Metanorma supports collections: groupings of individual Metanorma documents into a whole. In HTML, Metanorma collections are treated as a web site, with each Metanorma document a component web site. In PDF, they are treated as a single document, composed of the individual Metanorma documents.

Collections are of particular use when the component documents are tightly linked with each other; for example, multilingual documents (where each component document has the same content in a different language), or documents that extensively cross-reference each other (e.g. definitions of an informational model spanning multiple documents.)

Collections are a work in progress as of this writing, and their functionality is in flux as they are starting to be applied to flavours. Currently collections are used in the BIPM flavour, and are in the process of being applied to ISO standards.

Compiling collections

Collections are compiled using the metanorma executable, as follows:

  metanorma collection FILENAME [..options]

  -x, [--extensions=EXTENSIONS]     # Type of extension to generate
  -w, [--output-folder=FOLDER]      # Folder to generate collection in
  -c, [--coverpage=COVERPAGE]       # Cover page as Liquid template for collection (currently HTML only)

The FILENAME argument to metanorma collection is the collection manifest. This is a YAML file, outlining the structure of the collection, including the location and identifier for each of the component files. The COVERPAGE argument is a Liquid template for the index page to the collection in HTML. Both are described further below.

So the command:

bundle exec metanorma collection si-brochure-bilingual.yml \
  -x xml,html,presentation,pdf \
  -w bilingual-brochure \
  -c collection_cover.html

compiles the collection described in the collection manifest si-brochure-bilingual.yml. It generates the collection as XML, Presentation XML, HTML and PDF, in the folder bilingual-brochure, and using the HTML index page template collection_cover.html.

Currently the metanorma collection executable presupposes that the individual metanorma collections are already compiled into XML. The compilation of the collection resolves any cross-references between files in the collection in preprocessing, so that they become simple hyperlinks. The output folder contains those preprocessed individual files, and files named collection in the target formats.

Specifying collections

Collection Manifest

The collection manifest contains:

  • Directives on how the collection should be generated (directives):

    • documents-inline indicates that the files should be concatenated into a single XML file for processing; if absent, the files are kept separate in processing (documents-external).

  • Metadata about the collection (bibdata, in Relaton format).

    • Currently docid/type is mandatory, as this is how the processor establishes the flavour of the collection. Currently a collection can only contain documents of one flavour.

  • A manifest listing the documents contained in the collection, in nested hierarchy (manifest)

    • level names the current hierarchical level of the manifest

    • title gives the title of the current level of the manifest

    • docref lists the documents to be used at that level of the manifest: it is a list of file paths relative to the manifest file (fileref) and document identifiers (identifier).

    • manifest can appear recursively in a manifest.

  • Content to put at the beginning of the collection container (prefatory-content)

  • Content to put at the end of the collection container (final-content)

The following is an example collection manifest:

  - documents-inline
    type: title-main
    language: en
    content: ISO Collection 1
  type: collection
    type: iso
    id: ISO 12345
  edition: 1
    - type: created
      value: "2020"
    - type: issued
      value: "2020"
      name: International Organization for Standardization
      abbreviation: ISO
    from: "2020"
  level: collection
  title: ISO Collection
    - level: subcollection
      title: Standards
        - fileref:
          identifier: ISO 17301-1:2016
        - fileref: dummy.xml
          identifier: ISO 17302
        - fileref:
          identifier: ISO 1701:1974
    - level: subcollection
      title: Amendments
        identifier: ISO 17301-1:2016/Amd.1:2017

  == Clause
  Welcome to our collection

  == Exordium
  Hic explicit

Index page template

The HTML index page template is currently realised as a Liquid template, which forms a sidebar for the display of the HTML content of each file. The following fields are defined:

  • doctitle, docnumber, etc.: Information derived from the Relaton YAML description in the manifest. The field names are as defined for Liquid templates in Metanorma: see Metadata and Boilerplate.

  • navigation: A nested list giving hyperlinks to the constituent documents, following the specification in the manifest field of the collection manifest.


Direct cross-references

A source document can link to a target document in the same collection, or a specific location within the target document. Documents are processed one document at a time; so such a link is encoded as a bibliographical reference, to an external document, as described in Bibliography. That means that we need to define a bibliographic entry for each hyperlinked document in the same collection; those bibliographic entries will be suppressed from display in the collection. (If the documents are to be used in isolation, those bibliographic entries still need to be displayed: otherwise, the reference cannot be made sense of.)

