R Client for the MTurk Requester API

Provides programmatic access to the Amazon Mechanical Turk (MTurk) < https://www.mturk.com/mturk/welcome> Requester API.

MTurkR provides programmatic access to the crowdsourcing functionality of the Amazon Mechanical Turk (MTurk) Requester API.

Amazon provides a helpful chart for deciding whether an API client (like MTurkR) is useful for your project.

Possible applications of the package include:

  • Survey or experimental research using MTurk workers as human subjects
  • Large-scale human coding or categorization of text, images, video, audio, or other files
  • Collection of unstructured data from the web
  • Audio and video transcription
  • Preprocessing of "messy" data prior to statistical procedures

The MTurkR documentation files contain minimal examples for all functions. Further examples of how to use MTurkR are provided in the MTurkR GitHub wiki. Users can contribute their own examples or further documentation there, or via pull requests to the GitHub repository.

Using MTurkR requires setting two environment variables: AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY. Specify them on the command-line before initializing the R session, or use Sys.setenv() within R, or place these values in an .Renviron or Rprofile.site file. (Note: Environment variables are more secure than the credentials() function or options("MTurkR.keypair") from earlier versions of MTurkR.)

As a supplemental feature to MTurkR, there is also a graphical user interface available for the package. A simple, command-line version of this is available using the wizard.simple() function in MTurkR. A more advanced GUI, built using Tcl/Tk, is available in MTurkRGUI package.

If you experience problems using MTurkR, you can:


CRAN Version Downloads Travis-CI Build Status codecov.io

To install the latest version from CRAN, simply use:


To install the latest development version of MTurkR from GitHub:

# latest stable version
install.packages("MTurkR", repos = c(getOption("repos"), "http://cloudyr.github.io/drat"))
# latest (unstable) version from GitHub
if (!require("ghit")) {

cloudyr project logo




  • GetAssignments() now attempts to convert non-character variables to an appropriate R vector type via type.convert(, as.is = TRUE).
  • Some statistics have been deprecated for the GetStatistic() and RequesterReport(), including "NumberAssignmentsAvailable", "NumberAssignmentsAccepted", "NumberAssignmentsReturned", "TotalFeePayout", and "TotalRewardAndFeePayout". This is reflected in the return value of ListStatistics().


  • Added two papers (one from The Political Methodologist and one from The R Journal) as supplemental PDF documentation, accessible via ? MTurkR. (#93)
  • Documentation now indicates that GrantBonus() allows the reason argument to contain newlines and tabs. (#109)


  • Question identifiers used in a Qualification AnswerKey are only checked if CreateQualificationType() is called with validate.answerkey = TRUE. (#117, h/t Darrell Penta)
  • Fixed a bug in BulkCreateFromHITLayout() that caused the function to fail. (#114, h/t Tyler Scott)
  • Fixed a bug in trying to parse already parsed response for GetBonuses(). (#113, h/t Andrew Brown)
  • The simple wizard now correctly respects AWS credentials specified as environment variables instead of the deprecated credentials() function. (h/t Kim Gross)
  • Functions should now all return data frames with character class columns (unless otherwise noted) with sequentially numbered rownames. This corrects potentially some confusing behavior introduced during XML parsing. (#112)



  • GenerateExternalQuestion now sets a default value of frame.height = 400.
  • The AWS Access Key ID used for requests is now redacted from the error message printing when an API request fails to reduce the risk of accidentally disclosing the credential. (#105)
  • AWS removed the Categorization and Photo Moderation QualificationTypes on December 18th, 2015. Use of these QualificationTypeIds will trigger warnings in some cases and the "generic" Masters QualificationTypeId will be substituted automatically. Users should not need to update any code, though they may be receive warnings. These have also been removed from ListQualificationTypes(). (#104)
  • The use of credentials(), and options("MTurkR.keypair") are completely deprecated. A warning will now be issued if trying to supply credentials in this way. AWS credentials should be specified in environment variables AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY. Messages are triggered on package load if environment variables are not set. (#102)
  • HITStatus() will now display any non-empty RequesterAnnotation field for a HIT. (#101)
  • The deprecated argument print, which was removed in v0.5, has been fully removed. Use verbose instead. (#97)
  • The deprecated functions mturkrhelp and APIReference have been removed. (#96)
  • MTurkR.Wizard has been removed from MTurkR and has been recreated as a separate package called MTurkRGUI. (#95)
  • HITStatus now reports condensed column names better suited for narrow displays.
  • Correctly handle multiple response.group arguments passed to HIT-related functions. (#89)
  • authenticate, which was deprecated in v0.5, is removed.
  • Use of RCurl has been replaced by curl. This necessitated a new dependency, base64enc. (#86)
  • In lieu of credentials, AWS credentials should be specified via environment variables AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY. credentials will be deprecated.
  • SufficientFunds() reflects the new commission structure effective July 21, 2015. (#87)
  • BulkCreateFromTemplate and BulkCreateFromURLs now have a frame.height argument with a default of 450.


