fosslight-util 2.1.35__tar.gz → 2.1.37__tar.gz

{fosslight_util-2.1.35/src/fosslight_util.egg-info → fosslight_util-2.1.37}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fosslight_util
-Version: 2.1.35
+Version: 2.1.37
 Summary: FOSSLight Util
 Home-page: https://github.com/fosslight/fosslight_util
 Download-URL: https://github.com/fosslight/fosslight_util

{fosslight_util-2.1.35 → fosslight_util-2.1.37}/setup.py

@@ -14,7 +14,7 @@ with open('requirements.txt', 'r', 'utf-8') as f:
 if __name__ == "__main__":
     setup(
         name='fosslight_util',
-        version='2.1.35',
+        version='2.1.37',
         package_dir={"": "src"},
         packages=find_packages(where='src'),
         description='FOSSLight Util',

{fosslight_util-2.1.35 → fosslight_util-2.1.37}/src/fosslight_util/exclude.py

@@ -7,8 +7,14 @@ import os
 import fnmatch
 from typing import List
 
-EXCLUDE_DIRECTORY = ["test", "tests", "doc", "docs"]
+EXCLUDE_DIRECTORY = ["test", "tests", "doc", "docs", "intermediates"]
 PACKAGE_DIRECTORY = ["node_modules", "venv", "Pods", "Carthage"]
+EXCLUDE_FILENAME = ["changelog", "config.guess", "config.sub", "changes", "ltmain.sh",
+                    "configure", "configure.ac", "depcomp", "compile", "missing", "makefile",
+                    'fosslight_bin', 'fosslight_bin.exe']
+EXCLUDE_FILE_EXTENSION = ['qm', 'xlsx', 'pdf', 'pptx', 'jfif', 'docx', 'doc', 'whl',
+                          'xls', 'xlsm', 'ppt', 'mp4', 'pyc', 'plist', 'dat',
+                          "m4", "in", "po", "class"]
 
 
 def excluding_files(patterns: List[str], path_to_scan: str) -> List[str]:
@@ -98,13 +104,12 @@ def is_exclude_dir(rel_path: str) -> tuple:
     return False, False
 
 
-def get_excluded_paths(path_to_scan: str, custom_excluded_paths: list = [], exclude_file_extension: list = []) -> tuple:
+def get_excluded_paths(path_to_scan: str, custom_excluded_paths: list = [], custom_exclude_extension: list = []) -> tuple:
     path_to_exclude = []
     path_to_exclude_with_dot = []
     excluded_files = set()  # Use set for O(1) operations
     abs_path_to_scan = os.path.abspath(path_to_scan)
     custom_excluded_normalized = [p.replace('\\', '/') for p in custom_excluded_paths]
-    exclude_extensions_lower = [ext.lower().lstrip('.') for ext in exclude_file_extension]
     cnt_file_except_skipped = 0
 
     for root, dirs, files in os.walk(path_to_scan):
@@ -124,23 +129,30 @@ def get_excluded_paths(path_to_scan: str, custom_excluded_paths: list = [], excl
             file_path = os.path.join(root, file_name)
             rel_path = os.path.relpath(file_path, abs_path_to_scan).replace('\\', '/')
             should_exclude = False
-            has_dot = False
+            except_info_sheet = False
             if not _has_parent_in_exclude_list(rel_path, path_to_exclude):
+                file_ext = os.path.splitext(file_name)[1].lstrip('.').lower()
                 if rel_path in custom_excluded_normalized:
                     should_exclude = True
                 elif file_name in custom_excluded_normalized:
                     should_exclude = True
                 elif file_name.startswith('.'):
                     should_exclude = True
-                    has_dot = True
-                elif exclude_extensions_lower:
-                    file_ext = os.path.splitext(file_name)[1].lstrip('.').lower()
-                    if file_ext in exclude_extensions_lower:
-                        should_exclude = True
+                    except_info_sheet = True
+                elif file_ext and file_ext in custom_exclude_extension:
+                    should_exclude = True
+                elif file_name.lower() in EXCLUDE_FILENAME:
+                    should_exclude = True
+                    except_info_sheet = True
+                    cnt_file_except_skipped += 1
+                elif file_ext and file_ext in EXCLUDE_FILE_EXTENSION:
+                    should_exclude = True
+                    except_info_sheet = True
+                    cnt_file_except_skipped += 1
 
                 if should_exclude:
                     path_to_exclude.append(rel_path)
-                    if has_dot:
+                    if except_info_sheet:
                         path_to_exclude_with_dot.append(rel_path)
                     excluded_files.add(rel_path)
             else:

{fosslight_util-2.1.35 → fosslight_util-2.1.37}/src/fosslight_util/get_pom_license.py

@@ -8,6 +8,12 @@ import logging
 import urllib.request
 from urllib.error import URLError, HTTPError
 from defusedxml.ElementTree import fromstring as xml_fromstring
+import ssl
+# certifi is optional: if unavailable, use the default SSL context
+try:
+    import certifi  # type: ignore
+except Exception:  # pragma: no cover - optional dependency
+    certifi = None
 import fosslight_util.constant as constant
 
 logger = logging.getLogger(constant.LOGGER_NAME)
@@ -19,6 +25,15 @@ def get_license_from_pom(group_id: str = None,
                          pom_path: str = None,
                          check_parent: bool = True) -> str:
 
+    def get_ssl_context():
+        try:
+            if certifi is not None:
+                return ssl.create_default_context(cafile=certifi.where())
+            return ssl.create_default_context()
+        except Exception as e:
+            logger.debug(f"Failed to create SSL context: {e}")
+            return None
+
     def build_urls(g, a, v):
         group_path = g.replace('.', '/')
         name = f"{a}-{v}.pom"
@@ -27,11 +42,25 @@ def get_license_from_pom(group_id: str = None,
         return [repo1, google]
 
     def fetch_pom(g, a, v):
+        ssl_ctx = get_ssl_context()
         for url in build_urls(g, a, v):
             try:
+                if ssl_ctx is not None:
+                    with urllib.request.urlopen(url, context=ssl_ctx) as resp:
+                        return resp.read().decode('utf-8')
                 with urllib.request.urlopen(url) as resp:
                     return resp.read().decode('utf-8')
-            except (HTTPError, URLError):
+            except ssl.SSLError as e:
+                logger.warning(
+                    f"SSL certificate verification failed for {url}. "
+                    f"Please fix system certificates or use certifi. (error: {e})"
+                )
+                continue
+            except (HTTPError, URLError) as e:
+                logger.warning(f"Failed to fetch POM from {url}: {e}")
+                continue
+            except Exception as e:
+                logger.warning(f"Unexpected error fetching POM from {url}: {e}")
                 continue
         return None
 
@@ -66,10 +95,12 @@ def get_license_from_pom(group_id: str = None,
         visited.add(key)
         content = fetch_pom(g, a, v)
         if not content:
+            logger.warning(f"Failed to obtain POM content for {g}:{a}:{v} from remote sources.")
             return ''
         try:
             root = xml_fromstring(content)
-        except Exception:
+        except Exception as e:
+            logger.warning(f"Failed to parse POM for {g}:{a}:{v}: {e}")
             return ''
         licenses = extract_licenses(root)
         if licenses:

{fosslight_util-2.1.35 → fosslight_util-2.1.37/src/fosslight_util.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fosslight_util
-Version: 2.1.35
+Version: 2.1.37
 Summary: FOSSLight Util
 Home-page: https://github.com/fosslight/fosslight_util
 Download-URL: https://github.com/fosslight/fosslight_util