Security News
GitHub Removes Malicious Pull Requests Targeting Open Source Repositories
GitHub removed 27 malicious pull requests attempting to inject harmful code across multiple open source repositories, in another round of low-effort attacks.
splainer-search
Advanced tools
Angular JS Solr and Elasticsearch and OpenSearch Diagnostic Search Services
Splainer Search is an Angular Solr and OpenSearch and Elasticsearch Search library focussed on relevance diagnostics. It's used in the relevancy tuning tools Quepid and Splainer. Its available for anyone to use (see license).
Splainer search utilizes a JSONP wrapper for communication with Solr. Elasticsearch and OpenSearch communicate with simple HTTP and JSON via CORS. All fields are explained and highlighted if requested. A friendly interface is provided to specify the arguments in terms of a Javascript object. See below for basic examples.
Splainer-search will perform the specified search against Solr attempting to highlight and extract explain info. To request highlighting on a specific field, prefix the fieldname with "hl:"
i.e: hl:overview
.
// searcher that searches id, title, body, author
var searcher = searchSvc.createSearcher(
['id', 'title', 'hl:body', 'author'],
'http://localhost:8983/solr/select',
{
'q': ['*:*'],
'fq': ['title:Moby*', 'author:Herman']
}
);
searcher.search()
.then(function() {
angular.forEach(searcher.docs, function(doc) {
console.log(doc.source().title);
// highlights. You need to pass id as that's how Solr
// organizes the explain. See below for a friendlier/higher-level
// interface with normalDocs
console.log(doc.highlight(doc.source().id, 'title', '<b>', '</b>');
// explain info
console.log(doc.explain(doc.source().id);
});
});
Note: For now, we have a set of es*.js
files that support both search engines
Splainer-search supports these search engines, using the same API, and passing the query DSL in the same way ES expects it:
var searcher = searchSvc.createSearcher(
['id:_id', 'title', 'body', 'author'],
'http://localhost:9200/books/_search',
{
'query': {
'match': {
'title': '#$query##'
}
}
}
);
Paging is done by asking the original searcher for another searcher. This searcher is already setup to get the next page for the current search results. Tell that searcher to search()
just like you did above.
var results = [];
searcher.search()
.then(function() {
angular.forEach(searcher.docs, function(doc) {
results.push(doc.source().title));
});
// once results returned, get a new searcher for the next
// page of results, just rerun the search later exactly as
// its run here
searcher = searcher.pager();
});
// sometime later we page...
searcher.search()
.then(function() {
});
Let's say you have performed a search for tacos
and you get a bunch of results, but the chef comes back to you and says:
Hey! My new creation "La Bomba" is not showing up, fix it!!!!
So you are puzzled as to why it is not showing up, since it's clearly marked as a taco
in the db. Wouldn't it be nice if splainer-search
gave you some help?
Don't worry, we've got your back :)
So assuming you already have something like this:
var options = {
fields: ['id', 'title', 'price'],
url: 'http://localhost:8983/solr/select',
args: { 'q': ['#$query##'] },
query: 'tacos',
config: {},
searchEngine: 'solr'
};
var searcher = searchSvc.createSearcher(options.fields, options.url, options.args, options.query, options.config, options.searchEngine);
searcher.search();
You would want to create a new searcher with the same options/context, and use the explainOther()
function:
var fieldSpec = fieldSpecSvc.createFieldSpec(options.fields);
var explainSearcher = searchSvc.createSearcher(options.fields, options.url, options.args, options.query, options.config, options.searchEngine); # same options as above
# assuming that we know "El Bomba" has id 63148
explainSearcher.explainOther('id:63148', fieldSpec);
The explainOther()
function returns the same promise as the search()
function so you can you retrieve the results in the same way.
In ES, the explainOther()
function behaves the same way, except that it does not need a fieldSpec
param to be passed in.
var options = {
fields: ['id', 'title', 'price'],
url: 'http://localhost:9200/tacos/_search',
args: {
'query': {
'match': {
'title': '#$query##'
}
}
},
query: 'tacos',
config: {},
searchEngine: 'es'
};
var searcher = searchSvc.createSearcher(options.fields, options.url, options.args, options.query, options.config, options.searchEngine);
searcher.search();
var explainSearcher = searchSvc.createSearcher(options.fields, options.url, options.args, options.query, options.config, options.searchEngine); # same options as above
# assuming that we know "El Bomba" has id 63148
explainSearcher.explainOther('id:63148');
The explainOther()
function returns the same promise as the search()
function so you can you retrieve the results in the same way.
This library was originally written for dealing with debug tools such as Quepid and Splainer. As such, it provides a lot of help taking a user specified list of fields and associated roles, then once search is done turning the raw docs out of the Solr searcher into something more normalized based on that config (a normalDoc).
