Transcript Automatic-Page-Copies

Automatic Page Copies
A publish trigger and web services
case study
Daniel Horwitz and
Darren Schaumann,
ProHealth Care
Where’s Darren?
In Tahiti on a large boat sipping
margaritas and loving life
OR
In labor and delivery waiting for my
second child to be born.
Detailed Information
• Information on downloading the detailed
speaker’s notes, Java code, style
sheets and build configuration will be
made available at the end of this
presentation.
Introduction
•
•
•
Our policies and procedures system
was difficult to use, and even harder to
maintain.
Using a publish trigger and web
services, we were able to simplify our
system.
Only one copy of each policy and
procedure is updated.
Problem Statement
• Policies and procedures defined
• Current situation
• Challenge
Definition
• Policies and procedures are the
documented processes and guidelines
that the corporation and people are
required to follow.
Definition
• Policies can apply to multiple entities
within the ProHealth Care system.
Definition
• Each policy and procedure document
should contain either a policy section, a
procedure section or both.
Current Situation
• Users note that searching on the current
site is difficult to use.
Current Situation
• Each entity has a directory structure to
browse, but they all have different
structures.
Current Situation
• Policies were written in Microsoft
Word™ and then converted to PDF
files.
• They have many different, and
inconsistent, formats.
Challenge
• Simple search and site hierarchy for
finding policies.
• Use content management for archiving
and workflow.
• Single source policy applying to many
places in the hierarchy.
• Have all the policies use the same
format.
First Attempt
• Display document in multiple places
using ‘references’.
• Unable to get navigation to be based on
user’s current location in site. Need
separate index block generation for
each folder where the policy appears.
ProHealth Solution
•
•
•
•
Build policies out of a series of HTML
blocks.
Allow content editors to edit only the source
policies.
Include entity and category information as
metadata.
When source policy is published, create or
update copies of the policy at all the
appropriate places in the site hierarchy,
then publish the copies.
User View
•
•
•
•
•
•
Organization
Home page
Search
Left Navigation
Right Navigation
Policy Display
User View - Organization
• Policies are organized by entity and
category.
Policies and
Procedures
Entity:
Lawrence Center
Category:
Administration
Category:
Admitting
Entity:
ProHealth Care
Category:
Corporate
Compliance
Entity:
Waukesha Memorial
Hospital
Uncategorized
Policy
User View – Home Page
• Policies and procedures home page has
left-hand navigation and search.
• Both dynamically display all entities.
User View - Search
• Using Google™ search appliance.
• Finds policies under each applicable entity
and category.
User View – Left Navigation
• Left-hand navigation
block allows users to
browse for policies in
the site hierarchy.
User View – Right Navigation
• Right navigation displays policies
immediately under entity or policies
within current category.
User View – Policy Display
• Displayed as series of HTML blocks.
Content Editors
• Content editors see only the source
copy of the policy document.
• The different pieces of the document
are all edited as blocks.
Content Editors
• Defining entities and categories to
which the policy applies.
Content Editors
• Editor must enter policy, definitions and
procedure into separate blocks.
Content Editors
• When the content editor publishes the
source, all the copies are published.
• Content editors can publish the folders
containing the copies.
• When creating a new policy, the creator
selects a different page type for a policy
only, a procedure only or a combined
policy and procedure.
Site Administrators
• Only site administrators can see the
copies.
• Site administrators set up entity and
category selection blocks.
• Site administrators create folders and
index pages for all the entity/category
combinations.
Style Sheets
• The style sheets take advantage of
copies of all the policies being in the
correct place.
• For navigation, the “current” attribute is
the key to locating the system folder
which is the user’s current location in
the site.
Style Sheets – Left Nav
• Left-hand navigation displays the user’s
current location in the site.
• Style sheet called Left PandP Parent
Children, because it displays both the
parent (entity) information and the
children (category) information.
Style Sheets – Left Nav
• If current is an entity, displays all the entities,
highlights the current and displays the
categories of the current entity.
• If current is a category, displays all the
entities, highlights the current and displays
just the current category.
• If no current, then displays all the entities,
with no current entity highlighted.
Style Sheets - Search
• Search, on the right hand side on the
home page, displays the search block.
• Style sheet is PandP Search.
• Handling of two columns loosely based
on Home Left Navigation.
Style Sheets – Right Nav
• When user’s current location is an
entity, display that entity’s policies that
do not have a category.
• Style sheet called PandP Children
because it displays an entity’s children.
Style Sheets – Right Nav
• When user’s current location is a
category, display the category’s
policies.
