====== Similarity Index API ======



===== branch01.picalike.corpex-kunden.de:9095/index_query =====

loadbalancer: http://similarity-index.lb.int.picalike.corpex-kunden.de:80/\\
url (legacy oder upload): http://branch01.picalike.corpex-kunden.de:9095/index_query\\
host: **branch01.picalike.corpex-kunden.de**\\
port: **9095**\\
service name: **/index_query**\\
method: POST with a JSON


===== reload an index =====

all workers: http://branch01.picalike.corpex-kunden.de:9095/index_reload\\
worker one by one:\\
* **tower01**: http://tower01.picalike.corpex-kunden.de:4238/reload\\
* **dev01**: http://dev01.picalike.corpex-kunden.de:4238/reload\\
method: POST with a JSON\\
* **index_name** = The name of the indexes to be reloaded


===== git =====

picalike@sg01.picalike.corpex-kunden.de/home/picalike/repositories/v5/picalike_v5.git


===== deployment =====

SSH auf die Server, picalike@dev01 und neuron_cloud@tower01\\
screen -r index\\
es gibt zwei screen tabs:\\



  * **frontend** = uwsgi –ini uwsgi_config_index_frontend.ini\\

  * **backend** = python3 feature_similarity_service_all_shops.py\\



===== parameters =====


  * **index_name** = The name of the indexes used for the loopup. Default:“all_images”. For every shop_id, there is an Index with the same name. A list of names is also possible.
  * **k** = Number of images used for the similarity search


  * **shop_id** = The shop the root image is in
  * **prod_id** = The product if of the root image
  * **or**
  * **features** = list of features
  * **or if branch01 is called**
  * **myfile** = an image coded as base64 string
  * **return_metadata** = if this is set to any value, the full documents from the meta-data-db are returned for each product
  * **return_negativ** = if this is set to any value, all products are additionally returned that did not match the filter or have no metadata but were close in the index
  * **meta_data_filter** = A dict in mongo style. Filters results in the 'metadata' field. There might be less then k products left after filtering. Examples:


  * 1 Shop:

<code>
{"shop_id":"witt_de_feed"}
</code>

  * N shops:

<code>
{"shop_id":{"$in":["witt_de_feed", "peterhahn_de_crawler"]}}
</code>

  * last changed timestamp:

<code>
{"timestamp" :{
        "$gte": 1544199370.0,
        "$lt": 1544199371.0
    }
}
</code>

  * Minimal-Preis kleiner 50.00€ oder Maximal-Preis größer 10.00€

<code>
{"$or":[
    {"metadata.min_price" :{
            "$lt": 5000
        }
    },
    {"metadata.max_price" :{
            "$gte": 1000
        }
    }
    ]
}
</code>

  * Von Witt und zwischen x und y verändert:

<code>
{"$and":[
    {"shop_id":"witt_de_feed"},
    {"timestamp" :{
            "$gte": 1544199370.0,
            "$lt": 1544199371.0
        }
    }
    ]
}
</code>

===== out =====

'root_images' → the reference image\\
'close_images_filtered_in' → the closest images, that match the filter\\
'hostname_worker' → the machine that did the job\\
'last_index_update' → last time the index in the worker was updated last_index_update\\
'errors' → list of happend errors\\


<code>
{
    "_time_1_overall_ms": 7.35,
    "_time_2_get_args_ms": 0.15,
    "_time_3_get_root_ms": 2.14,
    "_time_4_feature_request_ms": 1.1,
    "_time_5_get_prod_ids_ms": 0.98,
    "_time_6_get_meta_data_ms": 2.98,
    "_time_7_sort_meta_data_ms": 0.02,
    "close_images_filtered_in": [
        {
            "distance": 0.0,
            "int_id": 5,
            "product_data": [
                {
                    "matches_db_filter": true,
                    "picalike_id": "9690328#asos_de_crawler",
                    "prod_id": "9690328",
                    "shop_id": "asos_de_crawler"
                }
            ],
            "url": "https://images.asos-media.com/products/asos-design-marineblaue-touchscreen-handschuhe/9690328-1-navy"
        },
        {
            "distance": 1.0188307896896731e-05,
            "int_id": 717894,
            "product_data": [
                {
                    "matches_db_filter": true,
                    "picalike_id": "9707579#asos_de_crawler",
                    "prod_id": "9707579",
                    "shop_id": "asos_de_crawler"
                }
            ],
            "url": "https://images.asos-media.com/products/asos-design-touch-screen-handschuhe-aus-wieder-aufbereitetem-polyester/9707579-1-navy"
        }
    ],
    "product_data_flat": [
        {
            "distance": 0.0,
            "matches_db_filter": true,
            "picalike_id": "9690328#asos_de_crawler",
            "prod_id": "9690328",
            "shop_id": "asos_de_crawler"
        },
        {
            "distance": 1.0188307896896731e-05,
            "matches_db_filter": true,
            "picalike_id": "9707579#asos_de_crawler",
            "prod_id": "9707579",
            "shop_id": "asos_de_crawler"
        }
    ],
    "root_images": [
        {
            "picalike_id": [
                "9690328#asos_de_crawler"
            ],
            "prod_id": [
                "9690328"
            ],
            "shop_id": [
                "asos_de_crawler"
            ],
            "url": "https://images.asos-media.com/products/asos-design-marineblaue-touchscreen-handschuhe/9690328-1-navy"
        }
    ]
}
</code>