fosslight-util 1.4.34__py3-none-any.whl → 2.1.28__py3-none-any.whl

fosslight_util/oss_item.py

@@ -5,41 +5,43 @@
 
 import logging
 import os
-from fosslight_util.constant import LOGGER_NAME
+import hashlib
+from fosslight_util.constant import LOGGER_NAME, FOSSLIGHT_SCANNER
+from fosslight_util.cover import CoverItem
+from typing import List, Dict
 
 _logger = logging.getLogger(LOGGER_NAME)
+CHECKSUM_NULL = "0"
 
 
 class OssItem:
-    def __init__(self, value):
-        self._name = ""
-        self._version = ""
+
+    def __init__(self, name="", version="", license="", dl_url=""):
+        self.name = name
+        self.version = version
         self._license = []
-        self._copyright = ""
+        self.license = license
+        self.download_location = dl_url
+        self.exclude = False
         self.comment = ""
-        self._exclude = False
         self.homepage = ""
-        self.relative_path = value
-        self._source_name_or_path = []
-        self.download_location = ""
-        self._yocto_recipe = []
-        self._yocto_package = []
-        self.is_binary = False
+        self._copyright = ""
 
     def __del__(self):
         pass
 
     @property
-    def copyright(self):
-        return self._copyright
+    def license(self):
+        return self._license
 
-    @copyright.setter
-    def copyright(self, value):
+    @license.setter
+    def license(self, value):
         if value != "":
-            if isinstance(value, list):
-                value = "\n".join(value)
-            value = value.strip()
-        self._copyright = value
+            if not isinstance(value, list):
+                value = value.split(",")
+            self._license.extend(value)
+        self._license = [item.strip() for item in self._license]
+        self._license = list(set(self._license))
 
     @property
     def exclude(self):
@@ -53,13 +55,18 @@ class OssItem:
             self._exclude = False
 
     @property
-    def name(self):
-        return self._name
+    def copyright(self):
+        return self._copyright
 
-    @name.setter
-    def name(self, value):
-        if value != "":
-            self._name = value
+    @copyright.setter
+    def copyright(self, value):
+        if value:
+            if isinstance(value, list):
+                value = list(set(value))
+            else:
+                value = set(value.split("\n"))
+            value = "\n".join(value).strip()
+        self._copyright = value
 
     @property
     def version(self):
@@ -73,107 +80,161 @@ class OssItem:
             self._version = ""
 
     @property
-    def license(self):
-        return self._license
-
-    @license.setter
-    def license(self, value):
-        if not isinstance(value, list):
-            value = value.split(",")
-        self._license.extend(value)
-        self._license = [item.strip() for item in self._license]
-        self._license = list(set(self._license))
-
-    @property
-    def source_name_or_path(self):
-        return self._source_name_or_path
+    def comment(self):
+        return self._comment
 
-    @source_name_or_path.setter
-    def source_name_or_path(self, value):
+    @comment.setter
+    def comment(self, value):
         if not value:
-            self._source_name_or_path = []
+            self._comment = ""
         else:
-            if not isinstance(value, list):
-                value = value.split(",")
-            self._source_name_or_path.extend(value)
-            self._source_name_or_path = [item.strip() for item in self._source_name_or_path]
-            self._source_name_or_path = list(set(self._source_name_or_path))
+            if self._comment:
+                self._comment = f"{self._comment} / {value}"
+            else:
+                self._comment = value
+
+
+class FileItem:
+    def __init__(self, value):
+        self.relative_path = value
+        self.source_name_or_path = ""
+        self._exclude = False
+        self._comment = ""
+        self.is_binary = False
+        self.oss_items: List[OssItem] = []
+        self.checksum = CHECKSUM_NULL
+
+    def __del__(self):
+        pass
 
     @property
-    def yocto_recipe(self):
-        return self._yocto_recipe
+    def exclude(self):
+        return self._exclude
 