The normalDoc provides a friendlier, more standard interface. This includes friendlier parsing of explain information as needed.
var userFieldSpec = "id:uuid, title, body, authors"
var fs = fieldSpecSvc.createFieldSpec(userFieldSpec)
var searcher = searchSvc.createSearcher(
fs.fieldList(),
'http://localhost:8983/solr/select',
{
'q': ['*:*'],
'fq': ['title:Moby*', 'authors:Herman']
}
);
searcher.search()
.then(function() {
var bestScore = 0;
angular.forEach(searcher.docs, function(doc) {
var normalDoc = normalDocSvc.createNormalDoc(fs, doc);
// access unique id and title
// (above specified to be uuid and title)
console.log("ID is:" + normalDoc.id);
console.log("Title is:" + normalDoc.id);
// snippets -- best try to highlight the field
angular.forEach(normalDoc.subSnippets, function(snippet, fieldName) {
console.log('hopefully this is a highlight! ' + snippet);
});
// prettier and heavily sanitized explain info:
// (the explain modal on Splainer shows this)
console.log(normalDoc.explain());
// hot matches contains the most important matches
// this drives the horizontal graph bars in Quepid/Splainer
var matches = normalDoc.hotMatches();
// Give hotMatchesOutOf a maximum score (for all docs returned) and you'll
// get the hot matches as a percentage of thewhole
if (normalDoc.score() > bestScore) {
bestScore = normalDoc.score();
}
var normalDoc.matchesOutOf(bestScore);
// a link to the document in Solr is handy:
console.log(normalDoc._url())
})
});
Most of what splainer-search does should be compatible with all versions of Solr and Elasticsearch. There are times though where one of these projects introducing a breaking change and it becomes necessary to specify the version number used.
For example, ES deprecated the fields
parameter in favor of stored_fields
(https://www.elastic.co/guide/en/elasticsearch/reference/current/breaking_50_search_changes.html#_literal_fields_literal_parameter). So it's necessary to tell splainer-search which version you are using in order to send the appropriate request.
To do so you only need to specify the version number in the config
param when constructing a new searcher:
var options = {
fields: ['id', 'title', 'price'],
url: 'http://localhost:9200/tacos/_search',
args: {
'query': {
'match': {
'title': '#$query##'
}
}
},
query: 'tacos',
config: { version: 5.1 },
searchEngine: 'es'
};
var searcher = searchSvc.createSearcher(options.fields, options.url, options.args, options.query, options.config, options.searchEngine);
searcher.search();
And splainer-search will take care of using the correct name in the parameters.
NB: The default behavior will be that of 5.x, so if you are on that version you do not need to do anything, whereas if you are on a previous version number you should provide the version number.
If the individual search result field is a string then it is automatically highlighted.
However, if the selected value is an array or a JSON object, it doesn't coerce it to a string (and as a result doesn't highlight it, either).
Secondly, if any component in the selected path results in array, the rest of the path is spread over the array value. To explain:
Data: { "variants": [ { "name": "red" }, { "name": "blue" } ] }
Path (or _field name_): "variants.name"
Result: [ "red", "blue" ]
Sometimes we want to understand what queries are being sent to the search engine, and it can be a bit opaque if we are going through a API or if we have parameters being appended inside the search engine (think Paramsets in Solr or templates in ES).
Consule the searcher.parsedQueryDetails
property to get a search engine specific JSON data structure.
For Solr we check if the responseHeader.params
array exists, and return that.
Send echoParams=all
to Solr to trigger this behavior.
There doesn't appear to be an equivalent feature.
Frequently we want to understand what the search engine is doing to the raw query input.
Consult the searcher.parsedQueryDetails
property to get a search engine specific JSON data structure.
For Solr we filter through all the keys in the debug
section of the output, filtering out the
keys var keysToIgnore = ['track', 'timing', 'explain', 'explainOther'];
. Everything else is
added to the searcher.parsedQueryDetails
property.
In ES we default profile=true
property, and nest everything under the profile
key in the
response is to the searcher.parsedQueryDetails
property.
Querqy is a query rewriting library. It helps you to tune your search results for specific search terms. Understanding what Querqy is doing to your queries is critical to achieving great search results.
The searcher.parsedQueryDetails
property surfaces all the debugging information about what rewriting Querqy is doing to the input query. Assuming you are also requesting the details on what rules are being matched via the querqy.infoLogging=on
query parameter, then
you will also see that information in the searcher.parsedQueryDetails
structure.
Splainer-search is written using AngularJS project. It requires npm
and grunt
:
npm install -g grunt-cli
To run the tests:
npm install
npm test
Tip: add an f
in front of any describe
or it
in your unit tests to run just that unit test.
We need to build a splainer-search.js
file as part of the build.
npm run-script build
We use NP to publish splainer-search to npmjs.org.
You need to update the
CHANGELOG.md
with your new version and the date, but you don't need to touch package.json
,
the np
script bumps that file! Check that file in.
Now install the 'np' script if you don't have it, and run it to create the release:
npm install --global np
np --no-2fa
CHANGELOG.md
into the Whats Changed section.Development for this library is done primarily by OpenSource Connections for search relevance tools Splainer and Quepid
Primary author is Doug Turnbull
FAQs
Angular JS Solr and Elasticsearch and OpenSearch Diagnostic Search Services
The npm package splainer-search receives a total of 28 weekly downloads. As such, splainer-search popularity was classified as not popular.
We found that splainer-search demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago. It has 3 open source maintainers collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
GitHub removed 27 malicious pull requests attempting to inject harmful code across multiple open source repositories, in another round of low-effort attacks.
Security News
RubyGems.org has added a new "maintainer" role that allows for publishing new versions of gems. This new permission type is aimed at improving security for gem owners and the service overall.
Security News
Node.js will be enforcing stricter semver-major PR policies a month before major releases to enhance stability and ensure reliable release candidates.