When it comes to integrating Apache Solr with Drupal there are currently two different modules that can be used, Search API, and the Apache Solr module. While both are valid options, for this series we've chosen to focus on the Search API module because amongst other things it's generally more flexible, and based on conversations with people in the community who are working on Solr integration it is currently seeing more focused development efforts and will likely superseded the Apache Solr module sometime in the future.
This tutorial provides some background information on the Search API module and why we've chosen to use it. We'll look at how the Search API module bridges the gap between Solr and Drupal, and explain some of the commonly used terms we'll encounter in the module's UI and codebase.
By the end of this lesson you should be able to explain the Search API module's terminology, requirements, and position in the Drupal ecosphere, as well as be able to make a good case for why someone should choose the Search API module as a starting point for creating better search tools in Drupal.
Additional resources
Before we can start building a search application we need some sample data that we can index and use for testing, not to mention a site we can use to test this all out on. In this tutorial we'll walk through installing Drupal 7 and importing some sample data.
In order for this to work I built a Drupal 7 site with a content type named Fish, and then imported a whole bunch of descriptions of various fish from Wikipedia. You should be able to use the provided database dump in order to get up and running with a sample Drupal site pre-populated with some sample data.
If you're not planning on following along and building the fish finder application in the Search API and Solr series, or are planning on implementing Solr search on your own site instead you can probably skip this tutorial. Just note that the rest of the tutorials in the series assume you've got a working Drupal 7 site with some content.
By the end of this tutorial you should have a working Drupal 7 site with sample content running on your localhost for playing with while watching the rest of the series.
Solr is an application that runs on it's own, independent of Drupal. So before we can integrate our Drupal site with the Solr server we first need to install Solr and get it running. In this tutorial we'll walk through the requirements for installing Solr and then look at a quick and easy way to get it running on our localhost for development purposes. This setup is best for when you're working on a site locally, but have it hosted elsewhere and someone else is managing your Solr install for you.
The basic process is to download the Solr application, unpack it somewhere on your localhost, and then run the start.jar file that comes with the example application: java -jar start.jar. To ensure it's working you can connect to the Solr web UI on your localhost at http://localhost:8983/solr/.
Note: When using Solr with Drupal and the Search API module we currently need to use the most recent 4.x version of Solr, despite the fact that there is a 5.x version that just came out. At the time these tutorials where recorded the Search API module had not yet been updated to work with Solr >= 5.x. The general installation instructions for localhost development however are not likely to change much at all between Solr 4 and 5 so even if you're using Solr 5 this should still be relevant information.
By the end of this tutorial you should be able to run Solr on your localhost and verify that it is working.
Additional resources
If you're installing Solr on a production server there are some additional considerations you'll want to be aware of. The most important thing is that instead of just running the Solr application directly you'll likely want to run it inside of a J2EE server container like Tomcat or Jetty. In this tutorial we'll look a installing Solr on Ubuntu using the same process that we would recommend for a production installation. We'll do this by starting with a Vagrant VM with nothing but Ubuntu installed and walk through the steps required to get Solr up and running.
The steps are going to be similar, but different depending on your OS of choice, and it would be impossible for us to cover them all. So here's a general outline of what you'll want to do:
- Install Java. Note, you need the full JDK for Solr not just the JRE.
- Install a J2EE application server like Tomcat or Jetty.
- Download the version of Apache Solr you want to use.
- Configure your J2EE application server to run Apache Solr.
- Configure your system to automatically start/stop your application server.
We'll be following along with the instructions provided in this excellent blog post by Ben Chavet over on Lullabot.com, though some of the commands will be a little different since software versions have changed, etc. If you prefer to follow along with written instructions, or want to have the list of commands next to you when it comes time to set things up for real, give that post a read.
A note about security. We will NOT be going over how to restrict access to the Solr instance on the Ubuntu server as that's outside the scope of this tutorial. However, it's important that you do so. At a minimum you'll want to configure your firewall to prevent access to the Solr host from anything other then the web server where you Drupal site lives.
By the end of this tutorial you should be able to list the requirements and considerations that need to be evaluated when installing Solr on a production server.
Things you might want to copy/paste if you're following along:
Content of Tomcat context file - /usr/local/tomcat/conf/Catalina/localhost/solr.xml :
<Context docBase="/usr/local/tomcat/webapps/solr.war" debug="0" crossContext="true">
<Environment name="solr/home" type="java.lang.String" value="/usr/local/tomcat/solr" override="true" />
</Context>
Content of the Solr definition file - /usr/local/tomcat/solr/solr.xml :
<?xml version="1.0" encoding="UTF-8" ?>
<solr persistent="false">
<cores adminPath="/admin/cores">
<core name="drupal" instanceDir="drupal" />
</cores>
</solr>
Additional resources
In this tutorial we'll walk through downloading and installing the Search API module, the Search API Solr module, and their dependencies. Then we'll look at using the Search API Solr configuration files with our Solr server. These configuration files are specially crafted to help with indexing data contained in a Drupal site and allow Solr to have a better understanding of Drupal's entities, fields, and the data that they contain. For example, mapping a Drupal Field API body field on a page node type to the appropriate field type in Solr.