-    @yocto_recipe.setter
-    def yocto_recipe(self, value):
-        if not isinstance(value, list):
-            value = value.split(",")
-        self._yocto_recipe.extend(value)
-        self._yocto_recipe = [item.strip() for item in self._yocto_recipe]
-        self._yocto_recipe = list(set(self._yocto_recipe))
+    @exclude.setter
+    def exclude(self, value):
+        if value:
+            self._exclude = True
+        else:
+            self._exclude = False
+        for oss in self.oss_items:
+            oss.exclude = value
 
     @property
-    def yocto_package(self):
-        return self._yocto_package
-
-    @yocto_package.setter
-    def yocto_package(self, value):
-        if not isinstance(value, list):
-            value = value.split(",")
-        self._yocto_package.extend(value)
-        self._yocto_package = [item.strip() for item in self._yocto_package]
-        self._yocto_package = list(set(self._yocto_package))
-
-    def set_sheet_item(self, item):
-        if len(item) < 9:
-            _logger.warning(f"sheet list is too short ({len(item)}): {item}")
-            return
-        self.source_name_or_path = item[0]
-        self.name = item[1]
-        self.version = item[2]
-        self.license = item[3]
-        self.download_location = item[4]
-        self.homepage = item[5]
-        self.copyright = item[6]
-        self.exclude = item[7]
-        self.comment = item[8]
+    def comment(self):
+        return self._comment
+
+    @comment.setter
+    def comment(self, value):
+        if not value:
+            self._comment = ""
+        else:
+            if self._comment:
+                self._comment = f"{self._comment} / {value}"
+            else:
+                self._comment = value
+            for oss in self.oss_items:
+                oss.comment = value
 
     def get_print_array(self):
         items = []
-        if len(self.source_name_or_path) == 0:
-            self.source_name_or_path.append("")
-        if len(self.license) == 0:
-            self.license.append("")
 
-        exclude = "Exclude" if self.exclude else ""
+        for oss in self.oss_items:
+            exclude = "Exclude" if self.exclude or oss.exclude else ""
+            lic = ",".join(oss.license)
 
-        for source_name_or_path in self.source_name_or_path:
-            lic = ",".join(self.license)
-            items.append([os.path.join(self.relative_path, source_name_or_path), self.name, self.version, lic,
-                          self.download_location, self.homepage, self.copyright, exclude, self.comment])
+            oss_item = [os.path.join(self.relative_path, self.source_name_or_path), oss.name, oss.version, lic,
+                        oss.download_location, oss.homepage, oss.copyright, exclude, oss.comment]
+            items.append(oss_item)
         return items
 
     def get_print_json(self):
-        json_item = {}
-        json_item["name"] = self.name
-
-        json_item["version"] = self.version
-        if len(self.source_name_or_path) > 0:
-            json_item["source name or path"] = self.source_name_or_path
-        if len(self.license) > 0:
-            json_item["license"] = self.license
-        if self.download_location != "":
-            json_item["download location"] = self.download_location
-        if self.homepage != "":
-            json_item["homepage"] = self.homepage
-        if self.copyright != "":
-            json_item["copyright text"] = self.copyright
-        if self.exclude:
-            json_item["exclude"] = self.exclude
-        if self.comment != "":
-            json_item["comment"] = self.comment
-
-        return json_item
+        items = []
+
+        for oss in self.oss_items:
+            json_item = {}
+            json_item["name"] = oss.name
+            json_item["version"] = oss.version
+
+            if self.source_name_or_path != "":
+                json_item["source path"] = self.source_name_or_path
+            if len(oss.license) > 0:
+                json_item["license"] = oss.license
+            if oss.download_location != "":
+                json_item["download location"] = oss.download_location
+            if oss.homepage != "":
+                json_item["homepage"] = oss.homepage
+            if oss.copyright != "":
+                json_item["copyright text"] = oss.copyright
+            if self.exclude or oss.exclude:
+                json_item["exclude"] = True
+            if oss.comment != "":
+                json_item["comment"] = oss.comment
+            items.append(json_item)
+        return items
+
+
+def get_checksum_sha1(source_name_or_path) -> str:
+    checksum = CHECKSUM_NULL
+    try:
+        checksum = str(hashlib.sha1(source_name_or_path.encode()).hexdigest())
+    except Exception:
+        try:
+            f = open(source_name_or_path, "rb")
+            byte = f.read()
+            checksum = str(hashlib.sha1(byte).hexdigest())
+            f.close()
+        except Exception as ex:
+            _logger.info(f"(Error) Get_checksum: {ex}")
+
+    return checksum
 
