fosslight-binary 5.1.31__tar.gz → 5.1.32__tar.gz

{fosslight_binary-5.1.31 → fosslight_binary-5.1.32}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fosslight_binary
-Version: 5.1.31
+Version: 5.1.32
 Summary: FOSSLight Binary Scanner
 Author: LG Electronics
 License-Expression: Apache-2.0
@@ -29,7 +29,7 @@ Requires-Dist: py-tlsh
 Requires-Dist: pytz
 Requires-Dist: XlsxWriter
 Requires-Dist: PyYAML
-Requires-Dist: fosslight_util>=2.2.8
+Requires-Dist: fosslight_util>=2.2.14
 Requires-Dist: truststore
 Requires-Dist: python-magic-bin; sys_platform == "win32"
 Requires-Dist: python-magic; "darwin" in sys_platform

{fosslight_binary-5.1.31 → fosslight_binary-5.1.32}/pyproject.toml

@@ -7,7 +7,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "fosslight_binary"
-version = "5.1.31"
+version = "5.1.32"
 description = "FOSSLight Binary Scanner"
 readme = "README.md"
 license = "Apache-2.0"
@@ -36,7 +36,7 @@ dependencies = [
     "pytz",
     "XlsxWriter",
     "PyYAML",
-    "fosslight_util>=2.2.8",
+    "fosslight_util>=2.2.14",
     "truststore",
     "python-magic-bin; sys_platform == 'win32'",
     "python-magic; 'darwin' in sys_platform",

fosslight_binary-5.1.32/src/fosslight_binary/_exclude.py

@@ -0,0 +1,15 @@
+# Copyright (c) 2026 LG Electronics Inc.
+# SPDX-License-Identifier: Apache-2.0
+
+"""Binary-scanner filename excludes."""
+
+from fosslight_util.exclude import is_excluded_filename
+
+# Scanner output binaries that should not be re-analyzed.
+EXCLUDE_FILENAME_BINARY = frozenset({
+    "fosslight_bin", "fosslight_bin.exe",
+})
+
+
+def is_excluded_binary_filename(file_path: str) -> bool:
+    return is_excluded_filename(file_path, EXCLUDE_FILENAME_BINARY)

{fosslight_binary-5.1.31 → fosslight_binary-5.1.32}/src/fosslight_binary/_jar_analysis.py

@@ -19,9 +19,16 @@ from fosslight_util.oss_item import OssItem
 
 logger = logging.getLogger(constant.LOGGER_NAME)
 
-_CENTRAL_SEARCH_URL = "https://search.maven.org/solrsearch/select"
+# Both hosts serve the same Central Solr index and the same g/a/v fields, but
+# search.maven.org answers only about one query in six: the TCP connect always
+# succeeds and then the response never starts, so no timeout budget rescues it.
+# The Portal host answers every query in about half a second, so it goes first.
+_CENTRAL_SEARCH_URLS = (
+    "https://central.sonatype.com/solrsearch/select",
+    "https://search.maven.org/solrsearch/select",
+)
 _REQUEST_TIMEOUT = 10          # seconds - used for HEAD / POM download
-_CENTRAL_SEARCH_TIMEOUT = 2.5  # seconds - tight timeout for Search API (retried on timeout)
+_CENTRAL_SEARCH_TIMEOUT = 2.5  # seconds - per-host budget, retried on timeout
 _MAVEN_JAR_HTTP_TIMEOUT = (2, 2)  # match Util probe timeouts for multi-repo jar checks
 _MAX_RETRY = 3                 # maximum Central API retry attempts per JAR
 _central_network_warned = False  # Flag to suppress repeated network-unavailable warnings within one run
@@ -116,16 +123,37 @@ def _warn_network_once(context=""):
 
