|
@@ -6,12 +6,13 @@ async = require "async"
|
|
|
settings = require("settings-sharelatex")
|
|
settings = require("settings-sharelatex")
|
|
|
request = require("request")
|
|
request = require("request")
|
|
|
crypto = require("crypto")
|
|
crypto = require("crypto")
|
|
|
|
|
+RangeManager = require("./RangeManager")
|
|
|
thirtySeconds = 30 * 1000
|
|
thirtySeconds = 30 * 1000
|
|
|
|
|
|
|
|
module.exports = DocArchive =
|
|
module.exports = DocArchive =
|
|
|
|
|
|
|
|
archiveAllDocs: (project_id, callback = (err, docs) ->) ->
|
|
archiveAllDocs: (project_id, callback = (err, docs) ->) ->
|
|
|
- MongoManager.getProjectsDocs project_id, {include_deleted: true}, {lines: true, rev: true, inS3: true}, (err, docs) ->
|
|
|
|
|
|
|
+ MongoManager.getProjectsDocs project_id, {include_deleted: true}, {lines: true, ranges: true, rev: true, inS3: true}, (err, docs) ->
|
|
|
if err?
|
|
if err?
|
|
|
return callback(err)
|
|
return callback(err)
|
|
|
else if !docs?
|
|
else if !docs?
|
|
@@ -26,21 +27,26 @@ module.exports = DocArchive =
|
|
|
archiveDoc: (project_id, doc, callback)->
|
|
archiveDoc: (project_id, doc, callback)->
|
|
|
logger.log project_id: project_id, doc_id: doc._id, "sending doc to s3"
|
|
logger.log project_id: project_id, doc_id: doc._id, "sending doc to s3"
|
|
|
try
|
|
try
|
|
|
- options = DocArchive.buildS3Options(doc.lines, project_id+"/"+doc._id)
|
|
|
|
|
|
|
+ options = DocArchive.buildS3Options(project_id+"/"+doc._id)
|
|
|
catch e
|
|
catch e
|
|
|
return callback e
|
|
return callback e
|
|
|
- request.put options, (err, res)->
|
|
|
|
|
- if err? || res.statusCode != 200
|
|
|
|
|
- logger.err err:err, res:res, project_id:project_id, doc_id: doc._id, statusCode: res?.statusCode, "something went wrong archiving doc in aws"
|
|
|
|
|
- return callback new Error("Error in S3 request")
|
|
|
|
|
- md5lines = crypto.createHash("md5").update(JSON.stringify(doc.lines), "utf8").digest("hex")
|
|
|
|
|
- md5response = res.headers.etag.toString().replace(/\"/g, '')
|
|
|
|
|
- if md5lines != md5response
|
|
|
|
|
- logger.err responseMD5:md5response, linesMD5:md5lines, project_id:project_id, doc_id: doc?._id, "err in response md5 from s3"
|
|
|
|
|
- return callback new Error("Error in S3 md5 response")
|
|
|
|
|
- MongoManager.markDocAsArchived doc._id, doc.rev, (err) ->
|
|
|
|
|
- return callback(err) if err?
|
|
|
|
|
- callback()
|
|
|
|
|
|
|
+ DocArchive._mongoDocToS3Doc doc, (error, json_doc) ->
|
|
|
|
|
+ return callback(error) if error?
|
|
|
|
|
+ options.body = json_doc
|
|
|
|
|
+ options.headers =
|
|
|
|
|
+ 'Content-Type': "application/json"
|
|
|
|
|
+ request.put options, (err, res) ->
|
|
|
|
|
+ if err? || res.statusCode != 200
|
|
|
|
|
+ logger.err err:err, res:res, project_id:project_id, doc_id: doc._id, statusCode: res?.statusCode, "something went wrong archiving doc in aws"
|
|
|
|
|
+ return callback new Error("Error in S3 request")
|
|
|
|
|
+ md5lines = crypto.createHash("md5").update(json_doc, "utf8").digest("hex")
|
|
|
|
|
+ md5response = res.headers.etag.toString().replace(/\"/g, '')
|
|
|
|
|
+ if md5lines != md5response
|
|
|
|
|
+ logger.err responseMD5:md5response, linesMD5:md5lines, project_id:project_id, doc_id: doc?._id, "err in response md5 from s3"
|
|
|
|
|
+ return callback new Error("Error in S3 md5 response")
|
|
|
|
|
+ MongoManager.markDocAsArchived doc._id, doc.rev, (err) ->
|
|
|
|
|
+ return callback(err) if err?