 
 def invalid(cmd):
     _logger.info('[{}] is invalid'.format(cmd))
+
+
+class ScannerItem:
+    def __init__(self, pkg_name, start_time=""):
+        self.cover = CoverItem(tool_name=pkg_name, start_time=start_time)
+        self.file_items: Dict[str, List[FileItem]] = {pkg_name: []} if pkg_name != FOSSLIGHT_SCANNER else {}
+        self.external_sheets: Dict[str, List[List[str]]] = {}
+
+    def set_cover_pathinfo(self, input_dir, path_to_exclude):
+        self.cover.input_path = input_dir
+        self.cover.exclude_path = ", ".join(path_to_exclude)
+
+    def set_cover_comment(self, value):
+        if value:
+            if self.cover.comment:
+                self.cover.comment = f"{self.cover.comment} / {value}"
+            else:
+                self.cover.comment = value
+
+    def get_cover_comment(self):
+        return [item.strip() for item in self.cover.comment.split(" / ")]
+
+    def append_file_items(self, file_item: List[FileItem], pkg_name=""):
+        if pkg_name == "":
+            if len(self.file_items.keys()) != 1:
+                _logger.error("Package name is not set. Cannot append file_item into ScannerItem.")
+            else:
+                pkg_name = list(self.file_items.keys())[0]
+        if pkg_name not in self.file_items:
+            self.file_items[pkg_name] = []
+        self.file_items[pkg_name].extend(file_item)
+
+    def get_print_array(self, scanner_name):
+        items = []
+        for file_item in self.file_items[scanner_name]:
+            items.extend(file_item.get_print_array())
+        return items
+
+    def get_print_json(self, scanner_name):
+        items = []
+        for file_item in self.file_items[scanner_name]:
+            items.extend(file_item.get_print_json())
+        return items
+
+    def __del__(self):
+        pass

fosslight_util/output_format.py

@@ -3,11 +3,17 @@
 # Copyright (c) 2021 LG Electronics Inc.
 # SPDX-License-Identifier: Apache-2.0
 import os
-from fosslight_util.write_excel import write_result_to_excel, write_result_to_csv, remove_empty_sheet
+import platform
+from fosslight_util.write_excel import write_result_to_excel, write_result_to_csv
 from fosslight_util.write_opossum import write_opossum
 from fosslight_util.write_yaml import write_yaml
+from fosslight_util.write_spdx import write_spdx
+from fosslight_util.write_cyclonedx import write_cyclonedx
+from typing import Tuple
 
-SUPPORT_FORMAT = {'excel': '.xlsx', 'csv': '.csv', 'opossum': '.json', 'yaml': '.yaml'}
+SUPPORT_FORMAT = {'excel': '.xlsx', 'csv': '.csv', 'opossum': '.json', 'yaml': '.yaml',
+                  'spdx-yaml': '.yaml', 'spdx-json': '.json', 'spdx-xml': '.xml',
+                  'spdx-tag': '.tag', 'cyclonedx-json': '.json', 'cyclonedx-xml': '.xml'}
 
 
 def check_output_format(output='', format='', customized_format={}):
@@ -26,7 +32,7 @@ def check_output_format(output='', format='', customized_format={}):
         format = format.lower()
         if format not in list(support_format.keys()):
             success = False
-            msg = 'Enter the supported format with -f option: ' + ', '.join(list(support_format.keys()))
+            msg = '(-f option) Enter the supported format: ' + ', '.join(list(support_format.keys()))
         else:
             output_extension = support_format[format]
 
