Created
April 16, 2014 20:38
-
-
Save AndrewO/10930544 to your computer and use it in GitHub Desktop.
Revisions
-
AndrewO renamed this gist
Apr 16, 2014 . 1 changed file with 0 additions and 0 deletions.There are no files selected for viewing
File renamed without changes. -
Apr 16, 2014 .There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters. Learn more about bidirectional Unicode charactersOriginal file line number Diff line number Diff line change @@ -0,0 +1,211 @@ # We have a couple of indices that have been created independently. # Now we want to search across them, with highlighting (and ideally # suggestors as well) without breaking exisiting services. # # The mappings do not have consistent field names, but they have fields that can # be semantically mapped into "title" or "body" (and some others) fields. # # From the size of the data, I'm told that creating a new purpose-build index is # out of the question--even something that would duplicate data in another field # in the same index is a stretch. # # Which is what lead me to look at `index_name` and `copy_to`. PUT /restaurants # For this type, `name` and `description` map 1:1 to `title` and `body` PUT /restaurants/fast_food/_mapping { "fast_food" : { "properties": { "name" : { "type": "string", "index_name": "title"}, "description" : { "type": "string", "index_name": "body"} } } } PUT /restaurants/fast_food/1 { "name": "Trough King", "description": "French fries injected directly into your heart!" } PUT /restaurants/fast_food/2 { "name": "Charles Cheese Family Fun Fest", "description": "Lots and lots of cheese. Also, robotic rats wearing funny hats." } # For this type, we have two fields that we want to treat as `body`, so we're going to treat # `summary` as the "main" `body` field, and `letter_from_chef` as a "secondary" one. PUT /restaurants/fancy/_mapping { "fancy" : { "properties": { "name" : { "type": "string", "index_name": "title"}, "summary" : { "type": "string", "index_name": "body"}, "letter_from_chef" : { "type": "string", "copy_to": "body"} } } } PUT /restaurants/fancy/1 { "name": "Chez Fromage", "summary": "French food masterfully cooked by some guy in a funny hat. Also, great cheese.", "letter_from_chef": "Bonjour! Enjoy my cheese!" } # This example's only here to test multi-index searching PUT /hotels PUT /hotels/hotel/_mapping { "hotel": { "properties": { "hotelName" : { "type": "string", "index_name": "title"}, "hotelDescription": { "type": "string", "index_name": "body"}, "hotelRating": {"type": "integer"} } } } PUT /hotels/hotel/1 { "hotelName": "Sleep Already", "hotelDescription": "Okay, I guess. Room smelled like cheese, but no rats, so can't complain, right?", "hotelRating": 2 } # Check that everything looks right... GET /restaurants,hotels/_mapping # Note that all of the fields are returned by mapping name, not index name (which makes sense). GET /restaurants,hotels/_search {"query": {"match_all": {}}} # Let's make sure our existing hotel search still works... GET /hotels/_search { "highlight": { "fields": { "hotelName": {}, "hotelDescription": {} } }, "query": { "filtered": { "query": { "multi_match": { "query": "cheese", "fields": [ "hotelName^2", "hotelDescription" ] } }, "filter": { "range": { "hotelRating": { "gte": 2} } } } } } # Good. As expected, searching by mapping names still works. # A highlighted multi-index search: GET /restaurants,hotels/_search { "highlight": { "fields": { "title": {}, "body": { "number_of_fragments": 3 } } }, "query": { "multi_match": { "query": "cheese", "fields": [ "title^2", "body" ] } } } # Pretty good: it queries our fields by "index_name" and highlights them. # However, restaurant/fancy/1's body highlight only contains a match from `summary`. # Maybe the copy_to isn't working? GET /restaurants,hotels/_search { "highlight": { "fields": { "title": {}, "body": { "number_of_fragments": 3 } } }, "query": { "multi_match": { "query": "bonjour", "fields": [ "title^2", "body" ] } } } # Nope, it matched on the `letter_from_chef` field, but no highlighting... # Can we restrict the source fields by index_name? GET /restaurants,hotels/_search { "highlight": { "fields": { "title": {}, "body": { "number_of_fragments": 3 } } }, "query": { "multi_match": { "query": "cheese", "fields": [ "title^2", "body" ] } }, "_source": ["title", "body"] } # No. Neither the "copy_to" or "index_name" fields show up. # Okay, so highlighting doesn't work as I thought it would here. How about suggestors? # Suggestors can use copied fields GET /restaurants,hotels/_search { "highlight": { "fields": { "title": {}, "body": { "number_of_fragments": 3 } } }, "query": { "multi_match": { "query": "eat my ches", "fields": [ "title^2", "body" ] } }, "suggest": { "text": "eat my ches", "my-suggestor": { "phrase": { "size": 3, "field": "body" } } } } # Interesting. The suggestor works fine with a copy_to field. (Not shown: works with index_name ones too. Try 'funn hat').