• Style sheet called PandP Grandchildren
because it displays an entity’s
grandchildren.
Style Sheets - Navigation
• The navigation style sheets use an
index block as their source. Generating
the XML for an index block takes some
time. This seems to be the main
bottleneck for publishing.
• Navigation must be rendered based on
user’s current location so that all the
links are correct.
Style Sheets - Policies
• The policy_procedure style sheet
displays the selected policy.
• The policy is simply a list of blocks,
which are displayed in order.
• The entities block is built from the “Site”
dynamic-metadata element, which is a
sibling of the system-data-structure
element.
Publish Trigger
•
•
•
•
•
•
•
•
•
Notes
Vocabulary
High-level Algorithm
Web service calls used
Publish trigger interface
Web services interface
WebservicesUtils
Building an Asset for a psuedo-policy
Build Notes
Publish Trigger - Notes
• In order to publish an asset, that asset
must exist.
• When an asset is published by the user,
or via the web service, it is placed on a
queue.
• A single thread in the Cascade server
takes an item from the queue and
publishes it.
Publish Trigger - Vocabulary
• Actual policy: The source policy, which
is edited by the content editors.
• Pseudo-policy: A copy of the source
policy, which is used to publish copies
of the actual policy.
High-level Algorithm - Notes
• Low-level details are in the Java code to
be supplied later.
• Publish trigger is called for every
destination of every document being
published.
• Upon return from the trigger, the
triggering document is published by
Cascade to the specified destination.
High-level Algorithm
• I) Examine the entity type and entity
path from the publish information. If not
a policy, do nothing.
• II) Read the policy being published.
• III) Look at the metadata. If a pseudopolicy, do nothing.
High-level Algorithm - IV
• IV) Using the applicable entity and
category lists from the metadata, loop
through each pseudo-policy associated
with the actual policy.
• A) Try to read the pseudo-policy.
High-level Algorithm - IV
• B) If the pseudo-policy doesn’t exist:
• 1) Create the pseudo-policy.
• 2) Remove non-administrator access to
the page.
• 3) Publish the indices to its parent and
grandparent folders.
High-level Algorithm - IV
• C) If the pseudo-policy exists but the
actual policy has been updated later:
• 1) Update the pseudo-policy to reflect
the actual policy’s content.
• 2) Remove non-administrator access to
the page.
High-level Algorithm - IV
• D) Publish the pseudo-policy.
Web Service Calls Used
• Only assets being manipulated are
pages.
• read()
• create()
• edit()
• editAccessRights()
• publish()
Publish Trigger Interface
• See “Publish Triggers” in the knowledge
base for details.
Publish Trigger Interface
• SDK contained in
CascadePublishTrigger.zip
• Use publishtrigger-5.5.jar
• Interface is part of Cascade installation.
Don’t include interface classes in the
publish trigger archives.
Web Services Interface
• See “Web Services Skeleton Project” in
the knowledge base for details.
Web Services Interface
• Download axis 1.4 from http://ws.apache.org/axis.
• Make the jar files in
axis-1_4/webapps/axis/WEB-INF/lib available.
• Get a current copy of the wsdl file using a browser.
Cascade server must be running. URL is
<server-base>/ws/services/AssetOperationService?wsdl.
• Build interface classes using
org.apache.axis.tools.ant.wsdl.Wsdl2javaAntTask.
• Careful, AssetOperationService_address is based on
how the wsdl file was addressed by the browser.
Production interface must be built from production
wsdl file.
WebservicesUtils
• Get
com.cms.webservices.WebservicesUtils
from Cascade+Webservices.zip.
• Pages contain a number of elements
where an identifier or path must be
supplied, but not both.
WebservicesUtils
• To set up a page for create() or edit(),
read the page, call nullPageValues(),
then perform application-specific
editing.
• If you don’t use nullPageValues(),
beware of dragons.
WebservicesUtils
• Since WebservicesUtils has been
written, the web services have been
updated.
• Before line 84, add:
pConf[i].setId( null );
Building Asset for Pseudo-Policy
• An asset must be passed to create() or
edit().
• Read actual policy page, run
nullPageValues(), make an asset for the
updated page and update the metadata
fields.
Building Asset for Pseudo-Policy
• Update metadata by getting dynamic fields
list from the page’s metadata and go through
the fields.
• Each field has a name and a list of values.
• If the name is “Actual”, set to single false
value.
• If the name is “pathToActual”, set to single
string which is the path to the actual policy.
• If any other name, leave alone.
Build Notes
• Use Java 1.5 JDK for compatibility with
the Cascade server.
