1. Paul Taylor, Solutions Architect October 29, 2014
SEARCH G2
Paul Taylor, Solutions Architect October 29, 2014

2.      Search Agenda Resources What is Search G2? Your Feedback
How do we request it? 
DEMO: Implementing Search G2 

3. What is Search G2?

4. SEARCH G2 - INTRODUCTION
WHAT IS SEARCH G2?
Highly scalable, fault tolerant, industry standard indexing & query platform, built in the cloud…..oh, and it’s fast!!

5. WHAT IS CROWNPEAK SEARCH G2?
BENEFITS
Highly available & scalable
Higher crawl frequency
Less obtrusive than Classic
Customisable Site Crawling at Implementation Time
Crawl website content
Crawl custom data
Supports Basic, NTLM & Digest Authentication
Easy Adoption for Customers & Partners
Industry Standard Lucene Query Language
Search G2 SDK
Supports UTF-8, double-byte & multi- byte character sets
Enhanced End-User Search Experience
Auto Language Detection (31 Languages)
Stemming, Spell-Checking & Suggestion Capability

6. WHAT IS CROWNPEAK SEARCH G2?
LOGICAL ARCHITECTURE
Customer Website / Browser
Search G2 SDK
Search G2 WS Endpoint
Search G2 Crawler
Search G2 Repository
http(s)://searchg2.crownpeak.net/{collectionName}/select/?q=*:*

7. WHAT IS CROWNPEAK SEARCH G2?
SEARCH G2 SDK
Makes implementation simple for Customers & Partners
Wraps up standard Lucene features offered via Search G2
Provided in both JavaScript & .NET Flavours
JavaScript – Available via CrownPeak CDN
Minified version available at production-time
.NET – Will be available in early 2015
Both versions will be maintained by CrownPeak, and will keep version numbers

8. WHAT IS CROWNPEAK SEARCH G2?
SEARCH G2 SDK - SHOWCASE
Showcase provided within JavaScript SDK
Highlights all features currently supported within the Search G2 platform
Implementation examples on CrownPeak Connect
KnockOutJS
jQuery
JavaScript (plain)

9. How Do We
Request It?

10. SEARCH G2 - INTRODUCTION
HOW DO WE REQUEST IT?
Part of the CrownPeak Subscription
Enabled by CrownPeak Support
Send support request
Live URL (e.g.
Dev, Authoring, Stage URLs (e.g. stage.crownpeak.com)
Initial seed for each collection (e.g.
Authentication information (for protected sites)
Lead-time 5 days
You will receive a custom Search G2 WS endpoint back!
http(s)://searchg2.crownpeak.net/{collectionName}/select/?q=*:*

11. NOW WHAT? HOW DO WE REQUEST IT? Customise crawling rules
By default Search G2 will crawl all content within any pages found as children of the initial seed – all navigation will appear
Search G2 uses CSS engine to allow you to pick what is indexed
Submit to CrownPeak Support for implementation
Build your search client
Use one of the Search G2 SDK options
JavaScript (examples available for KnockOutJS, jQuery & JavaScript (plain))
Handles cross-domain security issues
ASP.NET Web Forms & ASP.NET MVC
Cut your own in whatever language you choose
Deploy through the CMS to your publishing target(s)

