1

I know there is already a lot on this subject out there, but none of the questions do help me going on. I have a File-Upload-System via GridFS which generates documents like this:

    { "_id" : ObjectId( "4f3723607b5c0b8d01060000" ),
  "filename" : "7326_150970547928_746052928_2521002_2164327_n.jpg",
  "filetype" : "image/jpeg",
  "title" : "Indexed File 1",
  "author" : "chewbacca",
  "document" : "Rechnungen",
  "keywords" : { "0" : "Darth",
    "1" : "Vader",
    "2" : "Haut",
    "5" : "Putz",
    "6" : "Pushy" },
  "uploadDate" : Date( 1329013600519 ),
  "length" : 61423,
  "chunkSize" : 262144,
  "md5" : "c73cce0bb6f349007635751f9a1a7ddd" }

As you can see I have a field 'keywords' which is an array of keywords. I want to build a search-option to search this field comparable to a fulltext search. Therefor I indexed the field 'keywords' seperately.

db.fs.files.ensureIndex({ keywords : 1 })

Now the problem is, that this works sometimes. Or to say yesterday it worked on some files, but on some it won't find anything.

Assuming I did the Indexing like above, I would think that 

> db.fs.files.find({keywords : "Vader"})

would give me the document printed above. Or am I missing something??

(My only explanation why this could be, is: it takes a lot of time to create indexes and it ain't ready yet, which is practically impossible right, or that there is some problem with the options 'background', 'dropDups', 'unique' etc...

I tried everything. I dropped the Indexes with; 

> db.fs.files.dropIndexes()

And created them again. Always controlling via

> db.fs.files.getIndexes()

But no, I can't get any results...

I also tried to make the indexing via PHP just after I saved the file in the database. For that I use the following code:

    $gridFS->ensureIndex(array('keywords' => 1), array("unique" => true));

Or also without the unique option.

As I said sometimes this works and I can query the keywords, but sometimes I can't. Or some keywords are found, but those of other documents not. Could it be that indexes ain't made for every document equally???

Please somebody help me on that, I really don't get the issue here!

Thanks already.

Merc
  • 4,241
  • 8
  • 52
  • 81

1 Answers1

3

I suggest you to use a true array in the keywords:

"keywords" : ["Darth", "Vader", "Haut", "Putz", "Pushy"],

So, the following is expected to work:

db.fs.files.ensureIndex({ keywords : 1 })
db.fs.files.find({keywords : "Vader"})
Sony Santos
  • 5,435
  • 30
  • 41
  • What do you mean with 'true array'? In the meantime I went on trying, and again achieved something, that partially works. On some files I do get the results properly, on some I don't... – Merc Feb 12 '12 at 18:12
  • In JSON, `{ 0 : "Darth", 1 : "Vader"}` is an object, not an Array; `["Darth", "Vader"]` is a *true* array; you can check it thru [instanceof Array](http://stackoverflow.com/questions/767486/how-do-you-check-if-a-variable-is-an-array-in-javascript). Please, give us the examples that don't work. – Sony Santos Feb 12 '12 at 21:53