It just depends how you want to have your project set up. That is how I usually set things up, a separate provider for the nav (with
trackUrlState turned off) and a search provider for the main search results page (with trackUrlState turned on). That's a totally legit setup.
Jason, I am a Software Engineer with NASA and I am working on a unique tool for NASA called insight. My manager has notice your search-ui as being very similar to what he is looking for, in terms of the build out of this tool. He has tasked me to look into it just a bit deeper, as he is not yet totally convinced that REACT is the right tool for the job. I, on the other hand, and fighting hard for the use of REACT because of its fantastic component use capabilities. The question that he has, and I, is in terms of data. I am not sure how large the data sets that you are searching are, but, the datasets, as I am sure you can imagine, are very large here. How do you deal with large datasets in terms of searching, not to bog down the tool and cause slow load times. My manager was indicating that facebook does some of this across the server. Wondering if you might have any ideas in terms of speeding up load time with large datasets using a tool like the one have created here?
@JasonStoltz
Hi @JasonStoltz , a quick question on buildRequest.js, if set as
bool: {
must: [match],
...(filter && { filter })
}it doesn't seem to do phrase match. match_phrase from https://www.elastic.co/guide/en/elasticsearch/reference/7.x/full-text-queries.html doesn't seem to be supported - is there any way to include phrase matches? Thanks!
@damonponder Search UI is just a tool for composing the front-end, the size of the data set has no impact on search-ui. It gathers user input, ships the data in a query to the backend, and then renders the results. That's it... it doesn't matter to search-ui whether your data set is 100 records or 100 million. That is a concern of whatever backend you choose. For questions on the scalability of App Search or Elasticsearch, I suggest checking out the docs (https://www.elastic.co/guide/index.html) or community forums (https://discuss.elastic.co/).
@sophijka The example we give in buildRequest.js is just a generic search query to get your started. Every Elasticsearch user will have different search needs. If you need that to support
match_phrase, simply adjust it as needed per the docs that you posted.
Thanks, @JasonStoltz! I also added synonyms, and it's really fun how they are highlighted in case query terms are not found. I do have a quick UI question, though. Paragraphs I've indexed tend to be long, and hence I updated
fragment_size in highlight. That said, I've often see a fragment where the found term is the last one and the right context is completely lost. Is there perhaps any way to center a fragment around a term (especially in cases of a single occurrence)? Many thanks, as always.
Good question @sophijka. You're getting out of my area of expertise here, have you checked out the Elasticsearch docs on highlighting? https://www.elastic.co/guide/en/elasticsearch/reference/7.10/highlighting.html. You might be able to adjust offset or something like that.
Gotta look it up
OK, so you can add a filter as a "Global Filter", in the "searchQuery" configuration option, and it will be applied to all queries, but have no impact on search state (i.e., it won't show up in the address bar). elastic/search-ui#473
Give it a shot, I need to add that to the docs as well.
Thanks, @JasonStoltz. One more question I have is on facets, or better say, running aggregate queries before executing the main query rather than post-filtering. This would be the case of listing all unique values for a particular field X, selecting a value from a drop-down list and passing it as an additional constraint when executing a query (a combination of a query typed by a user and the value of field X). I've tried to change buildRequest, yet I'm not sure if SearchUI supports it. Thanks again!
Hey sophijka. I think post filtering (or the disjunctive approach that is already implemented in onSearch) will give you the behavior you are looking for.
listing all unique values for a particular field X
That is what faceting / aggregations is for.
passing it as an additional constraint when executing a query
This is what post filter is used for (or alternatively, passing it as a regular filter and using the already implemented "disjunctive faceting" approach).
Search UI supports whatever queries you want to run. You can do literally anything you want to do in onSearch. You can run multiple queries, a single query, you can return static data ... literally anything you want. You just need to provide a function that accepts Request State and returns Response State .... https://github.com/elastic/search-ui/blob/master/ADVANCED.md#event-handlers
The provided code in the Elasticsearch example is just that, an example. You'll need to dig into it and get your hands dirty to customize it.
thanks, @JasonStoltz. This is more or less what I've implemented now (two queries, one aggregated to dynamically populate a drop down list, and another is the modification of search on search terms + selected value from the drop down list). Moving on to result export :-) Thanks for your support & happy holidays!
