I'm using Solr 4.10.2 and Drupal 7.X, I have the Apache Solr Module Framework operating and sending the requests to Solr From Drupal. Currently when we perform a search, Drupal builds the query and sends it to Solr. Solr just executes the query and returns the results without using it's internal handlers which can be configured through SolrConfig.xml.
I would like to know if there is a way to just send the searched terms (without building a query) from Drupal and let Solr use the internal handlers declared in SolrConfig.xml to handle the request, build the query and then return the data?
The reason for this is we have been working on trying to boost some results when we perform a search (we want exact match first & fuzzy search results after) by changing the "weight" of some fields.
We know that from Back Office we can use the "Bias" function to boost some fields but this is too limited for what we are trying to achieve.
We also know we can change the query sent from Drupal directly from code side by using hook_apachesolr_modify_query() but we prefer changing as little code as possible and using the SolrConfig.xml /handlers which we already have configured to return the results as we want.
Ok, we figured out how to do this:
In order to choose the handler that is being used by Solr while sending a request from Drupal, we have to edit the "hook_apachesolr_query_alter" function and add the following code:
$query->addParam(‘qt’, ‘MyHandlerName’);
We did some extra coding to allow us to change the Handler directly from back office in order to be able to switch handlers without touching the code.
Related
I am building a Drupal 8 site and am new to the twig templating engine. For one specific content type I would like to make a call to an external restful api and render some of the returned data as fields in the twig template.
I have an internal id to call out to the API and I would like to embed in the template:
The api call
set a number of variables from the call
render the result (with some logic if it does not exist)
Is this something that is easy to do with twig and drupal 8?
As a secondary question, is this secure?
The alternative at this stage is to write small Drupal 8 module but as there is no user input on the page, just rendering from the returned api call, I thought it would be easier to have it all in one place.
In Drupal 7 it was possible, but a poor design, to put arbitrary PHP into the template. In Drupal 8 it was made hard to do intentionally. You should not attempt to execute arbitrary PHP in your Twig files or make remote API calls that late in the processing of a request.
You should call the API and gather the data before you reach twig. You should create a custom module that handles that API interaction and places the response in a field, block, or another structure for rendering in the appropriate context (often a custom block works well for things like this, but exactly which approach makes the most sense depends on your project). You should also keep in mind that any page requiring a remote API call is likely to be slow unless that API call is very simple and very very fast. The BigPipe module can help you address those kinds of speed issues, but involve an additional learning curve.
If you want the browser to handle the API call, you will want to create a div (or similar markup) to place the results, and attach the JavaScript to the structure and make the actual API call after most of the page load is complete.
As for security: it is as secure as you make it. Drupal will provide some help to avoid the most common security mistakes, but you can still do things that would make it insecure (like sharing data with an untrusted third party or assuming the response data is always safe).
I am working on a Web API service for our Web application (current project). One controller in the Web API will be responsible to retrieve a list of entities of a certain type. So far so good. The problem is that we have a request to filter the list based on search criteria. This search/filter criteria has about a dozen different parameters, some of which can be null. So I figured I create a custom class (let's call it "EntityFilterCriteria") and I instantiate it on the Web application side with whatever filtering fields the user enters (I leave the ones that the user do not enter set to null). Now how do I pass this object to my Web API method? I do not want to build an URL with all parameters because it's going to be a huge URL, plus some parameters may be missing. I can't have a body in the GET HTTP command to serialize my EntityFilterCriteria object so what do I use? POST? It is not really a POST because nothing is updated on the server side. It is really a GET, as in "get me all the records that match this search criteria". What is the common approach in such situations?
Thanks,
Eddie
Serialize the class to JSON and post it to your server, the JSON string will be your post body. Once it is posted you can then deserialize into a class with the same signature. Most languages have either built-in support or free 3rd party modules that can do this for you.
Well the initial danger of using GET for such a request is that the longest url you can send which is guarenteed to work across all web browsers and servers is about 1400 bytes in length.