@@ -42,11 +48,12 @@ def check_output_format(output='', format='', customized_format={}):
                 if format:
                     if output_extension != basename_extension:
                         success = False
-                        msg = f"Enter the same extension of output file(-o:'{output}') with format(-f:'{format}')."
+                        msg = f"(-o & -f option) Enter the same extension of output file(-o:'{output}') \
+                                with format(-f:'{format}')."
                 else:
                     if basename_extension not in support_format.values():
                         success = False
-                        msg = 'Enter the supported file extension: ' + ', '.join(list(support_format.values()))
+                        msg = '(-o option) Enter the supported file extension: ' + ', '.join(list(support_format.values()))
                 if success:
                     output_file = basename_file
                     output_extension = basename_extension
@@ -56,29 +63,150 @@ def check_output_format(output='', format='', customized_format={}):
     return success, msg, output_path, output_file, output_extension
 
 
-def write_output_file(output_file_without_ext, file_extension, sheet_list, extended_header={}):
+def check_output_formats(output='', formats=[], customized_format={}):
     success = True
     msg = ''
+    output_path = ''
+    output_files = []
+    output_extensions = []
+
+    if customized_format:
+        support_format = customized_format
+    else:
+        support_format = SUPPORT_FORMAT
+
+    if formats:
+        # If -f option exist
+        formats = [format.lower() for format in formats]
+        for format in formats:
+            if format not in list(support_format.keys()):
+                success = False
+                msg = '(-f option) Enter the supported format: ' + ', '.join(list(support_format.keys()))
+            else:
+                output_extensions.append(support_format[format])
 
-    is_not_null, sheet_list = remove_empty_sheet(sheet_list)
-    if is_not_null:
-        if file_extension == '':
-            file_extension = '.xlsx'
-        result_file = output_file_without_ext + file_extension
+    if success:
+        if output != '':
+            basename_extension = ''
+            if not os.path.isdir(output):
+                output_path = os.path.dirname(output)
 