 
 def _search_central_by_sha1(sha1, timeout=None):
+    """Look up Maven coordinates by JAR checksum, trying each Central host in turn.
+
+    An answer from any host is authoritative for the whole index, so a miss ends
+    the search. ``timed_out`` is reported only when no host answered at all.
+    """
     if not sha1:
         return {}, False
     _timeout = timeout if timeout is not None else _CENTRAL_SEARCH_TIMEOUT
-    try:
-        params = {"q": f"1:{sha1}", "rows": 1, "wt": "json"}
-        resp = requests.get(_CENTRAL_SEARCH_URL, params=params, timeout=_timeout)
-        resp.raise_for_status()
-        docs = resp.json().get("response", {}).get("docs", [])
+    params = {"q": f"1:{sha1}", "rows": 1, "wt": "json"}
+    any_timeout = False
+    unreachable = False
+
+    for url in _CENTRAL_SEARCH_URLS:
+        try:
+            resp = requests.get(url, params=params, timeout=_timeout)
+            resp.raise_for_status()
+            docs = resp.json().get("response", {}).get("docs", [])
+        except requests.exceptions.Timeout:
+            logger.debug(f"Maven Central SHA-1 search timed out ({sha1}) at {url} - will retry")
+            any_timeout = True
+            continue
+        except Exception as ex:
+            if _is_network_error(ex):
+                unreachable = True
+            else:
+                logger.debug(f"Maven Central SHA-1 search failed ({sha1}) at {url}: {ex}")
+            continue
+
         if not docs:
             return {}, False
+
         doc = docs[0]
         groupId = doc.get("g", "")
         artifactId = doc.get("a", "")
@@ -140,15 +168,11 @@ def _search_central_by_sha1(sha1, timeout=None):
             "artifactId": artifactId,
             "version": version,
         }, False
-    except requests.exceptions.Timeout:
-        logger.debug(f"Maven Central SHA-1 search timed out ({sha1}) - will retry")
-        return {}, True
-    except Exception as ex:
-        if _is_network_error(ex):
-            _warn_network_once(f"SHA-1 search: {sha1[:10]}...")
-        else:
-            logger.debug(f"Maven Central SHA-1 search failed ({sha1}): {ex}")
-        return {}, False
+
+    # Warn only once every host has failed: a fallback answer is not a failure.
+    if unreachable:
+        _warn_network_once(f"SHA-1 search: {sha1[:10]}...")
+    return {}, any_timeout
 
 
 def _download_pom_to_tempfile(group_id, artifact_id, version, timeout=None):
@@ -367,7 +391,8 @@ def _process_one_jar(jar_path, rel_path, sha1, search_timeout=None, skip_central
 
 
 def _has_valid_jar_oss(oss_list):
-    return any(oss.name and oss.license for oss in oss_list)
+    """True if any entry is identified. A missing license is reported as empty."""
+    return any(oss.name for oss in oss_list)
 
 
 def _store_jar_result(jar_items, sha1, result):

{fosslight_binary-5.1.31 → fosslight_binary-5.1.32}/src/fosslight_binary/_simple_mode.py

@@ -4,10 +4,10 @@
 # Copyright (c) 2024 LG Electronics Inc.
 # SPDX-License-Identifier: Apache-2.0
 import os
-import re
 import logging
 import zipfile
 import tarfile
+import magic
 import yaml
 import fosslight_util.constant as constant
 from fosslight_util.set_log import init_log
@@ -18,7 +18,14 @@ logger = logging.getLogger(constant.LOGGER_NAME)
 
 
 def is_compressed_file(filename):
-    return zipfile.is_zipfile(filename) or tarfile.is_tarfile(filename)
+    if zipfile.is_zipfile(filename) or tarfile.is_tarfile(filename):
+        return True
+
+    try:
+        file_type = magic.from_file(filename).lower()
+        return 'archive' in file_type or 'compressed' in file_type
+    except Exception:
+        return False
 
 
 def is_jar_file(filename):
@@ -35,9 +42,9 @@ def exclude_bin_for_simple_mode(binary_list):
 
         path = bin.bin_name_with_path
 
-        if is_jar_file(path) and not re.search(r".*sources\.jar", path.lower()):
+        if is_jar_file(path) and not path.lower().endswith('-sources.jar'):
             bin_list_exclude_compressed.append(path)
-        elif re.search(r".*sources\.jar", path.lower()) or is_compressed_file(path):
+        elif path.lower().endswith('-sources.jar') or is_compressed_file(path):
             compressed_list.append(path)
         else:
             bin_list_exclude_compressed.append(path)

{fosslight_binary-5.1.31 → fosslight_binary-5.1.32}/src/fosslight_binary/binary_analysis.py

