Copying License dependencies for Pyinstaller

Question

This is related to a previous question but is more specifically about implementation.

To recap the previous question, I have a Pyinstaller app with 3rd party dependencies. As @AKX answered, the license text from those dependencies should be included in the Pyinstaller distribution.

Pyinstaller makes no attempt to handle this.

Therefore I am looking for a way to automate this process.

My project uses a virtual env, and the dependencies are located at:

\venv\Lib\site-packages

One option would be to recursivley search through this folder, and find any files that match a specific pattern (e.g LICENSE.txt). However, some of the packages do not ship with a license file, and instead just name the license in their metadata. For example, I am using openpyxl

there is a folder

\venv\Lib\site-packages\openpyxl-3.0.0-py3.7.egg-info

which contains a PKG-INFO file. This lists the license (MIT/Expat) but does specifically contain the license text. In this case would it be acceptable to include a copy of the PKG-INFO file, or some copy of the standard MIT license text?

Has anyone faced this issue and implemented a solution in an elegant way? @AKX (in the previous question linked to a tool which outputs the license information as text/json etc.. but doesn't actually attempt to move any files.

Answer 1

If anyone has a similar question, this is the solution I implemented, which seems to work pretty well for me:

At the top of my .spec file I added this

matches = ["LICENSE.txt","METADATA","PKG-INFO"]
lics = []
print("Find 3rd party dependency license files")
for root, dir, files in os.walk("venv\Lib\site-packages"):
    for file in files:
            if file in matches:
               src = f"{root}/{file}"
               dest = f"licenses/{os.path.basename(root)}"
               lics.append((src,dest))
               print(f"\tLicense file: {root}/{file}")
print(f"{len(lics)} dependency licenses found. Copying to /license folder in distribution")

which builds a list of tuples containing the source of the license or metadata, and the destination to copy it to in the distribution.

To do the actual copying, I made use of the datas keyword in the Analysis object, and simply expanded the list using *lics.

This could be adapted to include more matches or be smarter and parse the files, or something like that. It is also probably a bit wasteful and copies some files multiple times, but this seems like a minimal overhead for a tiny text file.

If anyone sees this and has suggestions or comments then I'd be happy to be further informed on this subject!