12. DEMO: IMPLEMENTING
SEARCH G2 ON

13. DEMO: IMPLEMENTING SEARCH G2 ON www.crownpeak.com
Deploy a Search Page
Use the Search G2 SDK (JavaScript)
Use plain JavaScript implementation style
0. Let's start a new Web project, then create a new html page, and give it a nice title.
1. We can add a little initial UI to our page, just so we have something to use.
2. Next we'll need to add a reference to the Search G2 JS core.
There are numbered versions available, plus an edge version that's unnumbered.
//searchg2-assets.crownpeak.net/crownpeak.searchg2.js
//searchg2-assets.crownpeak.net/crownpeak.searchg js
You can do your initial work with the edge version if you want, but change to the latest numbered before go-live.
All are available in minified form, with an accompanying source map.
//searchg2-assets.crownpeak.net/crownpeak.searchg min.js
It has no dependencies at all - apart from JavaScript.
You can use your favourite templating or binding libraries.
We provide a showcase that uses KnockoutJS which I'll show you later.
3. Before we can actually get the search to do its work, we need to configure it.
CrownPeakSearch singleton object is available.
Has a number of properties, implemented as methods due to no cross-browser property setters/getters.
We will set the collection to be
Usually the URL of your site, but you'll be advised of this.
4. Now that we're configured, we can go ahead and call the query method of the CrownPeakSearch object.
Takes the query text, page number and array of filter queries to apply.
Returns a promise (not Promises/A) which has .status property and .done(), .fail() and .always() methods.
Here we're just interested in .done which will give us the JSON data back from the search.
Let's just log it out to the console so that we can see what we're going to be working with.
If we take a look at the log, we can see that the result contains three main classes:
pager - nice easy-to-bind array of pages, including the current page and the total.
you don't have to use this - you can roll your own, or decorate this.
responseHeader - time the query took, parameters that were used, and status
response - the important one - tells you how many results you got, and the first 10
you can change the number with CrownPeakSearch.rows(n)
docs is array of matching documents
key fields: title, content, url, contentLength, type, normalizedScore
also date = modified date from header if set, otherwise crawl date
5. Let's add a div to contain our results, and we'll start with a small template to show the results.
6. Obviously it's not much of a template system if it can only contain hard-coded text, so we'll make a quick format function.
Then we can update our template to include a place holder for the value, and pass it in.
7. Now we can make a new template for an individual result.
For now, we'll just include space for the title.
We'll loop through the results that we got, and build up some html.
Then we can add another item to the results template, and put the html in there.
8. So that's working nicely with the title, but that's probably not enough for us.
Let's add space for a precis, and then make the title into a link to the result.
9. That looks pretty good. The first result looks like it's missing its content, and the others are a bit big.
We can do better there, so let's make a function to show a nicer precis, given a document.
It's pretty simplistic, but it does the trick.
It's worth noting here that because Search G2 does language detection for you, you get some benefits:
The spell checking is in the language of the content.
You get word stemming in the language too - so test vs testing vs tested, etc.
10. But Search G2 can do more to help us with this. One nice feature is hit highlighting.
First we need to configure highlight(true).
Then we can modify our precis function to show the highlights if they're present.
Highlights are provided on all matching fields, so we can do the title/url as well if we wanted.
11. Obviously we're going to need a pager as well, otherwise we'll only ever see the first 10 results.
We'll make a new template for our pager, and make room for it under the result count.
It's time to refactor our query into its own function so we can call it more easily.
We'll pick some nice defaults so that we don't have to pass everything in every time.
12. Search G2 also offers spelling corrections, giving us the ability to do something like "Did you mean".
Configure spellcheck(true).
Next we'll add a new template for our Did You Mean results, and include space for it in the results template.
Now we'll make a quick helper function to produce the output for us.
Each result has a .display and a .query property, so it's very easy to use.
13. Another common thing that you might like to do is faceting, so you can break down results into categories.
First you need to configure facets(field) - we'll use the title, and Search G2 will pick words from it for us.
We'll make a new template to show our facets, and put another placeholder into the results template.
Again, we'll make a helper function.
Facets look like data.facet_counts.facet_fields.fieldName then pair = facet, count.
So we need to loop through and increment by 2 each time.
That gives us links - when we click on one, we filter the results to those matching the facet selection.
Small problem - currently if you click again, it resets the filter.
So you'll need to store the current state of your filters, and append to it each time.
Also how do you get rid of a filter once you've applied it. Your UI needs to do this too.
14. Finally I'll show you our AutoComplete offering. We have that in another JavaScript file.
Again there's an edge version and numbered ones, plus minified versions.
//searchg2-assets.crownpeak.net/crownpeak.searchg2.autocomplete js
We'll add the script reference to our <head> element.
Next we can add a call to CrownPeakAutocomplete.init()
Takes the id of the element we want it to run on. Or the element itself.
Works, but nothing happens when you click on an item.
15. Firstly we can make it look a little nicer. The UL tag is classed autocomplete-popup, so that's easy to reference.
We'll turn off the list style, padding and margin of the UL.
For handling the click, the CrownPeakAutocomplete.init() takes a second options argument.
One property on that is a callback property, which takes a function(text)
We'll just use that callback to fire off a new query.
That was a quick run-through of the Search G2 JavaScript SDK. There are few more features available at the minute, and we have plans to add more as we go. There's documentation on Connect, and of course you can just go look at the source code.
For exercising all of the features, we have a showcase page that we can supply. This uses KnockoutJS for easy data binding, and gives you examples of everything that we've thought of so far.
If you have any questions:

14. Resources

15. SEARCH G2 - INTRODUCTION
RESOURCES
Playbooks – Available on CrownPeak Connect JavaScript SDK – Available on CrownPeak CDN .NET SDK – Available in early 2015 SDK Showcase – Available on CrownPeak Connect

16. Before we go…
Your Feedback 

17. We are here to hear your ideas and requests.
PARTNER SUMMIT
- FEEDBACK
We want your feedback!
We are here to hear your ideas and requests.
You have our ear - here at the conference, through support, through Connect.
We want to open a dialogue that starts here and continues well beyond
So please share your feedback cos we are always looking for ways to improve our product.
Product improvements come from 3 channels-
Product Management
Innovations from Customer Success team-
And of course Feedback and contributions from Developers like you
Through conversation at the conference or by posting on Connect
Product Management
Customer Success
Community Feedback
Developers

18. Forums on Connect - FEEDBACK Template development
PARTNER SUMMIT
- FEEDBACK
Twitter
@CrownPeak
#CrownPeakPartners
For ongoing discussions with the CrownPeak community on
Template development
User & Developer Experience
Solutions and Extensions
Product features & roadmap
Hosting & Publishing
Integrations
Forums on Connect
Twitter hashtag and QR Code
<a href=" class="twitter-hashtag-button" data-related="CrownPeakHelp,CrownPeak">Tweet #CrownPeakPartners</a>
<script>!function(d,s,id){var js,fjs=d.getElementsByTagName(s)[0],p=/^ 'script', 'twitter-wjs');</script>
<img src=" d=https%3A%2F%2Ftwitter.com%2Fintent%2Ftweet%3Fbutton_hashtag%3DCrownPeakPartners" alt="QRCode"/>
Forums QR code
<img src=" alt="QRCode"/>

19.    Access to Customer Success and Product Teams - FEEDBACK
PARTNER SUMMIT
- FEEDBACK
Access to Customer Success and Product Teams
You have access the Product and Customer Success teams to give direct feedback on features and best practices
Meet the team during the presentations and Q&A sessions
Make suggestions – we are capturing it. Team members are armed with Post-its!
Ask questions – We want to start a conversation that last well beyond the summit
Any Questions? 
MEET THE TEAM 
MAKE
SUGGESTIONS 
ASK QUESTIONS 
Partners and Developers

20. Partners and Developers
PARTNER SUMMIT
- FEEDBACK
FEEDBACK? 
Partners and Developers

21. CrownPeak + Web Experience Management
CrownPeak + Web Experience Management