Unofficial Draft 11 May 2022
Latest editor’s draft: https://ocfl.io/draft/spec/
Editors:
Additional Documents:
Previous version:
Repository:
This document is licensed under a Creative Commons Attribution 4.0 License. OCFL logo: “hand-drive” by Patrick Hochstenbach is licensed under CC BY 2.0.
This section is non-normative.
This Oxford Common File Layout (OCFL) specification describes an application-independent approach to the storage of digital objects in a structured, transparent, and predictable manner. It is designed to promote long-term access and management of digital objects within digital repositories.
The OCFL initiative began as a discussion amongst digital repository practitioners to identify well-defined, common, and application-independent file management for a digital repository’s persisted objects and represents a specification of the community’s collective recommendations addressing five primary requirements: completeness, parsability, versioning, robustness, and storage diversity.
The OCFL recommends storing metadata and the content it describes together so the OCFL object can be fully understood in the absence of original software. The OCFL does not make recommendations about what constitutes an object, nor does it assume what type of metadata is needed to fully understand the object, recognizing those decisions may differ from one repository to another. However, it is recommended that when making this decision, implementers consider what is necessary to rebuild the objects from the files stored.
One goal of the OCFL is to ensure objects remain fixed over time. This can be difficult as software and infrastructure change, and content is migrated. To combat this challenge, the OCFL ensures that both humans and machines can understand the layout and corresponding inventory regardless of the software or infrastructure used. This allows for humans to read the layout and corresponding inventory, and understand it without the use of machines. Additionally, if existing software were to become obsolete, the OCFL could easily be understood by a light weight application, even without the full feature repository that might have been used in the past.
Another need expressed by the community was the need to update and change objects, either the content itself or the metadata associated with the object. The OCFL relies heavily on the prior art in the [Moab] Design for Digital Object Versioning which utilizes forward deltas to track the history of the object. Utilizing this schema allows implementers of the OCFL to easily recreate past versions of an OCFL object. Like with objects, the OCFL remains silent on when versioning should occur recognizing this may differ from implementation to implementation.
The OCFL also fills the need for robustness against errors, corruption, and migration. The versioning schema ensures an OCFL object is robust enough to allow for the discovery of human errors. The fixity checking built into the OCFL via content addressable storage allows implementers to identify file corruption that might happen outside of normal human interactions. The OCFL eases content migrations by providing a technology agnostic method for verifying OCFL objects have remained fixed.
Finally, the community expressed a need to store content on a wide variety of storage technologies. With that in mind, the OCFL was written with an eye toward various storage infrastructures including cloud object stores.
This normative specification describes the nature of an OCFL Object (the “object-at-rest”) and the arrangement of OCFL Objects under an OCFL Storage Root. A set of recommendations for how OCFL Objects should be acted upon (the “object-in-motion”) can be found in the [OCFL-Implementation-Notes]. The OCFL editorial group recommends reading both the specification and the implementation notes in order to understand the full scope of the OCFL.
This specification is designed to operate on storage systems that employ a hierarchical metaphor for presenting data to users. On traditional disk-based storage this may take the form of files and directories, and this is the terminology we use in this specification since it is widely known. However, it may equally apply to object stores, where namespaces, containers, and objects present a similar organization hierarchy to users.
As well as sections marked as non-normative, all authoring guidelines, diagrams, examples, and notes in this specification are non-normative. Everything else in this specification is normative.
The key words MAY, MUST, MUST NOT, SHOULD, and SHOULD NOT are to be interpreted as described in [RFC2119].
Content Path: The file path of a file on disk or in an object store, relative to the OCFL Object Root. Content paths are used in the Manifest within an Inventory.
Digest: An algorithmic characterization of the contents of a file conforming to a standard digest algorithm.
Extension: Extensions are used to collaborate, review, and publish additional non-normative functions related to OCFL. Extensions are intended to be informational and cite-able, but outside the scope of the normal specification process. Registered extensions may be found in the OCFL Extensions repository.
Inventory: A file, expressed in JSON, that tracks the history and current state of an OCFL Object.
Logical Path: A path that represents a file’s location in the logical state of an object. Logical paths are used in conjunction with a digest to represent the file name and path for a given bitstream at a given version.
Logical State: A grouping of logical paths tied to their corresponding bitstreams that reflect the state of the object content for a given version.
Logs Directory: A directory for storing information about the content (e.g., actions performed) that is not part of the content itself.
Manifest: A section of the Inventory listing all files and their digests within an OCFL Object.
OCFL Object: A group of one or more content files and administrative information, that together have a unique identifier. The object may contain a sequence of versions of the files that represent the evolution of the object’s contents.
OCFL Object Root: The base directory of an OCFL Object, identified by a [NAMASTE] file “0=ocfl_object_1.1”.
OCFL Storage Root: A base directory used to store OCFL Objects, identified by a [NAMASTE] file “0=ocfl_1.1”.
OCFL Version: The state of an OCFL Object’s content which is constructed using the incremental changes recorded in the sequence of corresponding and prior version directories.
Registered Extension Name: The registered name of an extension is the name provided in the Extension Name property of the extension’s definition in the OCFL Extensions repository.
An OCFL Object is a group of one or more content files and administrative information, that are together identified by a URI. The object may contain a sequence of versions of the files that represent the evolution of the object’s contents.
A file is defined as a content bitstream that can be stored and transmitted. Directories (also called “folders”) allow for the organization of files into tree-like hierarchies. The content of an OCFL Object is the files and the directories they are organized in that are stored within the hierarchy layout described in this specification.
An OCFL Object includes administrative information that identifies a directory as an OCFL Object, and also provides a means of tracking changes to the contents of the object over time.
An OCFL Object is therefore:
A conceptual gathering of all files (data and metadata), the directories they are organized in, and their changes over time which together form the digital representation of an entity that need to be managed, in preservation terms, as a single coherent whole (i.e., content); and
A file and directory layout and administrative information on a storage medium that provides a defined structure for the storage of this content, and through which these files and their changes may be understood (i.e., structure).
A key goal of the OCFL is the rebuildability of a repository from an OCFL Storage Root without additional information resources. Consequently, a key implementation consideration should be to ensure that OCFL Objects contain all the data and metadata required to achieve this. With reference to the [OAIS] model, this would include all the descriptive, administrative, structural, representation and preservation metadata relevant to the object.
A central feature of the OCFL specification is support for versioning. This recognizes that digital objects will change over time, through new requirements, fixes, updates, or format shifts. The specification takes no position on what constitutes a version or a versionable action, but it is recommended that implementers have a clear position on this within their local storage policies.
