Update logical location kinds to accommodate XML and JSON paths

[michaelcfanning]

'kind' for results grouping, XML and JSON domains, and can we find a magical general term that absorbs function/member

[michaelcfanning]

Received some feedback from a team implementing a JSON-driven analysis

• Some assistance with logical locations would be useful
• In this scenario, the leaf node is not super interesting, the parent objects are what's interesting. Our design currently requires that a logicalIndex point into the detail object.
• The scan team has two paths, one is qualified by actual array indices, e.g., object[5].property. They have another, more 'stable' form where they elide the actual index, e.g., object[].property. This data removal is analogous to eliding line location details (where document churn might invalidate them).

[ghost]

EBALLOT PROPOSAL: add a set of logical location kinds to the open-ended logicalLocation.kind property value set that help users locate paths within JSON, XML or other mark-up documents.

API IMPACT: no literal API impact, as this is an open-ended string. Producers should favor one of our set of 'blessed' terms if they deem them relevant. Terms that will be referenced in the specification and in schema comments will now also include:

• For JSON:
  • object
  • array
  • property
  • value
• For XML:
  • element
  • text
  • attribute
  • comment
  • dtd
  • processingInstruction

[kupsch]

Have to state how to represent a dtd in the logical location as it appears that there is not an XPath representation for the DTD. Perhaps /!DOCTYPE

[kupsch]

To make it clear, it should be noted that the line number are not part of the text. The line number could also be stylized differently such as color.

[kupsch]

In the JSON example there are a couple of errors in the SARIF data:

results/0/locations/0/fullQualifiedLogicalName should be "/orders/0/productIds/1"
results/0/locations/0/logicalLocationIndex should be 3
results/1/locations/0/logicalLocationIndex should be 4

[ghost]

@kupsch Thanks Jim. I noted those changes in this comment in #342, and will fix it up after the ballot ends.

[ghost]

Approved in e-ballot #2.