Remove copyright of GPL family license text

Signed-off-by: Wonjae Park <[email protected]>

Author: JustinWonjaePark

src/fosslight_source/_parsing_scancode_file_item.py

@@ -30,6 +30,35 @@
 SPDX_REPLACE_WORDS = ["(", ")"]
 KEY_AND = r"(?<=\s)and(?=\s)"
 KEY_OR = r"(?<=\s)or(?=\s)"
+GPL_LICENSE_PATTERNS = [
+    r'gpl',           # gpl, gpl-2.0, gpl-3.0
+    r'lgpl',          # lgpl, lgpl-2.1, lgpl-3.0
+    r'agpl',          # agpl, agpl-3.0
+    r'gnu.*general.*public.*license',    # GNU General Public License
+    r'gnu.*lesser.*general.*public.*license',  # GNU Lesser General Public License
+    r'gnu.*affero.*general.*public.*license'   # GNU Affero General Public License
+]
+
+
+def is_gpl_family_license(licenses: list) -> bool:
+    if not licenses:
+        return False
+
+    for license_name in licenses:
+        if not license_name:
+            continue
+
+        license_lower = license_name.lower()
+        for pattern in GPL_LICENSE_PATTERNS:
+            if re.search(pattern, license_lower):
+                logger.debug(f"GPL family license detected: {license_name}")
+                return True
+
+    return False
+
+
+def should_remove_copyright_for_gpl_license_text(licenses: list, is_license_text: bool) -> bool:
+    return is_license_text and is_gpl_family_license(licenses)
 
 
 def get_error_from_header(header_item: list) -> Tuple[bool, str]:
@@ -100,8 +129,6 @@ def parsing_scancode_32_earlier(scancode_file_list: list, has_error: bool = Fals
                                 pass
                             copyright_value_list.append(copyright_data)
 
-                    result_item.copyright = copyright_value_list
-
                     # Set the license value
                     license_detected = []
                     if licenses is None or licenses == "":
@@ -165,14 +192,21 @@ def parsing_scancode_32_earlier(scancode_file_list: list, has_error: bool = Fals
                     if len(license_detected) > 0:
                         result_item.licenses = license_detected
 
+                        if is_manifest_file(file_path):
+                            result_item.is_license_text = True
+
+                        # Remove copyright info for license text file of GPL family
+                        if should_remove_copyright_for_gpl_license_text(license_detected, result_item.is_license_text):
+                            logger.debug(f"Removing copyright for GPL family license text file: {file_path}")
+                            result_item.copyright = []
+                        else:
+                            result_item.copyright = copyright_value_list
+
                         if len(license_expression_list) > 0:
                             license_expression_list = list(
                                 set(license_expression_list))
                             result_item.comment = ','.join(license_expression_list)
 
-                        if is_manifest_file(file_path):
-                            result_item.is_license_text = True
-
                         if is_exclude_file(file_path, prev_dir, prev_dir_value):
                             result_item.exclude = True
                         scancode_file_item.append(result_item)
@@ -227,7 +261,6 @@ def parsing_scancode_32_later(
                         except Exception:
                             pass
                         copyright_value_list.append(copyright_data)
-                result_item.copyright = copyright_value_list
 
                 license_detected = []
                 licenses = file.get("license_detections", [])
@@ -263,6 +296,20 @@ def parsing_scancode_32_later(
                                             license_list[lic_matched_key] = lic_info
                                     license_detected.append(found_lic)
                 result_item.licenses = license_detected
+
+                result_item.exclude = is_exclude_file(file_path)
+                result_item.is_license_text = file.get("percentage_of_license_text", 0) > 90 or is_notice_file(file_path)
+
+                if is_manifest_file(file_path) and len(license_detected) > 0:
+                    result_item.is_license_text = True
+
+                # Remove copyright info for license text file of GPL family
+                if should_remove_copyright_for_gpl_license_text(license_detected, result_item.is_license_text):
+                    logger.debug(f"Removing copyright for GPL family license text file: {file_path}")
+                    result_item.copyright = []
+                else:
+                    result_item.copyright = copyright_value_list
+
                 if len(license_detected) > 1:
                     license_expression_spdx = file.get("detected_license_expression_spdx", "")
                     license_expression = file.get("detected_license_expression", "")
@@ -271,12 +318,6 @@ def parsing_scancode_32_later(
                     if license_expression:
                         result_item.comment = license_expression
 
