VersionOne Python SDK

Officially distributed via PyPi (pip) as: v1pysdk
An older version of this package, which flows with this version numbering, was distributed as 'v1pysdk-unoffical'

The VersionOne Python SDK is an open-source and community supported client for the VersionOne API.

As an open-sourced and community supported project, the VersionOne Python SDK is not formally supported by VersionOne.

That said, there are a number of options for getting your questions addressed:

• StackOverflow: For asking questions of the VersionOne Development Community.
• GitHub Issues: For submitting issues that others may try to address.

In general, StackOverflow is your best option for getting support for the VersionOne Python SDK.

The source code for the VersionOne Python SDK is free and open-source, and we encourage you to improve it by submitting pull requests!

Overview

Instantiating a connection

To interact, you must first create an instance of the V1Meta object. This requires you to specify how to connect to the server.

There are two options, specifying the full URL to your instance directly, or specifying individual details.

from v1pysdk import V1Meta

with V1Meta(
  instance_url = 'http://localhost/VersionOne',
  # any instance, scheme, or address values will be ignored
  username = 'admin',
  password = 'admin'
  ) as v1:

Alternatively

from v1pysdk import V1Meta

with V1Meta(
  address = 'localhost',
  instance = 'VersionOne',
  scheme = 'http', #optional, defaults to https
  username = 'admin',
  password = 'admin'
  ) as v1:

To authenticate, two methods are provided, username and password as demonstrated above, or Access Tokens. Tokens are created by logging in to VersionOne via the web interface, going to the user's profile, going to Applications, and creating a new application. This will provide an Access Token that looks something like 1.2cFHe7NkoO1kOV/x8WLpw1NasJg=. KEEP THIS SECRET since it's the secret API access key for your specific user on your instance.

To use an access token, you use it as your password and set the flag indicating it's really an Access Token. This will no longer require a username to be present.

from v1pysdk import V1Meta

with V1Meta(
  address = 'localhost',
  instance = 'VersionOne',
  password = '1.2cFHe7NkoO1kOV/x8WLpw1NasJg=',
  use_password_as_token=True,
  ) as v1:

Dynamic reflection of all V1 asset types:

Just instantiate a V1Meta. All asset types defined on the server are available as attributes on the instance. The metadata is only loaded once, so you must create a new instance of V1Meta to pick up metadata changes. Each asset class comes with properties for all asset attributes and operations.

from v1pysdk import V1Meta

with V1Meta(
  instance_url = 'http://localhost/VersionOne',
  username = 'admin',
  password = 'admin'
  ) as v1:

  user = v1.Member(20) # internal numeric ID

  print user.CreateDate, user.Name

Simple access to individual assets:

Asset instances are created on demand and cached so that instances with the same OID are always the same object. You can retrieve an instance by passing an asset ID to an asset class:

      s = v1.Story(1005)

Or by providing an OID Token:

      s = v1.asset_from_oid('Story:1005')

      print s is v1.Story(1005)   # True

Lazyily loaded values and relations:

NOTE: Making requests synchronously for attribute access on each object is costly. We recommend using the query syntax to select, filter, aggregate, and retrieve values from related assets in a single HTTP transaction.

Asset instances are created empty, or with query results if available. The server is accessed for attributes that aren't currently fetched. A basic set of attributes is fetched upon the first unfound attribute.

      epic = v1.Epic(324355)

      # No data fetched yet.
      print epic  #=>  Epic(324355)

      # Access an attribute.
      print epic.Name  #=> "Team Features"

      # Now some basic data has been fetched
      print epic       #=> Epic(324355).with_data({'AssetType': 'Epic',
                           'Description': "Make features easier for new team members", 'AssetState': '64',
                           'SecurityScope_Name': 'Projects', 'Number': 'E-01958', 'Super_Number': 'E-01902',
                           'Scope_Name': 'Projects', 'Super_Name': 'New Feature Development',
                           'Scope': [Scope(314406)], 'SecurityScope': [Scope(314406)],
                           'Super': [Epic(312659)], 'Order': '-24', 'Name': 'Team Features'})

  # And further non-basic data is available, but will cause a request.

      print epic.CreateDate   #=>  '2012-05-14T23:45:14.124'

The relationship network can be traversed at will, and assets will be fetched as needed.

    # Freely traverse the relationship graph
    print epic.Super.Scope.Name  #=> 'Products'

Since the metadata is modeled as data, you can find the list of "Basic" attributes:

    basic_attr_names = list( v1.AttributeDefinition
                               .where(IsBasic = "true")
                               .select('Name')
                               .Name
                           )

Operations:

Operations on assets can be initiated by calling the appropriate method on an asset instance:

      for story in epic.Subs:
        story.QuickSignup() 

The asset instance data will be invalidated upon success, and thus re-fetched on the next attribute access.

