Merge pull request #130 from fosslight/temp

Apply simple mode

Author: bjk7119

src/fosslight_binary/_simple_mode.py

@@ -0,0 +1,118 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# FOSSLight Binary analysis script
+# Copyright (c) 2024 LG Electronics Inc.
+# SPDX-License-Identifier: Apache-2.0
+import os
+import re
+import logging
+import zipfile
+import tarfile
+import fosslight_util.constant as constant
+from fosslight_util.write_txt import write_txt_file
+from fosslight_util.set_log import init_log
+
+REMOVE_FILE_EXTENSION_SIMPLE = ['ttf', 'otf', 'png', 'gif', 'jpg', 'bmp', 'jpeg']
+logger = logging.getLogger(constant.LOGGER_NAME)
+
+
+def is_compressed_file(filename):
+    if filename.lower().endswith('.jar'):
+        return False
+    return zipfile.is_zipfile(filename) or tarfile.is_tarfile(filename)
+
+
+def exclude_bin_for_simple_mode(binary_list):
+    bin_list = []
+    compressed_list = []
+
+    for bin in binary_list:
+        file_lower_case = bin.bin_name_with_path.lower()
+        extension = os.path.splitext(file_lower_case)[1][1:].strip()
+
+        if is_compressed_file(bin.bin_name_with_path):
+            compressed_list.append(bin.bin_name_with_path)
+            continue
+
+        remove_file_ext_list = REMOVE_FILE_EXTENSION_SIMPLE
+        if any(extension == remove_ext for remove_ext in remove_file_ext_list):
+            continue
+        if re.search(r".*sources\.jar", bin.bin_name_with_path.lower()) or bin.exclude:
+            continue
+
+        bin_list.append(bin.bin_name_with_path)
+    return compressed_list, bin_list
+
+
+def convert_list_to_str(input_list):
+    output_text = '\n'.join(map(str, input_list))
+    return output_text
+
+
+def check_output_path(output, start_time):
+    compressed_list_txt = ""
+    simple_bin_list_txt = ""
+    output_path = ""
+
+    if output != "":
+        if not os.path.isdir(output) and output.endswith('.txt'):
+            output_path = os.path.dirname(output)
+            basename = os.path.basename(output)
+            basename_file, _ = os.path.splitext(basename)
+            compressed_list_txt = f"{basename_file}_compressed_list.txt"
+            simple_bin_list_txt = f"{basename_file}.txt"
+        else:
+            output_path = output
+            compressed_list_txt = f"compressed_list_{start_time}.txt"
+            simple_bin_list_txt = f"binary_list_{start_time}.txt"
+    else:
+        compressed_list_txt = f"compressed_list_{start_time}.txt"
+        simple_bin_list_txt = f"binary_list_{start_time}.txt"
+
+    if output_path == "":
+        output_path = os.getcwd()
+    else:
+        output_path = os.path.abspath(output_path)
+
+    compressed_list_txt = os.path.join(output_path, compressed_list_txt)
+    simple_bin_list_txt = os.path.join(output_path, simple_bin_list_txt)
+
+    return output_path, compressed_list_txt, simple_bin_list_txt
+
+
+def init_simple(output_file_name, pkg_name, start_time):
+    global logger, _result_log
+
+    output_path, compressed_list_txt, simple_bin_list_txt = check_output_path(output_file_name, start_time)
+
+    log_file = os.path.join(output_path, f"fosslight_log_bin_{start_time}.txt")
+    logger, _result_log = init_log(log_file, False, logging.INFO, logging.DEBUG, pkg_name)
+
+    return _result_log, compressed_list_txt, simple_bin_list_txt
+
+
+def print_simple_mode(compressed_list_txt, simple_bin_list_txt, compressed_list, bin_list):
+    results = []
+    success = True
+    msg = ""
+    output_file = ""
+    if compressed_list:
+        success, error = write_txt_file(compressed_list_txt, convert_list_to_str(compressed_list))
+        if success:
+            output_file = compressed_list_txt
+        else:
+            msg = f"Error to write compressed list file for simple mode : {error}"
+        results.append(tuple([success, msg, output_file]))
+    if bin_list:
+        success, error = write_txt_file(simple_bin_list_txt, convert_list_to_str(bin_list))
+        if success:
+            output_file = simple_bin_list_txt
+        else:
+            msg = f"Error to write binary list file for simple mode : {error}"
+        results.append(tuple([success, msg, output_file]))
+    return results
+
+
+def filter_binary(bin_list):
+    compressed_list, bin_list = exclude_bin_for_simple_mode(bin_list)
+    return compressed_list, bin_list