The bibliographic reference for another document in the same collection is:

* [[[myanchor,repo:(current-metanorma-collection/docid)]]]

where docid is the document identifier as it appears in the collection manifest.

The location to link to in the target document can be specified as a clause number, as is normal in citations: e.g. <<myanchor,clause=3.1>>.

The processor will then navigate the target document, to try to resolve the reference. (Currently only one level of nesting of locations is implemented: the processor will not resolve references like clause=3.1,note-3.)

Alternatively, the location can be specified as an anchor, e.g. <<myanchor,anchor=ident>>. The hyperlink will then be made directly to the element with anchor ident in the the target document. That approach is to be preferred as simpler.

For example, we wish to link from the French BIPM Brochure to the English BIPM Brochure, and specifically to an example in the English document. We start by assigning the target document example an anchor identifier:

For example: The maximum electric potential difference is stem:[ii(U)_("max") = 1000 " "rm(V)]  but not stem:[ii(U) = 1000 " "rm(ii(V)_(max))]. The mass fraction of copper in the sample of silicon is stem:[w("Cu") = 1.3 xx 10^(-6)] but not stem:[1.3 xx 10^(-6) " "rm(w)//rm(w)].

We then define a citation in the source document, using that anchor:

Ce n’est que
lorsque l’on écrit le nom de l’unité en toutes lettres que l’on applique les règles
grammaticales ordinaires (voir un exemple en anglais page <<english-doc,anchor=english_example>>).

Finally, we define a bibliographic entry in the source document for the English-language target document:

== Bibliography

* [[[english-doc,repo:(current-metanorma-collection/si-brochure-en)]]] (Version anglaise de la brochure BIPM).

The identifier given to the target document needs to match that given in the collection manifest:

  level: brochure
  title: Brochure/Brochure
    - fileref: si-brochure-fr.xml
      identifier: si-brochure-fr
    - fileref: si-brochure-en.xml
      identifier: si-brochure-en

Indirect cross-references

In some documents, anchors (targets for cross-references) are inserted in various files in the collection, and we do not necessarily know at the time of authoring which files those anchors will end up in. A good example of that is computer-generated documentation of schemas: schema documentation is organised by entity, and the documentation of one entity can cross-reference attributes in a different entity. But at the time of authoring, we may not know which document the target entity will appear in, so we cannot supply a bibliographic entity naming that document.

To deal with that circumstance, Metanorma implements a special class of cross-references, which are namespaced and which use containers:

  • The namespace is provided to deal with the fact that such anchors can have different provenance, and they may have particular rendering requirements. (So if we are documenting two different schemas, we may want to differentiate their references, and render them differently.)

  • The container relies on the fact that such anchors can be grouped together in a target document, under a clause. (For example, a schema instance.) For efficient processing, we treat each of those container clauses as a single bibliographic reference, and use the identifier of that clause as the bibliographic anchor. We also assign the container clause the namespace as a type, again for efficiency and to enforce consistent rendering. This is mandatory.

  • The locality is the identifier of the particular component addressed within the container. It is an identifier in the target document, and will typically point to a subclause of the container clause.

  • The text is the text to be rendered for the cross-reference. If not provided, Metanorma will provide a clause reference for the target.

To give a worked example:

We are generating documentation for a set of schemas in the EXPRESS language as a Metanorma collection. We wish to point to the identifier basic_attribute_schema.id_attribute.identified_item from our source document. We do not know (or care) what document that identifier will turn up in: we will have collection processing deal with that.

basic_attribute_schema.id_attribute.identified_item is an identifier within the basic schema, and we are grouping the definitions of the basic schema together, under a single clause in the target document.

The target document will thus contain a container clause with identifier basic, containing all those definitions, including basic_attribute_schema.id_attribute.identified_item. The container clause is made to be of type express (because its content comes from that language, and we want to follow the conventions of that language in any processing).

=== Basic Schema


===== Identified Item

The cross-reference to that identifier, from either the same document or a different document in the same collection, is:

<<express:basic:basic_attribute_schema.id_attribute.identified_item,Identified Item>>

We do not need to indicate which document basic_attribute_schema.id_attribute.identified_item is in, unlike for direct cross-references. Because of the namespacing, we know that we are looking for the identifier basic_attribute_schema.id_attribute.identified_item inside a clause with id basic and type express: that narrows down our search while generating the collection. The basic collection identifier is actually optional; but if you don’t provide it, you will need to put [type=express] on any cross-reference target, and collection processing will be more expensive.