Add --no_merge option for source scanner

pyproject.toml

@@ -29,7 +29,7 @@ dependencies = [
     "pyyaml",
     "beautifulsoup4",
     "fosslight_util>=2.1.47,<3.0.0",
-    "fosslight_source>=2.2.16,<3.0.0",
+    "fosslight_source>=2.3.3,<3.0.0",
     "fosslight_dependency>=4.1.31,<5.0.0",
     "fosslight_binary>=5.1.17,<6.0.0",
 ]

src/fosslight_scanner/_help.py

@@ -66,6 +66,7 @@
     For 'all' or 'source' mode:
       --kb_url <url>       KB API URL for source analysis
       --kb_token <token>   KB API bearer token for source analysis
+      --no_merge           Keep source paths file-based without folder merge
 
     💡 Examples
     ────────────────────────────────────────────────────────────────────

src/fosslight_scanner/_parse_setting.py

@@ -28,12 +28,14 @@ def parse_setting_json(data):
     kb_token = data.get('kb_token', '')
     binary_simple = data.get('binary_simple', False)
     recursive_dep = data.get('recursive_dep', False)
+    no_merge = data.get('no_merge', False)
     str_lists = [mode, path, exclude_path]
     strings = [
         dep_argument, output, format, db_url,
         correct_fpath, link, selected_source_scanner, kb_url, kb_token
     ]
-    booleans = [timer, raw, no_correction, ui, source_write_json_file, source_print_matched_text, binary_simple, recursive_dep]
+    booleans = [timer, raw, no_correction, ui, source_write_json_file,
+                source_print_matched_text, binary_simple, recursive_dep, no_merge]
 
     is_incorrect = False
 
@@ -68,4 +70,4 @@ def parse_setting_json(data):
     return mode, path, dep_argument, output, format, link, db_url, timer, \
         raw, core, no_correction, correct_fpath, ui, exclude_path, \
         selected_source_scanner, source_write_json_file, source_print_matched_text, source_time_out, \
-        binary_simple, recursive_dep, kb_url, kb_token
+        binary_simple, recursive_dep, kb_url, kb_token, no_merge

src/fosslight_scanner/cli.py

@@ -16,7 +16,7 @@
 
 def set_args(mode, path, dep_argument, output, format, link, db_url, timer,
              raw, core, no_correction, correct_fpath, ui, setting, exclude_path,
-             recursive_dep, kb_url="", kb_token=""):
+             recursive_dep, kb_url="", kb_token="", no_merge=False):
 
     selected_source_scanner = "all"
     source_write_json_file = False
@@ -31,7 +31,7 @@ def set_args(mode, path, dep_argument, output, format, link, db_url, timer,
             s_mode, s_path, s_dep_argument, s_output, s_format, s_link, s_db_url, s_timer, s_raw, s_core, \
                 s_no_correction, s_correct_fpath, s_ui, s_exclude_path, \
                 s_selected_source_scanner, s_source_write_json_file, s_source_print_matched_text, \
-                s_source_time_out, s_binary_simple, s_recursive_dep, s_kb_url, s_kb_token = parse_setting_json(data)
+                s_source_time_out, s_binary_simple, s_recursive_dep, s_kb_url, s_kb_token, s_no_merge = parse_setting_json(data)
 
             # direct cli arguments have higher priority than setting file
             mode = mode or s_mode
@@ -51,6 +51,7 @@ def set_args(mode, path, dep_argument, output, format, link, db_url, timer,
             recursive_dep = recursive_dep or s_recursive_dep
             kb_url = kb_url or s_kb_url
             kb_token = kb_token or s_kb_token
+            no_merge = no_merge or s_no_merge
 
             # These options are only set from the setting file, not from CLI arguments
             selected_source_scanner = s_selected_source_scanner or selected_source_scanner
@@ -64,7 +65,7 @@ def set_args(mode, path, dep_argument, output, format, link, db_url, timer,
     return mode, path, dep_argument, output, format, link, db_url, timer, \
         raw, core, no_correction, correct_fpath, ui, exclude_path, \
         selected_source_scanner, source_write_json_file, source_print_matched_text, source_time_out, \
-        binary_simple, recursive_dep, kb_url, kb_token
+        binary_simple, recursive_dep, kb_url, kb_token, no_merge
 
 
 def main():
@@ -115,6 +116,8 @@ def main():
                         type=str, dest='kb_url', default="")
     parser.add_argument('--kb_token', help='KB API bearer token for source analysis',
                         type=str, dest='kb_token', default="")
+    parser.add_argument('--no_merge', help='Keep source paths file-based without folder merge',
+                        action='store_true', dest='no_merge', required=False, default=False)
 
     try:
         args = parser.parse_args()
@@ -128,17 +131,17 @@ def main():
     else:
         mode, path, dep_argument, output, format, link, db_url, timer, raw, core, no_correction, correct_fpath, \
             ui, exclude_path, selected_source_scanner, source_write_json_file, source_print_matched_text, \
-            source_time_out, binary_simple, recursive_dep, kb_url, kb_token = set_args(
+            source_time_out, binary_simple, recursive_dep, kb_url, kb_token, no_merge = set_args(
                 args.mode, args.path, args.dep_argument, args.output,
                 args.format, args.link, args.db_url, args.timer, args.raw,
                 args.core, args.no_correction, args.correct_fpath, args.ui,
                 args.setting, args.exclude_path, args.recursive_dep,
-                args.kb_url, args.kb_token)
+                args.kb_url, args.kb_token, args.no_merge)
 
