scanoss 1.35.0__tar.gz → 1.36.0__tar.gz

{scanoss-1.35.0/src/scanoss.egg-info → scanoss-1.36.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scanoss
-Version: 1.35.0
+Version: 1.36.0
 Summary: Simple Python library to leverage the SCANOSS APIs
 Home-page: https://scanoss.com
 Author: SCANOSS

{scanoss-1.35.0 → scanoss-1.36.0}/src/scanoss/__init__.py

@@ -22,4 +22,4 @@ SPDX-License-Identifier: MIT
   THE SOFTWARE.
 """
 
-__version__ = '1.35.0'
+__version__ = '1.36.0'

{scanoss-1.35.0 → scanoss-1.36.0}/src/scanoss/api/common/v2/scanoss_common_pb2_grpc.py

@@ -3,6 +3,7 @@
 import warnings
 
 import grpc
+import warnings
 
 GRPC_GENERATED_VERSION = '1.73.1'
 GRPC_VERSION = grpc.__version__

{scanoss-1.35.0 → scanoss-1.36.0}/src/scanoss/cli.py

@@ -59,7 +59,10 @@ from . import __version__
 from .components import Components
 from .constants import (
     DEFAULT_API_TIMEOUT,
+    DEFAULT_HFH_DEPTH,
+    DEFAULT_HFH_MIN_ACCEPTED_SCORE,
     DEFAULT_HFH_RANK_THRESHOLD,
+    DEFAULT_HFH_RECURSIVE_THRESHOLD,
     DEFAULT_POST_SIZE,
     DEFAULT_RETRY,
     DEFAULT_TIMEOUT,
@@ -869,6 +872,27 @@ def setup_args() -> None:  # noqa: PLR0912, PLR0915
         help='Filter results to only show those with rank value at or below this threshold (e.g., --rank-threshold 3 '
         'returns results with rank 1, 2, or 3). Lower rank values indicate higher quality matches.',
     )
+    p_folder_scan.add_argument(
+        '--depth',
+        type=int,
+        default=DEFAULT_HFH_DEPTH,
+        help=f'Defines how deep to scan the root directory (optional - default {DEFAULT_HFH_DEPTH})',
+    )
+    p_folder_scan.add_argument(
+        '--recursive-threshold',
+        type=float,
+        default=DEFAULT_HFH_RECURSIVE_THRESHOLD,
+        help=f'Minimum score threshold to consider a match (optional - default: {DEFAULT_HFH_RECURSIVE_THRESHOLD})',
+    )
+    p_folder_scan.add_argument(
+        '--min-accepted-score',
+        type=float,
+        default=DEFAULT_HFH_MIN_ACCEPTED_SCORE,
+        help=(
+            'Only show results with a score at or above this threshold '
+            f'(optional - default: {DEFAULT_HFH_MIN_ACCEPTED_SCORE})'
+        ),
+    )
     p_folder_scan.set_defaults(func=folder_hashing_scan)
 
     # Sub-command: folder-hash
@@ -887,6 +911,12 @@ def setup_args() -> None:  # noqa: PLR0912, PLR0915
         default='json',
         help='Result output format (optional - default: json)',
     )
+    p_folder_hash.add_argument(
+        '--depth',
+        type=int,
+        default=DEFAULT_HFH_DEPTH,
+        help=f'Defines how deep to hash the root directory (optional - default {DEFAULT_HFH_DEPTH})',
+    )
     p_folder_hash.set_defaults(func=folder_hash)
 
     # Output options
@@ -2456,6 +2486,9 @@ def folder_hashing_scan(parser, args):
             client=client,
             scanoss_settings=scanoss_settings,
             rank_threshold=args.rank_threshold,
+            depth=args.depth,
+            recursive_threshold=args.recursive_threshold,
+            min_accepted_score=args.min_accepted_score,
         )
 
         if scanner.scan():
@@ -2489,6 +2522,7 @@ def folder_hash(parser, args):
             scan_dir=args.scan_dir,
             config=folder_hasher_config,
             scanoss_settings=scanoss_settings,
+            depth=args.depth,
         )
 
         folder_hasher.hash_directory(args.scan_dir)

{scanoss-1.35.0 → scanoss-1.36.0}/src/scanoss/constants.py

@@ -13,4 +13,7 @@ DEFAULT_URL2 = 'https://api.scanoss.com'  # default premium service URL
 