  • A small test suite has been added to help ensure correct package functionality. It is only executed in the sandbox in the presence of valid AWS credentials. (#100)
  • The package now includes pages describing common use cases for MTurk (surveys, categorization, sentiment analysis, and webscraping). (#98)
  • Descriptions of global package options have been added to MTurkR-package.Rd. (#103)
  • The package now includes additional HTMLQuestion templates including examples for image categorization and sentiment rating. (#94)
  • Documentation for GenerateHITReviewPolicy and GenerateAssignmentReviewPolicy was expanded and tested, and a related HTMLQuestion HIT template is now installed.


  • Fixed a bug in checking for a valid AWS keypair in request(). (h/t Sean Murphy, #106)
  • When ttk widgets were added to the graphical wizard in v0.6, scrollbar calls retained tk-only options. Those have been removed. (h/t Justin Koch)
  • State-level locales are now correctly supported in QualificationRequirements, including in complex combinations with country locales with no subdivision (so mixing U.S. states with other countries is possible). This feature was added in v0.6.5 but had a critical bug. (h/t Ricky Bilakhia)
  • Closing the graphical wizard crashed RStudio due to bringToTop(-1). This is fixed. (h/t Carolina Kuepper-Tetzel)



  • Added BulkCreate, BulkCreateFromTemplate, BulkCreateFromURLs, and BulkCreateFromHITLayout functions that can be used to create multiple HITs of the same HITType with identical properties but different question content. The hitsfromtemplate function has been removed. (#76)
  • The graphical wizard now supports creating HIT- and Assignment-level ReviewPolicies during HIT creation. (#55)
  • CreateHIT now only warns about duplicated arguments (passing both hit.type and arguments for RegisterHITType) if verbose = TRUE.
  • The graphical wizard now allows users to create an Qualification AnswerKey Template (via AnswerKeyTemplate) from a manually entered Qualification Test. (#73)
  • GrantBonus now accepts a unique.request.token argument. (#75)
  • The unused format argument to GenerateNotification has been removed. (#71)


  • Examples have been added to demonstrate functionality of AnswerKeyTemplate.
  • Documentation for GrantQualification and RejectQualification has been merged to clarify the use of these functions. Documentation has been expanded with complete examples of both functions. (#72)
  • GenerateQualificationRequirement now notes that U.S. state locale values (following ISO 3166-2 format) are allows. (#74)
  • GenerateHITsFromTemplate now has a more complete example that includes an installed template HTML file. (#72)
  • Several small changes have been made to correct documentation and fix bad formatting in the PDF version of the manual. (#72)


  • Fixed a bug in the CreateQualification dialog of the graphical wizard related to destroying the dialog. (h/t Sean Murphy)
  • Fixed a bug in the AssignQualification dialog of the graphical wizard. (h/t Carolina Kuepper-Tetzel)
  • Loading MTurkR no longer overwrites existing MRurkR options. (#83)
  • GetStatistic now consistently returns a three-column data frame containing the name of the requested statistic, its value, and the date of the statistic. If the request fails, this data frame has zero rows. Printing of request information (if verbose = TRUE) has further been fixed to prevent incorrect printing when count was specified. (#78)
  • Fixed a bug in ExtendHIT wherein specifying both add.assignments and add.seconds caused only assignments to be incremented. (#79)