         run_main(mode, path, dep_argument, output, format, link, db_url, timer,
                  raw, core, not no_correction, correct_fpath, ui, exclude_path,
                  selected_source_scanner, source_write_json_file, source_print_matched_text,
-                 source_time_out, kb_url, kb_token, binary_simple, recursive_dep)
+                 source_time_out, kb_url, kb_token, binary_simple, recursive_dep, no_merge)
 
 
 if __name__ == "__main__":

src/fosslight_scanner/fosslight_scanner.py

@@ -152,7 +152,8 @@ def run_scanner(src_path, dep_arguments, output_path, keep_raw_data=False,
                 default_oss_name="", default_oss_version="", url="",
                 correct_mode=True, correct_fpath="", ui_mode=False, path_to_exclude=[],
                 selected_source_scanner="all", source_write_json_file=False, source_print_matched_text=False,
-                source_time_out=120, kb_url="", kb_token="", binary_simple=False, formats=[], recursive_dep=False):
+                source_time_out=120, kb_url="", kb_token="", binary_simple=False, formats=[],
+                recursive_dep=False, no_merge=False):
 
     final_excel_dir = output_path
     final_reports = []
@@ -230,6 +231,7 @@ def run_scanner(src_path, dep_arguments, output_path, keep_raw_data=False,
                                     kb_url=kb_url,
                                     kb_token=kb_token,
                                     formats=formats,
+                                    merge_by_folder=not no_merge,
                                     all_exclude_mode=_all_exclude_mode_for_scanner(
                                         excluded_path_with_default_exclusion,
                                         excluded_path_without_dot,
@@ -250,6 +252,8 @@ def run_scanner(src_path, dep_arguments, output_path, keep_raw_data=False,
                             command += f" --kb_url {shlex.quote(kb_url)}"
                         if kb_token:
                             command += f" --kb_token {shlex.quote(kb_token)}"
+                        if no_merge:
+                            command += " --no_merge"
                         command_result = subprocess.run(command, stdout=subprocess.PIPE, text=True)
                         logger.info(f"Source Analysis Result:{command_result.stdout}")
 
@@ -400,7 +404,8 @@ def run_main(mode_list, path_arg, dep_arguments, output_file_or_dir, file_format
              db_url, hide_progressbar=False, keep_raw_data=False, num_cores=-1,
              correct_mode=True, correct_fpath="", ui_mode=False, path_to_exclude=[],
              selected_source_scanner="all", source_write_json_file=False, source_print_matched_text=False,
-             source_time_out=120, kb_url="", kb_token="", binary_simple=False, recursive_dep=False):
+             source_time_out=120, kb_url="", kb_token="", binary_simple=False,
+             recursive_dep=False, no_merge=False):
     global _executed_path
 
     output_files = []
@@ -512,7 +517,8 @@ def run_main(mode_list, path_arg, dep_arguments, output_file_or_dir, file_format
                                                 default_oss_name, default_oss_version, url_to_analyze,
                                                 correct_mode, correct_fpath, ui_mode, path_to_exclude,
                                                 selected_source_scanner, source_write_json_file, source_print_matched_text,
-                                                source_time_out, kb_url, kb_token, binary_simple, formats, recursive_dep)
+                                                source_time_out, kb_url,
+                                                kb_token, binary_simple, formats, recursive_dep, no_merge)
 
                 if extract_folder:
                     shutil.rmtree(extract_folder)

tests/fixtures/setting.json

@@ -1,9 +1,16 @@
 {
-  "mode": ["binary", "source"],
-  "path": ["tests/fixtures/scan_project"],
+  "mode": [
+    "binary",
+    "source"
+  ],
+  "path": [
+    "tests/fixtures/scan_project"
+  ],
   "dep_argument": "",
   "output": "test_result_dir",
-  "format": ["excel"],
+  "format": [
+    "excel"
+  ],
   "link": "",
   "db_url": "",
   "timer": false,
@@ -12,13 +19,17 @@
   "no_correction": false,
   "correct_fpath": "",
   "ui": false,
-  "exclude": ["test", "sample_license.txt"],
+  "exclude": [
+    "test",
+    "sample_license.txt"
+  ],
   "selected_source_scanner": "scancode",
   "source_write_json_file": true,
   "source_print_matched_text": true,
   "source_time_out": 120,
   "kb_url": "",
   "kb_token": "",
   "binary_simple": false,
-  "recursive_dep": false
-}
+  "recursive_dep": false,
+  "no_merge": false
+}

tests/test__parse_setting.py

@@ -34,5 +34,5 @@ def test_parse_setting_json_valid_data():
     assert result == (
         ['test'], ['/some/path'], 'arg', 'output', 'json', 'http://example.com', 'sqlite:///:memory:', True,
         True, 4, True, '/correct/path', True, ['/exclude/path'], 'scanner', True, True, 60, True, False,
-        'http://kb.example.com', 'test_token'
+        'http://kb.example.com', 'test_token', False
     )

tests/test_cli.py

@@ -55,7 +55,7 @@ def mock_open(*args, **kwargs):
     expected = (
         ["test_mode"], ["test_path"], "test_dep_argument", "test_output", ["test_format"], "test_link", "test_db_url", True,
         True, 4, True, "test_correct_fpath", True, ["test_exclude_path"], "test_scanner", True, True, 100, True, False,
-        "http://kb.example.com", "test_token"
+        "http://kb.example.com", "test_token", False
     )
 
     assert result == expected