Hi Everyone.
We are running into scalability issues using Self Managed App Search and wondering whether someone here can help us.
One engine with 4,780 documents. We perform a search in our backend service but I am seeing that the API request to elasticsearch is taking 200-300ms every time. Is this normal? I was assuming it would be faster than this.
Also, our number of requests is increasing and under peak hrs (2000+ requests per second) we are getting “Service Unavailable” from the App Search endpoint.
Any help would be appreciated.
We are running into scalability issues using Self Managed App Search and wondering whether someone here can help us.
One engine with 4,780 documents. We perform a search in our backend service but I am seeing that the API request to elasticsearch is taking 200-300ms every time. Is this normal? I was assuming it would be faster than this.
Also, our number of requests is increasing and under peak hrs (2000+ requests per second) we are getting “Service Unavailable” from the App Search endpoint.
Any help would be appreciated.
Hello everyone. Thanks for the amazing work and resources that you are providing for everybody as open source <3
I am currently interested and experimenting with SearchUI and I was wondering if there are any projects/websites "built with SearchUI" that are publicly consultable besides the parks and video-games samples?
Some links would be very helpful.
I am currently interested and experimenting with SearchUI and I was wondering if there are any projects/websites "built with SearchUI" that are publicly consultable besides the parks and video-games samples?
Some links would be very helpful.
Maybe Meta Engines in App Search? https://www.elastic.co/guide/en/app-search/current/meta-engines-guide.html
It lets you query 2 separate engines as if they were one
Hello there.
I am building a search engine with App Search + Search UI and I hit a wall: Is there a way around in order to index files in full-text with App Search? the Documents API has a size limit of 100Kb and files' content that will be indexed in my use case can be bigger than that. I thought I could still index manually with a normal procedure or with FSCrawler but I can't even seem to find the indices created through App Search when accessing through Kibana for instance.
Is App Search ultimately not the solution I should be seeking for or did I just hit a cluelessness wall?
I am building a search engine with App Search + Search UI and I hit a wall: Is there a way around in order to index files in full-text with App Search? the Documents API has a size limit of 100Kb and files' content that will be indexed in my use case can be bigger than that. I thought I could still index manually with a normal procedure or with FSCrawler but I can't even seem to find the indices created through App Search when accessing through Kibana for instance.
Is App Search ultimately not the solution I should be seeking for or did I just hit a cluelessness wall?
I try to make a custom view to adjust the html structure
I created my own view and tried to replace the <Select> component from 'react-select' with my own html. But that doesn't work
I want to change the divs that the <Select> component creates to a <select> with multiple <input>
@dregearstrings_gitlab The configuration can be found in
config/enterprise-search.yml: https://www.elastic.co/guide/en/app-search/7.10/configuration.html#configuration-file
@IliasLG You need to be more specific than "that doesn't work". Please provide more information. In the meantime, here is an example that provides a custom view for sorting: https://codesandbox.io/s/search-ui-national-parks-example-with-sort-icons-q09dr
@IliasLG You need to be more specific than "that doesn't work". Please provide more information. In the meantime, here is an example that provides a custom view for sorting: https://codesandbox.io/s/search-ui-national-parks-example-with-sort-icons-q09dr
Already fixed it
Hi @JasonStoltz . A quick (and silly) question on ordering of fields in the results tab - I've seen a support ticket on that and checked the source code in
react-search-ui but couldn't find anything related to the order of fields. Is there a way to specify the desired field order (same for all documents in the result tab, say text, sector, ...) or would one need a custom component for that? Right now, fields seem to be ordered randomly and, with many metadata fields, the result doesn't look very user-friendly. Also, an id is needed for search (hence specified in buildRequest.jsrequest), but I'd rather hide it from the UI. Thanks & have a great weekend.
Looks like elasticsearch
I think you need to use a custom view for the result component
@harryct229 You can't disable it on individual queries. You can add analytics tags to queries though, so you can filter down to only queries that match a particular tag