# Updating Packages on DataONE

## Updating A DataONE Package

A DataONE package is a collection of datasets and other files that are described by a metadata file.

A DataPackage is a dataone R object that can contain a DataONE package that is either downloaded from DataONE or newly created locally in R and then uploaded to DataONE.

After a package has been uploaded to a DataONE Member Node, it may be determined by the package submitter or other interested parties that the package needs to be updated, for example to add a missing file, replace one file with another, or remove a package member from the package.

These types of modifications can be accomplished by downloading the package from DataONE using the getDataPackage method to create a local copy of the package in R, modifying the package contents locally, then uploading the modified package to DataONE.

The complete example script used to perform a package update is shown below:

library(datapack)
library(dataone)
d1c <- D1Client("STAGING", "urn:node:mnStageUCSB2")
packageId <- "resource_map_urn:uuid:a9aeefcf-228c-4534-b4ad-b480a937be7d"
pkg <- getDataPackage(d1c, identifier=packageId, lazyLoad=TRUE, limit="0MB", quiet=FALSE)
metadataId <- selectMember(pkg, name="sysmeta@formatId", value="eml://ecoinformatics.org/eml-2.1.1")
objId <- selectMember(pkg, name="sysmeta@fileName", value='Strix-occidentalis-obs.csv')
zipfile <- system.file("extdata/Strix-occidentalis-obs.csv.zip", package="dataone")
pkg <- replaceMember(pkg, objId, replacement=zipfile, formatId="application/octet-stream")
auxFile <- system.file("extdata/WeatherInf.txt", package="dataone")
auxObj <- new("DataObject", format="text/plain", filename=auxFile)
pkg <- addMember(pkg, auxObj, metadataId)
newPackageId <- uploadDataPackage(d1c, pkg, public=TRUE, quiet=FALSE)

This example script downloads the example package that was created and uploaded to DataONE in the vignette upload-data.

Each line of this script will be described in the following sections.

The first step in updating a package is to download the package from DataONE into R, so that it can be modified locally using methods in the dataone package. Modifications can be made to the package such as adding or removing members from the package, or changing the contents of a package member, as would be the case if the wrong file was initially uploaded.

The getDataPackage method downloads all files belonging to the package specified by the identifier parameter.

d1c <- D1Client("STAGING", "urn:node:mnStageUCSB2")
pkg <- getDataPackage(d1c, identifier=packageId, lazyLoad=TRUE, limit="0MB", quiet=FALSE)

Because packages might contain large files or a large number of files, it is possible to lazyLoad the package. This means that the metadata describing the files is downloaded downloaded, but the file contents, the data bytes, are not. An upper size limit can be specified when using lazyLoad by also specifying the limit parameter. Files that are larger than the limit value are not downloaded, only system metadata.

For the package update that will be shown, it is not necessary to download the data bytes of each package member, so lazyLoading is acceptable. An example when lazyloading would not be appropriate is when the files in a package will be used for local computational processing.

Note that metadata files, such as the EML in this example, are always downloaded regardless of the lazyLoad parameter value.

### 2. Review package contents.

The downloaded package can be viewed by typing the DataPackage object name at the console, which invokes the R show method for the object:

pkg

Note that the show output for a DataPackage is condensed to fit the width of the current R console. If the output is condensed and more detail is required, set the R console width to a larger value, for example, options(width=120), or if using Rstudio, widen the console window by clicking and dragging on the window boarder.

### 3. Modify DataObjects in the package

The original uploaded package included the file Strix-occidentalis-obs.csv. This file can be substituted for a different file, as would be necessary if it was determined that the zipped form of the file should have been used instead.

First, determine which DataObject in the DataPackage pkg contains the file to be replaced:

objId <- selectMember(pkg, name="sysmeta@fileName", value='Strix-occidentalis-obs.csv')

The selectMember method inspects every DataObject in the package pkg and checks for a match in the R S4 slot specified by the name argument for the value specified with the value argument. The identifier for any matching DataObject is returned.

The documentation for the DataObject R slots available can be viewed with the command help("DataObject-class"). As a SystemMetadata object is contained in each DataObject, the slots for SystemMetadata are available, with documentation viewable with help("SystemMetadata-class")

Next, update the DataObject in pkg to replace the file Strix-occidentalis-obs.csv with Strix-occidentalis-obs.csv.zip using the replaceMember method:

objId <- selectMember(pkg, name="sysmeta@fileName", value='Strix-occidentalis-obs.csv')
zipfile <- system.file("extdata/Strix-occidentalis-obs.csv.zip", package="dataone")
pkg <- replaceMember(pkg, objId, replacement=zipfile, formatId="application/octet-stream")

The replaceMember method updates the DataPackage pkg, replacing the data content of the DataObject with identifier objId, and updating the relevant system metadata slots such as size and checksum, to reflect the new contents of the DataObject.