• Axis web service interface classes
generate a lot of warnings.
• Copy publish trigger jar file to
../tomcat/webapps/ROOT/WEB-INF/lib
on each Cascade server.
Debugging
• Mostly done with Eclipse debugger.
• Use Log4j to log messages to Cascade log.
• Publish trigger interface for testing
implemented by class in Eclipse.
• Information hard-coded in the method that
drives test.
• Probable that org.apache.axis.utils.JavaUtils
will log warnings about DataHandler and
MimeMultipart. The warnings can be ignored.
Debugging SOAP Messages
• Prudent to see the SOAP messages.
• Hannon Hill often wants to see SOAP
messages.
• Hannon Hill recommends tcpmon. Get it from
https://tcpmon.dev.java.net.
• Web services interface code must explicitly
go through tcpmon. Update
AssetOperationService_address in
AssetOperationHandlerServiceLocator.java.
Change protocol:host:port section to
http://localhost:3333.
Debugging SOAP Messages
• Bring up tcpmon in a command window using
“java –jar tcpmon.jar”.
• Create new TCP Monitor Connection:
Local Port: 3333
Server Name: <Cascade-server-name>
Server Port: <Cascade-server-port>
• Run the test. tcpmon will capture the SOAP
traffic.
Cleaning up Orphaned Psuedo-policies
• Psuedo-policy with no associated actual
policy is an orphan. Caused when policy no
longer applies to entity or category.
• Orphaned policies appear in index blocks.
This causes them to appear in navigation
areas and allows them to be accessed by the
users.
• Finding orphaned policies is slow because all
the policy folders must be searched.
• Scheduled job runs weekly to clean up
orphans.
High-level Algorithm - Clean up Orphans
• Starting at root of policies, recursively go through that
folder and all its sub-folders.
• In each folder:
• A) Read the folder asset using read()
• B) Go through all the folder’s child assets:
• 1) If asset’s path ends in /index or /_cms, ignore it.
• 2) If asset is a folder, recursively process the folder.
• 3) If asset is a page, process potentially orphaned
policy by:
High-level Algorithm – Clean up Orphans
• i) Read the potentially orphaned policy.
• ii) If not a pseudo-policy, ignore it.
• iii) Read associated actual policy. If the
read fails, delete the pseudo-policy.
• iv) Look at entity and site in actual
policy’s metadata. If no combinations
match pseudo-policy, delete it.
Cleaning up Orphaned Psuedo-policies
• The application uses read() and delete()
web service calls.
• Activation and mailapi_1_1_3 jar files
must be made available to the
application.
• Directory on the server contains jar file,
batch file, jar files from Axis 1.4 library,
and activation and mailapi_1_1_3 jars.
Eclipse Organization
•
•
•
•
•
•
•
•
•
ProHealth uses Eclipse IDE and CVS.
Many ways to organize the Eclipse projects.
For reproducibility, copies of all third-party sources and libraries are in
CVS.
CascadePublishTriggerLib - publishtrigger-5.5.jar library from
CascadePublishTrigger.zip.
CascadeAxisLibs – Library jars for Axis 1-4.
CascadeWebServices – Java classes built here based on
AssetOperationService.wsdl from the Cascade server.
CascadePhcUtils – Project that builds the bulk of the publish trigger
and cleanup application.
CascadePublishTriggers – The publish trigger itself, ready for use on
the Cascade server.
CascadeCleanup – The application for deleting orphan pseudopolicies.
Gotchas
• The publish() web service call currently does
not allow target destinations to be defined CSNEW-132 requests that destinations be
supported by the publish() web service call.
• The generation of the index block, used for
generating the navigation, is rather intense.
This causes the publishing of policies to be
rather slow.
Good to Know
•
•
Hannon Hill has Professional Services that
can write triggers and web service clients.
Support model is very hands-on by Hannon
Hill. The number of customers writing web
service clients is small enough that
customers are supported better by people
than by extensive documentation that may
or may not address the issue.
Lessons Learned
•
•
Get technical portions working first
and then announce it to users.
Working with the web services is an
iterative process. The detailed
implementation can require changes
to the user process.
Future Development
• Add workflow for policy updating.
Detailed Information
• Additional information may be
downloaded from:
www.hannonhill.com/new-kb/Users-Conference/index.
Email Addresses
• [email protected]
• [email protected]
Summary
•
•
•
Need multiple copies of documents to
support location-specific navigation.
Allow content editors to edit source
documents and specify all display
locations.
Use publish trigger, that calls web
services, to create, update and publish
the copies.
Questions?