The OCFL Object structure organizes content files and administrative information in order to support content storage and object validation. The structure for an object with one version is shown in the following figure:
[object_root]
├── 0=ocfl_object_1.1
├── inventory.json
├── inventory.json.sha512
└── v1
├── inventory.json
├── inventory.json.sha512
└── content
└── ... content files ...
The OCFL Object Root MUST NOT contain files or directories other than those specified in the following sections.
The OCFL specification version declaration MUST be formatted according to the
[NAMASTE] specification. There MUST be exactly one version
declaration file in the base directory of the OCFL Object Root giving the OCFL version in the
filename. The filename MUST conform to the pattern T=dvalue
, where T
MUST be 0, and dvalue
MUST be ocfl_object_
,
followed by the OCFL specification version number. The text contents of the file MUST be the same as dvalue
, followed by a newline (\n
).
OCFL Object content MUST be stored as a sequence of one or more versions. Each
object version is stored in a version directory under the object root. Version directory names MUST be constructed by prepending v
to the version number. The version number MUST be taken from the sequence of positive, base-ten integers: 1, 2, 3, etc.. The version number
sequence MUST start at 1 and MUST be
continuous without missing integers.
Implementations SHOULD use version directory names constructed without
zero-padding the version number, ie. v1
, v2
, v3
, etc..
For compatibility with existing filesystem conventions, implementations MAY use zero-padded
version directory numbers, with the following restriction: If zero-padded version directory numbers are used then they
MUST start with the prefix v
and then a zero. For example, in an implementation
that uses five digits for version directory names then v00001
to v09999
are allowed, v10000
is not allowed.
The first version of an object defines the naming convention for all version directories for the object. All version directories of an object MUST use the same naming convention: either a non-padded version directory number, or a zero-padded version directory number of consistent length. The version naming convention MUST be consistent across all versions. In all cases, references to files inside version directories from inventory files MUST use the actual version directory names.
There MUST be no other files as children of a version directory, other than an inventory file and a inventory digest. The version directory SHOULD NOT contain any directories other than the designated content sub-directory. Once created, the contents of a version directory are expected to be immutable.
Version directories MUST contain a designated content sub-directory if the
version contains files to be preserved, and SHOULD NOT contain this sub-directory
otherwise. The name of this designated sub-directory MAY be defined in the inventory
file using the key contentDirectory
with the value being the chosen sub-directory name as a string,
relative to the version directory. The contentDirectory
value MUST represent a
direct child directory of the version directory in which it is found. As such, the contentDirectory
value MUST NOT contain the forward slash (/
) path separator and MUST NOT be either one or two periods (.
or ..
). If the key contentDirectory
is set, it
MUST be set in the first version of the object and MUST NOT change between versions of the same object.
If the key contentDirectory
is not present in the inventory file then the name of the designated content
sub-directory MUST be content
. OCFL-compliant tools (including any validators)
MUST ignore all directories in the object version directory except for the
designated content directory.
Every file within a version’s content directory MUST be referenced in the
manifest section of that version’s inventory. There MUST NOT be
empty directories within a version’s content directory. A directory that would otherwise be empty MAY be maintained by creating a file within it named according to local conventions, for example
by making an empty .keep
file.
A digest plays two roles in an OCFL Object. The first is that digests allow for content-addressable reference to files within the OCFL Object. That is, the connection between a file’s content path on physical storage and its logical path in a version of the object’s content is made with a digest of its contents, rather than its filename. This use of the content digest facilitates de-duplication of files with the same content within an object, such as files that are unchanged from one version to the next. The second role that digests play is provide for fixity checks to determine whether a file has become corrupt, through hardware degradation or accident for example.
For content-addressing, OCFL Objects MUST use either sha512
or sha256
, and
SHOULD use sha512
. The choice of the sha512
digest algorithm as default
recognizes that it has no known collision vulnerabilities and multiple implementations are available.
For storage of additional fixity values, or to support legacy content migration, implementers MUST choose from the following controlled vocabulary of digest algorithms, or from a list of additional algorithms given in the [Digest-Algorithms-Extension]. OCFL clients MUST support all fixity algorithms given in the table below, and MAY support additional algorithms from the extensions. Optional fixity algorithms that are not supported by a client MUST be ignored by that client.
Digest Algorithm Name | Note |
---|---|
md5 |
Insecure. Use only for legacy fixity values. MD5 algorithm and hex encoding defined by [RFC1321]. For example, the md5 digest of a zero-length bitstream is d41d8cd98f00b204e9800998ecf8427e . |
sha1 |
Insecure. Use only for legacy fixity values. SHA-1 algorithm defined by [FIPS-180-4] and MUST be encoded using hex (base16) encoding [RFC4648]. For example, the sha1 digest of a zero-length bitstream is da39a3ee5e6b4b0d3255bfef95601890afd80709 . |
sha256 |
Non-truncated form only; note performance implications. SHA-256 algorithm defined by [FIPS-180-4] and MUST be encoded using hex (base16) encoding [RFC4648]. For example, the sha256 digest of a zero-length bitstream starts e3b0c44298fc1c149afbf4c8996fb92427ae41e4... (64 hex digits long). |
sha512 |
Default choice. Non-truncated form only. SHA-512 algorithm defined by [FIPS-180-4] and MUST be encoded using hex (base16) encoding [RFC4648]. For example, the sha512 digest of a zero-length bitstream starts cf83e1357eefb8bdf1542850d66d8007d620e405... (128 hex digits long). |
blake2b-512 |
Full-length form only, using the 2B variant (64 bit) as defined by [RFC7693]. MUST be encoded using hex (base16) encoding [RFC4648]. For example, the blake2b-512 digest of a zero-length bitstream starts 786a02f742015903c6c6fd852552d272912f4740... (128 hex digits long). |
An OCFL Inventory MAY contain a fixity section that can store one or more blocks containing fixity values using multiple digest algorithms. See the section on fixity below for further details.
Non-normative note: Implementers may also store copies of their file digests in a system external to their OCFL Object stores at the point of ingest, to further safeguard against the possibility of malicious manipulation of file contents and digests.
Implementers should be aware that base16 digests are case insensitive. Different tools will generate digests in uppercase or lowercase, and this may lead to case differences between references to a digest and the digest itself within the inventory. If string-based methods are used to work with digests and inventories (as is the case in most common JSON libraries) then extra care must be taken to ensure case-insensitive comparisons are being made.