Iterating through all assets of a type

The asset class is iterable to obtain all assets of that type. This is equivalent to the "query", "select" or "where" methods when given no arguments.

      # WARNING: Lots of HTTP requests this way.
      members = list(v1.Member)                               # HTTP request to get the list of members.
      print "Members: " + ', '.join(m.Name for m in members)  # HTTP request per member to fetch the Name

      # A much better way, requiring a single HTTP access via the query mechanism.
      members = v1.Member.select('Name')
      print "Members: " + ', '.join(m.Name for m in members)  # HTTP request to return list of members with Name attribute.

      # There is also a shortcut for pulling an attribute off all the results
      members = v1.Member.select('Name')
      print "Members: " + ', '.join(members.Name)

      # Alternative to best way with more explicit indication of what's being done
      members = v1.Member.select('Name')
      members.queryAll()   # forces performing HTTP query to fetch all members' basic details
      print "Members: " + ', '.join(m.Name for m in members)

Queries

Query Objects

the select(), where(), and sort() methods on asset instances return a query object upon which you can call more .where()'s, .select()'s, and .sort()'s. Iterating through the query object will run the query.

the .first(), .queryAll(), and .reQueryAll() methods on a query object will run the query immediately and return the appropriate result.

the find() can be used to perform a server-side whole-word match on a field, though it's server intensive, can only match one field, and should be used sparing.

the page() can be used to limit results for the purposes of performing server-side paging.

the reQueryAll() can be used like the queryAll(), but will clear all previously cached data and re-run the HTTP query if any query options have been changed, allowing for easily repeating a query where only response limits such as page() have changed.

Simple query syntax:

Use .where(Attr="value", ...) to introduce "Equals" comparisons, and .select("Attr", ...) to append to the select list.

Non-"Equal" comparisons are not supported (Use the advanced query syntax instead).

      for s in v1.Story.where(Name='Add feature X to main product"):
          print s.Name, s.CreateDate, ', '.join([owner.Name for owner in s.Owners])

      # Select only some attributes to reduce traffic

      for s in v1.Story.select('Name', 'Owners').where(Estimate='10'):
          print s.Name, [o.Name for o in s.Owners]

Advanced query, taking the standard V1 query syntax.

The filter() operator will take arbitrary V1 filter terms.

      for s in (v1.Story
                  .filter("Estimate>'5';TotalDone.@Count<'10'")
                  .select('Name')):
          print s.Name

Limiting results from the server via paging

It can be easier on the client to have the server perform paging by limiting the number of results returned matching a query. Paging requires a limit on the number of items returned, and an index of the first item in the list to return.

The API allows the index to be left off, which assumes a default start index of 0.

    pageNum = 0
    pageSize = 3
    pageStart = 0
    while True:
        results = ( v1.Story
                     .select('Name')
                     .filter(str(myFilter))
                     .sort('-Name')
                     .page(size=pageSize, start=pageStart) 
                  ) # Requires a new query each time
        if not len(results):
            break;
        print("Page items = " + str(len(results)))
        pageNum += 1
        pageStart += pageSize
        print("Page " + str(pageNum) + " : " + ',   '.join(results.Name))

Alternatively the reQueryAll() can be used to force re-querying of the content based on updated query settings to make paging easier to implement.

    pageNum = 0
    pageSize = 3
    pageStart = 0
    results = ( v1.Story
                .select('Name')
                .filter(str(myFilter))
                .sort('-Name')
               )

    while True:
        results = results.page(size=pageSize, start=pageStart).reQueryAll()
        if not len(results):
            break;
        pageNum += 1
        pageStart += pageSize
        print("Page " + str(pageNum) + " : " + ',   '.join(results.Name))

Sorting

Sorting can be included in the query by specifying the order of the columns to sort on, and whether those columns should be sorted ascending or descending. The default sort order is ascending.

sort() operates like select(), where field names are listed in quotes and may be listed as separate arguments to a single sort call, separate sort calls, or a mixture of both. Sorting descending requires the field name to be prefaced with a dash, '-'.
Fields may only be listed in the sort order once, with repeats being ignored.

