bind4spider.go 5.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209
  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. bew.SetCurrentTabCssMark(cssMark)
  81. }
  82. // ExportEpubFile epub导出
  83. func (a *App) ExportEpubFile(bookname, filepath, code string) map[string]interface{} {
  84. sc := be.DataResults[code]
  85. qu.Debug("结果导出:", bookname, filepath, code)
  86. qu.Debug("EPUB 文件存储:", bookname, filepath)
  87. var msg string
  88. var errType int
  89. if sc != nil {
  90. if err := a.RunExportEpubFile(bookname, filepath, sc); err != nil {
  91. msg = err.Error()
  92. } else {
  93. msg = "导出成功"
  94. errType = 1
  95. }
  96. } else {
  97. msg = "导出失败,无法获取结果"
  98. }
  99. return map[string]interface{}{"err": errType, "msg": msg}
  100. }
  101. // ExportJsonFile json导出
  102. func (a *App) ExportJsonFile(filepath, code string) map[string]interface{} {
  103. sc := be.DataResults[code]
  104. qu.Debug("结果导出:", filepath, code)
  105. var msg string
  106. var errType int
  107. if sc != nil {
  108. if err := a.RunExportJsonFile(filepath, code, sc); err != nil {
  109. msg = err.Error()
  110. } else {
  111. msg = "导出成功"
  112. errType = 1
  113. }
  114. } else {
  115. msg = "导出失败,无法获取结果"
  116. }
  117. return map[string]interface{}{"err": errType, "msg": msg}
  118. }
  119. // ExportExcelFile excel导出
  120. func (a *App) ExportExcelFile(filepath, code string) map[string]interface{} {
  121. sc := be.DataResults[code]
  122. qu.Debug("结果导出:", filepath, code)
  123. var msg string
  124. var errType int
  125. if sc != nil {
  126. if err := a.RunExportExcelFile(filepath, code, sc); err != nil {
  127. msg = err.Error()
  128. } else {
  129. msg = "导出成功"
  130. errType = 1
  131. }
  132. } else {
  133. msg = "导出失败,无法获取结果"
  134. }
  135. return map[string]interface{}{"err": errType, "msg": msg}
  136. }
  137. // ImportSpiderConfigByExcelFile 通过excel文件导入爬虫配置
  138. func (a *App) ImportSpiderConfigByExcelFile(filepath string) string {
  139. db.BatchImport(filepath)
  140. return "ok"
  141. }
  142. // CountYestodayArts
  143. func (a *App) CountYestodayArts(url string, listDealy int64, trunPageDelay int64, headless bool, showImage bool) {
  144. exitCh = make(chan bool, 1)
  145. vm.CountYestodayArts(url, listDealy, trunPageDelay, headless, showImage, exitCh, currentSpiderConfig)
  146. }
  147. // LoadAllJobs 加载我的所有作业
  148. func (a *App) LoadAllJobs() be.Jobs {
  149. rs, err := bdb.LoadEntities[be.Job]("jobs")
  150. if err != nil {
  151. return make(be.Jobs, 0)
  152. }
  153. jobs := be.Jobs(rs)
  154. sort.Sort(jobs)
  155. return jobs
  156. }
  157. // SaveJob
  158. func (a *App) SaveJob(job *be.Job) string {
  159. err := bdb.SaveEntity[be.Job]("jobs", job.Code, job)
  160. if err != nil {
  161. return err.Error()
  162. }
  163. return "ok"
  164. }
  165. // DeleteJob
  166. func (a *App) DeleteJob(code string) string {
  167. err := bdb.DeleteEntity[be.Job]("jobs", code)
  168. if err != nil {
  169. return err.Error()
  170. }
  171. return "ok"
  172. }
  173. // LoadJob
  174. func (a *App) LoadJob(code string) *be.Job {
  175. job, _ := bdb.LoadEntity[be.Job]("jobs", code)
  176. return job
  177. }
  178. // RunJob
  179. func (a *App) RunJob(code string) string {
  180. go vm.RunJob(code)
  181. return "ok"
  182. }
  183. func (a *App) StopJob(code string) string {
  184. vm.StopJob(code)
  185. return "ok"
  186. }
  187. // ExportJobResult
  188. func (a *App) ExportJobResult(code string, filePath string) string {
  189. vm.ExportJobResult(code, filePath)
  190. return "ok"
  191. }