2_doc_lines_delete_from_project.coffee 5.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184
  1. Settings = require "settings-sharelatex"
  2. fs = require("fs")
  3. mongojs = require("mongojs")
  4. ObjectId = mongojs.ObjectId
  5. db = mongojs(Settings.mongo.url, ['projects', 'docs'])
  6. _ = require("lodash")
  7. async = require("async")
  8. exec = require("child_process").exec
  9. finished_projects_path = "/tmp/finished-projects-2"
  10. all_projects_path = "/tmp/all-projects-2"
  11. unmigrated_docs_path = "/tmp/unmigrated-2"
  12. printProgress = ->
  13. exec "wc #{finished_projects_path}", (error, results) ->
  14. setTimeout printProgress, 1000 * 30
  15. checkIfFileHasBeenProccessed = (project_id, callback)->
  16. exec "grep #{project_id} #{finished_projects_path}", (error, results) ->
  17. hasBeenProcessed = _.include(results, project_id)
  18. callback(error, hasBeenProcessed)
  19. loadProjectIds = (callback)->
  20. console.log "loading project ids from #{all_projects_path}"
  21. fs.readFile all_projects_path, "utf-8", (err, data)->
  22. ids = data.split("\n")
  23. console.log "loaded #{ids.length} project ids from #{all_projects_path}"
  24. callback err, ids
  25. getAndWriteProjectids = (callback)->
  26. console.log "finding all project id's - #{new Date().toString()}"
  27. db.projects.find {}, {_id:1}, (err, ids)->
  28. console.log "total found projects in mongo #{ids.length} - #{new Date().toString()}"
  29. ids = _.pluck ids, '_id'
  30. ids = _.filter ids, (id)-> id?
  31. fileData = ids.join("\n")
  32. fs.writeFile all_projects_path, fileData, ->
  33. callback(err, ids)
  34. markDocAsUnmigrated = (project_id, doc_id, callback)->
  35. console.log "#{project_id} #{doc_id} unmigrated"
  36. markProjectAsProcessed project_id, (err)->
  37. fs.appendFile unmigrated_docs_path, "#{project_id} #{doc_id}\n", callback
  38. markUnmigratedDocs = (project_id, docs, callback)->
  39. console.log docs.length, project_id, "unmigrated"
  40. jobs = _.map docs, (doc)->
  41. (cb)->
  42. markDocAsUnmigrated project_id, doc._id, cb
  43. async.series jobs, callback
  44. getProjectIds = (callback)->
  45. exists = fs.existsSync all_projects_path
  46. if exists
  47. loadProjectIds callback
  48. else
  49. getAndWriteProjectids callback
  50. markProjectAsProcessed = (project_id, callback)->
  51. fs.appendFile finished_projects_path, "#{project_id}\n", callback
  52. getAllDocs = (project_id, callback = (error, docs) ->) ->
  53. excludes = {}
  54. for i in [0..12]
  55. excludes["rootFolder#{Array(i).join(".folders")}.docs.lines"] = 0
  56. db.projects.findOne _id: ObjectId(project_id.toString()), excludes, (error, project) ->
  57. return callback(error) if error?
  58. if !project?
  59. console.log "no such project #{project_id}"
  60. return callback()
  61. findAllDocsInProject project, (error, docs) ->
  62. return callback(error) if error?
  63. return callback null, docs, project
  64. findAllDocsInProject = (project, callback = (error, docs) ->) ->
  65. callback null, _findAllDocsInFolder project.rootFolder[0]
  66. findDocInProject = (project, doc_id, callback = (error, doc, mongoPath) ->) ->
  67. result = _findDocInFolder project.rootFolder[0], doc_id, "rootFolder.0"
  68. if result?
  69. callback null, result.doc, result.mongoPath
  70. else
  71. callback null, null, null
  72. _findDocInFolder = (folder = {}, doc_id, currentPath) ->
  73. for doc, i in folder.docs or []
  74. if doc?._id? and doc._id.toString() == doc_id.toString()
  75. return {
  76. doc: doc
  77. mongoPath: "#{currentPath}.docs.#{i}"
  78. }
  79. for childFolder, i in folder.folders or []
  80. result = _findDocInFolder childFolder, doc_id, "#{currentPath}.folders.#{i}"
  81. return result if result?
  82. return null
  83. _findAllDocsInFolder = (folder = {}) ->
  84. docs = folder.docs or []
  85. for childFolder in folder.folders or []
  86. docs = docs.concat _findAllDocsInFolder childFolder
  87. return docs
  88. isDocInDocCollection = (doc, callback)->
  89. if !doc?._id? or doc._id.length == 0
  90. return callback(null, true)
  91. db.docs.find({_id: ObjectId(doc._id+"")}, {_id: 1}).limit 1, (err, foundDocs)->
  92. exists = foundDocs.length > 0
  93. callback err, exists
  94. getWhichDocsCanBeDeleted = (docs, callback = (err, docsToBeDeleted, unmigratedDocs)->)->
  95. docsToBeDeleted = []
  96. unmigratedDocs = []
  97. jobs = _.map docs, (doc)->
  98. return (cb)->
  99. isDocInDocCollection doc, (err, exists)->
  100. if exists
  101. docsToBeDeleted.push doc
  102. else
  103. unmigratedDocs.push doc
  104. cb(err)
  105. async.series jobs, (err)->
  106. callback err, docsToBeDeleted, unmigratedDocs
  107. whipeDocLines = (project_id, mongoPath, callback)->
  108. update =
  109. $unset: {}
  110. update.$unset["#{mongoPath}.lines"] = ""
  111. update.$unset["#{mongoPath}.rev"] = ""
  112. db.projects.update _id: ObjectId(project_id+''), update, callback
  113. removeDocLinesFromProject = (docs, project, callback)->
  114. jobs = _.map docs, (doc)->
  115. (cb)->
  116. findDocInProject project, doc._id, (err, doc, mongoPath)->
  117. whipeDocLines project._id, mongoPath, cb
  118. async.parallelLimit jobs, 5, callback
  119. processNext = (project_id, callback)->
  120. if !project_id? or project_id.length == 0
  121. return callback()
  122. checkIfFileHasBeenProccessed project_id, (err, hasBeenProcessed)->
  123. if hasBeenProcessed
  124. console.log "#{project_id} already procssed, skipping"
  125. return callback()
  126. console.log "#{project_id} processing"
  127. getAllDocs project_id, (err, docs, project)->
  128. if err?
  129. console.error err, project_id, "could not get all docs"
  130. return callback(err)
  131. else
  132. getWhichDocsCanBeDeleted docs, (err, docsToBeDeleted, unmigratedDocs)->
  133. if err?
  134. console.error err, project_id, "could not save docs into mongo"
  135. return callback(err)
  136. markUnmigratedDocs project_id, unmigratedDocs, (err)->
  137. removeDocLinesFromProject docsToBeDeleted, project, (err)->
  138. if err?
  139. return callback(err)
  140. markProjectAsProcessed project_id, (err)->
  141. setTimeout(
  142. -> callback(err)
  143. ,0)
  144. exports.migrate = (client, done = ->)->
  145. getProjectIds (err, ids)->
  146. printProgress()
  147. jobs = _.map ids, (id)->
  148. return (cb)->
  149. processNext(id, cb)
  150. async.series jobs, (err)->
  151. if err?
  152. console.error err, "at end of jobs"
  153. else
  154. console.log "finished"
  155. done(err)
  156. exports.rollback = (next)->
  157. next()