Recognize manifest file as License File (#210)

* Recognize manifest files as License File

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

Author: JustinWonjaePark

src/fosslight_source/_parsing_scancode_file_item.py

@@ -13,6 +13,7 @@
 from ._scan_item import is_exclude_file
 from ._scan_item import replace_word
 from ._scan_item import is_notice_file
+from ._scan_item import is_manifest_file
 from typing import Tuple
 
 logger = logging.getLogger(constant.LOGGER_NAME)
@@ -169,6 +170,9 @@ def parsing_scancode_32_earlier(scancode_file_list: list, has_error: bool = Fals
                                 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)
@@ -269,6 +273,10 @@ def parsing_scancode_32_later(
 
                 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}")

src/fosslight_source/_scan_item.py

@@ -14,6 +14,7 @@
 _notice_filename = ['licen[cs]e[s]?', 'notice[s]?', 'legal', 'copyright[s]?', 'copying*', 'patent[s]?', 'unlicen[cs]e', 'eula',
                     '[a,l]?gpl[-]?[1-3]?[.,-,_]?[0-1]?', 'mit', 'bsd[-]?[0-4]?', 'bsd[-]?[0-4][-]?clause[s]?',
                     'apache[-,_]?[1-2]?[.,-,_]?[0-2]?']
+_manifest_filename = [r'.*\.pom$', r'package\.json$', r'setup\.py$', r'pubspec\.yaml$', r'.*\.podspec$', r'Cargo\.toml$']
 _exclude_filename = ["changelog", "config.guess", "config.sub", "changes", "ltmain.sh",
                      "configure", "configure.ac", "depcomp", "compile", "missing", "makefile"]
 _exclude_extension = [".m4", ".in", ".po"]
@@ -137,6 +138,11 @@ def is_exclude_file(file_path: str, prev_dir: str = None, prev_dir_exclude_value
 
 def is_notice_file(file_path: str) -> bool:
     pattern = r"({})(?<!w)".format("|".join(_notice_filename))
-    file_path = file_path.lower()
     filename = os.path.basename(file_path)
-    return bool(re.match(pattern, filename))
+    return bool(re.match(pattern, filename, re.IGNORECASE))
+
+
+def is_manifest_file(file_path: str) -> bool:
+    pattern = r"({})$".format("|".join(_manifest_filename))
+    filename = os.path.basename(file_path)
+    return bool(re.match(pattern, filename, re.IGNORECASE))