  • GetReviewResults should now be able to return all results. This is untested due to the large page size of ReviewResults (65,535). (#54)
  • The graphical wizard now uses "themed" tk (ttk) widgets where possible, in order to improve the appearance. This creates a dependency on tcl >= 8.5.
  • The graphical wizard now allows retrieval of assignments by requester annotation (or batch) field.
  • The graphical and text-based wizards now allow you to preview the email that will be sent to workers using the ContactWorkers function. (#53)
  • The graphical and text-based wizards now allow users to assign qualifications to workers. (#61)
  • The graphical user interface (mturkr.wizard) has a significantly cleaned up code base aimed at simplifying future bug fixes. A user-visible side effect of this is modified aesthetics that should make the interface more attractive and more uniform. (#56)
  • mturkr.wizard now allows users to enter multiple WorkerIds for several functions by simply pasting one WorkerId per line into a multi-line textbox. (#52, h/t Justin Weigand)
  • GetReviewResultsForHIT now has pagenumber and pagesize arguments to control pagination of results. (#54)


  • Documentation for CreateHIT and CreateQualificationType now include examples using template files installed with the package. (#40)
  • Documentation of generating a Masters QualificationRequirement has been slightly clarified.
  • Examples of Amazon Simple Queue Service (SQS) notifications have been added, along with a detailed tutorial on the MTurk wiki. (#60)


  • Fixed a bug in the wizard that prevented the user from clearing and viewing QualificationRequirements that had been added to a HITType.
  • Numerous related bug fixes in the wizard related to populating HITId, HITTypeId, and QualificationTypeId fields were fixed. (#69)
  • Wizard balance checking message box now reports balance correctly. (#67)
  • Wizard dialogs for adding question data structures to a HIT did not close on accepting "OK". (#65)
  • GetAssignments only returned the first answer option when multiple selection answers were chosen by a worker. (#63, h/t Kyle Harms)
  • Fixed a possible bug in ContactWorkers batch mode that sent multiple emails to the same worker(s). (#58, h/t Andy Guess)
  • Fixed a few XML-related bugs in as.data frame.QuestionForm and as.data frame.AnswerKey.



  • GenerateReviewPolicy is removed in lieu of two new functions: GenerateHITReviewPolicy and GenerateAssignmentReviewPolicy, both of which have a simplified function API and include extensive examples in the documentation. The previous code never worked correctly, so this change should be unimportant. The use of CreateHIT remains unchanged. (#47)
  • HITStatus, ApproveAllAssignments, ChangeHITType, DisposeHIT, DisableHIT, ExpireHIT, ExtendHIT, GetAssignment, GetBonuses, SetHITAsReviewing now include an optional annotation argument that can be used to apply the function to all HITs of a given "batch" created in the online Requester User Interface. (#51)
  • GetFileUpload no longer has a file.ext argument. Instead the file extension is taken from the uploaded file itself. The format of the filename for the downloaded file has also been clarified. (#45)


  • Examples for all functions have been substantially expanded and checked for accuracy.
  • The MTurkR wiki (https://github.com/leeper/MTurkR/wiki) includes numerous updated code examples, including full documentation for both wizard.simple and mturkr.wizard, the MTurkR interactive interfaces.


  • Fixed a bug in SearchHITs that was not passing ... arguments ot GetQualificationType
  • Fixed a small bug in the batch (pagination) of SearchQualificationTypes.
  • Functions for QualificationTypes now return a data frame containing character variables rather than factor variables, as this could cause unintended coercion when using GenerateQualifiationType. (h/t Haotian Zhou)
  • Corrected several related bugs (API version number, LocaleValue specification) that prevented the "In" comparator in a QualificationRequirement from working with multiple LocaleValues in API version 2013-11-15. (#50, h/t Haotian Zhou)



