Change vendoring script to preserve license files

As pointed out by #2950, it is probably a good idea to keep the license
files for the vendored dependencies.

This is done by changing the `pavement.py` tasks.

1 files changed, 32 insertions, 0 deletions

diff --git a/pavement.py b/pavement.py
index 81ff6f12..1cebad79 100644
--- a/pavement.py
+++ b/pavement.py
@@ -1,6 +1,8 @@
 import re
 import sys
 import subprocess
+from itertools import chain
+from fnmatch import fnmatch
 
 from paver.easy import task, path as Path
 
@@ -52,12 +54,42 @@ def install(vendor):
         '-t', str(vendor),
     ]
     subprocess.check_call(install_args)
+    move_licenses(vendor)
     remove_all(vendor.glob('*.dist-info'))
     remove_all(vendor.glob('*.egg-info'))
     remove_all(vendor.glob('six.py'))
     (vendor / '__init__.py').write_text('')
 
 
+def move_licenses(vendor):
+    license_patterns = ("*LICEN[CS]E*", "COPYING*", "NOTICE*", "AUTHORS*")
+    licenses = (
+        entry
+        for path in chain(vendor.glob("*.dist-info"), vendor.glob("*.egg-info"))
+        for entry in path.glob("*")
+        if any(fnmatch(str(entry), p) for p in license_patterns)
+    )
+    for file in licenses:
+        file.move(_find_license_dest(file, vendor))
+
+
+def _find_license_dest(license_file, vendor):
+    basename = license_file.basename()
+    pkg = license_file.dirname().replace(".dist-info", "").replace(".egg-info", "")
+    parts = pkg.split("-")
+    acc = []
+    for part in parts:
+        acc.append(part)
+        for option in ("_".join(acc), "-".join(acc), ".".join(acc)):
+            candidate = Path(option)
+            if candidate.isdir():
+                return candidate / basename
+            if Path(f"{candidate}.py").isfile():
+                return Path(f"{candidate}.{basename}")
+
+    raise FileNotFoundError(f"No destination found for {license_file}")
+
+
 def update_pkg_resources():
     vendor = Path('pkg_resources/_vendor')
     install(vendor)