An OCFL Object Inventory MUST follow the JSON (defined by
[RFC8259]) structure described in this section with contents encoded in UTF-8, and MUST be named inventory.json
. The order of entries in both the JSON objects and arrays used in
inventory files has no significance. An OCFL Object Inventory MUST NOT contain
any keys not described in this specification.
The forward slash (/) path separator MUST be used in content paths in the manifest and fixity blocks within the inventory. Implementations that target systems using other separators will need to translate paths appropriately.
Non-normative note: A [JSON-Schema] for validating OCFL Object Inventory files is provided at inventory_schema.json.
Every OCFL inventory MUST include the following keys:
id
: A unique identifier for the OCFL Object. This MUST be unique in the local
context, MUST NOT change between versions of the same object, and SHOULD be a URI [RFC3986]. There is no expectation that a URI used is
resolvable. For example, URNs [RFC8141] MAY be used.
type
: A type for the inventory JSON object that also serves to document the OCFL specification version that the
inventory complies with. In the object root inventory this MUST be the URI of the
inventory section of the specification version matching the object conformance declaration. For the current
specification version the value is https://ocfl.io/1.1/spec/#inventory
.
digestAlgorithm
: The digest algorithm used for calculating digests for content-addressing within the OCFL Object and
for the Inventory Digest. This MUST be the algorithm used in
the manifest
and state
blocks, see the section on Digests for more information about algorithms.
head
: The version directory name of the most recent version of the object. This MUST be the version directory name with the highest version number.
There MAY be the following key:
contentDirectory
: The name of the designated content directory within the version directories. If not specified then
the content directory name is content
.In addition to these keys, there MUST be two other blocks present, manifest
and
versions
, which are discussed in the next two sections.
The value of the manifest
key MUST be a JSON object, and each key MUST correspond to a digest value key found in one or more state
blocks of the
current and/or previous version
blocks of the OCFL Object. The value for each key MUST be an array containing the content paths of files in the OCFL Object
that have content with the given digest. As JSON keys are case sensitive, for digest algorithms with case insensitive
digest values, there is an additional requirement that each digest value MUST
occur only once in the manifest block for any digest algorithm, regardless of case. Content paths within a manifest
block MUST be relative to the OCFL Object Root. The
following restrictions avoid ambiguity and provide path safety for clients processing the manifest
.
The content path MUST be interpreted as a set of one or more path elements
joined by a /
path separator.
Path elements MUST NOT be .
, ..
, or empty (//
).
A content path MUST NOT begin or end with a forward slash (/
).
Within an inventory, content paths MUST be unique and non-conflicting, so the content path for a file cannot appear as the initial part of another content path.
Non-normative note: If only one file is stored in the OCFL Object for each digest, fully de-duplicating the content, then there will be only one content path for each digest. There may, however, be multiple logical paths for a given digest if the content was not entirely de-duplicated when constructing the OCFL Object.
An example manifest object for three content paths, all in version 1, is shown below:
"manifest": { "7dcc35...c31": [ "v1/content/foo/bar.xml" ], "cf83e1...a3e": [ "v1/content/empty.txt" ], "ffccf6...62e": [ "v1/content/image.tiff" ] }
An OCFL Object Inventory MUST include a block for storing versions. This block
MUST have the key of versions
within the inventory, and it MUST be a JSON object. The keys of this object MUST
correspond to the names of the version directories used. Each value MUST be another JSON object that characterizes the version, as described in the 3.5.3.1
Version section.
A JSON object to describe one OCFL Version, which MUST include the following keys:
created
: The value of this key is the datetime of creation of this version. It MUST be expressed in the Internet Date/Time Format defined by [RFC3339]. This
format requires the inclusion of a timezone value or Z
for UTC, and that the time component be granular to the second
level (with optional fractional seconds).
state
: The value of this key is a JSON object, containing a list of keys and values corresponding to the logical
state of the object at that version. The keys of this JSON object are digest values, each of which
MUST exactly match a digest value key in the manifest of the
inventory. The value for each key is an array containing logical path names of files in
the OCFL Object’s logical state that have content with the given digest.
Logical paths present the structure of an OCFL Object at a given version. This is given as an array of values, with the following restrictions to provide for path safety in the common case of the logical path value representing a file path.
The logical path MUST be interpreted as a set of one or more path elements
joined by a /
path separator.
Path elements MUST NOT be .
, ..
, or empty (//
).
A logical path MUST NOT begin or end with a forward slash (/
).
Within a version, logical paths MUST be unique and non-conflicting, so the logical path for a file cannot appear as the initial part of another logical path.
Non-normative note: The logical state of the object uses content-addressing to map logical paths to their bitstreams, as expressed in the manifest section of the inventory. Notably, the version state provides de-duplication of content within the OCFL Object by mapping multiple logical paths with the same content to the same digest in the manifest. See [OCFL-Implementation-Notes].
An example
state
block is shown below:"state": { "4d27c8...b53": [ "foo/bar.xml" ], "cf83e1...a3e": [ "empty.txt", "empty2.txt" ] }
This
state
block describes an object with 3 files, two of which have the same content (empty.txt
andempty2.txt
), and one of which is in a sub-directory (bar.xml
). The logical state shown as a tree is thus:├── empty.txt ├── empty2.txt └── foo └── bar.xml
The JSON object describing an OCFL Version, SHOULD include the following keys:
message
: The value of this key is freeform text, used to record the rationale for creating this version. It MUST be a JSON string.
user
: The value of this key is a JSON object intended to identify the user or agent that created the current OCFL
Version. The value of the user
key MUST contain a user name
key, name
and SHOULD contain an address key, address
. The name
value is any
readable name of the user, e.g., a proper name, user ID, agent ID. The address
value SHOULD be a URI: either a mailto URI [RFC6068] with the e-mail address of the
user or a URL to a personal identifier, e.g., an ORCID iD.
An OCFL Object inventory MAY include a block for storing additional fixity information to
supplement the complete set of digests in the Manifest, for example to support legacy digests from a
content migration. If present, this block MUST have the key of fixity
within
the inventory, and its value MUST be a JSON object, which MAY be empty.
The keys within the fixity
block MUST correspond to the controlled vocabulary
of digest algorithm names listed in the Digests section, or in a table given in an
Extension. The value of the fixity block for a particular digest algorithm MUST follow the structure of the 3.5.2 Manifest block; that is, a key corresponding
to the digest value, and an array of content paths. The fixity
block for any digest algorithm
MAY include digest values for any subset of content paths in the object. Where included,
the digest values given MUST match the digests of the files at the corresponding
content paths. As JSON keys are case sensitive, for digest algorithms with case insensitive digest values, there is an
additional requirement that each digest value MUST occur only once in the
fixity
block for any digest algorithm, regardless of case. There is no requirement that all content files have a value
in the fixity
block, or that fixity values provided in one version are carried forward to later versions.
An example
fixity
withmd5
andsha1
digests is shown below. In this case themd5
digest values are provided only for version 1 content paths."fixity": { "md5": { "184f84e28cbe75e050e9c25ea7f2e939": [ "v1/content/foo/bar.xml" ], "c289c8ccd4bab6e385f5afdd89b5bda2": [ "v1/content/image.tiff" ], "d41d8cd98f00b204e9800998ecf8427e": [ "v1/content/empty.txt" ] }, "sha1": { "66709b068a2faead97113559db78ccd44712cbf2": [ "v1/content/foo/bar.xml" ], "a6357c99ecc5752931e133227581e914968f3b9c": [ "v2/content/foo/bar.xml" ], "b9c7ccc6154974288132b63c15db8d2750716b49": [ "v1/content/image.tiff" ], "da39a3ee5e6b4b0d3255bfef95601890afd80709": [ "v1/content/empty.txt" ] } }
Every occurrence of an inventory file MUST have an accompanying sidecar file
named inventory.json.ALGORITHM
stating its digest, where ALGORITHM
is the chosen digest algorithm for the object.
The ALGORITHM MUST match the value given for the digestAlgorithm
key in the
inventory. An example might be inventory.json.sha512
.
The digest sidecar file MUST contain the digest of the inventory file. This MUST follow the format:
DIGEST inventory.json
One or more whitespace characters (spaces or tabs) must separate DIGEST from the string inventory.json
; that is, the
name of the inventory file in the same directory.
The digest of the inventory MUST be computed only after all changes to the inventory have been made, and thus writing the digest sidecar file is the last step in the versioning process.
Every OCFL Object MUST have an inventory file within the OCFL Object Root,
corresponding to the state of the OCFL Object at the current version. Additionally, every version directory SHOULD include an inventory file that is an Inventory of all content for
versions up to and including that particular version. Where an OCFL Object contains inventory.json
in version
directories, the inventory file in the OCFL Object Root MUST be the same as the
file in the most recent version. See also requirements for the corresponding Inventory Digest.
In the case that prior version directories include an inventory file there will be multiple inventory files describing
prior versions within the OCFL Object. Each version
block in each prior inventory file MUST represent the same logical state as the corresponding version
block
in the current inventory file. Additionally, the values of the created
, message
and user
keys in each version
block in each prior inventory file SHOULD have the same values as the
corresponding keys in the corresponding version
block in the current inventory file.
Non-normative note: Storing an inventory for every version provides redundancy for this critical information in a way that is compatible with storage strategies that have immutable version directories.
Version directories in OCFL are intended to be immutable in that existing version directories do not change when a new
version directory is added. Each version directory within an OCFL Object MUST
conform to either the same or a later OCFL specification version as the preceding version directory. If inventories are
stored in the version directories then the OCFL specification version for a given version directory is apparent from the
type
attribute in that inventory.
The base directory of an OCFL Object MAY contain a directory named logs
, which MAY be empty. Implementers SHOULD use the logs
directory for storing files that contain a record of actions taken on the object. Since these logs
may be subject to local standards requirements, the format of these logs is considered out-of-scope for the OCFL Object.
Clients operating on the object MAY log actions here that are not otherwise captured.
Non-normative note: The purpose of the logs directory is to provide implementers with a location for storing local information about actions to the OCFL Object’s content that is not part of the content itself.
As an example, implementers may have different local requirements to store audit information for their content. Some may wish to store a log entry indicating that an audit was conducted, and nothing was wrong, while others may wish to only store a log entry if an intervention was required.
The base directory of an OCFL Object MAY contain a directory named extensions
for the
purposes of extending the functionality of an OCFL Object. The extensions
directory MUST NOT contain any files, and no sub-directories other than extension sub-directories.
Extension sub-directories SHOULD be named according to a registered extension
name in the OCFL Extensions repository.
Non-normative note: Extension sub-directories should use the same name as a registered extension in order to both avoid the possiblity of an extension sub-directory colliding with the name of another registered extension as well as to facilitate the recognition of extensions by OCFL clients. See also Documenting Local Extensions.
An OCFL Storage Root is the base directory of an OCFL storage layout.
An OCFL Storage Root MUST contain a Root Conformance Declaration identifying it as such.
An OCFL Storage Root MAY contain other files as direct children. These might include a human-readable copy of the OCFL specification to make the storage root self-documenting, or files used to document local extensions. An OCFL validator MUST ignore any files in the storage root it does not understand.
An OCFL Storage Root MUST NOT contain directories or sub-directories other than as a directory hierarchy used to store OCFL Objects or for storage root extensions. The directory hierarchy used to store OCFL Objects MUST NOT contain files that are not part of an OCFL Object. Empty directories MUST NOT appear under a storage root.
An OCFL Storage Root MAY contain a file named ocfl_layout.json
to describe the
arrangement of directories and OCFL objects under the storage root. If present, ocfl_layout.json
MUST be a JSON (defined by [RFC8259]) document encoded in UTF-8 and include the
following two keys in the root JSON object:
extension
- An extension name that identifies an arrangement of directories and OCFL objects under the storage root,
i.e. how OCFL object identifiers are mapped to directory hierarchies. The value of the extension
key MUST be the registered extension name for the extension
defining the arrangement under the storage root.
description
- A human readable description of the arrangement of directories and OCFL objects under the storage
root.
Although implementations may require multiple OCFL Storage Roots—that is, several logical or physical volumes, or multiple “buckets” in an object store—each OCFL Storage Root MUST be independent.
The following example OCFL Storage Root represents the minimal set of files and folders:
[storage_root]
├── 0=ocfl_1.1
├── ocfl_1.1.txt (human-readable text of the OCFL specification; optional)
└── ocfl_layout.json (description of storage hierarchy layout; optional)
The OCFL version declaration MUST be formatted according to the
[NAMASTE] specification. There MUST be exactly one version
declaration file in the base directory of the OCFL Storage Root giving the OCFL version in the
filename. The filename MUST conform to the pattern T=dvalue
, where T
MUST be 0, and dvalue
MUST be ocfl_
,
followed by the OCFL specification version number. The text contents of the file MUST be the same as dvalue
, followed by a newline (\n
).