  • GenerateQualificationRequirement now supports new comparators (given MTurk API update on 2014-07-17): "DoesNotExist", "In", and "NotIn". The "DoesNotExist" comparator allows requesters to effectively block workers who do not have, e.g., a requester-defined qualification. The "In" and "NotIn" comparators allow a boolean OR logical in QualificationRequirements (e.g., allowing a worker to be from the US or the UK) rather than having to create two separate HITs with different Locale restrictions.
  • The qual.number argument to GenerateQualificationRequirement has been removed.
  • The return.hit.dataframe argument to GetHITsForQualificationType has been removed.
  • Functions of the form *ToDataFrame have been replaced with as.data frame S3 methods, which are handled by a global as.data frame.MTurkResponse function. (#36)
  • ParseErrorCodes is removed (now used internally in request).
  • options('MTurkR.print') is deleted and replaced by options('MTurkR.verbose'). All functions will, temporarily, accept print arguments. All code should be updated to reflect the new terminology moving forward.
  • When validation.test=TRUE or browser=TRUE, functions now return an object of class "MTurkResponse" for the purpose of debugging.
  • The ability to execute requests in a web browser, via the options('MTurkR.browser') is removed from all functions except request. It can, however, still be passed via ... to request. (#38)
  • APIReference, mturkhelp, and genericmturkr are deprecated.
  • GetBonuses no longer supports the return.bonus.dataframe logical. All bonuses are automatically returned. To just retrieve details on number of bonuses per HIT, request only the first page with page.number=1 and verbose=TRUE.
  • MTurkR.Wizard now allows batch emails using ContactWorkers. WorkerIds can be specified as a comma-separated list. (#4)
  • The credentials function has now been changed to create and/or retrieve a global option MTurkR.keypair. This is a backwards compatible change that allows users to specify options(MTurkR.keypair=c(accesskey, secretkey)) directly without using credentials. All functions now point directly to the global option, so that credentials can be deprecated in a future release.
  • In many functions that return a data frame with a Valid column (e.g., ContactWorkers, etc.), Valid is now a factor. (#28)
  • Using ContactWorkers in batch mode now returns a data frame identical in structure to using it in regular mode, with one row per worker. (#32)
  • The signature argument to request now defaults to NULL and is generated automatically.
  • HITsToDataFrame (now as.data frame.HITs) now returns, for the QualificationRequirements list element, an empty data frame rather a single NA missing value if a given HIT does not have any QualificationRequirements attached to it.
  • request now generates the API authentication signature automatically, eliminating the need for authenticate, which is deprecated. request no longer accepts signature or timestamp arguments and has a new argument order. Arguments can be passed to request via a new ... argument in all functions.
  • request now uses curlPerform instead of getURL to gain finer control over API requests.
  • request now attempts to tryCatch any errors on writing the MTurkR log file. (#9)
  • request now returns an object of S3 class "MTurkResponse" and an associated print method has been added.


  • GenerateQualificationRequirement now correctly handles the "Exists" comparator by requiring an empty character string for the corresponding value argument.
  • HITStatus now prints correctly when only one HIT is retrieved.
  • ApproveAllAssignments now only attempts to approved assignments with AssignmentStatus "Submitted".
  • as.data frame.QuestionForm now returns "Overview" elements and "Question" elements as one data frame, distinguished by a new column: Element. (#8)
  • GetQualifications and GetQualificationRequests now respect the return.all argument.
  • The return.qual.dataframe argument in GetQualificationType and GetQualifications is removed.
  • Fixed a bug in GetBonuses whereby the default return.all=TRUE argument overwrote manually specified pagenumber and pagesize arguments. When GetBonuses is called with a hit.type, return.all is used recursively, such that return.all=FALSE returns the requested page for each HIT of the requested HITType. (h/t Dane Wendell, #37)
  • Corrected the API operation name in SendTestEventNotification.
  • Fixed a bug in GetBonuses using a HITTypeId that caused the function to fail catastrophically.


  • The return.qual.dataframe argument in GetQualificationType and GetQualifications is removed.
  • All functions now reflect the new ... argument, the contents of which are passed to request. (#38)
  • All references to the browser options are removed given removal of the feature from all functions.
  • Documentation for mturkhelp and APIReference now reflect deprecated status.
  • Documentation file for authenticate now reflect its deprecated status.
  • Documentation file for request has been updated to reflect its now non-exported status. request notes that it no longer accepts signature or timestamp arguments, which are now generated internally. Documentation for ParseErrorCodes is removed.
  • Work on the MTurkR wiki (https://github.com/leeper/MTurkR/wiki) has begun, with contributions from Kyle Hamilton.
  • The add.seconds parameter was incorrectly called add.time in the ExtendHIT documentation.



