EUDAT B2SHARE API - How to store and publish research data using the B2SHARE API

Editor's Notes

• #2: This presentation briefly introduces the B2SHARE data store and publication service of EUDAT.
• #3: B2SHARE is a user-friendly, reliable and trustworthy way for researchers, scientific communities and citizen scientists to store and publish research data coming from diverse contexts.
• #4: B2SHARE is part of the EUDAT CDI and is directly connected to the B2DROP and B2SAFE services for data import and export. Other services like B2HANDLE are used to store persistent identifiers. Additional annotation like provenance data is stored using B2NOTE. B2ACCESS is required for authentication and authorization in the future. The B2FIND service is used for metadata harvesting so that your records can be found in the more general search service of EUDAT.
• #5: B2SHARE has web-based GUI, but it also offers a REST API. The B2SHARE API is a set of instructions to interact with a B2SHARE service instance, for example the one of EUDAT. It provides direct interaction with the service without using a graphical user interface. Instead a command line tool or an application that integrates the communication is used. For example, the tools used in a research workflow might include automatic upload of final data in order to directly publish it. Most programming language have built-in functionality to interact with a remote service using HTTP requests. The B2SHARE REST API can be used by any user without registration and without limitations. The only functionality that requires registration is the creation and modification of records and communities.
• #6: Using the B2SHARE API, the user can automate the creation or draft records, uploading of files, alteration of metadata of draft records, changing the publication state of a draft record (in order to publish it), retrieval of record data and files and the removal of existing records. It enables precise replication of metadata into B2SHARE and large file uploads, which is more difficult to achieve using a web browser and retyping metadata field values. With the API, you can implement the sharing and publication of datasets into your own workflows or applications. You can make administration of existing records easier without having to use a browser and to apply changes to every record separately.
• #7: More specifically, using the B2SHARE API, the user can perform several important operations such as listing all existing records and communities that are currently defined in the service. When a user is looking for specific records and/or communities the query functionality can be used and filtered using several parameters. Once a community has been chosen, detailed community-specific information, including community metadata schema definitions and other requirements can be retrieved. New draft records can be created, including the upload of files and the addition of metadata in accordance to the metadata schema defined by a community or project. Once a draft record is completed, it can be published. Furthermore, existing published records can be modified in terms of metadata and published as a new versioned record. The old version will always be available and citable. Draft records can be deleted, as well as files uploaded in these draft records. Published records can only be deleted by the site administrator.
• #8: The next three slides may be common knowledge, so we can skip them if everybody knows about these. API stands for Application Programming Interface, and is a specification of a set of definitions, protocols and tools which allow to interact with a service, possibly from a remote location. An API provides an abstraction layer of the underlying service technologies used by the service itself and external applications to communicate with the technology without exactly knowing how it works. This allows machines to easily interact with the service and correctly handle responses and possible errors. When it comes to APIs for the Internet, communication is done through HTTP requests which return machine-readable structures of data representing the current state of a piece of information. The API can often used to alter information on the server side as well. The returned data is usually provided in specific formats like XML or JSON. In many cases this can be changed upon request. Your own browser is using the same mechanisms to get data from a server and present it to the user.
• #9: APIs are used by making HTTP requests. These are specific calls to a service using a supported method with address and parameters. The address is structured as a URL which consists of a protocol, hostname, port (often not necessary) and a path which possibly identifies the piece of information requested. The path is often called the API endpoint. Multiple endpoints can exists for a given service. Parameters are additional options given to the request which can further filter the return information or specify for example the return format. All requests always return a response, provided that the server is available, even when the request is malformed or some other error occurred. Depending on the success, the request contains a HTTP status code and a response text with further information. On success the current state of the requested piece of information is provided.
• #10: Each HTTP request is accompanied by a HTTP request method, such as the GET method which request data from a specified resource. Other possibilities (among others) are POST for data submits, PUT for uploading files and PATCH for alteration of existing data. Certain operations (like modification) are not specifically limited to a method, this entirely depends on the API specification and implementation. For B2SHARE the methods are specifically used for specific operations. Some operations require special privileges. Along with the request method, URI and authentication, the request also sends a payload if necessary. This can be text, binary data or even entire files.
• #11: To actually make a request, use a designated tool or application, or integrate it into your own application using a programming language which supports this. Examples are GUI applications like file transfer tools or web interfaces on websites. On the command line typical examples are curl and wget. Most programming languages support making requests over HTTP, but often require the inclusion of packages which provide this functionality. For more complex operations (like publication), a dedicated interface or command line application is required and therefore a file transfer tool often doesn’t suffice.
• #12: Here is an overview of how a request works in practice using a client-server model: A browser, tool or app/workflow creates a request with header, authentication and the actual request payloads (possibly with a reference to an object through a identifier) and sends it to the server (e.g. a B2SHARE instance) The server processes the request and determines whether it is understood, valid and allowed and if it applies to the right objects It sends a response back to the original sender together with a header, status code (indicating whether the request was successfully processed) and a response text, which may contain some information on what has changed, or the new state of the referred object in the original request To successfully render more complex webpages, as is common nowadays, many requests can be made from a browser to a server.
• #13: There are several important concepts used in B2SHARE which are relevant to using the API. First there are communities that curate datasets which are part of the scientific domain or a research project. Community administrators maintain the metadata schema of the community. Users have to select the community when creating new records in order to have it connected that community. Records contain data files and associated metadata and are always connected to a specific community. Communities actively curate records that are published under their name. When publishing under the EUDAT general community, no curation takes place. Metadata are a set of common fixed metadata fields and custom metadata blocks with additional fields. They are governed by fixed and community metadata schemas. When choosing a specific community, there might be additional metadata fields to be filled in. A state is the current condition of a record, either draft, submitted or published. They can be changed through the API. Only published records are visible in B2SHARE. Draft and submitted records are only visible to the owner and possibly the community administrator under which the record is published.