Root conformance indicates that the OCFL Storage Root conforms to this section (i.e. the OCFL Storage Root section) of the specification. OCFL Objects within the OCFL Storage Root also include a conformance declaration which MUST indicate OCFL Object conformance to the same or earlier version of the specification.
OCFL Object Roots MUST be stored either as the terminal resource at the end of a directory storage hierarchy or as direct children of a containing OCFL Storage Root.
A common practice is to use a unique identifier scheme to compose this storage hierarchy, typically arranged according to some form of the [PairTree] specification. Irrespective of the pattern chosen for the storage hierarchies, the following restrictions apply:
There MUST be a deterministic mapping from an object identifier to a unique storage path
Storage hierarchies MUST NOT include files within intermediate directories
Storage hierarchies MUST be terminated by OCFL Object Roots
Storage hierarchies within the same OCFL Storage Root SHOULD use just one layout pattern
Storage hierarchies within the same OCFL Storage Root SHOULD consistently use either a directory hierarchy of OCFL Objects or top-level OCFL Objects
The behavior of the storage root may be extended to support features from other specifications.
The base directory of an OCFL Storage Root MAY contain a directory named extensions
for
the purposes of extending the functionality of an OCFL Storage Root. The storage root extensions
directory MUST conform to the same guidelines and limitations as those defined for object
extensions.
Non-normative note: Storage extensions can be used to support additional features, such as providing the storage hierarchy disposition when pairtree is in use, or additional human-readable text about the nature of the storage root.
It is preferable that both Object Extensions and Storage Root Extenstions are documented and registered in the OCFL Extensions repository. However, local extensions MAY be documented by including a plain text document directly in the storage root, thus making the storage root self-documenting.
In order to maximize the compatibility of the OCFL with different filesystems, and thus improve the portability of OCFL Objects between different systems, some restrictions on the use of certain filesystem features are necessary. If the preservation of non-OCFL-compliant features is required then the content MUST be wrapped in a suitable disk or filesystem image format which OCFL can treat as a regular file.
Filesystem metadata (e.g. permissions, access, and creation times) are not considered portable between filesystems or preservable through file transfer operations. These attributes also cannot be validated in terms of fixity in a consistent manner. As such, the OCFL does not support the portability of these attributes.
Hard and soft (symbolic) links are not portable and MUST NOT be used within OCFL Storage hierachies. A common use case for links is storage deduplication. OCFL inventories provide a portable method of achieving the same effect by using digests to address content.
File paths and filenames in the OCFL are case sensitive. Filesystems MUST preserve the case of OCFL filepaths and filenames.
Transparent filesystem features such as compression and encryption should be effectively invisible to OCFL operations. Consequently, they should not be expected to be portable.
This section is non-normative.
The following example OCFL Object has content that is a single file (file.txt
), and just one version (v1
):
[object root]
├── 0=ocfl_object_1.1
├── inventory.json
├── inventory.json.sha512
└── v1
├── inventory.json
├── inventory.json.sha512
└── content
└── file.txt
The inventory for this OCFL Object, the same both at the top-level and in the v1
directory, might be:
{
"digestAlgorithm": "sha512",
"head": "v1",
"id": "http://example.org/minimal",
"manifest": {
"7545b8...f67": [ "v1/content/file.txt" ]
},
"type": "https://ocfl.io/1.1/spec/#inventory",
"versions": {
"v1": {
"created": "2018-10-02T12:00:00Z",
"message": "One file",
"state": {
"7545b8...f67": [ "file.txt" ]
},
"user": {
"address": "mailto:alice@example.org",
"name": "Alice"
}
}
}
}
The following example OCFL Object has three versions:
[object root]
├── 0=ocfl_object_1.1
├── inventory.json
├── inventory.json.sha512
├── v1
│ ├── inventory.json
│ ├── inventory.json.sha512
│ └── content
│ ├── empty.txt
│ ├── foo
│ │ └── bar.xml
│ └── image.tiff
├── v2
│ ├── inventory.json
│ ├── inventory.json.sha512
│ └── content
│ └── foo
│ └── bar.xml
└── v3
├── inventory.json
└── inventory.json.sha512
In v1
there are three files, empty.txt
, foo/bar.xml
, and image.tiff
. In v2
the content of foo/bar.xml
is
changed, empty2.txt
is added with the same content as empty.txt
, and image.tiff
is removed. In v3
the file
empty.txt
is removed, and image.tiff
is reinstated. As a result of forward-delta versioning, the object tree above
shows only new content added in each version. The inventory shown below details the other changes, includes additional
fixity information using md5
and sha1
digest algorithms, and minimal metadata for each version.
{
"digestAlgorithm": "sha512",
"fixity": {
"md5": {
"184f84e28cbe75e050e9c25ea7f2e939": [ "v1/content/foo/bar.xml" ],
"2673a7b11a70bc7ff960ad8127b4adeb": [ "v2/content/foo/bar.xml" ],
"c289c8ccd4bab6e385f5afdd89b5bda2": [ "v1/content/image.tiff" ],
"d41d8cd98f00b204e9800998ecf8427e": [ "v1/content/empty.txt" ]
},
"sha1": {
"66709b068a2faead97113559db78ccd44712cbf2": [ "v1/content/foo/bar.xml" ],
"a6357c99ecc5752931e133227581e914968f3b9c": [ "v2/content/foo/bar.xml" ],
"b9c7ccc6154974288132b63c15db8d2750716b49": [ "v1/content/image.tiff" ],
"da39a3ee5e6b4b0d3255bfef95601890afd80709": [ "v1/content/empty.txt" ]
}
},
"head": "v3",
"id": "ark:/12345/bcd987",
"manifest": {
"4d27c8...b53": [ "v2/content/foo/bar.xml" ],
"7dcc35...c31": [ "v1/content/foo/bar.xml" ],
"cf83e1...a3e": [ "v1/content/empty.txt" ],
"ffccf6...62e": [ "v1/content/image.tiff" ]
},
"type": "https://ocfl.io/1.1/spec/#inventory",
"versions": {
"v1": {
"created": "2018-01-01T01:01:01Z",
"message": "Initial import",
"state": {
"7dcc35...c31": [ "foo/bar.xml" ],
"cf83e1...a3e": [ "empty.txt" ],
"ffccf6...62e": [ "image.tiff" ]
},
"user": {
"address": "mailto:alice@example.com",
"name": "Alice"
}
},
"v2": {
"created": "2018-02-02T02:02:02Z",
"message": "Fix bar.xml, remove image.tiff, add empty2.txt",
"state": {
"4d27c8...b53": [ "foo/bar.xml" ],
"cf83e1...a3e": [ "empty.txt", "empty2.txt" ]
},
"user": {
"address": "mailto:bob@example.com",
"name": "Bob"
}
},
"v3": {
"created": "2018-03-03T03:03:03Z",
"message": "Reinstate image.tiff, delete empty.txt",
"state": {
"4d27c8...b53": [ "foo/bar.xml" ],
"cf83e1...a3e": [ "empty2.txt" ],
"ffccf6...62e": [ "image.tiff" ]
},
"user": {
"address": "mailto:cecilia@example.com",
"name": "Cecilia"
}
}
}
}
The following example OCFL Object inventory shows how content paths may differ from logical paths. The example object
has just one version, v1
, which has two files with logical paths a file.wxy
and another file.xyz
as shown in the
state
block. The corresponding content paths are v1/content/3bacb119a98a15c5
and v1/content/9f2bab8ef869947d
respectively, as shown in the manifest
. Except for location within the appropriate version directory, v1/content
in
this example, the OCFL specification does not constrain the choice of content paths used when creating or updating an
OCFL object. The choice might depend on particular limitations of, or optimizations for, the target storage system, or
on portability considerations. Any compliant implementation will be able to recover version state with the original
logical paths.