src/fosslight_binary/binary_analysis.py

@@ -18,6 +18,7 @@
 from ._binary_dao import get_oss_info_from_db
 from ._binary import BinaryItem, TLSH_CHECKSUM_NULL
 from ._jar_analysis import analyze_jar_file, merge_binary_list
+from ._simple_mode import print_simple_mode, filter_binary, init_simple
 from fosslight_util.correct import correct_with_yaml
 from fosslight_util.oss_item import ScannerItem
 import hashlib
@@ -41,7 +42,7 @@
 _REMOVE_DIR = [os.path.sep + dir_name + os.path.sep for dir_name in _REMOVE_DIR]
 _error_logs = []
 _root_path = ""
-_start_time = ""
+start_time = ""
 windows = False
 BYTES = 2048
 BIN_EXT_HEADER = {'BIN_FL_Binary': ['ID', 'Binary Path', 'OSS Name',
@@ -72,18 +73,9 @@ def get_checksum_and_tlsh(bin_with_path):
 
 
 def init(path_to_find_bin, output_file_name, formats, path_to_exclude=[]):
-    global _root_path, logger, _start_time
+    global logger, _result_log
 
     _json_ext = ".json"
-    _start_time = datetime.now().strftime('%y%m%d_%H%M')
-    _result_log = {
-        "Tool Info": PKG_NAME
-    }
-
-    _root_path = path_to_find_bin
-    if not path_to_find_bin.endswith(os.path.sep):
-        _root_path += os.path.sep
-
     success, msg, output_path, output_files, output_extensions, formats = check_output_formats_v2(output_file_name, formats)
 
     if success:
@@ -100,20 +92,20 @@ def init(path_to_find_bin, output_file_name, formats, path_to_exclude=[]):
                 if formats:
                     if formats[i].startswith('spdx'):
                         if platform.system() != 'Windows':
-                            output_files[i] = f"fosslight_spdx_bin_{_start_time}"
+                            output_files[i] = f"fosslight_spdx_bin_{start_time}"
                         else:
                             logger.warning('spdx format is not supported on Windows. Please remove spdx from format.')
                             to_remove.append(i)
                     else:
                         if output_extension == _json_ext:
-                            output_files[i] = f"fosslight_opossum_bin_{_start_time}"
+                            output_files[i] = f"fosslight_opossum_bin_{start_time}"
                         else:
-                            output_files[i] = f"fosslight_report_bin_{_start_time}"
+                            output_files[i] = f"fosslight_report_bin_{start_time}"
                 else:
                     if output_extension == _json_ext:
-                        output_files[i] = f"fosslight_opossum_bin_{_start_time}"
+                        output_files[i] = f"fosslight_opossum_bin_{start_time}"
                     else:
-                        output_files[i] = f"fosslight_report_bin_{_start_time}"
+                        output_files[i] = f"fosslight_report_bin_{start_time}"
         for index in sorted(to_remove, reverse=True):
             # remove elements of spdx format on windows
             del output_files[index]
@@ -127,7 +119,7 @@ def init(path_to_find_bin, output_file_name, formats, path_to_exclude=[]):
         logger.error(f"Format error - {msg}")
         sys.exit(1)
 
-    log_file = os.path.join(output_path, f"fosslight_log_bin_{_start_time}.txt")
+    log_file = os.path.join(output_path, f"fosslight_log_bin_{start_time}.txt")
     logger, _result_log = init_log(log_file, True, logging.INFO, logging.DEBUG,
                                    PKG_NAME, path_to_find_bin, path_to_exclude)
 
@@ -176,16 +168,30 @@ def get_file_list(path_to_find, abs_path_to_exclude):
                 bin_item.exclude = True
             elif extension in _EXCLUDE_FILE_EXTENSION:
                 bin_item.exclude = True
+            elif file.startswith('.'):
+                bin_item.exclude = True
             bin_list.append(bin_item)
             file_cnt += 1
     return file_cnt, bin_list, found_jar
 
 
 def find_binaries(path_to_find_bin, output_dir, formats, dburl="", simple_mode=False,
                   correct_mode=True, correct_filepath="", path_to_exclude=[]):
+    global start_time, _root_path, _result_log
 