To sort in reverse alphabetical order of names, then on Estimate time, then on Detailed Estimate time:

    results = v1.Story.select('Name').filter(str(myFilter)).sort('-Name','Estimate').sort('DetailedEstimate')
    print '\n'.join(results.Name)

Matched searching

Searching, while possible, is very server intensive and should be avoided as much as possible. Server-side searching can be whole-word matched within a single field. For this reason it should be significantly limited with appropriate filter/where commands.

    results = v1.Story.select('Name').filter(str(myFilter)).find('Get a', field='Name')
    print ', '.join(results.Name) #=> Get a handle on filtering, Get a toolkit for ease of use

Advanced selection, taking the standard V1 selection syntax.

The select() operator will allow arbitrary V1 "select" terms, and will add them to the "data" mapping of the result with a key identical to the term used.

    select_term = "Workitems:PrimaryWorkitem[Status='Done'].Estimate.@Sum"
    total_done = ( v1.Timebox
                     .where(Name="Iteration 25")
                     .select(select_term)
                 )
    for result in total_done:
      print "Total 'Done' story points: ", result.data[select_term]

Advanced Filtering and Selection

get a list of all the stories dedicated people are working on

      writer = csv.writer(outfile)
      results = (
        v1.Story
          .select('Name', 'CreateDate', 'Estimate', 'Owners.Name')
          .filter("Owners.OwnedWorkitems.@Count='1'")
          )
      for result in results:
          writer.writerow((result['Name'], ', '.join(result['Owners.Name'])))

Simple creation syntax:

GOTCHA: All "required" attributes must be set, or the server will reject the data.

      from v1pysdk import V1Meta
      v1 = V1Meta(username='admin', password='admin')
      new_story = v1.Story.create(
        Name = 'New Story',
        Scope = v1.Scope.where(Name='2012 Projects').first()
        )
      # creation happens immediately. No need to commit.
      print new_story.CreateDate
      new_story.QuickSignup()
      print 'Owners: ' + ', '.join(o.Name for o in story.Owners)

Simple update syntax.

Nothing is written until V1Meta.commit() is called, and then all dirty assets are written out.

      story = v1.Story.where(Name='Super Cool Feature do over').first()
      story.Name = 'Super Cool Feature Redux'
      story.Owners = v1.Member.where(Name='Joe Koberg')      
      errors = v1.commit()  # flushes all pending updates to the server
      if not errors:
          print("Successfully committed!")
      else:
          for e in errors:
              raise e 

The V1Meta object also serves as a context manager which will commit dirty object on exit.

      with V1Meta() as v1:
        story = v1.Story.where(Name='New Features').first()
        story.Owners = v1.Member.where(Name='Joe Koberg')
        
      print "Story committed implicitly."

Attachment Contents

Attachment file bodies can be fetched or set with the special "file_data" attribute on Attachment instances.

See the v1pysdk/tests/test_attachment.py file for a full example.

As Of / Historical Queries

Queries can return data "as of" a specific point in the past. The .asof() query term can take a list (or multiple positional parameters) of timestamps or strings in ISO date format. The query is run for each timestamp in the list. A single iterable is returned that will iterate all of the collected results. The results will all contain a data item 'AsOf' with the "As of" date of that item.
Note that the "As of" date is not the date of the previous change to the item, but rather is exactly the same date passed into the query.
Also note that timestamps such as "2012-01-01" are taken to be at the midnight starting that day, which naturally excludes any activity happening during that day. You may want to specify a timestamp with a specific hour, or of the following day. The timezone used when performing these comparisons is the timezone configured for the user specified in the V1Meta object, and the time comparison is performed based on the time as determined by the server.

      with V1Meta() as v1:
        results = (v1.Story
                     .select("Owners")
                     .where(Name="Fix HTML5 Bug")
                     .asof("2012-10-10", "2012-10-11")
                  )
        for result in results:
            print result.data['AsOf'], [o.Name for o in result.Owners]

Polling (TODO)

A simple callback api will be available to hook asset changes

      from v1meta import V1Meta
      from v1poll import V1Poller

      MAILBODY = """
      From: VersionOne Notification <notifier@versionone.mycorp.com>
      To: John Smith <cto@mycorp.com>

      Please take note of the high risk story '{0}' recently created in VersionOne.

      Link: {1}


      Thanks,

      Your VersionOne Software
      """.lstrip()

      def notify_CTO_of_high_risk_stories(story):
        if story.Risk > 10:
            import smtplib, time
            server = smtplib.SMTP('smtp.mycorp.com')
            server.sendmail(MAILBODY.format(story.Name, story.url))
            server.quit()
            story.CustomNotificationLog = (story.CustomNotificationLog +
                "\n Notified CTO on {0}".format(time.asctime()))

      with V1Meta() as v1:
        with V1Poller(v1) as poller:
          poller.run_on_new('Story', notify_CTO_of_high_risk_stories)

      print "Notification complete and log updated."

