Validate Third-Party Documentation with scancode

We should use scancode to systematically go through the third-party code in Qt, and verify that our documentation is correct and complete.

Scancode

Scancode is a command line tool to extract copyright, license information out of sources. It is slow, but it's results are pretty accurate.

Scancode Workbench

Scancode Workbench is a GUI tool to inspect the information generated by scancode.

Steps needed

For each repository of Qt, do

 - Extract sources, removing own license markers (lines between QT_BEGIN_LICENSE, QT_END_LICENSE)

• run scancode with options -n6 --license --copyright --json qtmodule.json path_to_qt_module/src , n6 being the number of CPU's to use
• Load generated qtmodule.json into Scancode Workbench
• Compare scancode results with Third-Party Code documented in the online documentation for the respective Qt modules
• Do we accurately report all third-party licenses?
• Do we accurately report all third-party Copyright statements in code that is not governed by the Qt licenses?
• Do we accurately report all files/directories that have different licenses?
• Report results, fix deviations in the respective qt_attribution.json file

How to compare licenses

Note that scancode identifies licenses through their identifier in DejaCode https://enterprise.dejacode.com/licenses/ . In Qt documentation we try to use the SPDX identifier and license name instead.

Anyhow, Scancode also reports the SPDX-Identifier (if available). If we hit cases where there's no SPDX identifier / yet we should fall back to the DejaCode URN and license name.
 

How to compare copyright statements

In Qt we try to sanitize copyright lines, sometimes aggregating multiple lines into one.

Copyright 2013-2015 Google
Copyright 2004 Google
Copyright 2008 Google

can be aggregated to

Copyright 2004, 2008, 2013-2015 Google

An internal script is available to help facilitate (parts of) these aggregations.

Note that some copyrights can also be ignored, e.g. because they affect parts of code that doesn't directly affect source code that is compiled into Qt. Examples are Copyrights for the Free Software Foundation (typically copyright for licenses).