-    _result_log, result_reports, output_extensions = init(
-        path_to_find_bin, output_dir, formats, path_to_exclude)
+    mode = "Normal Mode"
+    start_time = datetime.now().strftime('%y%m%d_%H%M')
+
+    _root_path = path_to_find_bin
+    if not path_to_find_bin.endswith(os.path.sep):
+        _root_path += os.path.sep
+
+    if simple_mode:
+        mode = "Simple Mode"
+        _result_log, compressed_list_txt, simple_bin_list_txt = init_simple(output_dir, PKG_NAME, start_time)
+    else:
+        _result_log, result_reports, output_extensions = init(
+            path_to_find_bin, output_dir, formats, path_to_exclude)
 
     total_bin_cnt = 0
     total_file_cnt = 0
@@ -201,7 +207,8 @@ def find_binaries(path_to_find_bin, output_dir, formats, dburl="", simple_mode=F
     if not os.path.isdir(path_to_find_bin):
         error_occured(error_msg=f"Can't find the directory : {path_to_find_bin}",
                       result_log=_result_log,
-                      exit=True)
+                      exit=True,
+                      mode=mode)
     if not correct_filepath:
         correct_filepath = path_to_find_bin
     try:
@@ -210,12 +217,21 @@ def find_binaries(path_to_find_bin, output_dir, formats, dburl="", simple_mode=F
     except Exception as ex:
         error_occured(error_msg=f"Failed to check whether it is binary or not : {ex}",
                       result_log=_result_log,
-                      exit=True)
-    total_bin_cnt = len(return_list)
+                      exit=True,
+                      mode=mode)
     if simple_mode:
-        bin_list = [bin.bin_name_with_path for bin in return_list]
+        try:
+            compressed_list, filtered_bin_list = filter_binary(return_list)
+            results = print_simple_mode(compressed_list_txt, simple_bin_list_txt, compressed_list, filtered_bin_list)
+            total_bin_cnt = len(filtered_bin_list)
+        except Exception as ex:
+            error_occured(error_msg=f"Failed to run simple mode: {ex}",
+                          result_log=_result_log,
+                          exit=True,
+                          mode="Simple mode")
     else:
-        scan_item = ScannerItem(PKG_NAME, _start_time)
+        total_bin_cnt = len(return_list)
+        scan_item = ScannerItem(PKG_NAME, start_time)
         scan_item.set_cover_pathinfo(path_to_find_bin, path_to_exclude)
         try:
             # Run OWASP Dependency-check
@@ -262,7 +278,7 @@ def find_binaries(path_to_find_bin, output_dir, formats, dburl="", simple_mode=F
                 logger.error(f"Fail to generate result file.:{writing_msg}")
 
     try:
-        print_result_log(success=True, result_log=_result_log,
+        print_result_log(mode=mode, success=True, result_log=_result_log,
                          file_cnt=str(total_file_cnt),
                          bin_file_cnt=str(total_bin_cnt),
                          auto_bin_cnt=str(db_loaded_cnt), bin_list=bin_list)
@@ -279,7 +295,7 @@ def return_bin_only(file_list, need_checksum_tlsh=True):
                 if need_checksum_tlsh:
                     file_item.checksum, file_item.tlsh, error_msg = get_checksum_and_tlsh(file_item.bin_name_with_path)
                     if error_msg:
-                        error_occured(error_msg=error_msg, exit=False)
+                        error_occured(modeerror_msg=error_msg, exit=False)
                 yield file_item
         except Exception as ex:
             logger.debug(f"Exception in get_file_list: {ex}")
@@ -317,21 +333,22 @@ def check_binary(file_with_path):
     return is_bin_confirmed
 
 
-def error_occured(error_msg, exit=False, result_log={}):
+def error_occured(error_msg, exit=False, result_log={}, mode="Normal mode"):
     global _error_logs
     _error_logs.append(error_msg)
     if exit:
-        print_result_log(success=False, result_log=result_log)
+        print_result_log(mode, success=False, result_log=result_log)
         sys.exit()
 
 
-def print_result_log(success=True, result_log={}, file_cnt="", bin_file_cnt="", auto_bin_cnt="", bin_list=[]):
+def print_result_log(mode="Normal Mode", success=True, result_log={}, file_cnt="", bin_file_cnt="", auto_bin_cnt="", bin_list=[]):
 
     if "Running time" in result_log:
-        start_time = result_log["Running time"]
+        starttime = result_log["Running time"]
     else:
-        start_time = _start_time
-    result_log["Running time"] = start_time + " ~ " + \
+        starttime = start_time
+    result_log["Mode"] = mode
+    result_log["Running time"] = starttime + " ~ " + \
         datetime.now().strftime('%Y%m%d_%H%M%S')
     result_log["Execution result"] = 'Success' if success else 'Error occurred'
     result_log["Binaries / Scanned files"] = f"{bin_file_cnt}/{file_cnt}"

src/fosslight_binary/cli.py

@@ -49,7 +49,7 @@ def main():
         print_package_version(_PKG_NAME, "FOSSLight Binary Scanner Version:")
         sys.exit(0)
 
-    if args.simple:
+    if args.simple:  # -s option
         simple_mode = True
 
     if args.path:   # -p option