imbo/imbo-metadata-search

Metadata search plugin for Imbo

, _(*1)

Metadata search plugin for Imbo

The metadata search event listener hooks onto metadata updates for your images and keeps the search backend of your choice up to date, and allows you to find images by querying its metadata., _(*2)

Installation

Setting up the dependencies

If you've installed Imbo through composer, getting the metadata search up and running is really simple. Simply add imbo/imbo-metadata-search as a dependency., _(*3)

In addition to the metadata search plugin you'll need a search backend client. Right now the plugin ships with support for elasticsearch only, so you'll want to add elasticsearch/elasticsearch as well in order to be able to use it as search backend., _(*4)

{
    "require": {
        "imbo/imbo-metadata-search": "dev-master",
        "elasticsearch/elasticsearch": "~2.1"
    }
}

Metadata search setup

In order for the metadata search plugin to be registered and actually do something usedful for your Imbo installation you need to add a config file which declares the routes, resource and event listeners., _(*5)

After installing with composer you will find a basic config file for the metadata search in vendor/imbo/imbo-metadata-search/config.dist.php. If you want to make changes to the file you should copy it to your config folder., _(*6)

Indexing

Updates in the search backend is triggered whenever one of the following events are fired; image.delete, images.post, image.post, metadata.post, metadata.put, metadata.delete., _(*7)

The image.delete event triggers a delete in the indexed object in the search backend, and the other ones trigger an update of the full object. When indexing, data in addition to metadata is provided to the search backend for indexin in order to sorting and such., _(*8)

The data provided to the backend are;, _(*9)

Querying

Querying is done by issuing an HTTP SEARCH request to /users/<user>/images if you want to search in the images of a single user, or /images if you want to search across multiple users. Supported query parameters are:, _(*10)

The query is sent in the request body., _(*11)

Examples

Querying one user, _(*12)

$ curl 'http://imbo/users/<user>/images?limit=1&metadata=1' -d '{"foo": "bar"}'

Querying multiple users, _(*13)

$ curl 'http://imbo/images?users[]=<user1>&users[]=<user2>&limit=1&metadata=1' -d '{"foo": "bar"}'

Both these requests results in a response that looks like this:, _(*14)

{
  "search": {
    "hits": 3,
    "page": 1,
    "limit": 1,
    "count": 1
  },
  "images": [
    {
      "added": "Mon, 10 Dec 2012 11:57:51 GMT",
      "updated": "Mon, 10 Dec 2012 11:57:51 GMT",
      "checksum": "<checksum>",
      "originalChecksum": "<originalChecksum>",
      "extension": "png",
      "size": 6791,
      "width": 1306,
      "height": 77,
      "mime": "image/png",
      "imageIdentifier": "<image>",
      "user": "<user>",
      "metadata": {
        "key": "value",
        "foo": "bar"
      }
    }
  ]
}

Imbo DSL

The query language used by Imbo Metadata Search is a subset of the MongoDB query DSL. The query is a JSON-encoded object including key => value matches and/or a combination of the supported operators, sent to Imbo in the request body. This section lists all operators and includes a number of examples showing you how to find images using the metadata query., _(*15)

Note: The results of the different queries might end up with slightly different results depending on the backend you use the for metadata., _(*16)

Key/value matching

The simplest form of a metadata query is a simple key => value match, where the expressions are AND-ed together if there is more than one key/value match in the query., _(*17)

 {"key":"value","otherkey":"othervalue"}

The above search would result in images that have the metadata key key set to value and otherkey set to othervalue, _(*18)

Greater than - $gt

This operator can be used to check for values greater than the value specified., _(*19)

{"age":{"$gt":35}}

Greater than or equal - $gte

Check for values greater than or equal to the value specified., _(*20)

 {"age":{"$gte":35}}

Less than - $lte

Check for values less than to the value specified., _(*21)

{"age":{"$lt":35}}

Less than or equal - $lte

Check for values less than or equal to the value specified., _(*22)

{"age":{"$lte":35}}

Not equal - $ne

Matches values that are not equal to the value specified., _(*23)

{"name":{"$ne":"christer"}}

In - $in

Look for values that appear in the specified set., _(*24)

{"styles":{"$in":["IPA","Imperial Stout","Lambic"]}}

Not in - $nin

Look for values that does not appear in the specified set., _(*25)

{"styles":{"$nin":["Pilsner"]}}

Conjunctions - $and

This operator can be used to combine a list of criteria that must all match. It takes an array of queries., _(*26)

{"$and": [{"name": {"$in": ["kristoffer", "morten"]}}, {"age": {"$lt": 30}}]}

Would find images where the key name is either kristoffer or morten and where the age key is less than 30., _(*27)

Disjunction - $or

This operator can be used to combine a list of criteria where at least one must match. It takes an array of queries., _(*28)

{"$or":[{"key":"value"},{"otherkey":"othervalue"}]}

Would fetch images that have a key named key with the value value and/or a key named otherkey which has the value of othervalue., _(*29)

Using several operators in one query

All the above operators can be combined into one query. Consider a collection of images of beers which have all been tagged with the name of the brewery, the name of the beer, the style of the beer and the ABV. If we wanted to find all images of beers within a set of styles, above a specific ABV, from two different breweries, and all images of beers from Nøgne Ø, regardless of style and ABV, but not beers called Wit, regardless of brewery, style or ABV, the query could look like this (formatted for easier reading):, _(*30)

{
    "name":
    {
        "$ne": "Wit"
    },
    "$or":
    [
        {
            "brewery": "Nøgne Ø"
        },

        {
            "$and":
            [
                {
                    "abv":
                    {
                        "$gte": 5.5
                    }
                },

                {
                    "style":
                    {
                        "$in":
                        [
                            "IPA",
                            "Imperial Stout"
                        ]
                    }
                },

                {
                    "brewery":
                    {
                        "$in":
                        [
                            "HaandBryggeriet",
                            "Ægir"
                        ]
                    }
                }
            ]
        }
    ]
}

Keep in mind that large complex queries against large image collections can take a while to finish, and might cause performance issues on the Imbo server(s)., _(*31)

License

Copyright (c) 2015, Kristoffer Brabrand and Morten Fangel, _(*32)

Licensed under the MIT License, _(*33)