@@ -22,6 +22,7 @@ from fosslight_util.time import current_timestamp_utc, format_running_time, time
 from fosslight_util.correct import correct_with_yaml
 from fosslight_util.oss_item import ScannerItem
 from fosslight_util.exclude import get_excluded_paths
+from ._exclude import EXCLUDE_FILENAME_BINARY, is_excluded_binary_filename
 import hashlib
 import tlsh
 from io import open
@@ -93,15 +94,14 @@ def init(path_to_find_bin, output_file_name, formats, path_to_exclude=[]):
         for i, output_extension in enumerate(output_extensions):
             if output_files[i] is None or output_files[i] == "":
                 if formats:
-                    if formats[i].startswith('spdx') or formats[i].startswith('cyclonedx'):
+                    if formats[i].startswith('spdx'):
                         if platform.system() == 'Windows':
                             logger.warning(f'{formats[i]} is not supported on Windows. Please remove {formats[i]} from format.')
                             to_remove.append(i)
                         else:
-                            if formats[i].startswith('spdx'):
-                                output_files[i] = f"fosslight_spdx_bin_{file_time}"
-                            elif formats[i].startswith('cyclonedx'):
-                                output_files[i] = f'fosslight_cyclonedx_bin_{file_time}'
+                            output_files[i] = f"fosslight_spdx_bin_{file_time}"
+                    elif formats[i].startswith('cyclonedx'):
+                        output_files[i] = f'fosslight_cyclonedx_bin_{file_time}'
                     else:
                         if output_extension == _json_ext:
                             output_files[i] = f"fosslight_opossum_bin_{file_time}"
@@ -147,7 +147,7 @@ def get_file_list(path_to_find, excluded_files):
         for file in files:
             bin_with_path = os.path.join(root, file)
             rel_path_file = os.path.relpath(bin_with_path, path_to_find).replace('\\', '/')
-            if rel_path_file in excluded_files:
+            if rel_path_file in excluded_files or is_excluded_binary_filename(rel_path_file):
                 continue
             file_lower_case = file.lower()
             extension = os.path.splitext(file_lower_case)[1][1:].strip()
@@ -200,10 +200,14 @@ def find_binaries(path_to_find_bin, output_dir, formats, kb_url="", kb_token="",
         excluded_path_with_default_exclusion, excluded_path_without_dot, excluded_files, cnt_file_except_skipped = all_exclude_mode
     elif simple_mode:
         excluded_path_with_default_exclusion, excluded_path_without_dot, excluded_files, cnt_file_except_skipped \
-            = get_excluded_paths(path_to_find_bin, path_to_exclude, REMOVE_FILE_EXTENSION_SIMPLE)
+            = get_excluded_paths(
+                path_to_find_bin, path_to_exclude, REMOVE_FILE_EXTENSION_SIMPLE,
+                exclude_filenames=EXCLUDE_FILENAME_BINARY)
     else:
         excluded_path_with_default_exclusion, excluded_path_without_dot, excluded_files, cnt_file_except_skipped \
-            = get_excluded_paths(path_to_find_bin, path_to_exclude)
+            = get_excluded_paths(
+                path_to_find_bin, path_to_exclude,
+                exclude_filenames=EXCLUDE_FILENAME_BINARY)
     logger.debug(f"Skipped paths: {excluded_path_with_default_exclusion}")
 
     if not os.path.isdir(path_to_find_bin):
@@ -273,7 +277,8 @@ def find_binaries(path_to_find_bin, output_dir, formats, kb_url="", kb_token="",
 
             for combined_path_and_file, output_extension, output_format in zip(result_reports, output_extensions, formats):
                 results.append(write_output_file(combined_path_and_file, output_extension, scan_item,
-                                                 BIN_EXT_HEADER, HIDE_HEADER, output_format))
+                                                 BIN_EXT_HEADER, HIDE_HEADER, output_format,
+                                                 scanner_covers=[scan_item.cover]))
 
         except Exception as ex:
             error_occured(error_msg=str(ex), exit=False)

{fosslight_binary-5.1.31 → fosslight_binary-5.1.32}/src/fosslight_binary.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fosslight_binary
-Version: 5.1.31
+Version: 5.1.32
 Summary: FOSSLight Binary Scanner
 Author: LG Electronics
 License-Expression: Apache-2.0
@@ -29,7 +29,7 @@ Requires-Dist: py-tlsh
 Requires-Dist: pytz
 Requires-Dist: XlsxWriter
 Requires-Dist: PyYAML
-Requires-Dist: fosslight_util>=2.2.8
+Requires-Dist: fosslight_util>=2.2.14
 Requires-Dist: truststore
 Requires-Dist: python-magic-bin; sys_platform == "win32"
 Requires-Dist: python-magic; "darwin" in sys_platform

{fosslight_binary-5.1.31 → fosslight_binary-5.1.32}/src/fosslight_binary.egg-info/SOURCES.txt

@@ -8,6 +8,7 @@ LICENSES/LicenseRef-3rd-party-licenses.txt
 src/fosslight_binary/__init__.py
 src/fosslight_binary/_binary.py
 src/fosslight_binary/_binary_dao.py
+src/fosslight_binary/_exclude.py
 src/fosslight_binary/_help.py
 src/fosslight_binary/_jar_analysis.py
 src/fosslight_binary/_simple_mode.py

{fosslight_binary-5.1.31 → fosslight_binary-5.1.32}/src/fosslight_binary.egg-info/requires.txt

@@ -9,7 +9,7 @@ py-tlsh
 pytz
 XlsxWriter
 PyYAML
-fosslight_util>=2.2.8
+fosslight_util>=2.2.14
 truststore
 
 [:"darwin" in sys_platform]