The gist of this tutorial is, locating the Solr configuration files in the search_api_solr/solr-conf/ directory, talking a little bit about what each one does, and then demonstrating how to copy those files into the configuration for your Solr server so that Solr will start using them.
After looking at the various configuration files, and then placing them into our Solr instance, we'll connect Drupal to our Solr server by creating a new Search API server configuration within Drupal's UI. This will allow us to confirm that our Apache Solr server, and Drupal, will be able to talk to one another.
By the end of this tutorial you should be able to configure Solr to work with Drupal, connect the two, and verify that the connection is working.
Additional resources
In order for Solr to provide search results we need to first send our Drupal content to the Solr server so that it can be indexed. In this tutorial we'll look at connecting the Search API module with the Solr server and creating an index that maps content in Drupal to data types in Solr. We'll also look at how the various configuration options effect the Solr index.
After creating a Search API index configuration we'll look at running the indexer, essentially queuing all the content on our site for indexing, and then telling Drupal to send the documents on our site to the Solr server for indexing. This can be done either via the UI or with Drush. You can choose to index content as it's created, or for sites with higher rates of new content you can send it to the indexer in periodic batches. Whichever you choose, making sure that you've got a system in place for periodically sending items queued for indexing to Solr is a critical step.
By the end of this tutorial you should be able to send content from Drupal to Solr for indexing, and verify that the content is showing up in the Solr server's index.
The Search API module by itself doesn't provide a UI for submitting a search query, or a page for displaying results. Instead, it exposes an API that other modules can use to provide those features. This makes it super flexible, but it also means we've got some extra work to do in order to allow someone to actually perform a search and see the results.
In this tutorial we'll look at using the Search API Pages module to create a simple search page with a form at the top and a list of results ordered by relevancy. Search API Pages is the quickest and easiest way to replace the Drupal core search module's functionality with a form that uses Solr for a search backend instead of MySQL.
When creating a new page with the Search API Pages module we can choose the view mode that we would like to use for displaying results. It works very nicely with Drupal's built-in view modes, as well as contributed modules like Display Suite, in order to allow for a high level of customization of view modes, and thus of the displayed results.
You can also configure the query type to use, choosing from one of: multiple terms, single term, or direct query. For integration with Solr you'll likely want to choose direct query, and allow Solr to handle the query parsing since it has a lot of advanced options that go far beyond what Search API handles on its own. However, we'll look at the different query type configurations, and demonstrate things we can do with direct query searches and the powerful Solr query syntax that we can't do with the other modes.
Finally, we'll look at the block that Search API Pages provides, and use it to replace the search form on the home page of our site with a form that points to our new Search API Pages search results page.
By the end of this tutorial you should be able to expose a page on your site that will allow your visitors to perform a search using the Solr index and have the results displayed in Drupal.
Additional resources
There are a couple of configuration options available when configuring a Search API index that we haven't looked at yet: adding additional fields, and using boost values to increase the relevance of a keyword when found in a specific field.
Solr allows you to index any number of additional fields, so we'll add a species and genus field to our index. This is one of the reasons using Search API to interface with Solr is so great. Through it's use of the Entity API, the Search API module has a deep understanding of all the content types on your site and the fields that are attached to them, without you having to write any code, or do anything other than configure things in the UI.
One of the benefits of creating your own search index is that you know your data better than anyone, and you know what people are hoping to find in your content. Solr allows you to configure a boosting value that can be used to increase the relevancy of keywords found depending on where in the data it's located. For example, when someone searches for a keyword we can probably assume that if the keyword is in the page title that the keyword is worth more relevancy points than if the keyword is found in the page body. With boosting we can affect the relevancy ranking of results and help our users more quickly find what they are looking for.
By the end of this tutorial you should be able to add additional fields to your Solr index so their content is available for searching, as well as assign a relevancy boosting value when keywords are found in specific fields.
The Search API module supports a handful of data alterations and processors; additional operations that can be performed on a document before it's indexed or during the display of search results. While Solr actually handles the majority of these for us already, this tutorial will look at the available options, talk about what each one does, and explain which ones are still relevant when using Solr as a backend.