|
|
|
|
|
+ callback()
|
|
|
|
|
|
|
|
unArchiveAllDocs: (project_id, callback = (err) ->) ->
|
|
unArchiveAllDocs: (project_id, callback = (err) ->) ->
|
|
|
MongoManager.getArchivedProjectDocs project_id, (err, docs) ->
|
|
MongoManager.getArchivedProjectDocs project_id, (err, docs) ->
|
|
@@ -60,26 +66,52 @@ module.exports = DocArchive =
|
|
|
unarchiveDoc: (project_id, doc_id, callback)->
|
|
unarchiveDoc: (project_id, doc_id, callback)->
|
|
|
logger.log project_id: project_id, doc_id: doc_id, "getting doc from s3"
|
|
logger.log project_id: project_id, doc_id: doc_id, "getting doc from s3"
|
|
|
try
|
|
try
|
|
|
- options = DocArchive.buildS3Options(true, project_id+"/"+doc_id)
|
|
|
|
|
|
|
+ options = DocArchive.buildS3Options(project_id+"/"+doc_id)
|
|
|
catch e
|
|
catch e
|
|
|
return callback e
|
|
return callback e
|
|
|
- request.get options, (err, res, lines)->
|
|
|
|
|
|
|
+ options.json = true
|
|
|
|
|
+ request.get options, (err, res, doc)->
|
|
|
if err? || res.statusCode != 200
|
|
if err? || res.statusCode != 200
|
|
|
logger.err err:err, res:res, project_id:project_id, doc_id:doc_id, "something went wrong unarchiving doc from aws"
|
|
logger.err err:err, res:res, project_id:project_id, doc_id:doc_id, "something went wrong unarchiving doc from aws"
|
|
|
return callback new Errors.NotFoundError("Error in S3 request")
|
|
return callback new Errors.NotFoundError("Error in S3 request")
|
|
|
- if !(lines instanceof Array)
|
|
|
|
|
- logger.err err:err, res:res, project_id:project_id, doc_id:doc_id, lines:lines, "doc lines from aws are not in array format, likely not JSON parsable"
|
|
|
|
|
- return callback(new Error("Error unpacking doc"))
|
|
|
|
|
- MongoManager.upsertIntoDocCollection project_id, doc_id.toString(), {lines}, (err) ->
|
|
|
|
|
- return callback(err) if err?
|
|
|
|
|
- logger.log project_id: project_id, doc_id: doc_id, "deleting doc from s3"
|
|
|
|
|
- request.del options, (err, res, body)->
|
|
|
|
|
- if err? || res.statusCode != 204
|
|
|
|
|
- logger.err err:err, res:res, project_id:project_id, doc_id:doc_id, "something went wrong deleting doc from aws"
|
|
|
|
|
- return callback new Errors.NotFoundError("Error in S3 request")
|
|
|
|
|
- callback()
|
|
|
|
|
|
|
+ DocArchive._s3DocToMongoDoc doc, (error, mongo_doc) ->
|
|
|
|
|
+ return callback(error) if error?
|
|
|
|
|
+ MongoManager.upsertIntoDocCollection project_id, doc_id.toString(), mongo_doc, (err) ->
|
|
|
|
|
+ return callback(err) if err?
|
|
|
|
|
+ logger.log project_id: project_id, doc_id: doc_id, "deleting doc from s3"
|
|
|
|
|
+ request.del options, (err, res, body)->
|
|
|
|
|
+ if err? || res.statusCode != 204
|
|
|
|
|
+ logger.err err:err, res:res, project_id:project_id, doc_id:doc_id, "something went wrong deleting doc from aws"
|
|
|
|
|
+ return callback new Errors.NotFoundError("Error in S3 request")
|
|
|
|
|
+ callback()
|
|
|
|
|
+
|
|
|
|
|
+ _s3DocToMongoDoc: (doc, callback = (error, mongo_doc) ->) ->
|
|
|
|
|
+ mongo_doc = {}
|
|
|
|
|
+ if doc.schema_v == 1 and doc.lines?
|
|
|
|
|
+ mongo_doc.lines = doc.lines
|
|
|
|
|
+ if doc.ranges?
|
|
|
|
|
+ mongo_doc.ranges = RangeManager.jsonRangesToMongo(doc.ranges)
|
|
|
|
|
+ else if doc instanceof Array
|
|
|
|
|
+ mongo_doc.lines = doc
|
|
|
|
|
+ else
|
|
|
|
|
+ return callback(new Error("I don't understand the doc format in s3"))
|
|
|
|
|
+ return callback null, mongo_doc
|
|
|
|
|
+
|
|
|
|
|
+ _mongoDocToS3Doc: (doc, callback = (error, s3_doc) ->) ->
|
|
|
|
|
+ if !doc.lines?
|
|
|
|
|
+ return callback(new Error("doc has no lines"))
|
|
|
|
|
+ json = JSON.stringify({
|
|
|
|
|
+ lines: doc.lines
|
|
|
|
|
+ ranges: doc.ranges
|
|
|
|
|
+ schema_v: 1
|
|
|
|
|
+ })
|
|
|
|
|
+ if json.indexOf("\u0000") != -1
|
|
|
|
|
+ error = new Error("null bytes detected")
|
|
|
|
|
+ logger.err {err: error, doc, json}, error.message
|
|
|
|
|
+ return callback(error)
|
|
|
|
|
+ return callback null, json
|
|
|
|
|
|
|
|
- buildS3Options: (content, key)->
|
|
|
|
|
|
|
+ buildS3Options: (key)->
|
|
|
if !settings.docstore.s3?
|
|
if !settings.docstore.s3?
|
|
|
throw new Error("S3 settings are not configured")
|
|
throw new Error("S3 settings are not configured")
|
|
|
return {
|
|
return {
|
|
@@ -88,6 +120,5 @@ module.exports = DocArchive =
|
|
|
secret: settings.docstore.s3.secret
|
|
secret: settings.docstore.s3.secret
|
|
|
bucket: settings.docstore.s3.bucket
|
|
bucket: settings.docstore.s3.bucket
|
|
|
timeout: thirtySeconds
|
|
timeout: thirtySeconds
|
|
|
- json: content
|
|
|
|
|
uri:"https://#{settings.docstore.s3.bucket}.s3.amazonaws.com/#{key}"
|
|
uri:"https://#{settings.docstore.s3.bucket}.s3.amazonaws.com/#{key}"
|
|
|
}
|
|
}
|