+                basename = os.path.basename(output)
+                basename_file, basename_extension = os.path.splitext(basename)
+            if basename_extension:
+                if formats:
+                    if basename_extension not in output_extensions:
+                        success = False
+                        msg = f"(-o & -f option) The format of output file(-o:'{output}') \
+                                should be in the format list(-f:'{formats}')."
+                else:
+                    if basename_extension not in support_format.values():
+                        success = False
+                        msg = '(-o option) Enter the supported file extension: ' + ', '.join(list(support_format.values()))
+                    output_extensions.append(basename_extension)
+                output_files = [basename_file for _ in range(len(output_extensions))]
+            else:
+                output_path = output
+    if not output_extensions:
+        output_extensions = ['.xlsx']
+
+    return success, msg, output_path, output_files, output_extensions
+
+
+def check_output_formats_v2(output='', formats=[], customized_format={}):
+    success = True
+    msg = ''
+    output_path = ''
+    output_files = []
+    output_extensions = []
+
+    if customized_format:
+        support_format = customized_format
+    else:
+        support_format = SUPPORT_FORMAT
+
+    if formats:
+        # If -f option exist
+        formats = [format.lower() for format in formats]
+        for format in formats:
+            if format not in list(support_format.keys()):
+                success = False
+                msg = '(-f option) Enter the supported format with -f option: ' + ', '.join(list(support_format.keys()))
+            else:
+                output_extensions.append(support_format[format])
+
+    if success:
+        if output != '':
+            basename_extension = ''
+            if not os.path.isdir(output):
+                output_path = os.path.dirname(output)
+
+                basename = os.path.basename(output)
+                basename_file, basename_extension = os.path.splitext(basename)
+            if basename_extension:
+                if formats:
+                    if basename_extension not in output_extensions:
+                        success = False
+                        msg = f"(-o & -f option) The format of output file(-o:'{output}') \
+                                should be in the format list(-f:'{formats}')."
+                else:
+                    if basename_extension not in support_format.values():
+                        success = False
+                        msg = '(-o option) Enter the supported file extension: ' + ', '.join(list(support_format.values()))
+                    output_extensions.append(basename_extension)
+                output_files = [basename_file for _ in range(len(output_extensions))]
+            else:
+                output_path = output
+    if not output_extensions:
+        output_extensions = ['.xlsx']
+    if not formats:
+        formats = []
+        for ext in output_extensions:
+            for key, value in support_format.items():
+                if value == ext:
+                    formats.append(key)
+                    break
+    return success, msg, output_path, output_files, output_extensions, formats
+
+
+def write_output_file(output_file_without_ext: str, file_extension: str, scan_item, extended_header: dict = {},
+                      hide_header: dict = {}, format: str = '', spdx_version: str = '2.3') -> Tuple[bool, str, str]:
+    success = True
+    msg = ''
+
+    if file_extension == '':
+        file_extension = '.xlsx'
+    result_file = output_file_without_ext + file_extension
+
+    if format:
+        if format == 'excel':
+            success, msg = write_result_to_excel(result_file, scan_item, extended_header, hide_header)
+        elif format == 'csv':
+            success, msg, _ = write_result_to_csv(result_file, scan_item, False, extended_header)
+        elif format == 'opossum':
+            success, msg = write_opossum(result_file, scan_item)
+        elif format == 'yaml':
+            success, msg, _ = write_yaml(result_file, scan_item, False)
+        elif format.startswith('spdx') or format.startswith('cyclonedx'):
+            if platform.system() == 'Windows' or platform.system() == 'Darwin':
+                success = False
+                msg = f'{platform.system()} not support spdx format.'
+            else:
+                if format.startswith('spdx'):
+                    success, msg, _ = write_spdx(output_file_without_ext, file_extension, scan_item, spdx_version)
+                elif format.startswith('cyclonedx'):
+                    success, msg, _ = write_cyclonedx(output_file_without_ext, file_extension, scan_item)
+    else:
         if file_extension == '.xlsx':
-            success, msg = write_result_to_excel(result_file, sheet_list, extended_header)
+            success, msg = write_result_to_excel(result_file, scan_item, extended_header, hide_header)
         elif file_extension == '.csv':
-            success, msg, result_file = write_result_to_csv(result_file, sheet_list)
+            success, msg, result_file = write_result_to_csv(result_file, scan_item, False, extended_header)
         elif file_extension == '.json':
-            success, msg = write_opossum(result_file, sheet_list)
+            success, msg = write_opossum(result_file, scan_item)
         elif file_extension == '.yaml':
-            success, msg, result_file = write_yaml(result_file, sheet_list, False)
+            success, msg, result_file = write_yaml(result_file, scan_item, False)
         else:
             success = False
-            msg = f'Not supported file extension({file_extension})'
-    else:
-        result_file = ""
-        msg = "Nothing is detected from the scanner so output file is not generated."
+            msg = f'(-f option) Not supported file extension({file_extension})'
 
     return success, msg, result_file

fosslight_util/parsing_yaml.py

@@ -8,8 +8,8 @@ import codecs
 import os
 import re
 import sys
-from .constant import LOGGER_NAME
-from .oss_item import OssItem
+from fosslight_util.constant import LOGGER_NAME
+from fosslight_util.oss_item import OssItem, FileItem
 
 _logger = logging.getLogger(LOGGER_NAME)
 SUPPORT_OSS_INFO_FILES = [r"oss-pkg-info[\s\S]*.ya?ml", r"sbom(-|_)info[\s\S]*.ya?ml"]
@@ -17,7 +17,7 @@ EXAMPLE_OSS_PKG_INFO_LINK = "https://github.com/fosslight/fosslight_prechecker/b
 
 
 def parsing_yml(yaml_file, base_path, print_log=True):
-    oss_list = []
+    fileitems = []
     license_list = []
     idx = 1
     err_reason = ""
@@ -38,37 +38,65 @@ def parsing_yml(yaml_file, base_path, print_log=True):
             err_reason = "empty"
             if print_log:
                 _logger.warning(f"The yaml file is empty file: {yaml_file}")
-            return oss_list, license_list, err_reason
+            return fileitems, license_list, err_reason
 
         is_old_format = any(x in doc for x in OLD_YAML_ROOT_ELEMENT)
 