Looking at data alterations in the Search API module also raises an important point about security. By default, Search API doesn't care about your content's access control settings. In order to prevent people from seeing results for their searches that contain data they shouldn't have access to we need to make sure we account for that in our configuration.
Here's a good list of the currently available data alterations and processors, though it's worth noting that not all of them are available for all search backends. Also, as we'll see, not all of them are recommended when using Solr even if they are available. Solr's tokenizer for example is much more full featured than the Search API tokenizer, so when using Solr as a backend it's best to keep the Search API tokenizer turned off and let Solr do its thing.
By the end of this lesson you should be able to use data alterations and processors to filter out specific content types from your Solr index and to highlight keywords found when displaying search results. You'll also be able to explain why some alterations and processors are better left off so that Solr can handle those tasks directly.
Additional resources
Being able to display search results using the Views module provides a huge amount of flexibility with respect to what is listed, what it looks like, and more. In this tutorial we'll look at using the Search API Views module, included in the Search API project, to create a view that allows users to search our Solr index and display the results as a table, or really, in any other way that Views can display content. We'll also cover some special considerations regarding access control and entity relationships that we need to keep in mind when using Views to display search results.
The biggest difference between creating a view that lists a bunch of nodes, and one that displays search results is that you need to use your Search API index as the base table from which you're building your view. Then, by default, the view only has access to the fields that are in the Solr index. This allows you to build the entire view without having to query the database. Or you can use the Views module's ability to define relationships to other buckets of content to query the database and pull in additional information. There's a huge amount of flexibility.
When building views from the Solr index you can optionally expose one or more filters. Essentially creating a form that allows someone to construct a search query. This can be as simple as exposing a keyword text field, or as complex as you would like to get. We'll look at using exposed filters to create a form that users can perform a search with and create a more complete search experience. We'll also look at how you can move those exposed filters into a block that can be displayed on the home page of our site, allowing us to replace the functionality provided to the Drupal core Search module with Views and Solr.
By the end of this tutorial you should be able to create a view that displays search results using the Search API Views module.
Additional resources
Using facets allows users of your search application to further narrow the results returned from a keyword search by selecting one or more attributes of the returned content and saying either show me only these, or show me everything but these. In this tutorial we'll take a look at some examples of faceted searching in practice, and then we'll use the Facet API module to expose facets for our genus and species fields.
One of the most common uses of facets is on e-commerce sites like Zappos.com that have huge collections of products that users can browse through, and narrow down, to focus in on exactly the pair of shoes they are after. In this example facets allow you do to things like narrow the results returned from your initial keyword search to just shoes for men, which are brown, size 10.5, and on sale. You can can also see faceting in action any time you perform a search on our site.
We'll use facets to allow users of the fish finder application to limit the results returned to just those of a specific species or genus. In doing so we'll also look at the options available for determining how facets should be displayed, whether or not we should show a facet that has zero documents in our result set, and how to combine multiple facets together into a single query using either AND or OR logic.
By the end of this tutorial you should be able to use the Facet API module in conjunction with Search API in order to provide facets that your users can use to further narrow and refine their search results.
Additional resources
Depending on the data that is being searched, some shorter general words, like "a", "the", or "is" can adversely effect search result relevancy. Consider the word "the", which in a standard description of a fish in our database could easily appear hundreds of times or more. When a search is performed, part of the algorithm that calculates the relevancy of any document in the index is to count the number of times a word appears in the text being searched. The more often it appears, the more relevant the document. Words like "the" however often have little to no real bearing on a document's actual relevancy. These words should instead be excluded from the ranking algorithm.
Stop words can also serve another purpose. You can filter out words that are so common in a particular set of data that the system can't handle them in a useful way. For example, consider the word "fish" in our dataset. It's probably very common. With only 500 fish being indexed it's not really going to make much difference, but what if we were indexing five million fish, and each one had the word "fish" in the description even just five times? That's 25 million occurrences of the word "fish". Eventually we might start to hit the upper limit of what Solr can handle. The word "fish" in this case is probably also not very useful in a search query. You're browsing a fish database. Are you really likely to search for the query fish and expect any meaningful results? Likely it would instead return every result. It would be like going to Drupal.org and searching for the word "drupal" and expecting to get something useful. Not going to happen.
Solr has the ability to read in a list of stop words, or words that should be ignored during indexing, so that those words do not clutter your index and are removed from influencing result relevancy. In this tutorial we'll take a look at configuring stop words for Solr.
First, we'll use the Solr web UI to see the most common terms in our index for the body field. Then, based on that list, and the list of common stop words provided by the Solr team, we'll configure our stopwords.txt file. Finally, we'll re-index all the content of our site so that it makes use of the new stop words configuration and re-examine the most common terms noting that our stop words no longer appear in the list.
