From e2a1e2cc835b917b2696b43fd6eae1f11314bff2 Mon Sep 17 00:00:00 2001
From: Al Grant <al.grant@arm.com>
Date: Sat, 1 Feb 2025 14:20:55 +0000
Subject: [PATCH] update source license checker

---
 check_for_license.py | 70 ++++++++++++++++++++++++++++++++------------
 1 file changed, 52 insertions(+), 18 deletions(-)
 mode change 100755 => 100644 check_for_license.py

diff --git a/check_for_license.py b/check_for_license.py
old mode 100755
new mode 100644
index 802b557..8ba4d10
--- a/check_for_license.py
+++ b/check_for_license.py
@@ -24,8 +24,13 @@
 
 import os, sys
 
+
+o_verbose = 0
+
+
 ## Distance we need to find the license text in
-LICENSE_MAX_LINE = 20
+LICENSE_MAX_LINE = 32
+
 
 ## Texts to search for, with a flag to say if it's the 'approved' license
 LICENSE_TEXT = {
@@ -35,14 +40,19 @@
 }
 
 
+# Source extensions.
+# Other files (e.g. makefiles) may be considered sources on an ad-hoc basis.
+SOURCE_EXTS = [".c", ".cpp", ".h", ".py", ".json", ".xml"]
+
+
 def fn_is_source(fn):
     """
     Test a filename to check that it's source code and needs a license.
     """
-    base, ext = os.path.splitext(fn)
-    if "makefile" in base:
-        return base
-    return ext in [".c", ".cpp", ".h", ".py", ".json", ".xml"]
+    (base, ext) = os.path.splitext(fn)
+    if base in ["makefile", "Makefile"]:
+        return True
+    return ext in SOURCE_EXTS
 
 
 def fn_is_licensed(fn):
@@ -50,24 +60,35 @@ def fn_is_licensed(fn):
     Check that a source file already has appropriate license text embedded in it.,
     within the first LICENSE_MAX_LINE lines.
     """
-    found = None
+    if o_verbose:
+        print("checking: %s..." % fn)
+    license_found = None
+    copyright_found = None
+    license_is_irregular = False
     f = open(fn)
     lno = 0
     for ln in f:
         lno += 1
-        for ltext, is_regular in LICENSE_TEXT.items():
+        for (ltext, is_regular) in LICENSE_TEXT.items():
             ix = ln.find(ltext)
             if ix >= 0:
-                found = lno
-                break
-            if found is not None:
+                license_found = lno
+                if lno > LICENSE_MAX_LINE:
+                    print("%s: license found at line %u, later than expected" % (fn, lno), file=sys.stderr)
+                if not is_regular:
+                    print("%s: irregular license waiver" % (fn), file=sys.stderr)
+                    license_is_irregular = True
                 break
-        if lno > LICENSE_MAX_LINE:
+        if "Copyright" in ln and ("ARM" in ln) or ("Arm" in ln):
+            copyright_found = lno
+        if license_found is not None and copyright_found is not None:
             break
-    if False:
-        if found is not None:
-            print("%s: found at line %u" % (fn, found))
-    return found is not None
+        if lno > LICENSE_MAX_LINE*5:
+            break
+    if o_verbose:
+        if license_found is not None:
+            print("%s: license found at line %u" % (fn, license_found))
+    return license_found is not None and ((copyright_found is not None) or license_is_irregular)
 
 
 def check_all_sources_licensed(dir):
@@ -76,9 +97,11 @@ def check_all_sources_licensed(dir):
     """
     n_files = 0
     n_unlicensed = 0
-    for root, dirs, files in os.walk(dir):
+    for (root, dirs, files) in os.walk(dir):
+        if root != "." and root.startswith("."):
+            continue
         for f in files:
-            fn = os.path.join(root, f)   
+            fn = os.path.join(root, f)
             if fn_is_source(fn):
                 n_files += 1
                 if not fn_is_licensed(fn):
@@ -88,5 +111,16 @@ def check_all_sources_licensed(dir):
 
 
 if __name__ == "__main__":
-    sys.exit(check_all_sources_licensed(os.path.dirname(os.path.realpath(__file__))) > 0)
+    import argparse
+    parser = argparse.ArgumentParser(description="check source files for licenses")
+    parser.add_argument("dirs", nargs="*", help="source directories to check")
+    parser.add_argument("-v", "--verbose", action="count", default=0, help="increase verbosity")
+    opts = parser.parse_args()
+    o_verbose = opts.verbose
+    if not opts.dirs:
+        opts.dirs = [os.path.dirname(os.path.realpath(__file__))]
+    n_unlicensed = 0
+    for dir in opts.dirs:
+        n_unlicensed += check_all_sources_licensed(dir)
+    sys.exit(n_unlicensed > 0)