. }
  142. // RunJob
  143. func (a *App) RunJob(code string) string {
  144. go vm.RunJob(code)
  145. return "ok"
  146. }
  147. func (a *App) StopJob(code string) string {
  148. vm.StopJob(code)
  149. return "ok"
  150. }
  151. // ExportJobResult
  152. func (a *App) ExportJobResult(code string, filePath string) string {
  153. vm.ExportJobResult(code, filePath)
  154. return "ok"
  155. }