{
"digestAlgorithm": "sha512",
"head": "v1",
"id": "http://example.org/diff-paths",
"manifest": {
"7545b8...f67": [ "v1/content/3bacb119a98a15c5" ],
"af318d...3cd": [ "v1/content/9f2bab8ef869947d" ]
},
"type": "https://ocfl.io/1.1/spec/#inventory",
"versions": {
"v1": {
"created": "2019-03-14T20:31:00Z",
"state": {
"7545b8...f67": [ "a file.wxy" ],
"af318d...3cd": [ "another file.xyz" ]
},
"user": {
"address": "mailto:admin@example.org",
"name": "Some Admin"
}
}
}
}
[BagIt] is a common file packaging specification, but unlike the OCFL it does not provide a mechanism for content versioning. Using the OCFL it is possible to store a BagIt structure with content versioning, such that when the logical state is resolved, it creates a valid BagIt ‘bag’. This example will illustrate one way this can be accomplished, using the example of a basic bag given in the BagIt specification.
[object root]
├── 0=ocfl_object_1.1
├── inventory.json
├── inventory.json.sha512
└── v1
├── inventory.json
├── inventory.json.sha512
└── content
└── myfirstbag
├── bagit.txt
├── data
│ └── 27613-h
│ └── images
│ ├── q172.png
│ └── q172.txt
└── manifest-md5.txt
If, for example, a new directory were added in a subsequent version, the OCFL Object would look like this:
[object root]
├── 0=ocfl_object_1.1
├── inventory.json
├── inventory.json.sha512
├── v1
│ ├── inventory.json
│ ├── inventory.json.sha512
│ └── content
│ └── myfirstbag
│ ├── bagit.txt
│ ├── data
│ │ └── 27613-h
│ │ └── images
│ │ ├── q172.png
│ │ └── q172.txt
│ └── manifest-md5.txt
└── v2
├── inventory.json
├── inventory.json.sha512
└── content
└── myfirstbag
├── data
│ └── 27614-h
│ └── images
│ ├── q173.png
│ └── q173.txt
└── manifest-md5.txt
The state of the object at version 2 would be the following BagIt object:
myfirstbag
├── bagit.txt
├── data
│ ├── 27613-h
│ │ └── images
│ │ ├── q172.png
│ │ └── q172.txt
│ └── 27614-h
│ └── images
│ ├── q173.png
│ └── q173.txt
└── manifest-md5.txt
The OCFL Inventory for this object would be as follows:
{
"digestAlgorithm": "sha512",
"head": "v2",
"id": "urn:uri:example.com/myfirstbag",
"manifest": {
"cf83e1...a3e": [ "v1/content/myfirstbag/bagit.txt" ],
"f15428...83f": [ "v1/content/myfirstbag/manifest-md5.txt" ],
"85f2b0...007": [ "v1/content/myfirstbag/data/27613-h/images/q172.png" ],
"d66d80...8bd": [ "v1/content/myfirstbag/data/27613-h/images/q172.txt" ],
"2b0ff8...620": [ "v2/content/myfirstbag/manifest-md5.txt" ],
"921d36...877": [ "v2/content/myfirstbag/data/27614-h/images/q173.png" ],
"b8bdf1...927": [ "v2/content/myfirstbag/data/27614-h/images/q173.txt" ]
},
"type": "https://ocfl.io/1.1/spec/#inventory",
"versions": {
"v1": {
"created": "2018-10-09T11:20:29.209164Z",
"message": "Initial Ingest",
"state": {
"cf83e1...a3e": [ "myfirstbag/bagit.txt" ],
"85f2b0...007": [ "myfirstbag/data/27613-h/images/q172.png" ],
"d66d80...8bd": [ "myfirstbag/data/27613-h/images/q172.txt" ],
"f15428...83f": [ "myfirstbag/manifest-md5.txt" ]
},
"user": {
"address": "mailto:someone@example.org",
"name": "Some One"
}
},
"v2": {
"created": "2018-10-31T11:20:29.209164Z",
"message": "Added new images",
"state": {
"cf83e1...a3e": [ "myfirstbag/bagit.txt" ],
"85f2b0...007": [ "myfirstbag/data/27613-h/images/q172.png" ],
"d66d80...8bd": [ "myfirstbag/data/27613-h/images/q172.txt" ],
"2b0ff8...620": [ "myfirstbag/manifest-md5.txt" ],
"921d36...877": [ "myfirstbag/data/27614-h/images/q173.png" ],
"b8bdf1...927": [ "myfirstbag/data/27614-h/images/q173.txt" ]
},
"user": {
"address": "mailto:somebody-else@example.org",
"name": "Somebody Else"
}
}
}
}
[Moab] is an archive information package format developed and used by Stanford University. Many of the ideas in Moab have been refined by the OCFL, and the OCFL is designed to give institutions currently using Moab an easy path to adoption.
Converting content preserved in a Moab object in a way that does not compromise existing Moab access patterns whilst
allowing for the eventual use of OCFL-native workflows requires a Moab to OCFL conversion tool. This tool uses the
Moab-versioning gem to extract deltas and digests of the Moab data directory for each Moab version and translate those
into version state
blocks in an OCFL inventory file, which would be placed in the root directory of the Moab object.