  • Functions that accept a hit.type argument can now generally accept a vector of HITTypeIds rather than just a single character string. See documentation.
  • GetBonuses now allows requests to be opened in the default browser, like most other API functions.
  • From version 0.3.7, global options (MTurkR.log,MTurkR.test,MTurkR.browser,and MTurkR.print) have been added to all functions. Defaults for these remain the same in all places.
  • From version 0.3.65, SearchHITs returns a named list of QualificationRequirements. This is a consequence of some significant changes to now the names of user-defined QualificationTypes are added to this list (moving that functionality from QualificationRequirementsToDataFrame to SearchHITs.
  • From version 0.3.6, a global option options('MTurkR.sandbox'=FALSE) has been added to allow easier reuse of code between testing in the sandbox and the live server. The default behavior remains to use the live server, but using an option, which is called by all functions, allows one to toggle the global option and then recycle code verbatim.
  • Per API changes on April 8, 2013, several built-in QualificationTypes were removed. See ListQualificationTypes for the available types.
  • Improved error, warning, and message handling by converting calls to cat to warning and message, where appropriate.
  • Most API query functions now include a validation.test parameter that, when TRUE, runs pre-query checks on the MTurkR call and returns the full API query URL. The call is not executed. This can be used for pre- or post-request debugging.
  • The 'MTurkRlog.tsv' file is now located according to a global option, specified by options(MTurkR.logdir = getwd()). The previous default behavior (to store the file in the working directory) is preserved but can now be globally modified for all requests. This might be useful for storing all requests (from all MTurk projects) in a single directory.


  • ContactWorkers now supplies additional information when attempting to contact ineligible workers (i.e., those who have not worked for the requester previously). These workers are indicated by a value of HardFailure in the Valid column of the response data frame. (#29)
  • All functions now explicitly convert factor arguments (e.g., for worker, qual, hit, etc.) to character in order to prevent some unintended side effects. (#26)
  • New versions of libcurl (>7.28) deprecate boolean values for SSL_VERIFYPEER and SSL_VERIFYHOST options. request was returning errors due to this and has been updated. Everything is backwards compatible to earlier versions of libcurl. (#31, h/t James Ben Taylor)
  • GenerateHITLayoutParameter now replaces any non-XML ampersands with &amp;, which had previously caused failures when creating HITs. (#30, h/t Eric Lin)
  • In GetAssignment, made a small change to prevent a warning and coercion of a one-assignment vector to a list. (h/t Eric Lin)
  • Two HITReviewPolicy names weres changed per a discussion on the AWS developer forum. Details here: https://forums.aws.amazon.com/thread.jspa?threadID=135285.
  • In HITStatus and HITsToDataFrame the correct capitalization of "Of" (from lower to upper) has been corrected throughout.
  • A series of related changes were made to GetAssignment, AssignmentsToDataFrame and QuestionFormAnswersToDataFrame to handle Answer data structures with differing QuestionIdentifiers. (h/t Robert Vesco and Solomon Messing)
  • Eliminated several response.group options from GetHITsForQualificationType due to personal communication with AWS.
  • Fixed bug in GetReviewableHITs that correctly queried the API but returned an incomplete data frame. (h/t Eric Lin)
  • Fixed bug in CreateHIT, saying type was not defined, that occurred when hit.type was non-NULL. (h/t Eric Lin)
  • Fixed bug in GetHIT, which trickled down through HITsToDataFrame and QualificationRequirementsToDataFrame that caused error when retrieving HITs from the sandbox with user-defined qualifications. (h/t Eric Lin)
  • Fixed bug in GenerateHTMLQuestion related to XML encoding. (h/t Alex Neustädter)
  • In request and several convenience functions, corrected shell.exec to browseURL to be platform-independent.
  • Added filters to request to handle a variety of problematic characters (that might emerge from FreeText responses in assignments). (h/t and contributions from Solomon Messing)
  • removed references to shell.exec and replaced with browseURL.


  • Where appropriate, functions using hit.type parameters are now described (correctly) as allowing vectors of HITTypeIds rather than just a single HITTypeId. See documentation.
  • Documentation for ContactWorkers now describes in greater detail what happens when attempting to contact ineligible workers. (#29)



  • CreateHIT now provides an option to validate the 'question' parameter for HTMLQuestion, ExternalQuestion, and QuestionForm HIT structures.
  • CreateQualificationType now provides options to validate the 'test' and/or 'answerkey' parameters for QuestionForm and AnswerKey structures, respectively.
  • GenerateAnswerKey was modified to accept slightly different inputs for the 'questions' parameter. A new function, AnswerKeyTemplate, produces a user-modifiable template (as a list) for that parameter based on a QuestionForm data structure for which an AnswerKey is desired.
  • When internal functions (described in the documentation of XMLToDataFrame) that normally return named lists have nothing to translate to a data frame, they now return named lists with NULL values rather than simply a single NULL.


