Search Results Rendering
The com.marklogic.widgets.searchresults widget supports pluggable rendering. This means if you know how to identify a particular class of document and know how that should be rendered in a search results page, you can add your own rendering support.
Out of the box is provided some default renderers. These make a best guess as to what should be used for a title or for a summary/snippet.
For JSON Documents, the following fields are checked for use as a title: title, then uri
For JSON Documents, the following fields are checked for use as a summary: summary, then tree representation of the JSON content
For XML Documents, the following fields are checked for use as a title: //title[1], //name[1], //h1[1], first text() element
For XML Documents, the following fields are checked for use as a summary: //summary[1], //synopsis[1], //description[1], //details[1], second text() element
For plain text Documents, the URI is used for the title and the text content used as a summary
Custom renderers
A custom renderer is a pair of methods - matcher(result) and processor(result). These are both passed the result object. This result object is a JSON object representing all information about this single result, including URI, score, and content (xml, json or text).
The matcher(result) function should return true if this renderer can be used to render the search result. Otherwise it must return false.
The processor(result) function must generate HTML and return this at the end of the function. Currently no dynamic javascript event handlers are supported within the HTML output, but that may be added in future.
To add your own renderer to a searchresults widget, simply call the addProcessor(name,matcher_func,processor_func) method.
Custom renderer examples
There are three custom renderers previously built for different demonstrations. These show a range of functionality. The first one is a simple example to showing a JSON movie document in a summary. The next one performs similar work for an XML document. This also shows conversion and using XPath in the browser to extract data.
The final example shows how to use the built in post-render lazy loading callback feature to paint additional information after search results are displayed. The example shown renders a description of a powerpoint slide's snippet content, then uses lazy loading to draw an image linking to the PNG format image thumbnail for the slide.
- Simple JSON movie example from page-mljstest-movies.js
- Custom XML song renderer from PoC project
- Advanced lazy loading in powerpoint
Advanced custom renderers
WARNING: The above functionality is still supported, but the default recommended mechanism will change in MLJS 1.8 (Oct 2014). The new mechanism allows you to selectively plug in and mix and match functionality, rather than replace the whole renderer.
This will allow, for example, every result to have a common set of actions on an action bar, related semantic facts, and extracted metadata whilst still also allowing you to just customise the preview/snippeting portion of the search result rendering area. This makes re-using and plugging new functionality much quicker.
An extension mechanism is also now supported in MLJS where you can simply drop in a JavaScript file to extend MLJS - no need for custom coding, calling methods on the search results widget to add new rendering components.
All this functionality will be documented in the Oct 2014 release.