 DEFAULT_API_TIMEOUT = 600
 
-DEFAULT_HFH_RANK_THRESHOLD = 5
+DEFAULT_HFH_RANK_THRESHOLD = 5
+DEFAULT_HFH_DEPTH = 1
+DEFAULT_HFH_RECURSIVE_THRESHOLD = 0.8
+DEFAULT_HFH_MIN_ACCEPTED_SCORE = 0.15

scanoss-1.36.0/src/scanoss/data/build_date.txt

@@ -0,0 +1 @@
+date: 20251013130805, utime: 1760360885

{scanoss-1.35.0 → scanoss-1.36.0}/src/scanoss/file_filters.py

@@ -269,162 +269,6 @@ DEFAULT_SKIPPED_EXT = {
     'sqlite3',
 }
 
-# TODO: For hfh add the .gitignore patterns
-DEFAULT_SKIPPED_EXT_HFH = {
-    '.1',
-    '.2',
-    '.3',
-    '.4',
-    '.5',
-    '.6',
-    '.7',
-    '.8',
-    '.9',
-    '.ac',
-    '.adoc',
-    '.am',
-    '.asciidoc',
-    '.bmp',
-    '.build',
-    '.cfg',
-    '.chm',
-    '.class',
-    '.cmake',
-    '.cnf',
-    '.conf',
-    '.config',
-    '.contributors',
-    '.copying',
-    '.crt',
-    '.csproj',
-    '.css',
-    '.csv',
-    '.dat',
-    '.data',
-    '.dtd',
-    '.dts',
-    '.iws',
-    '.c9',
-    '.c9revisions',
-    '.dtsi',
-    '.dump',
-    '.eot',
-    '.eps',
-    '.geojson',
-    '.gif',
-    '.glif',
-    '.gmo',
-    '.guess',
-    '.hex',
-    '.htm',
-    '.html',
-    '.ico',
-    '.iml',
-    '.in',
-    '.inc',
-    '.info',
-    '.ini',
-    '.ipynb',
-    '.jpeg',
-    '.jpg',
-    '.json',
-    '.jsonld',
-    '.lock',
-    '.log',
-    '.m4',
-    '.map',
-    '.md5',
-    '.meta',
-    '.mk',
-    '.mxml',
-    '.o',
-    '.otf',
-    '.out',
-    '.pbtxt',
-    '.pdf',
-    '.pem',
-    '.phtml',
-    '.plist',
-    '.png',
-    '.prefs',
-    '.properties',
-    '.pyc',
-    '.qdoc',
-    '.result',
-    '.rgb',
-    '.rst',
-    '.scss',
-    '.sha',
-    '.sha1',
-    '.sha2',
-    '.sha256',
-    '.sln',
-    '.spec',
-    '.sub',
-    '.svg',
-    '.svn-base',
-    '.tab',
-    '.template',
-    '.test',
-    '.tex',
-    '.tiff',
-    '.ttf',
-    '.txt',
-    '.utf-8',
-    '.vim',
-    '.wav',
-    '.woff',
-    '.woff2',
-    '.xht',
-    '.xhtml',
-    '.xml',
-    '.xpm',
-    '.xsd',
-    '.xul',
-    '.yaml',
-    '.yml',
-    '.wfp',
-    '.editorconfig',
-    '.dotcover',
-    '.pid',
-    '.lcov',
-    '.egg',
-    '.manifest',
-    '.cache',
-    '.coverage',
-    '.cover',
-    '.gem',
-    '.lst',
-    '.pickle',
-    '.pdb',
-    '.gml',
-    '.pot',
-    '.plt',
-    '.whml',
-    '.pom',
-    '.smtml',
-    '.min.js',
-    '.mf',
-    '.base64',
-    '.s',
-    '.diff',
-    '.patch',
-    '.rules',
-    # File endings
-    '-doc',
-    'config',
-    'news',
-    'readme',
-    'swiftdoc',
-    'texidoc',
-    'todo',
-    'version',
-    'ignore',
-    'manifest',
-    'sqlite',
-    'sqlite3',
-}
-
 
 class FileFilters(ScanossBase):
     """
@@ -707,9 +551,8 @@ class FileFilters(ScanossBase):
             bool: True if file should be skipped, False otherwise
         """
         file_name = os.path.basename(file_rel_path)
-
+        DEFAULT_SKIPPED_EXT_LIST = {} if self.is_folder_hashing_scan else DEFAULT_SKIPPED_EXT
         DEFAULT_SKIPPED_FILES_LIST = DEFAULT_SKIPPED_FILES_HFH if self.is_folder_hashing_scan else DEFAULT_SKIPPED_FILES
-        DEFAULT_SKIPPED_EXT_LIST = DEFAULT_SKIPPED_EXT_HFH if self.is_folder_hashing_scan else DEFAULT_SKIPPED_EXT
 
         if not self.hidden_files_folders and file_name.startswith('.'):
             self.print_debug(f'Skipping file: {file_rel_path} (hidden file)')

