DeltaCode Scoring

Delta Objects

A File-Level Comparison of Two Codebases

A Delta object represents the file-level comparison (i.e., the "delta") of two codebases, typically two versions of the same codebase, using ScanCode-generated JSON output files as input for the comparison process.

Based on how the user constructs the command-line input, DeltaCode's naming convention treats one codebase as the "new" codebase and the other as the "old" codebase:

deltacode -n [path to the 'new' codebase] -o [path to the 'old' codebase] [...]

Basic Scoring

A DeltaCode codebase comparison produces a collection of file-level Delta objects. Depending on the nature of the file-level change between the two codebases, each Delta object is characterized as belonging to one of the categories listed below.

Each category has an associated score intended to convey its potential importance to a user's analysis of the changes between the new and old codebases. In descending order of importance, the categories are:

1. added: A file has been added to the new codebase.
2. modified: The file is contained in both the new and old codebase and has been modified. This modification is demonstrated among other things by a change in the file's sha1 attribute.
3. moved: The file is contained in both the new and old codebase and has not been modified but has been moved.
4. removed: A file has been removed from the old codebase.
5. unmodified: The file is contained in both the new and old codebase and has not been modified or moved.

The score of a Delta object characterized as added or modified may be increased based on the detection of license- and/or copyright-related changes. See License Additions and Changes and Copyright Holder Additions and Changes below.

Delta Object Fields and Values

Each Delta object includes the following fields and values:

• factors: One or more strings representing the factors that characterize the file-level comparison and resulting score, e.g.,

      "factors": [
        "added",
        "license info added",
        "copyright info added"
      ],

• score: A number representing the magnitude/importance of the file-level change -- the higher the score, the greater the change.
• new: The ScanCode-based file attributes (path, licenses, copyrights etc.) for the file in the codebase designated by the user as new.
• old: The ScanCode-based file attributes for the file in the codebase designated by the user as old.

Note that an added Delta object will have a new file but no old file, while a removed Delta object will have an old file but not a new file. In each case, the value for the missing file will be null.

License Additions and Changes

Certain file-level changes involving the license-related information in a Delta object result in an increase in the object's score.

• For an added Delta object, if the new file contains one or more licenses(license info added), the object's score will be increased.
• A modified Delta object's score will be increased:
  • If the old file had at least one license and the new file has no licenses (license info removed).
  • If the old file had no licenses and the new file has at least one license (license info added).
  • If both the old and new file had at least one license and the license keys are not identical (e.g., the old file included an mit license and an apache-2.0 license and the new file includes only an mit license) (license change).
• For both an added and a modified Delta object, the object's score will be increased if the the new file contains any of the following license categories:

'Commercial'
'Copyleft'
'Copyleft Limited'
'Free Restricted'
'Patent License'
'Proprietary Free'

Copyright Holder Additions and Changes

[coming . . .]

Moved, Removed and Unmodified

[coming . . .]