  • Fixed bug in GetAssignments that prevented the return of assignments for an entire HITType or for multiple HITs. Specifically, previous version was retrieving assignments via the API but not returning them to the user. (h/t Robert Vesco)



  • To improve speed and reduce the number of API calls, HITsToDataFrame, which is called by SearchHITs, GetHIT, etc., now uses the convenience function ListQualificationTypes to return the name of QualificationRequirements when applicable.
  • Expanded the functionality of ContactWorkers to allow a 'batch' mode, wherein workers are contacted with generic (i.e., not customized) email messages and subject lines in batches of 100.
  • Text sent to console when print=TRUE modified slightly, so that iteration numbers are printed for each iteration (to monitor progress of mutli-item requests).
  • ResponseGroups updated to current API specifications for CreateHIT, GetAssignments, GetHIT, GetStatistic, GetWorkerStatistic, and SearchHITs. These now include an optional "Request" ResponseGroup to simply return information about the API call without executing that call.
  • AssignmentsToDataFrame now returns a "RequesterFeedback" variable, providing access to feedback supplied to workers when GetAssignments is called with the "AssignmentFeedback" ResponseGroup.
  • Updates were made to GenerateAnswerKey (see documentation) and a new function AnswerKeyToDataFrame was released.
  • QuestionFormToDataFrame created. See documentation for details.


  • MTurk API operation 'ChangeHITTypeofHIT' typo was fixed to 'ChangeHITTypeOfHIT'.
  • A small change was made to BlockWorkers to allow a single reason to be used for multiple workers, as originally intended.
  • Fixed bug in GetAssignments to allow retrieval of assignments by status, as originally intended.
  • Fixed bug in AssignmentsToDataFrame to correctly store 'ApprovalTime' or 'RejectionTime' in 'ApprovalRejectionTime'
  • Bug fixes to CreateQualificationType related to parsing of XML in 'test' and 'answerkey' parameters.
  • Modified credentials to pass CRAN checks.


  • Documentation for CreateQualificationType notes web browser-dependent constraints on URL length that may produce unintended behavior when 'browser=TRUE' and a 'test' parameter is specified.
  • Added documentation for AnswerKeyToDataFrame.
  • Various copyedits.



  • A completely new GUI interface for managing MTurkR, which now suggests library(tcltk), has been added. The previous, text-based wizard remains available by calling mturkr.wizard("simple").
  • Support for ReviewPolicies has been added, both creating them using GenerateReviewPolicy and retrieving their results using GetReviewResultsForHIT.
  • Support for AnswerKey data structures has been added, for use in CreateQualificationType.
  • Added simple checks for 'sortproperty', 'sortdirection', 'pagesize', and 'pagenumber' parameters in relevant functions.
  • Expanded support for QuestionForm data structures is currently under development and should be available in the next release.


  • A number of bug fixes have been corrected throughout the package. Most were minor and related to output, but some like the 'return.all' parameter in GetAssignments, were producing unintended behavior.
  • Various mislabeled function parameters (used within functions) have been corrected (some were creating notes on R CMD check).
  • request was modified to remove problematic whitespace characters when writing log entries (which made reading the log file into R problematic).


  • Fixed a number of minor errors in documentation.
  • Expanded documentation for functions associated with AnswerKey and ReviewPolicy data structures.
  • Expanded documentation for 'sortproperty' and 'sortdirection' parameters in relevant functions.

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.


0.8.0 by Thomas J. Leeper, 3 years ago


Report a bug at https://github.com/cloudyr/MTurkR/issues

Browse source code at https://github.com/cran/MTurkR

Authors: Thomas J. Leeper [aut, cre] , Solomon Messing [ctb] , Sean Murphy [ctb] , Jonathan Chang [ctb] , Andrew Brown [ctb]

Documentation:   PDF Manual  

Task views:

GPL-2 license

Imports utils, stats, curl, digest, base64enc, XML

Suggests testthat

See at CRAN