Mongodb——GridFS
本文内容纲要：
GridFS用于存储和恢复那些超过16M（BSON文件限制）的文件。
GridFS将文件分成大块，将每个大块存储为单独的文件.GridFS中限制chunk最大为256k。GridFS使用两个collection存储，一个存储chunks，一个存储元数据(metadata)。
fs.files和fs.chunks
WhenshouldIuseGridFS?
http://docs.mongodb.org/manual/faq/developers/#faq-developers-when-to-use-gridfs
fileCollection：具体形式如下
{
"_id":
"length":
"chunkSize":
"uploadDate":
"md5":
"filename":
"contentType":
"aliases":
"metadata":
}
Documentsinthefilescollectioncontainsomeorallofthefollowingfields.Applicationsmaycreateadditionalarbitraryfields:
files._id
TheuniqueIDforthisdocument.The_idisofthedatatypeyouchosefortheoriginaldocument.ThedefaulttypeforMongoDBdocumentsisBSONObjectID.
files.length
Thesizeofthedocumentinbytes.
files.chunkSize
Thesizeofeachchunk.GridFSdividesthedocumentintochunksofthesizespecifiedhere.Thedefaultsizeis256kilobytes.
files.uploadDate
ThedatethedocumentwasfirststoredbyGridFS.ThisvaluehastheDatetype.
files.md5
AnMD5hashreturnedfromthefilemd5API.ThisvaluehastheStringtype.
files.filename
Optional.Ahuman-readablenameforthedocument.
files.contentType
Optional.AvalidMIMEtypeforthedocument.
files.aliases
Optional.Anarrayofaliasstrings.
files.metadata
Optional.Anyadditionalinformationyouwanttostore.
ThechunksCollection：举例如下
{
"_id":
"files_id":
"n":
"data":
}
Adocumentfromthechunkscollectioncontainsthefollowingfields:
chunks._id
TheuniqueObjectIDofthechunk.
chunks.files_id
The_idofthe“parent”document,asspecifiedinthefilescollection.
chunks.n
Thesequencenumberofthechunk.GridFSnumbersallchunks,startingwith0.
chunks.data
Thechunk’spayloadasaBSONbinarytype.
GridFSIndex
GridFS使用chunks中files_id和n域作为混合索引，files_id是父文档的_id，n域包含chunk的序列号，该值从0开始。
GridFS索引支持快速恢复数据。
cursor=db.fs.chunks.find({files_id:myFileID}).sort({n:1});
如果没有建立索引，可以使用下列shell命令：
db.fs.chunks.ensureIndex({files_id:1,n:1},{unique:true});
ExampleInterface：
//returnsdefaultGridFSbucket(i.e."fs"collection)
GridFSmyFS=newGridFS(myDatabase);
//savesthefileto"fs"GridFSbucket
myFS.createFile(newFile("/tmp/largething.mpg"));
接口支持额外的GridFSbuckets
//returnsGridFSbucketnamed"contracts"
GridFSmyContracts=newGridFS(myDatabase,"contracts");
//retrieveGridFSobject"smithco"
GridFSDBFilefile=myContracts.findOne("smithco");
//savestheGridFSfiletothefilesystem
file.writeTo(newFile("/tmp/smithco.pdf"));
本文内容总结：
原文链接：https://www.cnblogs.com/bigbigtree/p/3156554.html