{scanoss-1.35.0 → scanoss-1.36.0}/src/scanoss/scanners/folder_hasher.py

@@ -6,6 +6,7 @@ from typing import Dict, List, Literal, Optional
 
 from progress.bar import Bar
 
+from scanoss.constants import DEFAULT_HFH_DEPTH
 from scanoss.file_filters import FileFilters
 from scanoss.scanoss_settings import ScanossSettings
 from scanoss.scanossbase import ScanossBase
@@ -15,8 +16,6 @@ from scanoss.utils.simhash import WordFeatureSet, fingerprint, simhash, vectoriz
 
 MINIMUM_FILE_COUNT = 8
 MINIMUM_CONCATENATED_NAME_LENGTH = 32
-MAXIMUM_FILE_NAME_LENGTH = 32
-
 
 class DirectoryNode:
     """
@@ -72,6 +71,12 @@ class FolderHasher:
 
     It builds a directory tree (DirectoryNode) and computes the associated
     hash data for the folder.
+
+    Args:
+        scan_dir (str): The directory to be hashed.
+        config (FolderHasherConfig): Configuration parameters for the folder hasher.
+        scanoss_settings (Optional[ScanossSettings]): Optional settings for Scanoss.
+        depth (int): How many levels to hash from the root directory (default: 1).
     """
 
     def __init__(
@@ -79,6 +84,7 @@ class FolderHasher:
         scan_dir: str,
         config: FolderHasherConfig,
         scanoss_settings: Optional[ScanossSettings] = None,
+        depth: int = DEFAULT_HFH_DEPTH,
     ):
         self.base = ScanossBase(
             debug=config.debug,
@@ -101,6 +107,7 @@ class FolderHasher:
 
         self.scan_dir = scan_dir
         self.tree = None
+        self.depth = depth
 
     def hash_directory(self, path: str) -> dict:
         """
@@ -123,7 +130,10 @@ class FolderHasher:
 
         return tree
 
-    def _build_root_node(self, path: str) -> DirectoryNode:
+    def _build_root_node(
+        self,
+        path: str,
+    ) -> DirectoryNode:
         """
         Build a directory tree from the given path with file information.
 
@@ -140,7 +150,7 @@ class FolderHasher:
         root_node = DirectoryNode(str(root))
 
         all_files = [
-            f for f in root.rglob('*') if f.is_file() and len(f.name.encode('utf-8')) <= MAXIMUM_FILE_NAME_LENGTH
+            f for f in root.rglob('*') if f.is_file()
         ]
         filtered_files = self.file_filters.get_filtered_files_from_files(all_files, str(root))
 
@@ -180,7 +190,7 @@ class FolderHasher:
         bar.finish()
         return root_node
 
-    def _hash_calc_from_node(self, node: DirectoryNode) -> dict:
+    def _hash_calc_from_node(self, node: DirectoryNode, current_depth: int = 1) -> dict:
         """
         Recursively compute folder hash data for a directory node.
 
@@ -189,12 +199,13 @@ class FolderHasher:
 
         Args:
             node (DirectoryNode): The directory node to compute the hash for.
+            current_depth (int): The current depth level (1-based, root is depth 1).
 
         Returns:
             dict: The computed hash data for the node.
         """
         hash_data = self._hash_calc(node)
-        
+
         # Safely calculate relative path
         try:
             node_path = Path(node.path).resolve()
@@ -204,13 +215,18 @@ class FolderHasher:
             # If relative_to fails, use the node path as is or a fallback
             rel_path = Path(node.path).name if node.path else Path('.')
 
+        # Only process children if we haven't reached the depth limit
+        children = []
+        if current_depth < self.depth:
+            children = [self._hash_calc_from_node(child, current_depth + 1) for child in node.children.values()]
+
         return {
             'path_id': str(rel_path),
             'sim_hash_names': f'{hash_data["name_hash"]:02x}' if hash_data['name_hash'] is not None else None,
             'sim_hash_content': f'{hash_data["content_hash"]:02x}' if hash_data['content_hash'] is not None else None,
             'sim_hash_dir_names': f'{hash_data["dir_hash"]:02x}' if hash_data['dir_hash'] is not None else None,
             'lang_extensions': hash_data['lang_extensions'],
-            'children': [self._hash_calc_from_node(child) for child in node.children.values()],
+            'children': children,
         }
 
     def _hash_calc(self, node: DirectoryNode) -> dict:
@@ -237,8 +253,6 @@ class FolderHasher:
 
         for file in node.files:
             key_str = file.key_str
-            if key_str in processed_hashes:
-                continue
 
             file_name = os.path.basename(file.path)
 