Personally I would use JSON to encode all of your filter parameters and submit them to the server in the body of a POST command due to the lack of size limit on the sent data.
While there is a semantic difference between GET and POST commands which was intentioned when the HTTP RFC's were written decades ago, those decades of practical use have shifted rather significantly (in the same way that barely anybody uses PUT or DELETE)
I don't see any drawbacks to use a POST method to execute your query and get the result back. For example, ElasticSearch uses this approach to execute queries against the database. See this link for example: http://exploringelasticsearch.com/searching_data.html. In REST, POST isn't necessary used to update data.
Hope it helps.
Thierry
I have an application of type asp.net mvc and web api.
I m little bit confused over http post and http put.
When to use what and what is the pros and cons of each.
I have gone through many blogs but no solid reason what is designed for what.
Use POST where you would have to create completely new record from scratch.
Use PUT where you would have to update existed record in your database
Here are Differences between PUT & POST
`POST is Not idempotent`-->
Means running POST operation again and again will create new instance everytime when you run call it.
`PUT is Idempotent`-->
PUT is Idempotent operation calling PUT again and again will result same result.
So POST is not idempotent while PUT is idempotent.
`There is also PATCH` -->
Use patch when you would have to update only few properties of your model.In other words Partial Updates.
Put simply (no pun intended):
POST is usually used to CREATE new objects.
PUT is usually used to UPDATE existing objects
Using the correct HTTP verbs allows you to publish a cleaner API and negates the need for encoding intent within the endpoint (url). For example, compare:
Using the correct verbs:
GET api/user/12345
POST api/user/12345
PUT api/user/12345
DELETE api/user/12345
Hacking the endpoint:
GET api/user/12345
POST api/user/12345/create
POST api/user/12345/update
POST api/user/12345/delete
I think the only Cons of using PUT etc are that not all developers are familiar with them and some third party software may not support them or at least it may not be as easy as using the more familiar verbs like GET & POST.
For example, I had a problem a few weeks ago when a proxy was placed in front of an API just before it was to go live and the proxy didn't support the HTTP PUT verb (maybe a config issue - but we didn't have access to the proxy to fix it) so we had to tweak the API and change it to POST at the last minute (which also meant we had to change the clients (mobile apps) that were using it).
Im working in a "home project solr"(aka PoC), using Solr as search engine.
I'm using master/slave configuration in two distinct servers, so(in my plans) my search would be avaiable even if the main server comes down but, the problem is:
Im currently doing my search at url
http://mainSOLRSERVER:8080/search/select/?q=*:*
Is there any built-in solr component to check if this url is OK, and if is not, change to another server
http://redundancySOLRSERVER:8080/search/select/?q=*:*?
I aint using Solrj in my application, its a simple ASP.Net application using xml parsing in Solr's results.
The only thing I came up with, would be send a ping request to Solr's main server and, if returns not ok, build the request with the redundancy URL, but that would be necessary in every single search request, is that the right approach?
Thanks in advance!
In production scenario, you would use a load balancer.
You don't need the ping. Directly execute the query on the main URL and if it returns an error, then query the backup URL. If that also throws error, then you are out of luck.
Hi I have a page in my java/jsp based web application which shows list of new products.
I want to provide an rss feed for this. So, what is the way to create an rss feed which other can use to subscribe?
I could find some java based feed creators. But, the question is how the feed will be self updating based on new products added to the system?
I'm not familiar with Java, so here's a general thought.
Your feed should be accessible via some URL, like http://mydomain.com/products/feeds/rss. When Feed Aggregator fetches this URL, the servlet (I believe this is how they are called in Java world) fetches a list of recent products from the DB or wherever, build RSS feed and then sends it back to the requester, which turns out to be a Feed Aggregator.
For performance reasons this particular servlet may not access the database each time it's executing. Rather, it can cache either resulting feed (recommended, HTTP allows for a very flexible caching) or a result of database query somewhere in memory/on disk.