Ignore some attributes when compare IDs

[andrew2net]

Sometimes we need to find last edition, release, or year of publication. Sometimes we need to select all parts collection. For example the method pubid_ref == pubid returns false if pubid_ref doesn't have edition but pubid does. We need a way to ignore some attributes if they undefined in pubid_ref. For example we can pass a list of attributes that should be ignored:

pubid_ref == pubid, [:edition, :year]

[mico]

@andrew2net how about we add an option to only compare elements which are defined? So we don't need to use list of attributes to ignore?

Something like: pubid_ref.eql?(pubid, only_defined: true)

[andrew2net]

@mico some undefined elements should be used in comparing. We have various cases across relaton-* gems when some elements should be ignored in comparing. So we need a way to exclude some elements from comparing but not all undefined elements.

[mico]

@mico some undefined elements should be used in comparing. We have various cases across relaton-* gems when some elements should be ignored in comparing. So we need a way to exclude some elements from comparing but not all undefined elements.

@andrew2net could you provide more details about cases when we need to exclude some elements but not all undefined?

[andrew2net]

We have many cases in relaton-* gem and the requirements are still changing. So we need some flexibility in IDs' comparing. Some cases are:

• In some cases users can omit edition in references and we need to select all the document's editions and return the latest.
• In some cases users can omit year in reference and we need to return the latest
• Users can make mistakes in year. The relaton-iso finds all the document's years and gives user a hint message.
• The relaton-iso has any_types_stages option. Types and stages should be ignored if the option is true.
• Some relaton-* gems support all_parts option. When the option is true, we need to ignore part and year.
• The relaton-bsi first tries to find exact ID. If fails, tries to ignore all undefined parts except amendment. Finally it tries to ignore amendment.
• The relaton-cen ignores part and year. It doesn't ignore amendment but amendment's year is ignored when omitted in a reference.
• The relaton-itu always ignores type and month. It also ignores year if omitted in a reference.
• The relaton-ecma ignores edition and value if they are omitted.
• The relaton-nist ignores stage, revision, and year if they are omitted.

[mico]

We have many cases in relaton-* gem and the requirements are still changing. So we need some flexibility in IDs' comparing.

Summarizing cases, we need functionality to:

• return latest year
• return identifiers for all years
• return latest edition
• return identifiers for all edition
• return identifiers for all edition month's (ignore month, but apply year)
• return documents with any types and stages
• return documents with any stages
• return documents with any parts
• return documents with any revisions
• return documents with any amendment's edition (year)
• return documents with any amendments

Did I miss something?

What is value in relaton-ecma?

Looking at that, I agree that excluding list could solve the problem we have. Also, I see here you need to fetch the latest year or edition sometimes, so we can add option for fetching latest version.

[andrew2net]

We have to scenarios in Relaton, fetching from Relaton repository and fetching from websites.

Relaton repositories have indexes, YAML files with ID => filename structure. Most indexes have ID as a string, but they are going to be a Hash with Pubid#to_h content (that what we need to_h function). So with IDs as Hash we can create a document Pubid (can we?) and compare it agains a reference Pubid. We need to select a list of document Pubids that match to all the attributes presented in the reference Pubid, ignoring attributes that omitted and listed in an ignore list (attr.nil? && ignore_list.include?(attr)). Next we can find the ID with the biggest attribute number, or create all_parts combined document, or do whatever we need.

Websites search engines return list of document IDs as strings. We need to parse these IDs and compare document Pubids with reference Pubid in same way as described above.

[andrew2net]

Summarizing cases, we need functionality to:

• return latest year
• return identifiers for all years
• return latest edition
• return identifiers for all edition
• return identifiers for all edition month's (ignore month, but apply year)
• return documents with any types and stages
• return documents with any stages
• return documents with any parts
• return documents with any revisions
• return documents with any amendment's edition (year)
• return documents with any amendments

Did I miss something?

These are parts of Relaton's internal functionality. We don't heed to hardcode each case in the Pubid. Cases can be changed later. So, in Relaton we need:

• Create Pubid objects from string ID
• Create Pubid objects from Hash (output Pubid#to_h method)
• Compare Pubid objects with ability to exclude some attributes from comparing. We should be able to exclude attributes from comparing.

What is value in relaton-ecma?

It's volume, sorry for mistake. Here is a part of ECMA index:

...
- :id:
    :id: ECMA-269
    :ed: '2'
  :file: data/ECMA-269-2.yaml
- :id:
    :id: ECMA-269
    :ed: '3'
    :vol: '1'
  :file: data/ECMA-269-3-1.yaml
- :id:
    :id: ECMA-269
    :ed: '3'
    :vol: '2'
  :file: data/ECMA-269-3-2.yaml
...

With ref ECMA-269 all 3 entries selected. With ref ECMA-269-2 only the first entry selected. With ref ECMA-269-3 last 2 entries selected.