{scanoss-1.35.0 → scanoss-1.36.0}/src/scanoss/scanners/scanner_hfh.py

@@ -29,7 +29,12 @@ from typing import Dict, Optional
 
 from progress.spinner import Spinner
 
-from scanoss.constants import DEFAULT_HFH_RANK_THRESHOLD
+from scanoss.constants import (
+    DEFAULT_HFH_DEPTH,
+    DEFAULT_HFH_MIN_ACCEPTED_SCORE,
+    DEFAULT_HFH_RANK_THRESHOLD,
+    DEFAULT_HFH_RECURSIVE_THRESHOLD,
+)
 from scanoss.cyclonedx import CycloneDx
 from scanoss.file_filters import FileFilters
 from scanoss.scanners.folder_hasher import FolderHasher
@@ -48,13 +53,16 @@ class ScannerHFH:
     and calculates simhash values based on file names and content to detect folder-level similarities.
     """
 
-    def __init__(
+    def __init__(  # noqa: PLR0913
         self,
         scan_dir: str,
         config: ScannerConfig,
         client: Optional[ScanossGrpc] = None,
         scanoss_settings: Optional[ScanossSettings] = None,
         rank_threshold: int = DEFAULT_HFH_RANK_THRESHOLD,
+        depth: int = DEFAULT_HFH_DEPTH,
+        recursive_threshold: float = DEFAULT_HFH_RECURSIVE_THRESHOLD,
+        min_accepted_score: float = DEFAULT_HFH_MIN_ACCEPTED_SCORE,
     ):
         """
         Initialize the ScannerHFH.
@@ -65,6 +73,9 @@ class ScannerHFH:
             client (ScanossGrpc): gRPC client for communicating with the scanning service.
             scanoss_settings (Optional[ScanossSettings]): Optional settings for Scanoss.
             rank_threshold (int): Get results with rank below this threshold (default: 5).
+            depth (int): How many levels to scan (default: 1).
+            recursive_threshold (float): Minimum score threshold to consider a match (default: 0.25).
+            min_accepted_score (float): Only show results with a score at or above this threshold (default: 0.15).
         """
         self.base = ScanossBase(
             debug=config.debug,
@@ -87,12 +98,15 @@ class ScannerHFH:
             scan_dir=scan_dir,
             config=config,
             scanoss_settings=scanoss_settings,
+            depth=depth,
         )
 
         self.scan_dir = scan_dir
         self.client = client
         self.scan_results = None
         self.rank_threshold = rank_threshold
+        self.recursive_threshold = recursive_threshold
+        self.min_accepted_score = min_accepted_score
 
     def scan(self) -> Optional[Dict]:
         """
@@ -102,8 +116,10 @@ class ScannerHFH:
             Optional[Dict]: The folder hash response from the gRPC client, or None if an error occurs.
         """
         hfh_request = {
-            'root': self.folder_hasher.hash_directory(self.scan_dir),
+            'root': self.folder_hasher.hash_directory(path=self.scan_dir),
             'rank_threshold': self.rank_threshold,
+            'recursive_threshold': self.recursive_threshold,
+            'min_accepted_score': self.min_accepted_score,
         }
 
         spinner = Spinner('Scanning folder...')
@@ -193,7 +209,7 @@ class ScannerHFHPresenter(AbstractPresenter):
                     }
                 ]
             }
-            
+
             get_vulnerabilities_json_request = {
                 'purls': [{'purl': purl, 'requirement': best_match_version['version']}],
             }
@@ -210,10 +226,10 @@ class ScannerHFHPresenter(AbstractPresenter):
                 error_msg = 'ERROR: Failed to produce CycloneDX output'
                 self.base.print_stderr(error_msg)
                 return None
-            
+
             if vulnerabilities:
                 cdx_output = cdx.append_vulnerabilities(cdx_output, vulnerabilities, purl)
-                
+
             return json.dumps(cdx_output, indent=2)
         except Exception as e:
             self.base.print_stderr(f'ERROR: Failed to get license information: {e}')

{scanoss-1.35.0 → scanoss-1.36.0/src/scanoss.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: scanoss
-Version: 1.35.0
+Version: 1.36.0
 Summary: Simple Python library to leverage the SCANOSS APIs
 Home-page: https://scanoss.com
 Author: SCANOSS

scanoss-1.35.0/src/scanoss/data/build_date.txt

@@ -1 +0,0 @@
-date: 20251007151647, utime: 1759850207