In this example, the file WeatherInf.txt was mistakenly omitted from the original package upload, so add it now:

First, the identifier for the metadata DataObject that describes the package members will be retrieved from the DataPackage:

metadataId <- selectMember(pkg, name="sysmeta@formatId", value="eml://ecoinformatics.org/eml-2.1.1")

The selectMember method returns the identifier of any DataObject in the package with an R slot name specified in the parameter name that matches the value specified in the argument value. In this particular package, there is only one DataPackage that has formatId of eml://ecoinformatics.org/eml-2.1.1, so only that identifier is returned.

Note that the getValue method can be used to retrieve the values for DataObject slots, for example:

getValue(pkg, name="sysmeta@formatId")

An R named list is returned, with the names being the identifiers of each DataPackage in the DataPackage and the values being the slot value.

Next, add a new package member that was omitted from the original package:

auxFile <- system.file("extdata/WeatherInf.txt", package="dataone")
auxObj <- new("DataObject", format="text/csv", filename=auxFile)
pkg <- addMember(pkg, auxObj, metadataId)

The modified package can be reviewed before updating to DataONE:

### 4. Upload the modified DataPackage

Now upload the modified package to DataONE. Each DataObject in the DataPackage will be inspected by uploadDataPackage and DataObjects that have been modified will be updated and DataObjects that have been added to the DataPackage will be uploaded.

newPackageId <- uploadDataPackage(d1c, pkg, public=TRUE, quiet=FALSE)

## Updating Metadata For A DataPackage

As DataPackage members are modified, the DataObject that holds the metadata that describes the package members may become outdated.

For example, it was shown above that a package member was updated to contain the file Strix-occidentalis-obs.csv.zip instead of Strix-occidentalis-obs.csv. After this change, the DataObject holding the EML metadata should be updated so that the EML element objectName for this dataset matched the new filename:

metadataId <- selectMember(pkg, name="sysmeta@formatId", value="eml://ecoinformatics.org/eml-2.1.1")
nameXpath <- '//dataTable/physical/objectName[text()="Strix-occidentalis-obs.csv"]'
newName <- basename(zipfile)
pkg <- updateMetadata(pkg, metadataId, xpath=nameXpath, replacement=newName)

The updateMetadata method updates a DataObject containing XML, substituting the value located in the document specified with the xpath argument with the value provided in the replacement argument. In this example, the XML contained in the DataObject with identifier metadataId is an EML metadata document. The EML element objectName is updated with the value Strix-occidentalis-obs.csv.zip.

Note the the xpath argument uses the XPath query language, which is used to locate elements within an XML document.

For EML metadata, another element that may need to be updated, if it is present in the original EML file, is the distribution url. This element contains the link that can be used to download the file from DataONE, which includes the DataONE identifier for the object. Because the identifier value is set when the DataObject is created, the EML metadata is out of date and needs to be updated with the new identifier value.

In this example, the section of the metadata that describes the dataset Strix-occidentalis-obs.csv.zip will be updated.

A portion of the EML from our example looks like this:

    <dataTable>
<entityName>Strix Occidentalis</entityName>
<entityDescription>A data file that contains only observations of Strix occidentalis</entityDescription>
<physical>
<objectName>Strix-occidentalis-obs.csv</objectName>
<size unit="byte">6017</size>
<dataFormat>
<externallyDefinedFormat>
<formatName>text/csv</formatName>
</externallyDefinedFormat>
</dataFormat>
<distribution id="1430343425153">
<online>
</online>
</distribution>
</physical>
<entityType>Other</entityType>
</dataTable>

The following lines get the current identifier for the DataObject that contains the correct file, and uses this identifier value to construct a standard DataONE access URL, using the base URL of the DataONE coordinating node.

objId <- selectMember(pkg, name="sysmeta@fileName", value='Strix-occidentalis-obs.csv.zip')
newURL <- sprintf("%s/%s/resolve/%s", d1c@cn@baseURL, d1c@cn@APIversion, objId)
newURL

The selectMember method checks slot specified in name argument, in this case sysmeta@fileName, of each DataObject in the DataPackage pkg and returns the identifier of any member that matches the specified value. This package has only one DataObject with the name Strix-occidentalis-obs.csv.zip so only one identifier is returned in objId.

Next, the metadata element for the distribution url will be updated in the DataObject that contains the EML metadata in the package pkg

metadataId <- selectMember(pkg, name="sysmeta@formatId", value="eml://ecoinformatics.org/eml-2.1.1")
xpathToURL <- "//dataTable/physical/distribution[../objectName/text()=\"OwlNightj.csv\"]/online/url"
pkg <- updateMetadata(pkg, do=metadataId, xpath=xpathToURL, replacement=newURL)

When the modified metadata file is updated to DataONE, a new identifier is required, as the modified metadata will replace the old one by created a new object in DataONE, and marking the old one as obsoleted by the new one. For this reason, the call to updateMetadata will assign a new identifier to the metadata DataObject in the DataPackage pkg, if necessary.