The content of the data
directory in the Moab version directories (and thus, the bitstreams that Moab is preserving)
is tracked by OCFL, via the contentDirectory
value. The contents of the Moab manifests
directories are not tracked,
as the intention is not to encapsulate a Moab object inside an OCFL object, but rather to migrate Moab’s preserved
bitstreams into an OCFL object without compromising legacy access patterns.
During the transitionary period the OCFL inventory file exists only in the root of the Moab object. Once OCFL-native object creation workflows have been completed, future versions of that object will be fully OCFL compliant - new versions will no longer have a manifests directory and will contain an OCFL inventory file. At this stage OCFL tools will be able to access all versions of the content originally preserved by Moab.
Consider the following sample Moab object:
[object root]
└── bj102hs9687
├── v0001
│ ├── data
│ │ ├── content
│ │ │ ├── eric-smith-dissertation-augmented.pdf
│ │ │ └── eric-smith-dissertation.pdf
│ │ └── metadata
│ │ ├── contentMetadata.xml
│ │ ├── descMetadata.xml
│ │ ├── identityMetadata.xml
│ │ ├── provenanceMetadata.xml
│ │ ├── relationshipMetadata.xml
│ │ ├── rightsMetadata.xml
│ │ ├── technicalMetadata.xml
│ │ └── versionMetadata.xml
│ └── manifests
│ ├── fileInventoryDifference.xml
│ ├── manifestInventory.xml
│ ├── signatureCatalog.xml
│ ├── versionAdditions.xml
│ └── versionInventory.xml
├── v0002
│ ├── data
│ │ └── metadata
│ │ ├── contentMetadata.xml
│ │ ├── embargoMetadata.xml
│ │ ├── events.xml
│ │ ├── identityMetadata.xml
│ │ ├── provenanceMetadata.xml
│ │ ├── relationshipMetadata.xml
│ │ ├── rightsMetadata.xml
│ │ ├── versionMetadata.xml
│ │ └── workflows.xml
│ └── manifests
│ ├── fileInventoryDifference.xml
│ ├── manifestInventory.xml
│ ├── signatureCatalog.xml
│ ├── versionAdditions.xml
│ └── versionInventory.xml
└── v0003
├── data
│ └── metadata
│ ├── contentMetadata.xml
│ ├── descMetadata.xml
│ ├── embargoMetadata.xml
│ ├── events.xml
│ ├── identityMetadata.xml
│ ├── provenanceMetadata.xml
│ ├── rightsMetadata.xml
│ ├── technicalMetadata.xml
│ ├── versionMetadata.xml
│ └── workflows.xml
└── manifests
├── fileInventoryDifference.xml
├── manifestInventory.xml
├── signatureCatalog.xml
├── versionAdditions.xml
└── versionInventory.xml
An OCFL inventory that tracks the data
directory would include a manifest comprised as follows. Note the absence of
the manifests
directory, as we are not encapsulating the Moab object in an OCFL object, and the presence of
contentDirectory
to specify data
as the preserved content directory:
{
"digestAlgorithm": "sha512",
"head": "v3",
"id": "druid:bj102hs9687",
"contentDirectory": "data",
"manifest": {
"98114a...588": [ "v0001/data/content/eric-smith-dissertation-augmented.pdf" ],
"7f3d87...15b": [ "v0001/data/content/eric-smith-dissertation.pdf" ],
"6d19f0...064": [ "v0001/data/metadata/technicalMetadata.xml" ],
"6e4be4...375": [ "v0001/data/metadata/provenanceMetadata.xml" ],
"d8a319...d0f": [ "v0001/data/metadata/descMetadata.xml" ],
"de823a...acc": [ "v0001/data/metadata/rightsMetadata.xml" ],
"080617...40c": [ "v0001/data/metadata/identityMetadata.xml" ],
"e15267...58d": [ "v0001/data/metadata/versionMetadata.xml" ],
"0d9e0b...9a2": [ "v0001/data/metadata/contentMetadata.xml" ],
"dd9289...31d": [ "v0001/data/metadata/relationshipMetadata.xml" ],
"7519c5...63f": [ "v0002/data/metadata/provenanceMetadata.xml" ],
"abda4c...622": [ "v0002/data/metadata/workflows.xml" ],
"76549e...b2b": [ "v0002/data/metadata/rightsMetadata.xml" ],
"bdc4d6...3b6": [ "v0002/data/metadata/events.xml" ],
"7b331c...f9b": [ "v0002/data/metadata/identityMetadata.xml" ],
"80ceac...b9c": [ "v0002/data/metadata/versionMetadata.xml" ],
"4853a2...fbe": [ "v0002/data/metadata/contentMetadata.xml" ],
"1d5090...f5f": [ "v0002/data/metadata/relationshipMetadata.xml" ],
"f209bf...ceb": [ "v0002/data/metadata/embargoMetadata.xml" ],
"dd9125...d4b": [ "v0003/data/metadata/technicalMetadata.xml" ],
"d9e177...477": [ "v0003/data/metadata/provenanceMetadata.xml" ],
"4f5908...4f5": [ "v0003/data/metadata/workflows.xml" ],
"e64db0...500": [ "v0003/data/metadata/descMetadata.xml" ],
"05fa51...818": [ "v0003/data/metadata/rightsMetadata.xml" ],
"d70dd8...5ad": [ "v0003/data/metadata/events.xml" ],
"509a2d...dc6": [ "v0003/data/metadata/identityMetadata.xml" ],
"548066...893": [ "v0003/data/metadata/versionMetadata.xml" ],
"93884e...aae": [ "v0003/data/metadata/contentMetadata.xml" ],
"4c5ab4...b02": [ "v0003/data/metadata/embargoMetadata.xml" ]
},
"type": "https://ocfl.io/1.1/spec/#inventory",
"versions": {
"v1": {
"created": "2019-03-14T20:31:00Z",
"state": {
"98114a...588": [ "content/eric-smith-dissertation-augmented.pdf" ],
"7f3d87...15b": [ "content/eric-smith-dissertation.pdf" ],
"6d19f0...064": [ "metadata/technicalMetadata.xml" ],
"6e4be4...375": [ "metadata/provenanceMetadata.xml" ],
"d8a319...d0f": [ "metadata/descMetadata.xml" ],
"de823a...acc": [ "metadata/rightsMetadata.xml" ],
"080617...40c": [ "metadata/identityMetadata.xml" ],
"e15267...58d": [ "metadata/versionMetadata.xml" ],