By the end of this tutorial you should be able to use the Solr web UI to get a list of the most common terms in your index, and know how to add terms to Solr's stopwords.txt file to prevent them from showing up in your index.
Additional resources
Solr provides the option to configure synonyms for use during both indexing and querying of textual data. A synonym is a word or phrase that means exactly or nearly the same thing as another word or phrase in the same language. For example, shut is a synonym of close. Synonyms, if not accounted for, can cause a dilution of search result relevancy when searching for a keywords that have lots of variations in your index.
Consider for example the words, "ipod", "i-pod", and "i pod". It's pretty easy to imagine a scenario in which the content of our site could contain all three variations of the word. When someone searches though they are likely just going to search for one, but expect results for all three. In order to not break those expectations we need to make sure we account for this scenario. Another example from the the Drupal world would be the terms "CMI" and "configuration management". Chances are if you search for one you would be happy to see results for the other.
In this tutorial we'll look at using the synonyms.txt file that is part of our Solr configuration in order to account for synonyms in our data. Of course the exact words you use will depend on the content of your site, but we can at least cover how they work and how to configure them.
By the end of this tutorial you should be able to configure Solr to be aware of synonyms in your data in order to improve the quality of your search results.
Additional resources
One of the benefits of building our own search application is that we have ultimate control over the ranking of items. Combined with our superior knowledge of our own content we can use this to ensure that when someone searches for a specific keyword we bubble our best content for that term to the top of the list, regardless of whatever Solr might rank it based on its internal algorithms. This is commonly referred to as promoted, or sponsored, results; the artificial boosting of a particular document to the top of the result list for a specific query.
A similar, but not exactly the same, example would be sponsored results on Google searches, where you can pay to have your page listed at the top of the results for a specific keyword or set of keywords. We are going to be doing all of this except for the part where we let people pay to promote results, though you could certainly build that part on your own if you need that.
Solr uses a configuration file named elevate.xml, in conjunction with a processor, to elevate results at the time a query is performed. We can promote specific documents in our Solr index by figuring out the unique Solr ID for a document and then adding it to the elevate.xml file along with some information about a query, or queries, this document should be promoted for.
In this tutorial we'll learn how to find a Solr document's unique ID, and then configure Solr to use an elevate.xml file that will promote the "How to Use the Fish Finder" page to the top of the results when someone searches for the term "fish". This configuration is all within the Solr application itself and doesn't really rely on Drupal in anyway. As such, the material in this tutorial should be applicable to your Solr search applications even if you're not building them with Drupal.
By the end of this lesson you should be able to configure promoted documents in your own Solr-based search application.
Additional resources
With our interface taken care of, now we can configure multilingual support for our content. That is, we need to be able to identify which language a given piece of content is written in and then create translations. So, we already have articles, and we need to add a knowledge base to the site as well. For both of these types of content, we’ll need to identify the language they are written in, as well as provide related, linked translated versions. In this lesson we'll enable the core Book module for our knowledge base, enable multilingual content, and start translating our knowledge base.
Additional resources
Not everyone wants or needs to install yet another module, and due to the caveats we mentioned about Localization client, it is still a good idea to be familiar with all of your options. In this lesson, we'll review the Locale module translation process, then enable Localization Client and start translating some text!
Additional resources
Localization Client project
Using Drupal, 2nd edition
Using Drupal source code
So far we’ve got a nice start with getting our site translated, but everything is not quite smooth yet. We have translations for some of our content and menu items, but we're seeing all the content at the same time. You will also see some stray interface text still in English. To take our multilingual site further and really make it shine, in this lesson we're going to turn to a package of modules called Internationalization (i18n). There is a central Internationalization module, which comes packaged with a handful of other modules designed to work together to extend core’s multilingual features.
Additional resources
Internationalization (i18n) project
Using Drupal, 2nd edition
Using Drupal source code
Translated content on a Drupal is all shown at the same time by default. The Internationalization module will let you display only content that is relevant to the language currently in use, which is referred to as "content selection." In this lesson we're going to configure our site's content selection and the test it out to make sure it working as expected.
Additional resources
Before we can translate the various site-wide variables, like the site name or slogan, we need to let Drupal know which ones we want to make translatable. To do this, we’ll need to use the Variable translation module from the Internationalization package. In this lesson we'll enable a multilingual variable so that we can see how to translate the site name.
Additional resources
We’ve translated the site name, but we still have other stubborn text on the site that we couldn't select on the Variable translation page. Our content types for the site are still using English for the content type name on the “Create content” page, and for field names when making new content. In this lesson, to fix this up, we'll walk through translating content type names, the Title label on the content type, and also see how to tackle the field labels as well.