Performance notes

An HTTP request is made to the server the first time each asset class is referenced.

Assets do not make a request until a data item is needed from them. Further attribute access is cached if a previous request returned that attribute. Otherwise a new request is made.

The fastest way to collect and use a set of assets is to query with the attributes you expect to use included in the select list. The entire result set will be returned in a single HTTP transaction if you manually call one of the methods that triggers a full query. These methods include __iter__() (e.g. .join() uses this), __len__(), queryAll(), and reQueryAll().

Writing to assets does not require reading them; setting attributes and calling the commit function does not invoke the "read" pipeline. Writing assets requires one HTTP POST per dirty asset instance.

When an asset is committed or an operation is called, the asset data is invalidated and will be read again on the next attribute access. Grouping your updates then calling queryAll() on a fresh query is a good way to enhance performance.

GOTCHA: reQueryAll() tracks the dirty state of the query object separately from the way asset data is invalidated following an update. Unless the terms of the query have been changed, the reQueryAll won't update the cached data and a new query will be generated for each invalidated data item accessed. To avoid this, adding and then restoring a query term on the query object can be used to cause the re-query to actually occur.

reQueryAll() can be very useful when implementing paging, changing the sorting, etc, but it should be used with care. It clears all cached data, so any fields that were not included in the original query and have since been retrieved are also cleared. Accessing those fields will prompt the same individual query as before. To avoid this problem, either include the extra field(s) in your initial query, or create a new query object for the updated query terms.

TODO

• Make things Moment-aware

• Convert types between client and server (right now everything is a string)

• Add debug logging

• Beef up test coverage

• Asset creation templates and creation "in context of" other asset

• Correctly handle multi-valued attributes including removal of values.

Installation

run python setup.py install, or just copy the v1pysdk folder into your PYTHONPATH.

Revision History

2018-07-02 v0.6.2 - Fix a critical memoization bug, error reponse printing, some HTTP/PUT calls, authentication error handling

A critical memoization bug caused by the decorator being used prevented the same field of more than one item of the same type from being updated in a single invocation of the Python intepreter; i.e. it's only possible to update the Title of one Story within a Python script, regardless of how many V1Meta objects are created. It also prevented the V1Meta objects from being created with separate credentials.

Bug in how HTTP 400 responses were handled caused an exception to be thrown during handling and raising of an exception, preventing the actual error response provided with the HTTP 400 from being printed.

Bug in how NTLM authentication was handled prevented the HTTP 401 authentication error from being raised and handled so the errors would silently fail without the GET/POST command completing.

A bug in the creation of the HTTP POST commands in Python3 caused a TypeError exception to be thrown when no data payload was needed. This prevent Operations with no arguments from being used on V1 objects.

Unittests were added to ensure some Operations work properly. Connection tests to ensure bad credentials result in an identifable failed connection were also added. Tests specifically to ensure separation of credentials between different V1Meta objects within the same tests produce different results, thereby checking that memoization is working properly on a per-V1Meta object basis were also added.

2018-06-21 v0.6.1 - Fix a new item creation bug and added unittests for creation

2018-06-21 v0.6 - Rebased to include some historical changes that were lost between 0.4 and 0.5.

Fixed the tests so they can be run and succeed, including adding tests that check functionality of connections and some basic querying.

Critical lost differences that were recovered: OAuth token support memoization fixes

2018-06-13 v0.5.1 - PyPi upload so it's available via pip as "v1pysdk".

2018-06-12 v0.5 - Dynamic Python3 support added.

Add page(), sort(), queryAll(), find(), max_length(), length(), and support for len() usage to the query objects.

Primary repository moved to a fork that's maintained.

2013-09-27 v0.4 - A correction has been made to the multi-valued relation setter code. It used the wrong value for the XML "act" attribute, so multi-value attributes never got set correctly. Note that at this time, there is no way to un-set a value from a multi-valued relation.

2013-07-09 v0.3 - To support HTTPS, A "scheme" argument has been added to the V1Meta and V1Client constructors.

An instance_url keyword argument was added to V1Meta and V1Client. This argument can be specified instead of the address, instance_path, scheme, and port arguments.

A performance enhancement was made to calls such as "list(v1.Story.Name)". The requested attribute is added to the select list if it's not present, thus preventing an HTTP GET for each matched asset.

Some poor examples were removed and logging cleaned up in places.

Fix some issues with NTLM and urllib2. (thanks campbellr)

Missing attributes now return a None-like object can be deferenced to any depth. (thanks bazsi)

License

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

• Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.

• Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.

• Neither the name of VersionOne, Inc. nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission of VersionOne, Inc.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY, AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.