Developing Marklogic Applications I

Developing Marklogic Applications I - Serach Concepts

This is continuation of my notes from the MarkLogic training course Developing Marklogic Applications I - XQuery

Day 3: Search Concepts

Search

Everything in the Information Studio can be scripted via the admin module

By Default, cts:search is filtered

cts:search(
  fn:collection(),
  cts:near-query((cts:word-query("cats"), cts:word-query("cats")), 2),
  "filtered"
)

By Default, Reindexing occurs as soon as a configuration value is changes. IE automatic reindexing.

xdmp:plan(cts:search(fn:collection("books"), cts:word-query("dog")))//qry:final-plan

GeoSpacial Search

Basis of geospacial is latitude and longditude. The basic Marklogic type is cts:point

Queries

cts:circle($radiusInMiles, cts:point(-33.8830, 151.2216)
cts:polygon($pointsSequence)
cts:box($lowerCorner, $upperCorner)
cts:element-pair-geospatial-query()
cts:element-geospatial-query()

Index Types

Element - <element> lat, long </element>
Element Child - <element> <child>lat, long</child> </element>
Element Pair - <element> <lat>lat</lat><long>long</long></element>
Attribute - <element lat="lat" long="long" />

Information Studio Workspaces are xml.

Examples

cts:search(
  fn:collection(),
  cts:element-pair-geospacial-query(
    xs:QName("place"), xs:QName("lat"), xs:QName("lon"),
    cts:circle( 200, cts:point(25.0, -80) )))

xdmp:http-get("some-url", options);

Snippets, Highlighting, Sorting and Pagination

Snippets

Search returns a snippeted results set. Results provide a search snippet that gives the context of the search results.

Included snippet elements

search:result
snippet:sni

Can provide a option on the serch request to apply transform

<options xmls="http://marklogic.com/appservices/search">
  <transform-results apply="snippet">
    <preferred-elements>
      <element ns="http://marklogic.com/mlu/top-songs" name="descr" />
    </preferred-elemetns>
  </transform-results>
</options>

Highlighting

Results set includesd a highligh that can then have a class wrapping the string.

Sort options

Sort options require range indexes

option node defined how the search results are sorted.

Optiizing Sort
- Ranged indexs can use range indexes.
- Ranged indexes cab improve performance.
- Adding <debug> to the options will show if range indexes are being used.

PAgniation

Default to 10 results
Modified in the results set.
search:response inclused total, start, and page-length

search:search("beatles", (), 11)

Faceted Navigation

Facets are grouped search results. by adding contraints to the query
Facets can be Bucketed contraints (eg Decade 2010-2020). These have upper and lower bounds.
Facets require ranged index
String range indexes have a collation. A collation is a uri to the string rules (eg diacritices, case sensitive
Facets are returned in the search results.

Creation of facets

Create a contraint (with collation if needed)
- This may include buckets for bucket contraints.
Configure the Facet options.

Updating Content and Transactions

MVCC Multi Version Concurrency Controll

Inserts and Updates
On Commit, Document is given a created transaction stamp on the document.
1. Doc and indexes are inserted into Forest in In-Memory Stand, and
2. An Entry has been made in Forest level Journal on disk
Once the stand reaches a point (doc count, mem count, /other configuratable ceiling) the docuemtns is pushed from the In-Memory Stand into the On-Disk Stand

Updates are the same except:
1. The document uri exists.
2. The old document received a delete timestamp equal to the new docs create.
3. Updates aquire a write lock on the document.
MarkLogic is an append database
```
xdmp:document-insert("uri", xml):
```
```
xdmp:node-replace(fn:doc("song1.xml")/top-song/title, <title>Trouble for Nothing</title)
```
```
xdmp:node-insert-child(fn:doc("title")/book, <chapter no="2">...</chapter>)
xdmp:node-insert-after(fn:doc("title")/book/title, <author>Herman Melvill</author>)
```
```
xdmp:node-insert-child(fn:doc("moby_dick.xml")/book/author, attribute dob {"1819-08-01"})
```
```
xdmp:node-insert-child(fn:doc("moby_dick.xml")/book/author/@dob, attribute dob {"unknown"})
```
Queries

Run on a timestamp, so the data from that time is returned. this means queries never need to lock documents, becuase its a snapshot.
Merge / Update transactions

Merge, merges stands together. Merge purges old deleted documents from the system. Only includes the current data.

Delete

Delete marks the document with a delete timestamp. Delete is a document update.

xdmp:node-delete(fn:doc("moby_dick.xml")//chapter[2])

xdmp:node-delete(fn:doc("moby_dick.xml")/book/author/@dob)