"0d9e0b...9a2": [ "metadata/contentMetadata.xml" ],
"dd9289...31d": [ "metadata/relationshipMetadata.xml" ]
}
},
"v2": {
"created": "2019-03-24T09:22:00Z",
"state": {
"98114a...588": [ "content/eric-smith-dissertation-augmented.pdf" ],
"7f3d87...15b": [ "content/eric-smith-dissertation.pdf" ],
"6d19f0...064": [ "metadata/technicalMetadata.xml" ],
"7519c5...63f": [ "metadata/provenanceMetadata.xml" ],
"d8a319...d0f": [ "metadata/descMetadata.xml" ],
"76549e...b2b": [ "metadata/rightsMetadata.xml" ],
"7b331c...f9b": [ "metadata/identityMetadata.xml" ],
"80ceac...b9c": [ "metadata/versionMetadata.xml" ],
"4853a2...fbe": [ "metadata/contentMetadata.xml" ],
"1d5090...f5f": [ "metadata/relationshipMetadata.xml" ],
"abda4c...622": [ "metadata/workflows.xml" ],
"bdc4d6...3b6": [ "metadata/events.xml" ],
"f209bf...ceb": [ "metadata/embargoMetadata.xml" ]
}
},
"v3": {
"created": "2019-04-02T11:07:00Z",
"state": {
"98114a...588": [ "content/eric-smith-dissertation-augmented.pdf" ],
"7f3d87...15b": [ "content/eric-smith-dissertation.pdf" ],
"dd9125...d4b": [ "metadata/technicalMetadata.xml" ],
"d9e177...477": [ "metadata/provenanceMetadata.xml" ],
"e64db0...500": [ "metadata/descMetadata.xml" ],
"05fa51...818": [ "metadata/rightsMetadata.xml" ],
"509a2d...dc6": [ "metadata/identityMetadata.xml" ],
"548066...893": [ "metadata/versionMetadata.xml" ],
"93884e...aae": [ "metadata/contentMetadata.xml" ],
"1d5090...f5f": [ "metadata/relationshipMetadata.xml" ],
"4f5908...4f5": [ "metadata/workflows.xml" ],
"d70dd8...5ad": [ "metadata/events.xml" ],
"4c5ab4...b02": [ "metadata/embargoMetadata.xml" ]
}
}
}
}
The following example OCFL Storage Root has an extension containing custom content. The OCFL Storage Root itself remains valid.
[storage root]
├── 0=ocfl_1.1
├── extensions
│ └── 0000-example-extension
│ └── file-example.txt
├── ocfl_1.1.txt
└── ocfl_layout.json
The following example OCFL Object has an extension containing custom content. The OCFL Object itself remains valid.
[object root]
├── 0=ocfl_object_1.1
├── inventory.json
├── inventory.json.sha512
├── extensions
│ └── 0000-example-extension
│ └── file1-draft.txt
└── v1
├── inventory.json
├── inventory.json.sha512
└── content
└── file.txt
[FIPS-180-4] FIPS PUB 180-4 Secure Hash Standard. U.S. Department of Commerce/National Institute of Standards and Technology. URL: https://nvlpubs.nist.gov/nistpubs/FIPS/NIST.FIPS.180-4.pdf
[NAMASTE] Directory Description with Namaste Tags. J. Kunze.9 November 2009. URL: https://confluence.ucop.edu/download/attachments/14254149/NamasteSpec.pdf
[RFC1321] The MD5 Message-Digest Algorithm. R. Rivest. IETF. April 1992. Informational. URL: https://www.rfc-editor.org/rfc/rfc1321
[RFC2119] Key words for use in RFCs to Indicate Requirement Levels. S. Bradner. IETF. March 1997. Best Current Practice. URL: https://www.rfc-editor.org/rfc/rfc2119
[RFC3339] Date and Time on the Internet: Timestamps. G. Klyne; C. Newman. IETF. July 2002. Proposed Standard. URL: https://www.rfc-editor.org/rfc/rfc3339
[RFC3986] Uniform Resource Identifier (URI): Generic Syntax. T. Berners-Lee; R. Fielding; L. Masinter. IETF. January 2005. Internet Standard. URL: https://www.rfc-editor.org/rfc/rfc3986
[RFC4648] The Base16, Base32, and Base64 Data Encodings. S. Josefsson. IETF. October 2006. Proposed Standard. URL: https://www.rfc-editor.org/rfc/rfc4648
[RFC7693] The BLAKE2 Cryptographic Hash and Message Authentication Code (MAC). M-J. Saarinen, Ed.; J-P. Aumasson. IETF. November 2015. Informational. URL: https://www.rfc-editor.org/rfc/rfc7693
[RFC8259] The JavaScript Object Notation (JSON) Data Interchange Format. T. Bray, Ed.. IETF. December 2017. Internet Standard. URL: https://www.rfc-editor.org/rfc/rfc8259
[BagIt] The BagIt File Packaging Format (V1.0). J. Kunze; J. Littman; E. Madden; J. Scancella; C. Adams. 17 September 2018. URL: https://datatracker.ietf.org/doc/html/rfc8493
[Digest-Algorithms-Extension] OCFL Community Extension 0001: Digest Algorithms. OCFL Editors.URL: https://ocfl.github.io/extensions/0001-digest-algorithms.html
[JSON-Schema] JSON Schema Validation: A Vocabulary for Structural Validation of JSON. A. Wright; H Andrews.20 September 2018. URL: https://json-schema.org/latest/json-schema-validation.html
[Moab] The Moab Design for Digital Object Versioning. Richard Anderson.15 July 2013. URL: https://journal.code4lib.org/articles/8482
[OAIS] Reference Model for an Open Archival Information System (OAIS), Issue 2. June 2012. URL: https://public.ccsds.org/pubs/650x0m2.pdf
[OCFL-Implementation-Notes] OCFL Implementation Notes. URL: https://ocfl.io/draft/implementation-notes
[PairTree] Pairtrees for Object Storage. J. Kunze; M. Haye; E. Hetzner; M. Reyes; C. Snavely. 12 August 2008. URL: https://confluence.ucop.edu/display/Curation/PairTree
[RFC6068] The ‘mailto’ URI Scheme. M. Duerst; L. Masinter; J. Zawinski. IETF. October 2010. Proposed Standard. URL: https://www.rfc-editor.org/rfc/rfc6068
[RFC8141] Uniform Resource Names (URNs). P. Saint-Andre; J. Klensin. IETF. April 2017. Proposed Standard. URL: https://www.rfc-editor.org/rfc/rfc8141