-                result_item.exclude = is_exclude_file(file_path)
-                result_item.is_license_text = file.get("percentage_of_license_text", 0) > 90 or is_notice_file(file_path)
-
-                if is_manifest_file(file_path) and len(license_detected) > 0:
-                    result_item.is_license_text = True
-
                 scancode_file_item.append(result_item)
             except Exception as ex:
                 msg.append(f"Error Parsing item: {ex}")

tests/test_tox.py

@@ -6,6 +6,16 @@
 import subprocess
 import pytest
 import shutil
+import sys
+
+# Add project root to sys.path for importing FL Source modules
+sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..', 'src'))
+
+# Import after sys.path modification to access our custom GPL license functions
+# flake8: noqa E402
+from fosslight_source._parsing_scancode_file_item import (
+    is_gpl_family_license, should_remove_copyright_for_gpl_license_text
+)
 
 remove_directories = ["test_scan", "test_scan2", "test_scan3"]
 
@@ -26,6 +36,69 @@ def run_command(command):
     return success, process.stdout if success else process.stderr
 
 
+def test_is_gpl_family_license():
+    gpl_licenses = [
+        ["gpl-2.0"],
+        ["gpl-3.0"],
+        ["lgpl-2.1"],
+        ["lgpl-3.0"],
+        ["agpl-3.0"],
+        ["GPL-2.0"],
+        ["LGPL-2.1"],
+        ["AGPL-3.0"],
+        ["gnu-general-public-license"],
+        ["gnu-lesser-general-public-license"],
+        ["gnu-affero-general-public-license"],
+        ["gpl-2.0", "mit"],
+        ["mit", "lgpl-3.0"]
+    ]
+
+    non_gpl_licenses = [
+        ["mit"],
+        ["apache-2.0"],
+        ["bsd-3-clause"],
+        ["mozilla-2.0"],
+        ["isc"],
+        [],
+        ["mit", "apache-2.0"]
+    ]
+
+    for licenses in gpl_licenses:
+        assert is_gpl_family_license(licenses), \
+            f"Should detect GPL family license: {licenses}"
+
+    for licenses in non_gpl_licenses:
+        assert not is_gpl_family_license(licenses), \
+            f"Should not detect GPL family license: {licenses}"
+
+
+def test_should_remove_copyright_for_gpl_license_text():
+    assert should_remove_copyright_for_gpl_license_text(["gpl-2.0"], True), \
+        "Should remove copyright for GPL license text file"
+    assert should_remove_copyright_for_gpl_license_text(["lgpl-3.0"], True), \
+        "Should remove copyright for LGPL license text file"
+    assert should_remove_copyright_for_gpl_license_text(["agpl-3.0"], True), \
+        "Should remove copyright for AGPL license text file"
+
+    assert not should_remove_copyright_for_gpl_license_text(["gpl-2.0"], False), \
+        "Should NOT remove copyright for GPL source file"
+    assert not should_remove_copyright_for_gpl_license_text(["lgpl-3.0"], False), \
+        "Should NOT remove copyright for LGPL source file"
+
+    assert not should_remove_copyright_for_gpl_license_text(["mit"], True), \
+        "Should NOT remove copyright for MIT license text file"
+    assert not should_remove_copyright_for_gpl_license_text(["apache-2.0"], True), \
+        "Should NOT remove copyright for Apache license text file"
+
+    assert not should_remove_copyright_for_gpl_license_text(["mit"], False), \
+        "Should NOT remove copyright for MIT source file"
+
+    assert not should_remove_copyright_for_gpl_license_text([], True), \
+        "Should NOT remove copyright for empty license list"
+    assert not should_remove_copyright_for_gpl_license_text([], False), \
+        "Should NOT remove copyright for empty license list"
+
+
 def test_run():
     scan_success, _ = run_command("fosslight_source -p tests/test_files -j -m -o test_scan")
     scan_exclude_success, _ = run_command("fosslight_source -p tests -e test_files/test cli_test.py -j -m -o test_scan2")