• #14: B2SHARE defines several request variables that function as identifiers for objects in B2SHARE. They are used in most HTTP request addresses as part of the path to access specific objects directly The most important variables are the community ID which uniquely identifies a community. Record IDs identify specific records, either in draft or published state. The file bucket ID is used to identify a set of files of a specific record
• #15: All records are published under a specific community and have metadata added according to the corresponding community metadata schema in case this is required. Metadata schemas contain the descriptions, vocabularies and expected structure and format of every metadata field in the schema. Furthermore, they define which fields are mandatory to fill in during the creation of new records. Community metadata schema definitions are publically available through the API and on the B2SHARE website.
• #16: B2SHARE supports two types of records: draft records and published records. Draft records are unpublished records that can be freely modified by the owner. This means new files and metadata can added, existing ones be changed or removed. Published records are final and therefore cannot have any files added or changed. Metadata changes are allowed but discouraged in order to keep the authenticity of the published record. B2SHARE supports versioning of records: published records have a draft record equivalent which can be freely changed again but is closely
• #17: There are two ways of adding metadata to a record using the API: upon creation of a draft record and by providing so-called JSON patches in a request. This can also be done to already published records, but will create a new version of that record with new PIDs and checksums. JSON patch requests modify the current state of the metadata by either changing, adding or removing fields and values. The structure of the data provided in the patch request must strictly follow the metadata schema of the community. You can make as many patch requests as necessary before publishing your draft record or new published version.
• #18: A JSON patch enables a user to modify the metadata of an existing draft record. It is a set of operations that alter an existing set of metadata fields based on another set of fields. It loosely equals the difference between two sets of metadata, i.e. the old and new contents of metadata of a given draft record. The operations can be add, remove, replace, copy, move and test. All these operations are valid in B2SHARE, but might not always be applicable or useful. In Python, the jsonpatch package can be loaded which automates the generation of patches that can be sent to the B2SHARE instance. The PATCH HTTP method is used to instruct the instance to apply the patch provided in the request data.
• #19: Files can be added during the draft phase of your new record using a PUT request. Files are uploaded into the file bucket of the draft record, not the draft record itself. Therefore use the file bucket ID found in the metadata of the draft record. All added files have their checksum calculated. Files are uploaded one-by-one in separate requests. You can upload as many files as necessary before publishing your draft record or published version.
• #20: Draft records are published by altering the value of the publication state in the metadata. Once your record is published, the included files can not be changed anymore and no new files can be added! Metadata can be changed after publication, but will create a new version of your published record. After your publication request, the persistent identifiers are automatically added to the record
• #21: We will now look at some examples of using the API. For many requests you need to authenticate yourself. As APIs are often used directly in applications, you don’t want to provide usernames and passwords for every request you make. Instead an access token is generated on the website which uniquely identifies yourself during a request. B2SHARE contains public or open access and restricted data. All open access published records and metadata are public and can be accessed by anyone, this mostly holds for the contained files as well. Therefore no access tokens are required. For privately shared records and your draft records and files an access token is required. Depending on the authorization and community settings, you may or may not be able to access these records. As the access token unique identifies yourself and allows alteration of your published and draft records, do not share it with anyone!
• #22: To generate an access token, go to the B2SHARE website, log in and navigate to the profile page. You can only create an access token using the B2SHARE website. Create a new token by entering a name, followed by a click on the new token button. Note that you will not be able to see this token again afterwards, so store it safely somewhere else. If you lose your token you can create a new one on this page. You can create as many as you like.
• #23: Here are some example GET requests using the B2SHARE training instance and the curl application. No access token is necessary yet, as all this information is publically available. Note that all command are a single line only, here they are displayed on multiple lines for clarity. Examples: To retrieve all existing records and list all communities examples do not use a variable in the address To search for specific records of a community, use the community ID in the query parameter q. Community IDs can be found in the listing of all communities
• #24: Examples: To get a specific record, use the record ID of that record in the address To get all your draft records you now need to add your access token as a parameter. You also need to set parameter ‘drafts’ to value 1 To list all the files contained by a record, get the file bucket ID from the record metadata and use it in the address directly in the path
• #25: The full publication workflow using the B2SHARE REST API is as follows: Identify a target community to place your new record under. You need the community ID of that community and depending on the community, only members of that community can publish under their name. Get the metadata schema definition of that community using the community ID. Now you know the required fields for your publication Create a draft record and upload your files and add metadata. This can all be done in multiple separate steps and for files one-by-one. When adding files, checksums will be generated for each file. Finally publish your record. Depending on the community’s settings, your record may need approval of the community before it is shown in B2SHARE. A persistent identifier is added to the record so that it can be uniquely identified.
• #26: For clarity, here there is a complete overview of the full publication workflow using the B2SHARE REST API in a diagram. For every step (blue boxes) the corresponding request and HTTP method has been added in which the variables need to be filled in. The only exception is the ‘community approve’ step which cannot be done through the API yet. A record can be in three states: draft, submitted or published (red boxes). Draft and submitted records are not visible openly, but only to the user or community. Checksums are calculated for every file added and a PID is added once the draft record is committed (green boxes).
• #27: Have a look at our website for more information regarding B2SHARE User documentation is also available here B2SHARE hands-on training can be found on GitHub. Currently only API access using Python is covered. In the future more modules will be added.
• #28: Thank you!