+        filepath_list = []
         for root_element in doc:
             oss_items = doc[root_element]
             if oss_items:
                 if not isinstance(oss_items, list) or 'version' not in oss_items[0]:
                     raise AttributeError(f"- Ref. {EXAMPLE_OSS_PKG_INFO_LINK}")
                 for oss in oss_items:
-                    item = OssItem(relative_path)
-                    if not is_old_format:
-                        item.name = root_element
-                    for key, value in oss.items():
-                        if key:
-                            key = key.lower().strip()
-                        set_value_switch(item, key, value, yaml_file)
-                    oss_list.append(item)
-                    license_list.extend(item.license)
-                    idx += 1
+                    source_paths = get_source_name_or_path_in_yaml(oss)
+                    for source_path in source_paths:
+                        if os.path.join(relative_path, source_path) not in filepath_list:
+                            filepath_list.append(os.path.join(relative_path, source_path))
+                            fileitem = FileItem(relative_path)
+                            fileitem.source_name_or_path = source_path
+                            fileitems.append(fileitem)
+                        else:
+                            fileitem = next((i for i in fileitems if i.source_name_or_path == source_path), None)
+                        ossitem = OssItem()
+                        if not is_old_format:
+                            ossitem.name = root_element
+                        for key, value in oss.items():
+                            if key:
+                                key = key.lower().strip()
+                            set_value_switch(ossitem, key, value, yaml_file)
+                        fileitem.oss_items.append(ossitem)
+                        license_list.extend(ossitem.license)
+                        idx += 1
     except AttributeError as ex:
         if print_log:
             _logger.warning(f"Not supported yaml file format: {yaml_file} {ex}")
-        oss_list = []
+        fileitems = []
         err_reason = "not_supported"
     except yaml.YAMLError:
         if print_log:
             _logger.warning(f"Error to parse yaml - skip to parse yaml file: {yaml_file}")
-        oss_list = []
+        fileitems = []
         err_reason = "yaml_error"
-    return oss_list, set(license_list), err_reason
+
+    return fileitems, set(license_list), err_reason
+
+
+def get_source_name_or_path_in_yaml(oss):
+    source_name_or_path = []
+    find = False
+    for key in oss.keys():
+        if key in ['file name or path', 'source name or path', 'source path',
+                   'file', 'binary name', 'binary path']:
+            if isinstance(oss[key], list):
+                source_name_or_path = oss[key]
+            else:
+                source_name_or_path.append(oss[key])
+            find = True
+            break
+    if not find:
+        source_name_or_path.append('')
+    return source_name_or_path
 
 
 def find_sbom_yaml_files(path_to_find):
@@ -101,8 +129,6 @@ def set_value_switch(oss, key, value, yaml_file=""):
         oss.download_location = value
     elif key in ['license', 'license text']:
         oss.license = value
-    elif key in ['file name or path', 'source name or path', 'file', 'binary name']:
-        oss.source_name_or_path = value
     elif key in ['copyright text', 'copyright']:
         oss.copyright = value
     elif key == 'exclude':
@@ -111,10 +137,6 @@ def set_value_switch(oss, key, value, yaml_file=""):
         oss.comment = value
     elif key == 'homepage':
         oss.homepage = value
-    elif key == 'yocto_package':
-        oss.yocto_package = value
-    elif key == 'yocto_recipe':
-        oss.yocto_recipe = value
     else:
         if yaml_file != "":
             _logger.debug(f"file:{yaml_file} - key:{key} cannot be parsed")