Exclude package dirs with directory name

requirements.txt

@@ -6,4 +6,6 @@ PyYAML
 wheel>=0.38.1
 intbitset
 fosslight_binary>=5.0.0
-scancode-toolkit>=32.0.2
+scancode-toolkit>=32.0.2
+fingerprints==1.2.3
+normality==2.6.1

src/fosslight_source/_parsing_scancode_file_item.py

@@ -14,6 +14,7 @@
 from ._scan_item import replace_word
 from ._scan_item import is_notice_file
 from ._scan_item import is_manifest_file
+from ._scan_item import is_package_dir
 from typing import Tuple
 
 logger = logging.getLogger(constant.LOGGER_NAME)
@@ -99,6 +100,13 @@ def parsing_scancode_32_earlier(scancode_file_list: list, has_error: bool = Fals
                     copyright_list = file.get("copyrights", [])
 
                     result_item = SourceItem(file_path)
+                    is_pkg, pkg_path = is_package_dir(os.path.dirname(file_path))
+                    if is_pkg:
+                        result_item.source_name_or_path = pkg_path
+                        if not any(x.source_name_or_path == result_item.source_name_or_path for x in scancode_file_item):
+                            result_item.exclude = True
+                            scancode_file_item.append(result_item)
+                        continue
 
                     if has_error and "scan_errors" in file:
                         error_msg = file.get("scan_errors", [])
@@ -235,6 +243,13 @@ def parsing_scancode_32_later(
                     continue
 
                 result_item = SourceItem(file_path)
+                is_pkg, pkg_path = is_package_dir(os.path.dirname(file_path))
+                if is_pkg:
+                    result_item.source_name_or_path = pkg_path
+                    if not any(x.source_name_or_path == result_item.source_name_or_path for x in scancode_file_item):
+                        result_item.exclude = True
+                        scancode_file_item.append(result_item)
+                    continue
 
                 if has_error:
                     error_msg = file.get("scan_errors", [])

src/fosslight_source/_parsing_scanoss_file.py

@@ -8,6 +8,7 @@
 import fosslight_util.constant as constant
 from ._scan_item import SourceItem
 from ._scan_item import is_exclude_file
+from ._scan_item import is_package_dir
 from ._scan_item import replace_word
 from typing import Tuple
 
@@ -45,6 +46,13 @@ def parsing_scanResult(scanoss_report: dict, path_to_scan: str = "", path_to_exc
         if any(os.path.commonpath([abs_file_path, exclude_path]) == exclude_path for exclude_path in abs_path_to_exclude):
             continue
         result_item = SourceItem(file_path)
+        is_pkg, pkg_path = is_package_dir(os.path.dirname(file_path))
+        if is_pkg:
+            result_item.source_name_or_path = pkg_path
+            if not any(x.source_name_or_path == result_item.source_name_or_path for x in scanoss_file_item):
+                result_item.exclude = True
+                scanoss_file_item.append(result_item)
+            continue
 
         if 'id' in findings[0]:
             if "none" == findings[0]['id']:

src/fosslight_source/_scan_item.py

@@ -22,6 +22,7 @@
 _exclude_directory = [os.path.sep + dir_name +
                       os.path.sep for dir_name in _exclude_directory]
 _exclude_directory.append("/.")
+_package_directory = ["node_modules", "venv", "Pods", "Carthage"]
 MAX_LICENSE_LENGTH = 200
 MAX_LICENSE_TOTAL_LENGTH = 600
 SUBSTRING_LICENSE_COMMENT = "Maximum character limit (License)"
@@ -146,3 +147,13 @@ 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))
+
+
+def is_package_dir(dir_path: str) -> bool:
+    path_parts = dir_path.split(os.path.sep)
+    for pkg_dir in _package_directory:
+        if pkg_dir in path_parts:
+            pkg_index = path_parts.index(pkg_dir)
+            pkg_path = os.path.sep.join(path_parts[:pkg_index + 1])
+            return True, pkg_path
+    return False, ""

src/fosslight_source/run_scancode.py

@@ -127,6 +127,8 @@ def run_scan(
                                 result_list, key=lambda row: (''.join(row.licenses)))
 
                             for scan_item in result_list:
+                                if os.path.isdir(scan_item.source_name_or_path):
+                                    continue
                                 if check_binary(os.path.join(path_to_scan, scan_item.source_name_or_path)):
                                     scan_item.exclude = True
             except Exception as ex: